Datasets:

mimiklee

/

masterthesis-longt5-sum-8k

like 0

Introduction Social networks and social cohesion affect health [1,2]. The perceptions of general health and overall quality of life are influenced by the received social support [3]. Individuals connected to supportive social networks have better general and oral health related quality of life (OHRQoL) [4]. The current concepts of social networks focus on how structural arrangements of social institutions shape resources available to individuals, and hence, their behavioral and emotional responses [1]. The structure of network ties influences people's health by providing different types and levels of support. Lower social support is associated with more symptoms of depression [5][6][7][8] and poor social support is linked to higher mortality rates [9][10][11]. Berkman and Kawachi argued that social networks operate at the behavioral level through social support and social influence, which affects social engagement and attachment and access to resources and material goods [1]. The concepts of social networks and social supports are intrinsically interconnected and overlap [12]. However, social networks are the structure through which social support is provided [13]. Social support is generally defined in terms of the availability of people who individuals trust, and on whom they can rely on and who will care for them [1]. Research on social support emphasizes the importance of types, frequency, intensity and extent of social networks and on the effects of variation of the individual's social environment [14] as well as on the contexts for developing social networks [1]. The main mechanism that might explain why social support operates via social networks and enhances quality of life is the existence of positive social relationships. Social networks can enhance mood, provide people with a sense of identity, enhance coping strategies and be a source of companionship for sharing activities [4]. Lack of social support is an important risk factor for maternal well-being and quality of life during pregnancy, and has adverse effects on pregnancy outcomes [15]. Some studies on the relationship between social support and health in pregnant women have focused on social support interventions; others were related to family support [16,17]. Women with low social support are more likely to report postnatal depression and lower quality of life than well-supported women [18]. Pregnant women with poor social networks were at high risk for emotional and behavioral problems both to mothers and their children [19]. As stated earlier, the contexts for developing social networks affect the quality and quantity of social support. Employed women are healthier than those not employed [20,21]. That suggests that work colleagues can be an important network of social relationships and social support. They are likely to confer health benefits [22]. Social processes in women's daily activities may affect their subjective perceptions of health. In a study of Japanese women workers, poor social networks at work were associated with worse self-perceived health, mainly among older women. Older workers with social networks mainly at work reported better health than those with better social networks at home [23]. Furthermore, there was a positive association between lack of social networks outside the work environment and worse general health among middle-aged women [23]. There is a positive relationship between work-related psychosocial factors such as decision latitude, job demands and social support, and the health of workers [24]. Women in the labor market may perform tasks involving high demand and over which they have little control. That may lead to stress and poorer health. In addition, they may be less intellectually and socially stimulated; aspects considered harmful to health [25,26]. Oral health conditions are associated with social networks and social support [27][28][29]. The use of dental services was associated with better levels of social networks and social support [27,28]. Men who had more social supports and those reporting having at least one close friend and those who participated in religious activities were less likely to develop periodontitis [30]. Whereas there are numerous studies showing that dental status affects OHRQoL [31][32][33][34], there are very few on the relationship between social networks, using social support as a measure of support, and domains of OHRQoL [35]. There are very few studies on OHRQoL in pregnant women. In two studies the prevalence of negative impacts of pregnancy on OHRQoL was about 25% [36,37]. Oral pain during pregnancy had a negative effect on women's quality of life. The most frequently mentioned effects were difficulty in maintaining emotional balance, difficulty eating and difficulty cleaning teeth [36]. As studies showed that social support during pregnancy affected their health and other outcomes, it was considered important to test whether social support from the supportive relationships in the predominant environments of pregnant women, namely home or work contexts, affected their OHRQoL. The study focused on the different domains of social support that women get predominantly from work-related networks compared to those from home-based networks, rather than on the elaboration of the structural aspects of social networks. The objective of the main study [38], of which this is a part, was that social support and social network affect positively women's health. The specific hypothesis for this study was that predominantly home-based social network women with low social support had poorer perceived OHRQoL than those whose social networks were work-based and had high social support. The objective was to assess whether there were differences in OHR-QoL between women connected to either predominantly home-based and work-based social networks. The research sets out to provide insights into the possible associations of predominantly occupational contexts, home or work, linked to social support and OHRQoL in pregnant and post-partum women. --- Methods A follow-up prevalence study was carried out in two middle-sized cities in the State of Rio de Janeiro, Brazil, to test the relationship of social determinants with pregnancy outcomes and oral health measures [39]. All pregnant women enrolled in a fixed cohort who sought prenatal care at the four main public health care units administered by the National Health Care System ("Sistema Unico de Saude -SUS") were selected and invited to participate in this study. They were a representative sample of 95% of the women who were pregnant during the study period in both cities. The sample size was estimated as 1059 subjects based on the prevalence of 59.5% of the impact of oral health on quality of life, considering OHIP>1 [32,40] to detect a 5% of the differences between groups, with a significance level of 5% and power of 95% [41]. A study with 20% of losses during follow-up required 1270 participants. Primary data were collected through face-to-face individual structured interviews between October 2008 and December 2009. The information was obtained at baseline (first trimester of pregnancy) and during the 30 days postpartum period (follow-up). The selection criteria were women in the first trimester of pregnancy and living at their current address for at least 12 months. The latter criterion was used because social networks and social support tend to be stable after some months. First, the interviewers inspected the medical notes and chose pregnant women according to the selection criteria. All eligible pregnant women were invited to participate. They were informed about the objectives of the study. One of the interviewers requested their participation. After obtaining their consent, the women were interviewed. The study was approved by the Committee of Ethics and Research of the National School of Public Health -ENSP/FIO-CRUZ (protocol no. 158/06). --- Definition of occupational context The main exposure was the occupational context, which was considered to be composed of different characteristics of way of life and characteristics related to occupational status. --- Groups of comparison Participants were allocated to two groups: 1. the workbased social network group were employed women with paid work. 2. the home-based social network group were women with no paid work, housewives or unemployed women. Measures of social support and social network were evaluated to characterize the occupational context. --- Social network and social support measures Social networks was considered as the "web" of social relationships surrounding the individual as well as their characteristics, or groups of people who have contact with, or with some form of participation [42]. The questionnaire used to assess social networks consisted of 5 questions concerning the person's relationship with family and friends, and their participation in social groups. The instrument has adequate psychometric properties for the Brazilian population [43,44]. Social support was considered as a system of formal and informal relationships through which individuals receive emotional support, material or information to cope with stressful emotional situations [45]. Social support was evaluated using a questionnaire consisting of 19 items comprising five dimensions of functional social support: material (4 questions -provision of practical resources and support material), emotional (3 questions -physical expressions of love and affection), emotional (4 questions -expressions of positive affection, understanding and feelings of confidence), positive social interaction (4 questions -availability of people to have fun or relax), and information (4 questions -availability of people to obtain advice or guidance) [14]. For each item, the women indicated how often they experienced each type of available support: never, rarely, sometimes, often or always. This questionnaire had good reliability for the Brazilian population [44]. --- The impact of oral health on quality of life The outcome was the impact of oral health on quality of life, which reflects the perception of people about dysfunction, discomfort and disability related oral conditions. The validated version of Oral Health Impacts Profile (OHIP-14) for Brazilian population was used to evaluate the experience of impact on oral health on quality of life in the preceding 6 months [32,40]. OHIP-14 is composed of 14 items, aggregated in 7 dimensions (two items per dimension) as following: functional limitation (items 1 and 2), physical pain (items 3 and 4), psychological discomfort (items 5 and 6), physical disability (items 7 and 8), psychological disability (items 9 and 10), social disability (items 11 and 12) and handicap (items 13 and 14). The overall score was computed by additive method, which is the sum of the individual scores of all items. For each item, the score varied from 0 to 4: "never" = 0, "hardly ever" = 1, "occasionally" = 2, "often" = 3, and "very often" = 4. A high score indicates a negative influence of oral health on quality of life. --- Covariates The covariates were demographic and socioeconomic characteristics, health related behaviors previous and during pregnancy, dental pain in the last 6 months and number of teeth (<unk>10 teeth versus <unk>10 teeth). Demographic data were maternal age, ethnicity and number of children. Socioeconomic characteristics were marital status, educational level (years of schooling), familial income, head of the family, housing conditions and social class. In this study the term social class refers to the social and economic factors that influence what position(s) individuals and groups hold within the structure of society [46]. A standard social class classification commonly used in Brazil was used [47]. This is an economic classification based on market power comprising a group of specific indicators such as number of bathrooms, number of full-time domestic servants, number of cars owned by the family, possession of domestic items such as television sets, radio sets, VCRs, vacuum cleaners, washing machine, fridges, freezers; and level of education of the head of household. A set of points is assigned to these indicators and a final score defines the socioeconomic groups; A (highest), B, C, D, and E (lowest). Those with the highest scores represented the highest socioeconomic groups. The health behaviors, assessed before pregnancy, were smoking, cigarette consumption and alcohol consumption. In addition, the Brazilian version of T-ACE questionnaire, based on 5 questions concerning selfperception of drinking habits, was used to assess risky alcohol drinking before pregnancy [48]. --- Pilot study The interviewers were trained to conduct structured and standardized interviews. After training the interviewers, a pilot study was performed to test understanding and layout of questionnaires. Examiners interviewed 40 pregnant selected women at the same health care units of the main study but who were not included in the main study. --- Main study Data collection was performed by 20 trained interviewers and four fieldwork supervisors. The baseline was conducted in the prenatal health care units to collect occupational context data, social network, social support, demographic and socioeconomic characteristics, number of teeth and health related behaviors. During the baseline interview different strategies were established to reduce the losses to follow-up. First, two telephone numbers were requested. Second, the full current address was registered, including the zip code. Third, contact telephone numbers of the fieldwork supervisors were provided for all women. They were requested to telephone one of the supervisors when admission to the maternity unit or discharge from it was arranged. In addition, they were asked to report if they moved home or changed their telephone number. The follow-up study was performed in the post partum period immediately after the delivery to collect data on the impact of oral health on quality of life and dental pain in the last 6 months. The interview was conducted in the maternity hospital wards or at the mother's house up to 30 days after discharge. Women who moved home were excluded. In addition, those who had a miscarriage (pregnancy interrupted before the 20 th gestational week) or abortion were not re-interviewed. --- Data analysis All variables were computed for each participant and then for each group. The normal distribution of continuous variables was tested using the Kolmogorov-Smirnov test. Since the continuous variables were not normally distributed, the comparison of groups was performed by Mann-Whitney test. Categorical variables were analyzed by Chi-square test. Internal consistencies for the OHIP scale and its domains were evaluated by the Cronbach's <unk> coefficient. Cronbachs' <unk> removing each domain of the OHIP were also assessed. The relationship between occupational context and the impact of oral health on quality of life was tested using multinomial logistic regression. The sample was categorized into 3 groups according to the prevalence and the median (the median of OHIP = 3) of the number of impacts of OHIP: OHIP = 0 (No impact); OHIP 1-3 (Scores from 1 to 3); OHIP <unk>4 (Scores <unk>4). In addition, the sample was grouped concerning the dimensions of social support. Subjects with low levels of impacts were those with scores equal to zero, moderate level subjects were those between zero and the median, and high level subjects were those above the median. First, a comparison was made between social support dimensions and types of social networks in the workbased and home-based groups. Social support and social network variables that were statistically different between occupational context groups were included in the bivariate analysis. The crude Odds Ratio (OR) and Confidence Intervals of 95% were calculated between occupational context and covariates and OHIP groups. Second, multinomial logistic regression was performed to obtain adjusted OR of occupation context, affectionate support, positive social interaction and social network/friends with OHIP adjusted for age, ethnicity, family income, schooling, marital status and social class (Model 1). To test the statistical significance of interaction between occupational context and potential modifying factors (social support dimensions and social network) the occupational context and covariates were first added to the regression model. After that, the interaction terms 'occupational context X affectionate support','occupational context X positive social interaction' and 'occupational context X social network/friends' where added to the model (Model 2). Model 1 (without interaction terms) and Model 2 (with interaction terms) were compared using Likelihood Ratio tests. All statistical analyses were performed using the SPSS (Statistical Package for Social Sciences, version 13.0). The significance level for all analysis was 5% (P = 0.05). --- Results Initially 1750 pregnant women were invited to participate. The acceptance rate was 96%. Of the 1680 women interviewed at baseline, 12 (0.7%) declined to participate in the follow-up, 160 (9.5%) were excluded because they had moved home and 105 (6.3%) were lost in the follow-up (miscarriage or moved home without informing the fieldwork supervisor). The final sample was 1403 women, 83.5% of the baseline sample. Of the 1403 women, 580 (41.3%) were women in paid employment (work-based social network group) and 823 (58.7%) were unemployed women or those not doing paid work (home-based social network group). Among the women in paid work, 25 (4.3%) were civil servants, 342 (59.0%) were employees, 210 (36.2%) were selfemployed, and only 3 (0.5%) were employers. Demographic data, socioeconomic and housing conditions characteristics of the occupational context groups are presented in Table 1. The average age of the sample was 25.2 <unk> 6.3 years; 42.8% were Brown. The participants were predominantly from low socioeconomic status, married (70.6%) and in their first pregnancy (47.3%). Even though most were living in adequate housing conditions, 42.8% reported lack of sewage and 18.4% had water supply outside the house. Women from the work-based social network group were older, had more years of schooling and higher family income compared with women in the homebased social network group. The work-based social network group had more married women, and women who were head of family and from higher (B and C) social classes. The home-based social network group had a higher proportion of women living in houses without general drainage (P <unk> 0.001) and more residents per room (P <unk> 0.001) (Table 1). The comparison between oral health measures and health related behaviors in work-based women and those with home-based social networks is presented in Table 2. Women from the work-based social network group had lower OHIP-14 scores than those from home-based social network group (3.5 versus 4.0), but the statistical significance was borderline. There was no difference in the proportions of women in the two groups with dental pain in the last six months and with 10 teeth or more. The frequency of alcohol intake, alcoholism and smoking was similar in the two groups (Table 2). There were marked differences in the social support dimensions between occupational context groups (Table 3). Affectionate support and positive social interaction scores were statistically higher in the work-based social network group compared with those in the home-based social network (P <unk> 0.005). There was a borderline association between emotional support and informational support and being in the work-based social network group. Material support scores were similar in the two groups. Different types of social network were assessed. Women in the work-based social network group were more likely to have more friends that they felt comfortable with and could talk to about something (P = 0.001). The work-based social network group tended to have a higher proportion of women who participated in religious activities in the past 12 months (P = 0.064). Other types of social networks did not differ between groups (Table 3). The mean OHIP-14 was 3.8 <unk> 7.5. The Cronbach <unk> coefficient of OHIP-14 was 0.92. Cronbach <unk> coefficients if OHIP-14 dimensions deleted varied from 0.90 to 0.92. The OHIP-14 scores were statistically associated with dental pain in the last six months (P <unk> 0.05), and were not associated with the presence of 10 teeth or more. The association of home-based social network, demographic and socioeconomic characteristic with the impact of oral health on quality of life was initially tested using unadjusted risk estimates [Odds Ratio (OR)] (Table 4). In that analysis, women with OHIP = 0 (Control group) were the reference category, and the increased odds of having a OHIP score of 1-3 (Group 1) and OHIP score <unk>4 (Group 2) was estimated. The frequencies of women with OHIP score 1-3 and OHIP score <unk>4 were higher in the home-based social network group compared with those in the work-based social network group (60.7% versus 39.3% and 62.5% versus 37.5%). Occupational context, affectionate support, positive social interaction, social network/friends, demographic and socioeconomic characteristics were not associated with OHIP scores 1-3. Women with home-based social networks had significantly higher odds of OHIP score <unk>4 [OR 1.32 (95% CI: 4). The results of the final model (Model 2) of the multinomial regression analysis of the association between occupational context and independent variables with the impact of oral health on quality of life are presented in Table 5. In the fully adjusted model (Model 1), social network/friends was not associated with home-based social network. Although social network/friends was not statistically associated with home-based social network, the interaction term 'occupational context X social network/friends' was also tested and no association with home-based social network was detected. The association of home-based social network [OR 1.34 (95% CI: 0.98-1.85)] and moderate positive social interaction [OR 1.31 (95% CI: 0.99-1.72)] with OHIP score <unk>4 was of borderline significance. Low positive social interaction [OR 1.51 (95% CI: 1.01-2.27)] was significantly associated with OHIP score <unk>4 prior to adding the interaction terms (occupational context X positive social interaction). The interaction term when added to this model (Model 2) was significant, suggesting that positive social interaction modified the occupational context: OHIP relationship (Table 5). Women with work-based --- Discussion The main finding of this study was the positive association between work-based social networks and better oral health quality of life. The identified interaction between occupational context and social support also showed a gradient in the final model of OHIP. The lower the social support, the higher the odds of having more negative oral impacts on quality of life. In addition, home-based social network women with moderate positive social interaction had significantly higher odds of having poorer oral health quality of life. The stratified analysis illustrated the modifying effect of social support. The odds of occupational context on OHIP was higher among women with higher levels of positive social interaction compared with those with lower levels of positive social interaction. It appears that in women with high social support, the importance of occupational context on oral health is more relevant than for those with low social support. This study shows that being employed is not a sufficient condition for having lower impacts on oral health on quality of life. A combination of a higher social support (positive social interaction) and work-based social network appears to be needed to have better OHRQoL. The observed link between occupational context and social support with oral health was probably related to the marked differences in two social support dimensions in occupational context groups. Higher scores of affectionate support and positive social interaction, domains of social support, were positively associated with predominantly work-based social network women. Even though the findings reflect and reinforce the theory that work improves and facilitates formation of stable social relationships, social support acted as an effect modifier on the relationship between occupational context and OHIP [49]. Based on the theory of benefits of work on health, social support originating from partnerships at work can provide benefits to health and decrease risks of diseases [50]. It has been hypothesized that work environment can offer greater opportunities to build self-esteem and improve confidence in the decision making processes. Employed female workers also have more social support and working increases experiences that enhances satisfactions with life [50]. The positive or negative impacts of formal work on physical and mental health are still a subject of debate. In general health, the relationship between work-related psychosocial factors, such as job control, job strain (high demand and low control), insecurity, and social support and workers' health has been widely reported [51][52][53][54][55]. Most of studies have focused on the association between health and the ways that work is structured in terms of hours of work, continuing education and flexibility to manage work and home demands [24]. On the other hand, occupations with high strain and lack of support at work were closely associated with psychological distress [54]. Concerning oral health outcomes, Marcenes and Sheiham (1992) addressed the relationship between work-related mental demand and periodontal disease [56]. The association of flexibility in working hours with oral health related behaviors and gingival health has also been investigated [57]. Previous evidence suggests that social connections are powerful predictors, and probably affects subjective well-being [58]. The OHIP-14 questionnaire was used as a subjective measure of the impacts of oral disorders and conditions on quality of life. The OHIP aims to evaluate the positive and negative impact of oral health on well-being, considering the social, psychological and biological dimensions. OHIP has been widely used in oral epidemiology studies to evaluate subjective oral health [59]. Our findings highlight the importance of the extent to which oral health problems are experienced by women in different occupational contexts. This study suggests possible mechanisms of how social connections and social support are important and may influence women's quality of life. Our findings agree with those of Hanson et al. (1994), who found that oral health related conditions were associated with social support [29]. Unfavorable socioeconomic circumstances have been associated with poor oral health outcomes regardless the indicator used or the level of analysis [60]. Paid work is the key mechanism through which people obtain important material resources to health, especially income, which in turn, is related to better diet, adequate housing and other material goods [61,62]. In this study, home-based social network women had lower levels of family income and poorer social network compared with work-based social network women. It appears that predominantly home-based women did not have enough social networks to provide sufficient social support. We can hypothesize that home-based social network women are clustered in socially excluded groups; low income and less educated women. Because of that, they would have a higher OHIP. However, it can also be argued that the poor oral health quality of life might exclude them from the labor market, thereby excluding them from social interactions, and as a result, provide less social support. The positive aspects of the present study were the use of questionnaires with adequate psychometric properties for the Brazilian population concerning social support, social network and the impact of oral conditions on quality of life. OHIP presented good psychometrics properties. Furthermore, the data collection was standardized and collected by trained interviewers. In addition, the response rate was high and losses to follow-up were low. The time sequence of the exposure and outcome in this study provides relevant evidence on the potential benefit of work related social networks on oral health. Although a robust sample was used in this study, our findings are limited to pregnant and post-partum women. Previous studies have shown that social support is higher in pregnant and post-partum women compared to general women in general [16,17]. In addition, our findings suggest that social support (positive social interaction) mediates the association between homebased social network and OHIP scores. Therefore, the results should not be generalized. There is scope for more comprehensive studies on the relationship between work-based social networks and oral health. Detailed information concerning work environment should be collected in future studies, including job quality, hours spending at work and job demand. As the women in the labor market usually perform tasks over which they have little control and high demand, it would be relevant to consider women's sense of coherence in future investigations [50]. Even though paid work had a positive association with oral health, future studies can offer a better understanding about social networks at work and health. For example, the levels of social networks may vary among different types of jobs, and coping strategies may play an important role on health among those under worst job conditions or in workers with low social network at labor market. --- Conclusions Being in paid employment and having good social support was positively related to oral health related quality of life of women during pregnancy and the post-partum period. --- Authors' contributions GL was involved in design of the study, acquisition of data, analysis and interpretation of the data, interpretation of the results and drafted the manuscript. M do CL helped design the study, interpreted the data and reviewed the manuscript. ATTL was involved in the analysis and interpretation of data and contributed to writing the manuscript. AS was involved with interpretation of the data and revising the manuscript. MV was involved with the conception and design of the study, developed the statistical framework for data analysis, and drafted the manuscript. All authors read and approved the final manuscript. --- Competing interests The authors declare that they have no competing interests.

Background: Individuals connected to supportive social networks have better general and oral health quality of life. The objective of this study was to assess whether there were differences in oral health related quality of life (OHRQoL) between women connected to either predominantly home-based and work-based social networks. Methods: A follow-up prevalence study was conducted on 1403 pregnant and post-partum women (mean age of 25.2 ± 6.3 years) living in two cities in the State of Rio de Janeiro, Brazil. Women were participants in an established cohort followed from pregnancy (baseline) to post-partum period (follow-up). All participants were allocated to two groups; 1. work-based social network group -employed women with paid work, and, 2. home-based social network groupwomen with no paid work, housewives or unemployed women. Measures of social support and social network were used as well as questions on sociodemographic characteristics and OHRQoL and health related behaviors. Multinomial logistic regression was performed to obtain OR of relationships between occupational contexts, affectionate support and positive social interaction on the one hand, and oral health quality of life, using the Oral Health Impacts Profile (OHIP) measure, adjusted for age, ethnicity, family income, schooling, marital status and social class. Results: There was a modifying effect of positive social interaction on the odds of occupational context on OHRQoL. The odds of having a poorer OHIP score, ≥4, was significantly higher for women with home-based social networks and moderate levels of positive social interactions [OR 1.64 (95% CI: 1.08-2.48)], and for women with home-based social networks and low levels of positive social interactions ] compared with women with work-based social networks and high levels of positive social interactions. Black ethnicity was associated with OHIP scores ≥4 [OR 1.73 (95% CI: 1.23-2.42)]. Conclusions: Pregnant and post-partum Brazilian women in paid employment outside the home and having social supports had better OHRQoL than those with home-based social networks.

Introduction Dating violence is a serious problem facing today's adolescents, with 10% of high school students reporting victimization by physical dating violence within the past year (Centers for Disease Control [CDC], 2014). Furthermore, 24% of 10 th grade students reported being victimized by verbal dating abuse (Haynie et al., 2013). Dating violence has been linked with prior exposure to violence in the family, including harsh parenting (Jouriles, Mueller, Rosenfield, McDonald, & Dodson, 2012) and interparental conflict (Tschann et al., 2009). However, the developmental mechanisms through which family violence leads to dating violence are poorly understood. It is likely that family violence serves as a distal risk factor for later adolescent dating violence, with its effects mediated by more proximal individual and peer factors (Olsen, Parra, & Bennett, 2010). For instance, predictors of dating violence include individual attitudes toward dating violence (Ali, Swahn, & Hamburger, 2011), aggressive behavior (Capaldi, Dishion, Stoolmiller, & Yoerger, 2001), and deviant peer affiliations (Miller, Gorman-Smith, Sullivan, Orpinas, & Simon, 2009). However, there are still several notable gaps in existing research investigating the development of dating violence. Specifically, few studies have integrated family, individual, and peer-level risk factors into a comprehensive model of dating violence development, nor have they simultaneously accounted for potential sex differences in the development of dating violence (Champion, Foley, Sigmon-Smith, Sutfin, & DuRant, 2008;Feiring, Deblinger, Hoch-Espada, & Haworth, 2002;Williams, Connolly, Pepler, Craig, & Laporte, 2008). Thus, this study tests a developmentally-informed pathway model linking preadolescent family violence with late adolescent dating violence, while examining sex differences in these relationships. --- Theoretical Background Two types of family violence figure prominently in the literature on risk factors for dating violence-harsh discipline, defined as use of physical punishment such as spanking, slapping, hitting, and hitting with an object (Ge, Conger, Lorenz, & Simons, 1994), and witnessing interparental violence. The associations between childhood exposure to family violence and adolescent dating violence perpetration and victimization can be explained by several theoretical perspectives. Consistent with the Social Learning Theory (Bandura, 1977), which posits that individuals imitate the actions of others they observe in their environment, children who received harsh discipline or observed interparental violence may learn to view abusive interactions as a normal feature of relationships, as well as a means of eliciting compliance from others. Thus, adolescents may imitate those interactions in their own relationships, including dating relationships. Likewise, beliefs that violence is a normal part of relationships and previous experiences of victimization by harsh discipline or observations of parents being victimized during interparental violence may increase the likelihood of enduring victimization in later peer and intimate relationships. Furthermore, similarities in pro-violent beliefs and aggressive behavior are strong determinants of children's and adolescents' friendships. Thus, exposure to harsh discipline and interparental violence early in childhood may contribute to pro-aggressive beliefs and aggressive behavior in peer relationships, which may facilitate the development of friendships with other children who accept and engage in more antisocial behavior. Because adolescence marks a time when youth begin to spend less time with their families and more time with peers (Collins & Steinberg, 2006), peer relationships represent a key domain for the application of learned behaviors and beliefs about relationship interactions (Connelly & Goldberg, 1999). As a result, associations with deviant friends may provide additional modeling and reinforcement of antisocial behavior that may extend to early romantic interactions and later dating relationships. Thus, affiliations with deviant peers are likely to perpetuate patterns of family violence perpetration and victimization experienced earlier in development (Tyler, Brownridge, & Melander, 2011). --- Family Violence as a Precursor of Dating Violence Based on these theoretical underpinnings, it is evident that interactions within the family serve as a powerful influence on adolescent behavior. In fact, the co-occurrence of harsh discipline and interparental violence has been documented in numerous studies (e.g., McDonald, Jouriles, Tart, & Minze, 2009;Renner & Slack, 2006;Zolotor, Theodore, Coyne-Beasley, & Runyan, 2007). Furthermore, adolescents and young adults who perpetrated or were victimized by dating violence are more likely to report being harshly disciplined or maltreated as children (Chiodo et al., 2012;Gover et al., 2011;Jouriles et al., 2012;Wekerle & Wolfe, 1998). However, retrospective reports of harsh parenting may be biased by later involvement in dating violence, given that memories of past events may be distorted or biased by later experiences (Brewin, Andrews, & Gotlib, 1993;Henry, Moffitt, Caspi, Langley, & Silva, 1994;Tajima, Herrenkohl, Huang, & Whitney, 2004;Widom & Shepard, 1996). This potential for bias underscores the need for prospective studies. Longitudinal studies have linked harsh discipline with perpetration of dating violence three to five years later (Lavoie et al., 2002;Simons, Lin, & Gordon, 1998), but no prospective studies have evaluated the link between harsh discipline and later dating victimization. It is important to note that, although some risk factors predict both dating violence perpetration and victimization, other factors are specific to either perpetration or victimization alone (e.g., positive attitudes toward violence vs. hopelessness; Champion et al., 2008;Sears, Byers, & Price, 2007). Thus, harsh discipline may have different relationships with later dating violence perpetration vs. victimization. Compared to their peers, adolescents and young adults involved in dating violence perpetration or victimization are more likely to report witnessing interparental violence in childhood than their peers (Arriaga & Foshee, 2004;Gover et al., 2011;Jouriles et al., 2012). Among the few longitudinal studies addressing these links, some have replicated the results of cross-sectional studies (Tschann et al., 2009), while others have not found significant relationships between interparental violence exposure and later dating violence perpetration (Lavoie et al., 2002;Simons, Lin, & Gordon, 1998). The lack of association in the latter studies could be due to longer intervals between assessments of interparental conflict and dating violence (i.e., more than a year) and exclusion of girls, who perpetrate more dating violence than boys (Champion et al., 2008;Feiring et al., 2002). Thus, it is not clear whether witnessing interparental conflict in childhood prospectively predicts dating violence in adolescent boys and/or girls. --- A Dual Pathway Model To elucidate the relationships between family violence and later dating violence, as well as clarify the roles of individual and peer factors, we proposed a dual pathway model (see Figure 1). The overall dual pathway model is comprised of a "general violence" pathway and an "early romantic aggression" pathway. Each pathway first links exposure to family violence in pre-adolescence with early adolescent pro-violent beliefs and/or aggressive behavior at the level of the individual child. In both pathways, pro-violent beliefs may reinforce aggressive behaviors between same-sex and opposite-sex peers, as well as strengthen bonds with deviant peers (Thornberry, Lizotte, Krohn, Farnworth, & Jang, 1994). In the last part of both pathways, aggressive behavior and peer deviance in early adolescence may contribute directly to late adolescent dating violence perpetration and victimization. Significant associations are also expected between harsh discipline and interparental violence; beliefs about general violence and beliefs about dating violence; and perpetration of general peer aggression and opposite-sex aggression. Parts of the proposed model have received ample empirical support, particularly those associated with the "general violence" pathway. Exposure to harsh parenting is associated with beliefs condoning violence to resolve conflicts (Pardini, Loeber, & Stouthamer-Loeber, 2005); such beliefs predict physical aggression among adolescents (Poteat, Kimmel, & Wilchins, 2011); physical aggression is related to affiliation with deviant peers (Tyler et al., 2011); and deviant peer affiliations predict later dating violence perpetration and victimization (Capaldi et al., 2001;Gagné, Lavoie, & Hébert, 2005;Miller et al., 2009). Some of the proposed mediating links have also received support in previous work. For instance, antisocial behavior partly mediated the effects of harsh parenting (Tyler et al., 2011) and deviant peer associations (Miller et al., 2009) on dating violence perpetration. Also, consistent with the proposed model, studies comparing family vs. peer risk factors for dating violence have identified deviant peer affiliations as a stronger (and more proximal) risk factor for dating violence than family risk factors (Arriaga & Foshee, 2004;Gagné, Lavoie, & Hébert, 2005). Yet, the impact of some parental behaviors (e.g., monitoring) on adolescent dating aggression continues to persist alongside peer influences (Connolly, Friedlander, Pepler, Craig, & Laporte, 2010;Leadbeater, Banister, Ellis, & Yeung, 2008). In support of the "early romantic aggression" pathway, beliefs in favor of dating violence have been linked with actual dating violence (Josephson & Proulx, 2008) and have mediated the link between marital violence exposure and dating violence perpetration (Kinsfogel & Grych, 2004;Lichter & McCloskey, 2004). Additionally, beliefs approving dating violence and deviant peer affiliations were related to early adolescents' perpetration of or victimization by aggression toward opposite-sex peers (Windle & Mrug, 2009). However, no studies have examined whether aggression or victimization in early adolescents' relationships with opposite sex peers predicts later dating violence. In summary, parts of the proposed dual pathway model have received extensive support, whereas others have not yet been evaluated (see Figure 2 for a summary of mediational links examined). Thus, more research is needed to clarify the distinction and overlap of the two proposed pathways. Longitudinal examination of the multi-domain risk factors for dating violence can provide more nuanced and targeted approaches to youth dating violence intervention and prevention. --- Sex Differences in the Dual Pathway Model Sex differences have been reported in dating violence perpetration and/or victimization. Overall, more females than males report perpetrating dating violence (Champion et al., 2008;Feiring et al., 2002). Females perpetrate more verbal abuse and less severe physical violence, whereas males perpetrate more severe physical dating violence (Tyler et al., 2011). Although similar proportions of males and females report dating victimization (O' Keefe, 2005), females experience more severe and frequent physical injury from dating violence than males (Arriaga & Foshee, 2004). The salience of family and peer risk factors for dating violence may also vary by sex, especially among youth from urban lower-SES and single parent families where relationship violence is more common (Foshee, Benefield, et al, 2004;Foshee, Benefield, et al., 2009). Female adolescents in these families may be more likely to identify with their mothers and their use of harsh discipline or interparental violence (e.g., Jankowski, Leitenberg, Henning, & Coffey, 1999), leading to stronger links between family violence and dating violence perpetration among girls than boys. Similarly, males growing up in families headed by a single mother may become accustomed to a dominant female presence in the household, relative to missing or transient male partners. Thus, males may carry the notion of female household dominance into their own dating relationships and be more likely to accept victimization by females in their dating relationships. However, sex differences in cross-sectional studies linking interparental violence with dating violence perpetration have been inconsistent. Several studies found stronger links for females compared to males (e.g., Jouriles et al., 2012;Wolf & Foshee, 2003), but another study found a relationship for males but not for females (Kinsfogel & Grych, 2004). Moreover, Wekerle and Wolfe (1998) found that child maltreatment was predictive of physical and verbal dating abuse perpetration for adolescent males, but less so for females, while maltreatment was predictive of victimization for both males and females. Thus, more research is needed to better understand possible sex differences in the roles of risk factors for the development of dating violence. --- Hypotheses This study proposes and tests a comprehensive theoretical model that explains how family violence translates into dating violence. We hypothesize that higher levels of pre-adolescent exposure to harsh discipline and interparental violence will predict late adolescent dating violence perpetration and victimization. Additionally, we expect that the link between harsh discipline and dating violence will be mediated by the "general violence" pathway, following the developmental sequence of harsh discipline, general violence beliefs, general peer aggression and affiliation with deviant peers, and dating violence perpetration and victimization. In parallel, we expect that the relationship between interparental violence and dating violence will be mediated by the "early romantic aggression" pathway, from interparental violence to beliefs supporting dating violence, affiliation with deviant peers, involvement in aggression with opposite-sex peers, and dating violence perpetration and victimization. Finally, we expect some parts of the proposed model to vary by sex; however, these differences will be exploratory given insufficient evidence for directional hypotheses. --- Method --- Participants and Procedures Participants include 461 adolescents (51% female; 80% African American, 19% Caucasian, 1% other ethnicities) who took part in the Birmingham Youth Violence Study (BYVS). Adolescents were initially recruited from 5th grade classrooms in 17 Birmingham area schools selected through a two-stage probability sampling process. In the first stage, schools were randomly selected based on probabilities designed to achieve a representative sample of all students attending public schools in the Birmingham metropolitan area in terms of racial/ethnic, sex, and SES composition. Adolescents completed individual interviews at average ages 11.8 (Wave 1), 13.2 (Wave 2), and 18.0 (Wave 3) between 2003 and 2012. Primary caregivers also completed individual interviews at Waves 1 and 2 as part of the larger study; however, only child reports were used in the current study. Of the 704 youth participating in Wave 1 (W1; 42% recruitment rate), 603 (86%) returned at Wave 2 (W2) and 480 (68%) were interviewed at Wave 3 (W3). Despite intensive tracking efforts, some participants could not be located after the period between W2 and W3 (9% of W1 sample). Others did not participate because they had moved out of the area (5%), refused to participate or repeatedly did not come to scheduled interviews (18%), were deceased (0.6%), or were incarcerated (0.4%). At W3, same-sex intimate relationships were reported by 4% of the youth. Because this very small group faces more discrimination (Dank, Lachman, Zweig, & Yahner, 2013) that may contribute to qualitative differences in dating violence experiences, we excluded these adolescents from analyses. Thus, this report includes all youth who participated in W3 and reported on dating violence within heterosexual romantic relationships. Compared to W1 participants not included in this report, the current sample had more females (51% vs. 41%, <unk> 2 =6.16, p=.013) and members of racial minority groups (81% vs.72%, <unk> 2 =8.03, p=.005), as well as individuals from families with higher parent education and family income (i.e., higher SES; average z=.05 vs. -.09, t=2.02, p=.043). Prior to participants' enrollment, this study was approved by the Institutional Review Board at the University of Alabama at Birmingham. At each wave, parents and adolescents were given detailed information about the study and provided written informed consent (parents or adult participants) and assent (adolescents). Participants were informed about the voluntary nature of participation, including the option to refuse participation, skip any questions, or stop the interview. Interviews were administered by trained interviewers using Computer-Assisted-Personal-Interviews, with sensitive questions (including those about dating and peer deviance) completed privately by participants through Audio-Computer-Assisted-Self-Interview (ACASI). Participants received monetary compensation for their time. At W3, 82% (N=377) of the present study sample were enrolled in high school (72%) or college (10%). Of the remainder, 9% had dropped out of high school and 8% graduated from high school but did not attend college. Two-way ANOVAs (with academic status and gender) revealed no differences among the four groups in dating violence perpetration (F[3, 429]=1.26, p=.286) or victimization (F[3, 429]=2.34, p=.073). Although group means were not significantly different, youth who dropped out of high school reported the highest levels of both dating violence perpetration and victimization (5.59 and 6.39, respectively, compared to 4.42 -5.56 and 3.69 -4.75 for the other groups). These patterns are consistent with higher truancy and drop-out rates among youth victimized by general violence (Banyard & Cross, 2008;Dyregrov, 2004). Because peer group relationships may vary between adolescents in school and those who have graduated or dropped out, school enrollment was included as a covariate in analyses. --- Measures Dating violence perpetration and victimization.-At W3, youth responded to the Conflict in Adolescent Dating Relationships Inventory (CADRI; Wolfe et al., 2001), reporting on whether they had committed violent acts against a dating partner (perpetration) and whether a dating partner had committed violent acts against them (victimization) in the past 12 months. The overall scale consisted of 18 perpetration and 18 identical victimization items coded 1 for "yes" and 0 for "no". Because emotional abuse often co-occurs with physical dating violence (Sears & Byers, 2010), we included physical, threatening, and emotional/verbal abuse in the composite dating violence variable. The physical abuse scale consisted of 4 items (e.g., "I slapped him/her or pulled his/her hair" for perpetration). The threatening abuse scale consisted of 4 items (e.g., "He/she threatened to hurt me" for victimization). Finally, the emotional/verbal abuse scale consisted of 10 items (e.g., "I blamed him/her for the problem" for perpetration). Scores for overall dating violence perpetration and victimization scales were derived by summing the responses to the 18 individual items. Reliabilities for overall dating violence scales were adequate (<unk>=.88) and consistent with reliability estimates from other studies (<unk>=.77-.82 for perpetration; <unk>=.82-.86 for victimization ;Hopper, 2011). Construct validity of both scales was supported by positive correlations between self-reports of dating violence and observer ratings of dating couples' abusive interactions (Wolfe et al., 2001). Early romantic aggression.-At W1 and W2 (ages 11 and 13), youth reported on the frequency of perpetrating aggression towards and victimization by opposite-sex peers in the past 12 months (Foshee et al., 1996). Each scale (victimization and perpetration) consisted of 15 items assessing involvement in the same aggressive behaviors toward opposite-sex peers as a victim or perpetrator, from milder behaviors (e.g., "How many times did you push, grab, or shove a boy/girl?") to more serious behaviors (e.g., "How many times did a boy/girl try to choke you?"). At both waves, items were rated from 0 (Never happened) to 2 (Happened 3 or more times) and summed for each scale. Reliability for each scale was good (<unk>=.84-.87 for perpetration, <unk>=.82-.85 for victimization). Validity was supported by positive correlations with beliefs about dating violence (Windle & Mrug, 2009). Family violence.-Two types of family violence were examined in this study: Interparental violence and harsh discipline. Interparental violence.: At W1, youth completed two items from the Children's Perception of Interparental Conflict Scale (CPIC; Grych, Seid, & Fincham, 1992) and two items from the Conflict Tactics Scale (CTS; Straus, 1979) assessing parent-to-partner violent behaviors they had witnessed in the past 12 months. The four items included pushing and shoving during an argument; throwing or breaking things during an argument; mother hitting father when angry; and father hitting mother when angry. Adolescents could report on violence between parents or between a parent and a partner, so that youth from single-parent homes could still report witnessed violence (92% of participants provided data for this scale). Scores were obtained by summing the responses for the four individual items, which ranged from 0 (False) to 2 (True). This scale demonstrated good reliability in this study (<unk>=.84), consistent with previous research (<unk>=.78-.90; Grych, Seid, & Fincham, 1992). Convergent and construct validity have been supported by positive correlations of child reports of interparental violence with parental reports of the same construct, parent/teacher reports of child internalizing and externalizing problems, and observer ratings of child reactions to parental conflict (Grych et al., 1992). Harsh discipline.: Harsh discipline was measured through child report at W1. Youth reported on their experiences of harsh discipline using a three-item scale (Ge, Conger, Lorenz, & Simons, 1994). The items included frequency of yelling, spanking or slapping, and hitting with a belt. Scale scores were obtained by summing the responses for individual items, rated 1 (Never) to 5 (Always). Reliability was.67 in the present study, which is somewhat lower than the range from.78 to.81 reported in previous research (Kim & Ge, 2000). Validity of the scale has been supported by positive correlations with other measures of poor discipline, maternal psychopathology, and adolescents' antisocial behavior (Mrug, Loosier, & Windle, 2008;Murry, Simons, Simons, & Gibbons, 2013). Parental reports of harsh discipline was also measured using the same scale, but was uncorrelated with the child report (r=.05, p=.270), consistent with previous studies showing low agreement between parent and child reports of parenting behaviors (Gonzales, Cauce, & Mason, 1996). Given this lack of agreement, only the child report was used in this study because it is more valid and less subject to social desirability than parental reports (Morsbach & Prinz, 2006). Peer deviance.-Youth reported on affiliations with deviant peers at W1 and W2. At both waves, youth completed six items assessing their close friends' delinquent behaviors from the Denver Youth Survey (Institute of Behavioral Science, 1987). These items asked how many of their friends have purposely damaged public property; used alcohol, marijuana, or other illegal drugs; hit or threatened to hit someone; gotten into a physical fight; or carried a knife or gun. At W1, possible responses ranged from 1 (None of their friends) to 5 (All of their friends); responses were averaged. At W2, the scale score was computed as the average percentage of close friends who had engaged in each deviant behavior, which was then rescaled to a 1 -5 scale comparable to the one used at W1, from 1 (No friends) to 5 (All friends). The scale demonstrated adequate reliability (<unk>=.73-.74). Previous research demonstrated higher reliability for a scale with a larger number of items (<unk>=.89; Institute of Behavioral Science, 1987). Validity was supported by multi-informant reports of child's antisocial behavior (Mrug et al., 2008). General peer aggression.-At W2, youth reported on aggressive behavior using the Forms and Functions of Aggression measure (Little et al., 2003a;Little et al., 2003b). The scale consists of 18 items and includes pure overt aggression, reactive overt aggression, and instrumental overt aggression. To be consistent with the other measures in the study that primarily focused on physical aggression (i.e., harsh discipline, interparental violence, and beliefs about general and dating violence), only the overt aggression scales was used (e.g., "You are the kind of person who fights with others"). Items were rated from 1 (not at all) to 4 (completely true) and summed. The scale had a good reliability in this study (<unk>=.88) and in previous research (<unk>=.79-.84; Little et al., 2003b). Validity was demonstrated by positive correlations with other measures of antisocial behavior (Fite et al., 2008;Little et al., 2003a). Beliefs about general violence.-At W2, youth reported on beliefs about violence by completing 5 items about the acceptability of violence (Bosworth, Espelage, & Simon, 1999). Example items include "You don't need to fight because there are other ways to deal with being mad" and "It's okay to hit someone who hits you first." Items were rated from 1 (Strongly disagree) to 5 (Strongly agree) and summed. Reliability was.61, somewhat lower compared to previous research (<unk>=.71; Bosworth et al., 1999). Validity was supported by positive correlations with bullying (Bosworth et al., 1999). Beliefs about dating violence.-At W2, participants reported on dating violence beliefs by completing 12 items about the acceptability of dating violence and how they anticipate they might act if upset with a romantic partner. This measure was developed specifically for this study and was adapted from the Conflict in Adolescent Dating Relationships Inventory (CADRI; Wolfe et al., 2001), but in contrast to the CADRI, the items assessed responses to hypothetical dating conflicts, rather than actual events (e.g., "If you were upset or angry, would you push, shove, or shake [your boyfriend/girlfriend]?)" Possible responses ranged from 1 (Definitely would not) to 5 (Definitely would). Total scores were obtained by summing the scores on individual items. This scale demonstrated good reliability (<unk>=.88). Construct validity was supported by positive correlations with early romantic aggression perpetration (Windle & Mrug, 2009). Sociodemographic characteristics.-These characteristics include adolescent sex and ethnicity, obtained through parental reports at W1. Sex was coded as 0 for males and 1 for females. Given that small number of participants who were not Caucasian or African American, ethnicity was coded 0 for Caucasian and 1 for non-Caucasian. Family SES was assessed using a composite variable created from standardized parental reports of their highest education level (8-point scale) and family income (13-point scale) at W1 (r=.49, p<unk>. 001). Age was computed from participants' birth dates and dates of interview. Parents reported on family structure at W1, which was coded into two dummy variables indicating single-parent households and households headed by other relatives, vs. the reference group of two-parent families. Adolescents reported on school status at W3, which was coded as 0 for not in high school and 1 for in high school. All main analyses were performed using path modeling in Mplus (Muthén & Muthén, 2006). First, a direct effects model evaluated whether harsh discipline and interparental violence in pre-adolescence predicted dating violence in late adolescence. Paths were adjusted for covariates (age, sex, ethnicity, family SES, family structure, and school status). Then, we examined the model fit of the general violence pathway only. The "general violence" pathway linked pre-adolescent (W1) exposure to harsh discipline with early adolescent (W2) beliefs endorsing general violence, which were associated with W2 general peer aggression and deviant peer affiliations, as well as late adolescent (W3) dating violence perpetration and victimization. Next, the "early romantic aggression" pathway was added to the model to determine whether its addition resulted in an improved model fit. This pathway linked pre-adolescent (W1) exposure to interparental violence to early adolescent (W2) beliefs approving of dating violence, which were associated with W2 early romantic aggression victimization and perpetration, and ultimately with W3 dating violence perpetration and victimization. The two pathways were also connected through covariances at the levels of beliefs and perpetration of aggression. All paths were adjusted for covariates. W2 measures for which W1 equivalents were available (deviant peers, early romantic aggression perpetration and victimization) were also adjusted for their W1 levels to provide stronger support for the hypothesized directionality of relationships. Mediation was tested with bootstrapping using the MODEL INDIRECT statement in Mplus (Muthén & Muthén, 2006). Then, multi-group modeling was conducted to test sex differences. In each case, a constrained model (with all paths fixed to be equal across sex) was compared to an unconstrained model where all paths were freely estimated for each sex. A significant likelihood ratio test indicated that the unconstrained model fit the data better than the constrained model. Overall model differences were followed by tests of invariance in each structural path, using p<unk>.001 as the cutoff for significant sex differences given the high number of follow-up tests that were conducted. --- Results --- Preliminary Analyses A total of 2.9% of all data points were missing. Youth with any missing data were younger at W2 (M=13.02 vs. M=13.23; t=-1.99, p=.047), more likely to be Caucasian (34% vs. 30%; <unk> 2 =4.69, p=.030), and from higher-SES backgrounds (M=.16 vs. M=-.05; t=2.24, p=.026). These missing data were handled using Full-Information Maximum Likelihood (FIML) in all analyses (Wothke, 2000). Outliers were identified and truncated at 3.5 SD (Iglewicz & Hoaglin, 1993). All mediators and dependent variables in the model that were non-normally distributed were normalized with logarithmic transformations. Descriptive statistics for all variables, including overall means and sex differences in means, are displayed in Table 1. Pearson and point-biserial correlations among the main variables (log-transformed if nonnormal) are displayed in Table 2. At W3, 88% of adolescents reported having ever been on a date and 80% of adolescents reported having ever been in a romantic relationship, which lasted, on average, for 1.71 years. However, because previous literature has not clearly determined what characteristics constitute a formal adolescent dating "relationship" (e.g. length of time, intensity of relationship, risk involvement; Giordano, Soto, Manning, & Longmore, 2010), and because dating violence may occur outside of what youth would define as dating (e.g., as part of "hooking up"; Bogle, 2007), we did not consider past relationship experience as a prerequisite for reports of dating violence. Those who had ever been in a relationship reported more dating violence perpetration compared to those who had never been in a relationship (5.11 vs. 4.04, independent samples t=-2.05, p=.041), though the two groups did not differ on dating violence victimization. --- Main Analyses Family violence and dating violence.-The direct effects model predicting W3 dating violence perpetration and victimization from W1 family violence, adjusting for covariates, revealed that W1 harsh discipline predicted W3 dating violence perpetration (<unk>=.09, p=. 043), but not victimization (<unk>=.07, p=.164; see Figure 3). W1 interparental violence predicted both W3 dating violence perpetration (<unk>=.12, p=.011) and victimization (<unk>=.12, p=.014). These predictive relationships were somewhat weaker compared to the zero-order correlations, suggesting that the inclusion of demographic covariates attenuated these links. Dual pathway model.-Model fit indices for the general violence pathway alone revealed borderline adequate model fit (<unk> 2 (17)=60.953, p<unk>.001, CFI=.96, TLI=.83, RMSEA of.08, SRMR=.03). However, when the early romantic aggression pathway was added to this model, the fit of the overall mediation model improved substantially (<unk> 2 (44)=80.881, p<unk>. 001, CFI=.98, TLI=.94, RMSEA of.04, SRMR=.03; see Table 3 for R 2 values for key dependent variables). Notably, the R 2 values for dating violence perpetration and victimization increased when the early romantic aggression pathway was added to the model (from.22 to.24 for perpetration, and from.11 to.13 for victimization). Thus, only the combined model is interpreted further. The path coefficients (see Tables 3-4 andFigure 4) provided support for most of the links in the general violence pathway. Specifically, W1 harsh discipline predicted more general peer aggression and general violent beliefs at W2; general violent beliefs were concurrently associated with more general peer aggression and deviant peer affiliation; W2 general peer aggression predicted dating violence perpetration at W3; and W2 deviant peer affiliation predicted both dating violence perpetration and victimization at W3. Dating violence perpetration and victimization at W3 were strongly correlated. In addition to testing individual paths, we also tested the mediational pathways to see if any

Risk factors for adolescent perpetration of or victimization by dating violence stem from different levels of adolescents' social ecologies, including the family, individual, and peer domains. However, these multiple risk factors have not been fully integrated into a single comprehensive model of dating violence development. The present study examined prospective links between exposure to family violence in pre-adolescence; pro-violent beliefs, aggression, deviant peer affiliation, and aggression toward opposite-sex peers in early adolescence and dating violence in late adolescence. Using a longitudinal study of 461 youth (51% female; 80% African American, 19% Caucasian, 1% other ethnicities), path modeling evaluated a theoretically-developed dual pathway model involving a general violence pathway and an early romantic aggression pathway. Each pathway links exposure to family violence in pre-adolescence with early adolescent proviolent beliefs and/or aggressive behavior. In both pathways, pro-violent beliefs may reinforce aggressive behaviors between same-sex and opposite-sex peers, as well as strengthen bonds with deviant peers. In the last part of both pathways, aggressive behavior and peer deviance in early adolescence may contribute directly to late adolescent dating violence perpetration and victimization. The findings provided support for both pathways, as well as sex differences in the model.

(44)=80.881, p<unk>. 001, CFI=.98, TLI=.94, RMSEA of.04, SRMR=.03; see Table 3 for R 2 values for key dependent variables). Notably, the R 2 values for dating violence perpetration and victimization increased when the early romantic aggression pathway was added to the model (from.22 to.24 for perpetration, and from.11 to.13 for victimization). Thus, only the combined model is interpreted further. The path coefficients (see Tables 3-4 andFigure 4) provided support for most of the links in the general violence pathway. Specifically, W1 harsh discipline predicted more general peer aggression and general violent beliefs at W2; general violent beliefs were concurrently associated with more general peer aggression and deviant peer affiliation; W2 general peer aggression predicted dating violence perpetration at W3; and W2 deviant peer affiliation predicted both dating violence perpetration and victimization at W3. Dating violence perpetration and victimization at W3 were strongly correlated. In addition to testing individual paths, we also tested the mediational pathways to see if any indirect pathways from family violence to dating violence reached significance. In the general violence pathway, the sum of all indirect links between W1 harsh discipline and W3 dating violence perpetration reached significance (<unk>=.026, p=.018), although no single indirect pathway from W1 harsh discipline to W3 dating violence was statistically significant. The indirect pathways from harsh discipline to dating violence victimization were not significant (sum: <unk>=.015, p=.130). The early romantic aggression pathway received only partial support. W1 interparental violence did not predict beliefs in favor of dating violence at W2. However, dating violence beliefs were significantly associated with concurrent deviant peer affiliation and early romantic aggression perpetration and victimization. Deviant peer affiliation predicted both W3 dating violence perpetration and victimization, while neither W2 early romantic aggression perpetration nor victimization predicted W3 dating violence perpetration. However, W2 early romantic aggression victimization significantly predicted subsequent dating violence victimization at W3. In addition, W2 deviant peer affiliation was related to concurrent early romantic aggression perpetration and victimization. However, when we tested the mediational paths of the early romantic aggression pathway, none of the indirect paths (nor their sums) significantly mediated the relationship between W1 interparental violence and W3 dating violence. The following covariances linking the two pathways (see Table 3) were statistically significant: W1 harsh discipline with W1 interparental violence; W2 beliefs in favor of general violence with W2 beliefs in favor of dating violence; and W2 general peer aggression with W2 perpetration of early romantic aggression. Additionally, W1 harsh discipline predicted W2 beliefs in favor of dating violence, and W2 beliefs in favor of dating violence predicted concurrent general peer aggression. When the 20% of participants who reported not having been in a relationship in the past year were excluded from analyses, all results remained the same with one exception -the covariance between harsh discipline and interparental violence was no longer significant. Sex differences in the dual pathway model.-Multigroup modeling indicated significant sex differences in the overall dating violence model (<unk> 2 (81) =352.993, p<unk>.001). Follow-up analyses were conducted and their results are shown in Tables 34. In general, links in the general violence pathway were stronger for males, whereas links in the early romantic aggression pathway were stronger for females. --- Discussion Violence and abuse in romantic relationships are common experiences for many adolescents (CDC, 2014). Although dating violence has been linked with family, individual, and peer factors, the developmental mechanisms through which family violence leads to dating violence are poorly understood. This study aimed to fill these gaps in the dating violence literature by proposing and testing a dual pathway model of family violence leading to dating violence, accounting for mediating roles of individual and peer factors and moderation by sex. We hypothesized that harsh discipline and interparental violence would lead to dating violence through two distinct yet interrelated pathways: the "general violence" pathway and the "early romantic aggression" pathway. The results provided strong support for the general violence pathway and partial support for the early romantic aggression pathway. Multiple sex differences emerged, with the general violence pathway being more prominent for males and the early romantic aggression pathway being more notable for females. The prospective relationship between pre-adolescent reports of harsh discipline and adolescent dating violence perpetration was consistent with previous longitudinal research (Lavoie et al., 2002;Simons et al., 1998) and the Social Learning Theory (Bandura, 1977). While the links between interparental violence and both dating violence perpetration and victimization were consistent with previous findings (Ehrensaft et al., 2003;Jouriles et al., 2012), the lack of prospective relationship between harsh discipline and dating violence victimization contrasts with previous cross-sectional results (Chiodo et al., 2012;Gover et al., 2011;Windle & Mrug, 2009), and may be due to a weaker effect of harsh discipline on future victimization than perpetration of violence. It is also possible that youth who were harshly disciplined as children may have learned to resolve conflicts using the forceful methods characteristic of how they were disciplined; as such, these youth may have been more likely to use similar methods of communication with their later dating partners -hence the link between harsh discipline and dating violence perpetration but not necessarily victimization. --- Links from Family Violence to Dating Violence The results provided strong support for the "general violence" pathway as a developmental mechanism through which childhood harsh discipline translates into perpetration of dating violence in late adolescence. Specifically, the effect of harsh discipline on dating violence perpetration was mediated by beliefs in favor of violence, general peer aggression, and deviant peer affiliations in early adolescence. These results extend previous work on individual components of the pathway, such as harsh parenting predicting general proviolent beliefs and general peer aggression (Ehrensaft et al., 2003), general pro-violent beliefs and general peer aggression predicting deviant peer affiliations (Hurd, Zimmerman, & Reischl, 2011), and general peer aggression and deviant peer affiliation predicting dating violence perpetration (Ehrensaft et al., 2003). Notably, the positive relationship between two key constructs in the general violence pathway, general peer aggression and deviant peer affiliations, was consistent with past research documenting the role of deviant peers in escalating general peer aggression (e.g., Benson & Buehler, 2012). These findings are consistent with the peer selection and homophily theories and reports of similarity in proviolent beliefs and aggressive behavior facilitating friendships with other aggressive peers (Poulin & Boivin, 2000;Poulin et al., 1997). Furthermore, deviant peer affiliations predicted both dating violence victimization and perpetration five years later, extending prior cross-sectional and short-term longitudinal findings (Vézina et al., 2011;Williams et al., 2008). These results suggest that the peer context is a key contributor to the transmission of violence into dating relationships, as youth with more deviant peers or friends who perpetrate dating violence are more likely to perpetrate dating violence themselves (Foshee et al., 2011;Miller et al., 2009). Deviant peers also put adolescents in contact with potential dating partners who engage in antisocial behavior (Krueger, Moffitt, Caspi, Bleske, & Silva, 1998), which may foster dating violence perpetration and victimization. In contrast to the general violence pathway, the present study showed only partial support for the early romantic aggression pathway. Interparental violence did not predict beliefs in favor of dating violence, in contrast to previous research documenting pro-dating violence beliefs as mediators of the link between marital violence exposure and dating violence perpetration (Lichter & McCloskey, 2004). It is possible that the low levels of interparental violence reported by participants in this study were insufficient to elicit effects on dating violence beliefs. Deviant peer affiliations contributed to early adolescent aggression toward oppositesex peers (both as a perpetrator and a victim), consistent with past research (Windle & Mrug, 2009). However, involvement in early adolescent romantic aggression perpetration did not predict later dating violence perpetration or victimization. Thus, early adolescent romantic aggression perpetration may be more a function of situational influences than stable tendencies to aggress against other-sex peers. In contrast, early adolescent romantic victimization predicted later dating violence victimization. Thus victimization by oppositesex peers in early adolescence may put youth at risk for later dating victimization, consistent with broader research on victimization as a risk factor for other types of victimization (Finkelhor, Ormrod, & Turner, 2007). The strong association between late adolescent dating violence perpetration and victimization likely reflects concurrent involvement in dating violence perpetration and victimization (O'Leary, Slep, Avery-Leaf, & Cascardi, 2008) and suggests that violence in dating relationships is mutual. The greater magnitude of this correlation compared to early romantic aggression perpetration and victimization may be a function of greater stability, intimacy, and length of opposite-sex and romantic relationships in older youth (Collins, 2003). Indeed, youth in longer-term relationships are more likely to report bidirectional violence than those in shorter-term relationships (Palmetto, Davidson, Breitbart, & Rickert, 2013). The high residual correlation between dating violence perpetration and victimization, together with a substantial amount of unexplained variance in each variable, also suggest that some common determinants of both dating violence perpetration and victimization have been omitted from the model. These variables may involve distal contextual predictors (e.g., positive parenting practices, community violence), as well as more proximal predictors (e.g., norms in concurrent social networks). --- Sex Differences in the Development of Dating Violence A number of sex differences emerged in mean levels and relationships among variables. In early adolescence, males reported more victimization by girls, whereas females reported more perpetration of aggression against boys. In late adolescence, females continued to report higher levels of perpetrating dating violence, but males and females reported similar levels of dating violence victimization. These sex differences are consistent with studies reporting greater female perpetration of dating violence in adolescence (Champion et al., 2008;Feiring et al., 2002) and similar levels of victimization among middle adolescent males and females (Arriaga & Foshee, 2004;O'Keefe, 2005). Also, the generally higher use of physical aggression by male vs. female youth (Lansford et al., 2012) may make it more socially acceptable for girls to use physical aggression against male peers. Consistent with the predictions from harsh discipline, pro-violent beliefs predicted general peer aggression slightly more strongly for males and early romantic aggression (both perpetration and victimization) for females. Next, affiliating with deviant peers in early adolescence was a stronger predictor of concurrent early romantic aggression perpetration for males than females, consistent with past findings establishing deviant peer affiliation as a prominent influence on adolescent boys' perpetration of dating violence (Arriaga & Foshee, 2004;Ehrensaft et al., 2003). In addition, deviant peer affiliations predicted dating violence perpetration and victimization slightly more strongly among males than females, which contrasts with previous literature (e.g., Arriaga & Foshee, 2004;Miller et al., 2009). These sex differences may reflect greater males' vs. females' involvement in deviant peer groups, which may promote more antisocial behavior (including dating violence) to raise the males' status within the peer group (LaFontana & Cillessen, 2010). By contrast, early adolescent involvement in aggression with opposite-sex peers (as a victim or perpetrator) predicted subsequent dating violence perpetration or victimization more strongly for females than males, perhaps because girls' physical aggression may be more likely to occur in early romantic interactions. In fact, the covariance between perpetration of general peer aggression and early romantic aggression was significant for females, but not males, suggesting that males' aggression in early adolescence may be confined to same-sex peer relationships, whereas females' aggression may manifest across relationships with both boys and girls. However, many of these sex differences were of small magnitude, suggesting that many of the processes operate similarly in boys and girls. --- Implications Given that dating violence arises through processes across several ecological contexts, interventions may take place at multiple points along this dual pathway model. Because harsh discipline was a direct risk factor for later dating violence perpetration, and both harsh discipline and interparental violence contributed to involvement in violence through early adolescent beliefs, efforts should be made to prevent both types of family violence. In particular, several primary prevention programs have demonstrated success in promoting positive parenting, such as Incredible Years (Webster-Stratton, 2007) and the Nurse Family Partnership (Olds, 2006). Incredible Years is targeted toward parents of children ages 2 to 5 years old. Through a 12-to 14-week course, the program aims to improve parents' knowledge of basic parenting skills, such as giving praise, providing appropriate discipline, and playing with their children. Nurse Family Partnership incorporates nurse home visits for pregnant or postnatal low-income women to improve prenatal health, promote sensitive parenting, and support young mothers in finishing school and finding employment. Given the evidence base for these programs, they should be implemented as methods of primary prevention of poor parenting, which may lower the risk of child aggression, behavior problems, and ultimately dating violence over time. Interventions should also target the prevention of dating violence more directly. For example, the Safe Dates Project is a primary and secondary dating violence prevention program for adolescents that addresses gender role beliefs, dating violence norms, conflict resolution skills, and help-seeking strategies for those already perpetrating or being victimized by dating violence (Foshee et al., 1996;1998;Foshee, Bauman, & Greene, 2000;Foshee et al., 2004). Our findings suggest that it may be beneficial to administer dating violence programs earlier (e.g., in 6 th or 7 th grade when aggression toward the romantic partners increases, rather than in 8 th or 9 th grade when typically implemented) and to incorporate components targeting general aggressive behavior, deviant peer influences, and victimization by early romantic aggression. Indeed, several dating violence prevention programs target youth beginning in middle school. One program that has been particularly successful in raising awareness of dating violence among middle-school students is Shifting Boundaries (Taylor, Mumford, & Stein, 2015). Shifting Boundaries includes interventions at the classroom level (facilitated through a six-session structured curriculum on dating violence and associated laws), the building level (facilitated through informational posters hung around the school, monitoring of unsafe areas by faculty and staff), or both. Preliminary findings in testing the effectiveness of the program have shown that the intervention has increased student knowledge about dating violence, reduced the incidence of sexual harassment among students, and lowered perpetration and victimization of sexual and physical dating violence when assessed at sixmonth follow-up (Taylor, Mumford, & Stein, 2015). Given the success of programs like Shifting Boundaries, which target younger students and incorporate interventions at multiple levels, such programs should serve as models for interventions implemented in middle schools nationwide. Finally, the investigation of sex differences has illuminated areas of potential focus for males vs. females. To allow for more tailored attention to issues that are more salient for males vs. females, student groups could be split up by gender to allow for separate discussions of issues more salient for boys vs. girls. For example, since pro-violent beliefs were more strongly linked to early romantic aggression for females, discussions among females could focus on the importance of equal respect in relationships regardless of gender, as well as general expectations for behavior in relationships. Because deviant peer affiliations were more strongly linked to dating violence for males than females, discussions in the malesonly group could focus on peer pressure and overcoming the need to imitate behaviors of peers that are harmful to others. Including such sex-specific components to dating violence prevention programs may be a valuable addition to targeting specific vulnerabilities for males vs. females. --- Limitations and Future Directions The findings should be interpreted in the context of the study's limitations. First, the measurement of some constructs was limited. For example, harsh discipline was assessed with only three items, and the measure of beliefs about general violence had low reliability. These measurement limitations may have weakened associations between these and other variables. Also, all variables were informed only by adolescent self-report, which may lead to under-or over-reporting of certain variables (particularly family violence) and shared method bias. Thus, our results should be replicated with additional informants for certain constructs (e.g., peer or teacher reports of aggression and peer deviance; partner reports of dating violence). Also, general aggression and beliefs about general and dating violence were not measured at Wave 1; therefore we could not control for previous levels of these variables in the model. Furthermore, some of the directional paths were cross-sectional rather than lagged (i.e., the links from beliefs about violence to early romantic aggression, peer aggression, and deviant peers). Ideally, we would be able to more strongly test the causality of the directional relationships proposed if all paths were lagged. Thus, it is important to replicate the results of the current study using data spanning a greater number of assessment points. Moreover, our measure for interparental violence did not incorporate the sex of the parent committing violence. Because the downstream impact of witnessing interparental violence on adolescents' own violence may vary based on the sex of the parents involved as well as the directionality of violence (Temple, Shorey, Tortolero, Wolfe, & Stuart, 2013), lack of accounting for the parents' sex may have contributed to a lack of associations between interparental violence and mediating variables. Furthermore, some of the sex differences obtained for pathways and covariances were quite small and may not have great practical significance. Finally, the results may not generalize to racial/ethnic groups not included or underrepresented in our sample of primarily African American and Caucasian youth. --- Conclusion This is one of the first investigations to integrate risk factors at the family, individual, and peer domains into a cohesive model of dating violence development over a seven-year period from pre-adolescence to late adolescence. The results revealed that the effect of preadolescent exposure to harsh discipline on late adolescent dating violence perpetration can be explained by pro-violent beliefs, aggressive behavior, and affiliation with deviant peers; these paths were generally stronger for males. Other risk factors for dating violence included beliefs in favor of dating violence and victimization by opposite-sex peers in early adolescence, with these risk factors being more prominent for females. The proposed dual pathway model, which links pre-adolescent exposure to family violence with early adolescent violent beliefs and aggression, and ultimately with late adolescent dating violence, provides a comprehensive theoretical framework for the development of adolescent dating violence and can be used to guide future research and intervention development. Conceptual model of the proposed dual pathway model. The "general violence" pathway is outlined in black, while the "early romantic aggression" pathway is outlined in grey. Linkages between the two pathways are shown with dashed paths. Correlations between Independent and Dependent Variables 1. --- 2. 3. --- 4. 5. --- 6. 7. --- 8. --- 9. 10. --- 11. 12. 1. Harsh discipline (W1) -- --- Interparental violence (W1).18 * -- --- General violence beliefs (W2

Risk factors for adolescent perpetration of or victimization by dating violence stem from different levels of adolescents' social ecologies, including the family, individual, and peer domains. However, these multiple risk factors have not been fully integrated into a single comprehensive model of dating violence development. The present study examined prospective links between exposure to family violence in pre-adolescence; pro-violent beliefs, aggression, deviant peer affiliation, and aggression toward opposite-sex peers in early adolescence and dating violence in late adolescence. Using a longitudinal study of 461 youth (51% female; 80% African American, 19% Caucasian, 1% other ethnicities), path modeling evaluated a theoretically-developed dual pathway model involving a general violence pathway and an early romantic aggression pathway. Each pathway links exposure to family violence in pre-adolescence with early adolescent proviolent beliefs and/or aggressive behavior. In both pathways, pro-violent beliefs may reinforce aggressive behaviors between same-sex and opposite-sex peers, as well as strengthen bonds with deviant peers. In the last part of both pathways, aggressive behavior and peer deviance in early adolescence may contribute directly to late adolescent dating violence perpetration and victimization. The findings provided support for both pathways, as well as sex differences in the model.

Introduction Cultural beliefs and practices can influence how provision of social support (SS), in its many forms, affects SS providers. While cultural differences in the effects of SS receipt have been studied in recent years [1][2][3][4][5], the interaction between the dynamics of SS provision and culture has been left largely unexamined. Notable exceptions include investigations into SS provision [1,6]. In the present study, we investigated which, if any, of the recognized cross-cultural differences in SS receipt have parallels in SS provision, and how SS provision and culture interact. To accomplish this, it is necessary to first define SS, and to outline the ways in which SS receipt varies by culture. --- Social support Social support is critically important for health and well-being throughout the entire lifespan [7]. SS can come in practical or problem-focused forms including material or informational aid, or through emotional means, which make the recipient feel understood, cared for, and part of a mutually beneficial social network [8]. In addition to its emotional benefits, SS is also one of the most widely recognized buffers of physical illness and mortality [9]. Previous research indicates that individuals who receive very little SS, or who are socially isolated suffer a 50% to 91% greater risk of mortality, a factor that predicts mortality about as well as alcohol consumption and smoking [10]. For example, Yang and colleagues found that ratings of feelings of social connectedness predicted lower systolic blood pressure, body mass index, waist circumference, and C-reactive protein levels, all of which are well established biomarkers of physiological health [11]. Although the need for SS appears to be universal, the nuances of SS exchange, such as the type, frequency, intention of the provider, and identity of the recipient can all be colored by cultural context [12]. --- Cultural differences in SS exchange Culture is the system of meanings, folk beliefs, values, practices, and customs of a group of people. Living within a culture necessarily involves repeated engagement in culturally specific behaviors and cognitions, a pattern that affects psychology at every level, from social interaction to neuroanatomy [13]. Culture has also been shown to influence both SS-seeking and SSproviding behavior [6,7]. In particular, prior research suggests that cultural values and SS norms can influence the perceived appropriateness and ratings of effectiveness of various forms of SS. Perhaps the most commonly studied dimension on which cultures are compared along is the independence-interdependence continuum. This dimension captures the degree of importance that relationships and group membership play in one's self-construal, or view of one's self. In more interdependent contexts, individuals are more likely to think of themselves in terms of their role in important relationships ("I am Michael's sister"), or in terms of group identity ("I am an American." [14]). In more independent cultures, the individual is seen as more autonomous and the self as less "embedded" within relationships and the larger society. Values such as individualism and autonomy are emphasized, and the needs of the individual may be prioritized over those of the group. In independent cultures, social schemas in which the self serves as the primary referent of thought, action, and emotion are more dominant [15]. Independence is characteristic of modern Western cultures such as the United States, whereas interdependence is commonly associated with East Asian cultures like Japan and China. Interdependent cultures emphasize values like communalism, cooperation, and collectivism. Consequently, those in more interdependent cultures generally place greater emphasis on the maintenance of group harmony and prioritize problems of the group. The present study compares the relationship between specific emotions and SS provision in the highly independent culture of the United States and the relatively more interdependent culture of Singapore [16]. We expect that differences in culture will promote differences in the styles of SS that participants from each culture will provide. Although the primary focus of this study is to compare the dynamics of SS provision with those of SS receipt rather than to study cultural differences or speculate upon what mechanisms drive them, we suggest that differences in our Singaporean and American sample are, by definition, cultural differences that can be partially explained by the two cultures' differences along the interdependence-independence continuum. --- Emotion versus problem-focused support Social support is often categorized as either emotion-or problem-focused. Problem-focused support refers to SS that is intended to help the recipient eliminate the stressor [17]. Typically, problem-focused SS is provided by sharing information that helps resolve the stressor (advicegiving), or by providing instrumental assistance such as giving a ride to the airport or loaning money. In contrast, emotion-focused support refers to SS intended to assist the recipient in coping with the negative emotions and stress caused by the stressor. Emotion-focused SS includes supportive acts such as comforting, expressing affection and providing encouragement. Research on the relationship between SS and culture suggests that in more interdependent cultures, SS exchanges tend to be more problem-focused, that is, support that is more aimed at addressing the stressor itself. In contrast, SS exchange in more independent cultural contexts tends to focus on addressing recipients' emotional needs through approaches such as comforting and esteem-boosting [1,7]. In interdependent cultures, requiring emotional SS can be seen as disruptive to group cohesion, but problem-focused SS is not. In independent cultures, emotional SS can be used to bolster self-esteem, which is seen as an important trait in independent cultural contexts. In contrast, problem-focused SS in independent cultures can undermine an individual's sense of self-efficacy. --- Solicited versus unsolicited support One important characteristic of SS provision is whether or not the recipient asked for support. Taylor, Sherman, Kim, Jarcho, Takagi, & Dunagan found that Asians and Asian Americans requested less SS than their European American counterparts [4]. This finding was especially true of requests for emotion-focused SS. Taylor and colleagues suggested that Asians and Asian Americans may request less SS due to fear of straining relationships, feeling burdensome, loss of dignity, or disrupting group harmony. Additionally, Mojaverian and Kim found that Asian Americans reported more positive outcomes, such as higher self-esteem and less stress, when receiving unsolicited support than solicited support, whereas there was no difference in outcomes for European Americans when comparing receipt of solicited SS to receipt of unsolicited support [3]. Further, Taylor, Welch, Kim, and Sherman found that SS recipients from more interdependent contexts reported greater stress and negative emotions when support was requested than when they received unsolicited support [5]. Taken together, this research suggests that within interdependent cultures, it is less common and potentially less beneficial to receive SS that has been explicitly requested. --- Receipt versus provision Much of the research we have reviewed on cultural differences in SS exchange has emerged from research focused on SS recipients. In some cases, these findings provide clear insight into how SS provision may vary across cultures. For example, research indicates that within highly interdependent cultures, a greater proportion of SS receipt is unsolicited [5]. It stands to reason then that within the same culture, a greater proportion of SS provision must be unsolicited as well. However, the extent to which other SS receipt findings generalize to SS provision is less clear. For instance, researchers have found that within highly interdependent cultures, SS recipients are more likely to endorse feelings of burdensomeness and shame when they require or when they receive SS [18,19]. In this case, it is not obvious how recipients' feelings of burdensomeness and shame affect SS providers. They may mirror the negative emotions of recipients (i.e., feel that their recipients are an irritating burden and/or feel shame on their behalf), or the opposite could be true (i.e., they may feel especially pleased to be of service to their partners). It is also possible that cultural characteristics that lead SS recipients to feel anxious about burdening their support providers may have little effect on providers. To examine whether interactions between culture and SS receipt are paralleled in interactions of culture and SS provision, we have selected two variables along which cross-cultural differences in SS exchange have been observed and supported. Specifically, we examine the specific emotional correlates of SS providers who provided both solicited and unsolicited emotional SS and informational SS, a common form of problem-focused SS. --- Culture and specific emotions Both between-culture and within-culture factors contribute to the experience and expression of emotions. Kuppens, Ceulemans, Timmerman, Diener, and Kim-Prieto refer to these cultural factors that contribute to the experience and expression of emotions as dimensions of emotional experience [20]. They describe characteristics of emotional experience that operate at the individual level, such as individual differences in temperament and personality, as intracultural dimensions. In contrast, characteristics that operate at the cultural level are intercultural dimensions. For example, those from individualistic cultures are more likely than those from collectivistic cultures to report wanting to maximize experiences of positive affect and minimize experiences of negative affect [21]. Members of interdependent cultures also tend to rate negative emotions as less harmful than do those in individualistic contexts. Cross-cultural differences in the way certain emotions are viewed can result in differences in the actual frequency and degree of emotional experiences across cultures. For example, individuals from independent cultural contexts where positive emotions are more desirable to experience than negative emotions, report experiencing positive emotions more frequently than negative emotions because they are considered to be more favorable within that cultural context [22]. One reason why members of different cultures may experience certain emotions to different extents is through a mechanism called situational selection. Harmon-Jones, Harmon-Jones and Summerell note that individuals differ in how positively or negatively they judge different discrete emotions [23]. They posit that an individual with an extremely negative attitude toward anger may engage in emotional situational selection, or the deliberate avoidance of situations that are likely to cause them anger, and thereby experience less anger than someone who holds a more neutral attitude toward the emotion. Likewise, if a culture strongly favors (or disapproves of) a particular emotional state then the individual members of the culture may engage in emotional situation selection to change the odds of experiencing the emotion. In this way, entire cultures may engage in situational selection, resulting in patterns of crosscultural differences in the frequency and degree of experiencing certain discrete emotions. Similarly, in a cross-cultural study of values and emotion, Tamir and colleagues found that participants reported wanting to feel more of the specific emotions that corresponded with the values they endorsed [24]. For example, participants who endorsed the value of self-enhancement (a classically individualistic value) reported a stronger desire to feel the value-consistent emotions of anger and pride. These findings suggest that values precede and shape emotional experience, and that insofar as cultural context shapes individual values, it can also affect the desirability of certain emotions. People can experience a wide variety emotions, but Diener, Smith, and Fujita suggest that all possible emotions experienced fall under one of six discrete emotion categories [25]. They suggest love, joy, fear, anger, shame, and sadness capture the complete range of human emotion. Diener and colleagues derived these six categories from cognitive [26,27], biological/ evolutionary [28,29], and empirical [30,31] perspectives. For the current study, we will consider the emotions of affection, happiness, anxiety and worry, irritation/anger, shame and embarrassment, and sadness, closely mirroring Diener et al.'s six main emotional categories. --- Emotional outcomes of support provision Inagaki and Orehek suggest that as long as two boundary conditions are met, the provision of SS can be an inherently rewarding experience [7]. First, support must be given freely. That is, support must be given without coercion via interpersonal or societal pressures. Second, providers must believe that the support they are providing is effective. According to Inagaki and Orehek, when these two conditions are met, providers can experience emotional and physiological benefits similar to those enjoyed by the recipients of responsive SS. When considering whether cross-cultural differences in SS receipt can be used to inform expectations for SS providers, at least two possibilities emerge. The first is that cultural differences in provision will tend to mirror those of receipt. Culturally inappropriate or culturally non-normative styles of support may impose more stress upon providers, leading to worse emotional outcomes. Providing styles of SS deemed inappropriate within a provider's culture may also negatively influence providers' sense of the effectiveness of SS provision, leading to further negative emotions. The second possibility is that the cultural factors that shape the dynamics of SS receipt are not generally mirrored in the dynamics of SS provision, and instead operate through different structures. The present study seeks to explore whether the interaction of provider culture and SS type parallels cultural differences in SS receipt. --- The present study The present study examines the degree to which cross-cultural differences in SS provision mirror cross-cultural differences in SS receipt. Namely, we examine whether SS providers report more discrete negative emotions (anxiety, anger, shame, and sadness) and less discrete positive emotions (affection and happiness) at times when they provided forms of SS that prior research has generally identified as less culturally appropriate. The current study therefore examines differences in the patterns of SS provision and reports of specific emotions among college students from Singapore and the United States. If the dynamics of SS provision parallel those of SS receipt, it would be expected that participants in Singapore would provide more informational support, while participants in the United States would provide more emotional support. Similarly, participants in Singapore would be expected to provide unsolicited support while participants in the United States would be more likely to provide solicited support. In addition to examining differences in the characteristics of SS provision, we also tested whether the emotional correlates of SS provision varied by culture. If the emotional outcomes of providers parallel those of recipients' emotions, it would be expected that Singaporean participants would report more affection and happiness and less anxiety, anger, shame, and sadness when providing unsolicited SS and more informational SS. Likewise, we expected that American participants would report more positive and less negative emotions when they provided more solicited support and more emotional support. In addition to these main effects tests of social support characteristics on emotion, we also examined two way interactions between culture and social support solicitation, and between culture and the extent of informational/emotional SS provision. We also compared the emotional correlates of solicited and unsolicited emotional and informational SS. Finally, we examined three-way interactions testing whether cultural differences in emotional or informational SS provision were similar for solicited and unsolicited SS provision. Our reasoning for studying three-way interactions has to do with the nature of unsolicited SS provision. Because unsolicited SS is, by definition, provided without the recipient's asking, it is much more likely to meet the first of Inagaki and Orehek's first condition for mutually beneficial SS provision; that support must be given freely [7]. Without the pressure to acquiesce to recipient's requests for support, unsolicited SS is, by its very nature, given voluntarily. For this reason, we anticipated that the potential for "dysfunctional" SS provision-that is, SS provision that is associated with lower ratings of positive emotions and greater ratings of negative emotions-might be greater for instances of solicited SS. --- Method --- Participants Responses from 167 University students recruited from Psychology subject pools in the United States (52.7%) and Singapore (47.3%) received research credit for participation in this study. The sample was 73.1% women, and the gender ratio was similar in the United States and Singaporean sample. Mean age was 21.77 (SD = 2.77). The Singaporean sample was drawn from an urban private University, while the U.S. sample was drawn from a mid-sized public regional university in a small city in the Pacific Northwest. Despite these differences, there were no statistically significant differences in income or age between the two locations, though there was more variability in age in the U.S. sample. Although 179 University undergraduates participated in this study, 167 produced viable data for analyses. Because this study focused on the qualities of SS provision, we were unable to use data from the ten participants who did not report providing any SS. In addition, one participant was omitted because data collection errors made it impossible to match the responses provided on the two days of the study. Another participant only partially completed the study. This research was approved by the institutional review boards of Singapore Management University (IRB17-007-A001-117) and Western Washington University (secondary review). Written consent to participate was obtained from all participants. --- Procedure All participants took part in two different hour-long sessions in campus computer labs in Singapore and the United States. Participants took part on either a Tuesday/Wednesday or a Saturday/Sunday. Participants provided informed consent and used the Qualtrics research platform to respond to questions. Following Kahneman, Krueger, Schkade, Schwarz, & Stone's Day Reconstruction Method (DRM) procedure [32], all participants were asked to think of their previous day as a series of episodes, and complete a diary sheet listing all episodes that occurred. Participants noted the start and stop times of each episode, listed some descriptive features of the episode, and indicated whether they provided or received social support during the episode. After completing the diary sheet, participants notified a research assistant to help them initiate the Qualtrics questions related to social support provision and receipt. Next, participants responded to questions describing qualities of each episode from the prior day. In addition to the measures described in the measures section, participants indicated their main activities and social interactions for each episode, identified features of SS provision and receipt, and rated several emotions during that episode. SS was defined for participants as being networks of shared social relationships involving reciprocal caring and communication. In addition to examples of emotional and instrumental SS, participants were told that "Sometimes indirect ways of supporting another, such as keeping track of another person's situation or spending time with the other person are also forms of social support." Participants answered specific questions about the SS they provided, including the type of recipient of the SS (e.g. friend) who was subsequently referred to as the participant's "SS partner." Participants reported an average of 16.04 episodes over the two days. Social support provision was reported in 735 total episodes by 167 different participants (M = 4.34; SD = 2.62). --- Measures Descriptive statistics for the variables described in the sections that follow are shown in Table 1, together with the intraclass correlation coefficient, where appropriate. Emotional SS. This 3-item scale adapted from Maisel and Gable [33] asked participants to describe the extent to which they used emotional SS during the episode in which they had reported providing SS. Participants responded using a scale from 1 (not at all) to 7 (very much) to indicate the extent to which they tried to provide emotional SS to their episode partner (M = 5.13, SD = 1.45). Items from this scale include I tried to offer comforting and encouraging words, I tried to tell my partner how much I care about them, and I tried to understand my partner. Cronbach's alpha was.80 for participants from Singapore and.75 for participants from the United States, and the overall distribution was negatively skewed. Informational SS. This 2-item scale asked participants to describe the extent to which their social support provision during that episode involved providing informational SS. Participants responded using a scale from 1 (not at all) to 7 (very much) to indicate the extent to which they engaged in each informational social support behavior (M = 4.90, SD = 1.89). Items from this scale include I tried to give specific suggestions about how to solve the problem and I provided my partner with advice to help them deal with the problem. Cronbach's alpha was.95 for participants from Singapore and.96 for participants from the United States. The overall distribution was somewhat negatively skewed. SS request. This measure assessed whether episodes of social support provided by the participant had been requested by the social support recipient. Participants were asked to respond with Yes (coded 1) or No (coded 0) to the question Did the other person ask for support? Emotions. At the start of each episode participants reported the extent to which they had experienced 13 distinct emotions during the episode, using a scale from 1 (not at all) to 7 (very much). Only the eight measures that aligned with the emotion typology described by Diener, Smith, and Fujita were considered as part of this study [25]. Specifically, we evaluated the extent to which culture and SS characteristics predicted affection, happiness, irritation/anger, anxiety, worry, embarrassment, shame, and sadness. Note that this list of emotions has been previously used in cross-cultural research using the Day Reconstruction Method [32]. Because the distribution of each negatively valanced emotion was positively skewed, a natural logarithm was calculated to help reduce the effect of extreme negative emotions scores on the analyses. All tests of negative emotions were conducted both using the original metric and the logtransformed variables. The two positively valanced variables were not severely skewed, and no transformations were used. Two multilevel modeling analyses were used to test whether culture predicted differences in the continuous variables of informational and emotional SS provision. Location did not predict the amount of reported emotional SS provision (b = -.12; t(164) = -0.67, p =.505) or informational SS provision (b = -.18; t(164) = -0.88, p =.380). --- Results --- Cultural differences in the frequency and characteristics of SS provision --- Emotional correlates of SS provision Data analysis overview. All analyses predicting discrete emotions were conducted using multilevel modeling to account for the nested data structure. Specifically, variables associated with specific episodes (including emotions and all social support provision characteristics) were analyzed at Level 1, while characteristics of the individual (i.e., culture) were analyzed at Level 2. All Level 1 variables were group mean centered prior to analyses. For consistency, random effects for each variable were initially tested and were included in all subsequent analyses if the random effect was statistically significant, using p <unk>.10. Analyses considering emotional SS provision and informational SS provision were conducted separately for each emotion, yielding the 16 different combinations of SS type and emotional outcome (affection, happiness, irritation/anger, anxiety, worry, embarrassment, shame, and sadness). The large number of complex analyses that were conducted as part of this study raises concerns about capitalization on chance. For descriptive purposes, Tables 2 and3 provide an indication of regression coefficients that are statistically significant at p <unk>.01, p <unk>.05, and p <unk>.10. These values are useful for understanding patterns observed across multiple variables and may be of interest to the reader. However, because of the large number of analyses we only interpret results that are statistically significant at p <unk>.01. For parsimony, most of the regression coefficients and standard errors are presented only in the Tables, and the bulk of this section interprets the results without repeating the values that were not statistically significant. The formulas below summarize the main analyses that test the main effects of social support provision (informational or emotional), whether that support was requested, and location (Singapore or United States), as well as the two way interactions between SS provision and SS request, between location by SS provision, and between location by SS request, as well as the three-way interaction between location, SS provision, and SS request. Decisions about whether Level 1 predictor variables should be modeled as fixed or random effects were determined through preliminary analyses looking only at the Level 1 variables. Level 1 : Emotion ij 1<unk>4 p 0j <unk> p 1j <unk>Social Support<unk> <unk> p 2j <unk>Request<unk> <unk> p 3j <unk>Social Support x Request<unk> <unk> e ij Level 2 : p 0j 1<unk>4 b 00 <unk> b 01 <unk>Location<unk> <unk> r 0j p 0j 1<unk>4 b 10 <unk> b 11 <unk>Location<unk> <unk> r 1j p 0j 1<unk>4 b 20 <unk> b 21 <unk>Location<unk> <unk> r 2j p 0j 1<unk>4 b 30 <unk> b 31 <unk>Location<unk> <unk> r 3j In the Level 1 formula above, emotion ij, is person j's specific emotion at time i. That score was predicted by <unk> 0j, the person-level intercept for that emotion, by <unk> 1j, which is the effect of SS provision (emotional or informational) for person j, by <unk> 2j, the person-level effect of SS request, by <unk> 3j, which is the effect of and the centered social support by request interaction, and by e ij which is error for person j at time i. Variability at Level 2 is captured by estimates of the intercept (<unk> 00 ), the average of each level 1 effect across participants (<unk> 10 through <unk> 30 ; i.e., <unk> 10 is the cross-person average slope of SS provision on emotion), the effect of location on the intercept(<unk> 01 ), interactions between location and SS provision and SS request (<unk> 11 and <unk> 21 ), and the three-way interaction (<unk> 31 ), as well as error (r 0j through r 3j ). Note that the r 0j error term indicates individual variability in the intercept, while r 1j through r 3j capture between-person differences in the magnitude of the corresponding slope; these random factors were included only when preliminary analyses indicated between-person variability at p <unk>.10 (as shown by the underlined values in the Tables). The results of these analyses are summarized in Table 2 for emotional SS analyses and Table 3 for informational SS. Analyses were conducted using the log-transformed negative emotion variables. Unless otherwise noted, all results are similar when conducted without the logarithmic transformation of the dependent variable. To aid in interpretability, estimated values in the Figures were calculated using non-transformed variables. In addition a set of supplementary analyses tested whether the results remained consistent if the type of recipient (family member, friend, romantic partner, or acquaintance) of SS was statistically considered. Cultural differences in the target of the SS provision were considered as a possible alternative explanation for the observed cultural differences in social support. This step was important because even though most SS was provided to friends or to romantic partners in both samples, preliminary analyses indicated that there were cultural differences in the targets of the SS. Specifically, those in Singapore were relatively more likely to provide SS to family members, while those in the United States were especially likely to provide SS to acquaintances. Tests were conducted by using a set of three dummy coded variables to indicate whether the participant reported providing support to friends (the reference category), family members, romantic partners, or acquaintances. The dummy coded covariates were used to predict the <unk> 0j at level 2 in the formulas above. Because statistically considering SS recipient did not alter the effects described below and reported in Tables 2 and3, these tests are not presented in this manuscript. Details on these analyses are available upon request. Cultural differences in the effects of emotional SS provision and solicitation. Emotional SS provision predicted participant ratings of greater affection (b =.62, p <unk>.001) and happiness (b =.35, p <unk>.001), as well as less irritation/anger (b = -.06, p =.006). Unique relationships between emotional SS provision and anxiety, worry, and embarrassment were not statistically significant in these multivariate models. There was no indication that emotional SS provision predicted sadness or shame. Likewise, with the exception of anger, which needs to be interpreted in the context of a 3-way interaction, tests of SS request indicated that emotions differed based on whether or not the social support had been requested by the recipient, at p <unk>.01. Further, there was no suggestion that discrete emotional correlates of emotional SS provision differed based on whether that support was or was not solicited, as evidenced by the non-significant interactions between emotional SS provision and whether or not SS was requested. For irritation/anger only, the main effect of emotional SS provision and request (as well as their interaction) should be interpreted in the context of a 3-way interaction between culture, emotional SS provision, and whether the SS was solicited by the recipient (b = -.25, p =.002). As shown in Fig 1, more provision of emotional SS predicted less irritation/anger overall. However, at times when SS was requested, the emotional consequences of emotional SS provision varied by culture. For participants from Singapore, reports of irritation/anger were highest when participants reported few attempts to provide solicited emotional SS and irritation was lowest with more emotional SS provision (simple slope b = -.49, p <unk>.001). In contrast, for those in the U.S. sample, requested emotional SS provision did not predict irritation/anger (simple slope b = -.10, p =.362). Although this cross-over interaction pattern was not evident for unsolicited SS, unsolicited emotional SS provision was associated with relatively less anger both in Singapore (simple slope b = -.13, p <unk>.042) and in the U.S. sample (simple slope b = -.14, p =.018). Likewise, there were no other main effects or two-way interactions related to SS request and culture on other emotions. Although the 3-way interaction pattern was not observed for the effects of emotional SS provision on any emotion other than irritation/anger, as described below similar results were obtained in tests of the emotional correlates of informational SS provision. Cultural differences in the effects of informational SS provision and solicitation. A summary of the analyses of the provision of informational SS, as shown in Table 3, shows few consistent direct effects of culture, the provision of informational SS, or requests for SS on any of the emotional outcomes. However, a similar 3-way interaction to the one described above was observed for anger (b = -.19, p =.010) and embarrassment (b = -.08, p =.001), and a similar trend existed for anxiety (b = -.14, p =.039). Each of these effects was also statistically significant for tests of the emotional outcome without a logarithmic transformation. Although Fig 2 only shows the pattern of estimated effects for embarrassment, the direction of the effects is similar for irritation/anger. Specifically, in both cultural contexts, there was no link between unsolicited informational support provision and negative emotions. However, when support was requested, the emotional correlates (i.e., embarrassment and irritation) of informational SS provision varied by culture. In the Singapore sample, relatively more provision of solicited informational SS predicted less embarrassment (and less irritation), while in the U.S. sample more provision of solicited informational SS predicted more embarrassment and irritation. --- Discussion Our primary interest in this study was observing whether cultural differences in SS provision would follow similar patterns to those previously observed within SS receipt, or if patterns for SS providers would differ from the findings in previous research on SS recipients. If SS provision processes paralleled SS receipt processes, participants in Singapore would be expected to provide more informational support and more unsolicited support to others, and participants in the United States would provide more emotional support and more solicited support to others. Likewise, participants in Singapore would be expected to report more affection and happiness and less anxiety, anger/irritation, worry, embarrassment, shame, and sadness when providing informational and unsolicited support, and participants in the United States would report similar emotions in instances in which they provided emotional and solicited SS. Overall, the patterns of cross-cultural differences in SS provision of the present study do not mirror cross-cultural differences in patterns of SS receipt, and raise the possibility that cross-cultural differences in emotion responses to SS provision operate differently than for SS receipt. Rather than expecting provider and recipient dynamics to mirror each other, equity theory considers SS to be an equitable exchange, where each member of the dyad brings their own dynamics to the exchange, meaning the emotions and SS behaviors that each member of the dyad experiences during the interaction do not need to be exactly the same [34]. In the context of SS, this means that patterns of SS provision and patterns of SS receipt do not need to perfectly match for a SS exchange to be successful. Similarly, SS providers and SS recipients do not need to experience the same emotions for a SS exchange to be successful. The findings from this study support the idea that even within the same SS exchange, providers and recipients may have different emotional outcomes, and this may influence the perceived success of the SS exchange. When we examined only episodes in which SS was provided, our findings did not adhere to our expectations drawn from previous research. First, although the direction of the difference between solicited and unsolicited SS was anticipated, the proportion of support that was requested did not differ significantly between Singapore and the United States. Secondly, the amount of informational support provision and the amount of emotional SS provision did not differ between Singapore and the United States. These results differ from previous research that has suggested that SS exchange in interdependent cultures such as Singapore tends to emphasize problem-focused SS, such as informational SS, whereas SS exchange in more independent cultures such as the United States tends to emphasize emotional SS. Our findings suggested that there were some notable cultural differences in patterns of SS provision. First, participants in Singapore provided SS to others overall in more episodes than participants in the United States. This is in keeping with previous findings regarding SS exchange in more interdependent cultures. Though most SS was provided to either friends or to romantic partners in both Singapore and the United States, individuals in Singapore were relatively more likely to report providing SS to family members, while individuals in the United States were relatively more likely to provide SS to acquaintances. This is likely due to the fact that many college students in Singapore live with their families, while this is not necessarily the case in the United States. The Singaporean participants may have had more exposure to family members than the American participants, allowing them more opportunities to provide SS to family. Similarly, many college students in the United States have jobs or engage in multiple extracurricular activities outside of school, whereas this is less common in Singapore. Participants in the United States may interact with acquaintances on a more regular basis than participants in Singapore, allowing them more opportunities to provide SS to acquaintances. Our findings indicate that regardless of cultural context, when providing emotional SS, participants reported greater affection and happiness, as well as less irritation/anger and anxiety. Fostering happiness and affection and lessening negative emotions are key characteristics of emotional SS. Therefore, it makes sense that providing emotional support would be related to greater happiness and affection. If emotional SS is genuinely expressed, it is not surprising that providers experienced greater positive emotions and less negative emotions. After all, research on SS suggests that emotion-focused SS provision results in more positive outcomes than problem-focused SS provision [6]. However, regardless of cultural context, greater informational SS provision did not predict any of the emotional outcomes. It is possible that there is more contextual variability when it comes to providing informational SS, particularly if these emotional outcomes are not being considered in a cross-cultural context. This could be due to the fact that informational SS can be used to address a wide range of problems, from providing advice on how to resolve a conflict with a romantic partner to information on how to best prepare for a presentation at school or work. This means that informational support might include small, easy acts of providing information, or large acts that require a great amount of time or resources on the part of the provider. Informational support therefore can require varied

Most prior research on culture and the dynamics of social support has focused on the emotional outcomes for social support recipients. Though an existing body of research has identified cross-cultural differences in the emotional correlates of receiving different types of social support, researchers have seldom examined possible cultural differences in the experience of social support providers. This study used the Day Reconstruction Method to examine cultural differences in the emotional correlates of the provision of solicited and unsolicited and emotional and informational social support in the daily lives of Singaporean (n = 79) and American (n = 88) participants. Singaporean participants reported providing more social support overall. Regardless of culture, participants reported more positive emotion (affection, happiness) and less negative emotion (anger, anxiety) when they provided emotional social support. Also, multilevel modeling analyses revealed a 3-way interaction between culture, social support provision, and social support solicitation, indicating cultural differences in negative emotional responses to providing solicited social support. Specifically, results suggest that attempts to provide more solicited social support were associated with more negative emotions in the U.S. In contrast, provider negative emotions were highest in Singapore when the provider did not meet the recipient's request for support. Patterns of cultural differences in social support provision are dissimilar to-rather than simply mirroring-those found in published research on social support receipt, highlighting the importance of studying social support provision as a distinct phenomenon.

less irritation/anger and anxiety. Fostering happiness and affection and lessening negative emotions are key characteristics of emotional SS. Therefore, it makes sense that providing emotional support would be related to greater happiness and affection. If emotional SS is genuinely expressed, it is not surprising that providers experienced greater positive emotions and less negative emotions. After all, research on SS suggests that emotion-focused SS provision results in more positive outcomes than problem-focused SS provision [6]. However, regardless of cultural context, greater informational SS provision did not predict any of the emotional outcomes. It is possible that there is more contextual variability when it comes to providing informational SS, particularly if these emotional outcomes are not being considered in a cross-cultural context. This could be due to the fact that informational SS can be used to address a wide range of problems, from providing advice on how to resolve a conflict with a romantic partner to information on how to best prepare for a presentation at school or work. This means that informational support might include small, easy acts of providing information, or large acts that require a great amount of time or resources on the part of the provider. Informational support therefore can require varied levels of provider investment, likely involving different emotional outcomes. The positively and negatively valenced emotional outcomes for providers of informational SS may balance out across situations and people, resulting in null overall effects. When considering culture, specific emotions, SS requests, and SS provision together, the results indicated an interesting interactive relationship. Overall, episodes with more emotional SS were characterized by more affection and happiness and less irritation/anger and anxiety. However, when considering episodes in which emotional SS was requested, only ratings of irritation/anger varied significantly by culture. In Singapore, ratings of irritation/anger were lowest when participants reported more provision of solicited emotional SS and greatest when participants reported less provision of solicited emotional SS. In contrast, in the United States, irritation/anger was lower with less provision of solicited emotional SS and highest with more provision of solicited emotional SS. It could be that participants in Singapore who provided relatively more emotional SS in response to their partners requests did so because the requests seemed important. Therefore, there would not be a reason to feel particularly irritated/angry. However, if in Singapore, a SS partner requested emotional SS that the participants did not consider to be necessary or appropriate, the provider would not provide as much support, and might feel anger/irritation at the request. In contrast, within the U.S. sample, participants may have felt irritated/angry when asked to provide more emotional SS. This pattern makes sense within the context of what we know about cultural differences in SS. The goal of emotional SS is reassurance and esteem-building, but in an interdependent cultural context, this type of support can violate cultural expectations of modesty and appropriate emotional expression [35]. In addition, work on cultural differences in the importance of saving face suggests that negative emotions may arise when group members break face, meaning that they fail to appropriately follow norms related to hierarchy and societal expectations [36]. Losing face can be considered selfish, rude, and inappropriate, as it causes discomfort for all who are involved in the social interaction [37]. All of these factors could result in SS providers in Singapore feeling angry/irritated at being requested to provide a form of SS that both members of the SS exchange would recognize as being culturally inappropriate. However, an exception could be made in the instance of a request for emotional SS that seemed especially important, because knowing the inappropriateness of a request for emotional SS would probably make a SS recipient less likely to request it unless they felt that emotional SS was truly needed. Requests for emotional SS in the U.S. context may have made participants feel overburdened, and may have made the provision feel less voluntary, violating Inagaki and Orehek's first criterion for beneficial SS provision: support must be given freely [7]. This pattern did not hold true for unsolicited emotional SS, further supporting the idea that providing inappropriate or burdensome forms of SS would lead SS providers to experience more irritation/anger. Similarly, provision of unsolicited SS was not associated with negative emotions for providers in either culture, whereas informational SS that was requested was associated with negative emotions. This association between solicited informational SS provision and negative emotions varied by culture. In Singapore, greater provision of solicited informational SS was related to less irritation/anger and embarrassment. In the United States greater provision of solicited informational SS was related to more irritation/anger, anxiety, and embarrassment. Again, culture moderates the association between solicited SS provision (in this case informational SS provision) and worse emotional outcomes for providers. For both cultures, unsolicited SS provision did not carry the same costs to providers as solicited support provision. Alternatively, participants in Singapore may have felt fewer negative emotions when providing informational SS because informational support is a more culturally appropriate form of support. Participants in the United States may have experienced more irritation/anger and embarrassment when providing informational SS because they provided a less culturally appropriate type of support to their partners. Rather than expecting identical input and emotional outcomes for providers and recipients, equity theory posits that both parties should perceive the exchange of resources as being equal [38]. If a provider feels as if the SS receiver is requesting too much support, or requiring more of the provider than they are willing to give, the exchange is not equal and can have negative outcomes for the provider. This concern is less prevalent when SS is provided without the recipient requesting it, as there is less potential for the provider to feel coerced by their partner's request. The current results may differ from previous SS research because our measure of SS provision assessed the extent to which providers believed they had provided each type of SS. It is possible that what providers tend to categorize as a particular type of SS (informational vs. emotional or solicited vs. unsolicited) could be categorized differently by recipients. In this way, studying providers might have yielded seemingly different patterns of SS than have been observed in the largely recipient-focused literature. For these reasons, our findings did not neatly parallel those of the SS receipt studies. Alternatively, these findings may suggest that cultural differences are most apparent when SS exchanges are seen as inequitable or otherwise dysfunctional. Tolstoy noted of families, "All happy families are alike; each unhappy family is unhappy in its own way." Similarly, cultural differences may be most subtle when SS exchange is most beneficial and most apparent when SS exchange is dysfunctional. Making one's partner feel loved and supported may evoke the same positive emotions universally, whereas cultures may have unique patterns of response to unsuccessful or unfavorable SS exchanges. --- Limitations and future directions One potential limitation to this study is that participants were asked to self-report their experiences from only two previous days regardless of whether or not they had provided SS that day. Though Kahneman et al.'s Day Reconstruction Method [32] is an accepted procedure for collecting retrospective accounts of previous days, it is not without its limitations. Participants misremembering or intentionally omitting information is a possible concern with any selfreport study. In addition, although participants reported a large number of episodes, only an average of four episodes per person involved the provision of SS, thereby reducing statistical power, especially for tests of cross-level interactions. Future research might expand upon this study by obtaining more episodes involving SS provision and by asking participants to provide a brief explanation of the context in which SS was provided. The rich contextual information may allow a more complete understanding of potential cross-cultural differences in emotional responses to SS provision and the Day Reconstruction Method would be a useful tool for capturing this kind of information. Likewise, analyses involving more episodes of SS provision would provide a more powerful replication of these results; the results reported here should be considered tentative, pending replication. Future studies exploring cross-cultural SS should build upon this research by providing participants with a wider variety of SS options to report on. In the current study, we used a 2-item measure of problem-focused SS that focused exclusively on informational SS. Both questions regarding problem-focused SS only involved giving advice or providing suggestions to fix the problem, and failed to include other types of problem-focused SS such as instrumental assistance, like giving someone a ride or buying them groceries. It is possible that potential cultural differences were not observed because participants were not asked about instances of instrumental problem-focused SS. Future research should be sure to include a variety of measures regarding different types of SS in order to more thoroughly capture nuances in responses. Likewise, although the multi-item SS provision measures were reliable in each cultural context, we did not use multilevel approaches to ascertain cultural measurement invariance, and it is possible that factor structures may vary by culture [39]. One final potential limitation to this study comes from the sample characteristics. The study was designed to examine cross-cultural differences in SS provision and emotional outcomes among a sample of individuals living in a independent context and a sample of individuals living in an interdependent context. However, Singapore may not be an ideal example of an interdependent culture. Singapore is a relatively young, highly diverse country that has experienced a great deal of Western influence in recent decades [16]. While previous research has indicated that Singapore is higher in interdependence than the United States is [40], Singapore may be comparatively individualistic compared to classically and often studied nations in East Asia like Japan and China. Furthermore, by sampling exclusively from a young, urban, student population, it is plausible that the Singaporeans included in our sample are considerably more individualistic/Westernized than the general population. Compared to the U.S. sample, the Singaporean sample were more likely to live at home with family rather than with friends and roommates. Singaporean participants lived in a very densely populated urban environment versus the American's suburban or campus living. These differences are not necessarily cultural ones, and-to the extent that they drove differences in SS provision-may have limited our ability to study cross-cultural differences. Because the intent of this study was to focus on personal dynamics in the process of SS provision, culture was considered a backdrop of this study rather than a focus. For this reason, this study did not include a specific measure of interdependent and independent cultural values, potentially limiting the generalizability of these results. Because of these factors, future research should replicate and extend the study of SS provision in different interdependent contexts to better capture the cultural variability between highly independent cultures and highly interdependent cultures, as well as include specific measure of cultural values to help pinpoint the cultural components that may be contributing to these findings. --- Implications and conclusions Overall, this study highlights both cross-cultural differences and cross-cultural similarities in the emotional effects of SS for providers. Although much of the existing SS research has focused on the recipient, this study focuses on the experience of the SS provider. The findings from this study support the idea that it is extremely important to consider culture when studying SS processes. A deeper understanding of the role culture plays in SS interactions has a number of real-world implications. Due to globalization, the increasing ease of travel, and social media, engaging in regular cross-cultural interactions is easier and more common than it has ever been before. Because of this, individuals around the world are able to make connections and maintain relationships with people who exist in cultural contexts that are completely different from their own. The global expansion of social networks calls for an increased understanding of the mechanisms behind cross-cultural interactions. Similarly, people are more mobile, and many individuals have the option to relocate to a country they were not born in. Integrating into a new culture has numerous implications for stress and well-being and crosscultural SS may play an important role in mitigating the negative effects of culture shock. Finally, research on cross-cultural SS can help us to better understand the nuances of acculturation in a cross-generational context. Because it is relatively common to relocate, multiculturalism within families where children are growing up in a different cultural context than their parents did, is increasingly normalized. Understanding how individuals from different cultural contexts prefer to provide and receive SS can help facilitate mutually beneficial SS exchanges among loved ones with different cultural values. Overall, these findings highlight the importance of research on cross-cultural SS provision and emphasize the need for more research on SS provision in general. --- Data is held in a public repository, at https://osf.io/6g4pz/. --- Author Contributions Conceptualization: Kendall A. Lawley, Zachary Z. Willett, Christie N. Scollon, Barbara J. Lehman. Data curation: Kendall A. Lawley, Zachary Z. Willett, Christie N. Scollon, Barbara J. Lehman. Formal analysis: Barbara J. Lehman. Methodology: Christie N. Scollon, Barbara J. Lehman. Project administration: Christie N. Scollon. Supervision: Barbara J. Lehman. Writing -original draft: Kendall A. Lawley, Zachary Z. Willett, Barbara J. Lehman. Writing -review & editing: Kendall A. Lawley, Zachary Z. Willett, Christie N. Scollon, Barbara J. Lehman.

Most prior research on culture and the dynamics of social support has focused on the emotional outcomes for social support recipients. Though an existing body of research has identified cross-cultural differences in the emotional correlates of receiving different types of social support, researchers have seldom examined possible cultural differences in the experience of social support providers. This study used the Day Reconstruction Method to examine cultural differences in the emotional correlates of the provision of solicited and unsolicited and emotional and informational social support in the daily lives of Singaporean (n = 79) and American (n = 88) participants. Singaporean participants reported providing more social support overall. Regardless of culture, participants reported more positive emotion (affection, happiness) and less negative emotion (anger, anxiety) when they provided emotional social support. Also, multilevel modeling analyses revealed a 3-way interaction between culture, social support provision, and social support solicitation, indicating cultural differences in negative emotional responses to providing solicited social support. Specifically, results suggest that attempts to provide more solicited social support were associated with more negative emotions in the U.S. In contrast, provider negative emotions were highest in Singapore when the provider did not meet the recipient's request for support. Patterns of cultural differences in social support provision are dissimilar to-rather than simply mirroring-those found in published research on social support receipt, highlighting the importance of studying social support provision as a distinct phenomenon.

INTRODUCTION In recent years there has been increasing research interest in combining social and technical networks to improve performance of a team process. A socio-technical network lies at the confluence of people, process and technology. Conway's law [9] proposed a long time ago states that organizations have a tendency to design systems that mirror their communication structures. Other researchers have suggested that software should be conceived of as a socio-technical concept [2]. They argue that design, execution and productivity of software process models can be improved by mapping the social network among developers who contribute to components of a software application onto the network of connections among the software components. It was shown [6] that the failureproneness of a software component can be predicted more accurately by considering the combined socio-technical network of a software application. Research in healthcare [15] has shown that communication breakdowns among medical professionals can lead to adverse effects on surgical patients. These breakdowns result from poor handoffs involving verbal communications and ambiguities about responsibilities. Related research in the same stream of work [14] has shown that often the members of a surgical team that performs an operation have never even met each other leading to adverse outcomes. From these and other studies there is ample evidence to suggest that the nature of the social network among participants who collaborate in a team process does have an impact on the outcome and performance of the process in terms of quality, failure rate and other measures. However, it is not clear what specific social network factors affect key performance metrics in various domains. Thus, it is important to determine which specific features of the associated social network have the most significant impact on the performance of a business process. This can lead to a deeper understanding of how social networks affect a process performance and also to actionable guidelines for leveraging social networks in practice. First, these features can be utilized to optimize resource allocation. For instance, [6] reported that software failures can be predicted by over a dozen of socialtechnical networks metrics such as centrality, reachability, and hierarchy. However, the finding would be more useful if the study can pinpoint the most significant features, e.g. reachability or centrality. If, say, reachability is the most significant factor, optimizing it among resources can be a goal when assigning resources to software development tasks. Second, it is well accepted that the resource social networks need to be in synergy with technical structure of tasks to facilitate resource collaboration. Identifying the most significant features of social networks can help organizations evaluate the fitness of social networks and align them with the technical structure of tasks. Moreover, business processes in different domains have different technical structures and task dependencies. Hence, a finding from one domain may not apply to another, and the analysis has to be done case by case. Often statistical analysis is needed to discover significant features of social networks. Following the finding, rigorous mathematical algorithms can be used to leverage these features to lead to optimal resource allocation and optimize process performance. In this paper, we wish to test whether certain features or metrics of resources' social networks contribute to overall business process performance using a real example in IT Incident Management domain. Since there are many metrics of a social network, we also desire to understand which if any of them play a larger role. We study real data from incidents and evaluate how throughput times of incidents depend on various metrics such as closeness and density among resources that work on an incident. We would also like to gain a better understanding of the varying nature and intensity of interactions among teams since any business process is a team process involving the various resources that perform its different steps. Considering a process involving n resources, at the "low-interaction end" where each interaction occurs sequentially, there are only n-1 interactions. At the "high interaction" extreme, where every resource interacts with every other resource, there are n(n-1)/2 interactions. Typical examples at this extreme are medical surgery processes where intensive team collaboration is required. The IT incident management process may be positioned in the middle of this spectrum. Simple incidents may just need pairwise interaction during handoffs, but complex issues usually require intensive team interactions such as group chats. This paper is organized as follows. In section 2 we discuss social network preliminaries. Section 3 gives applications where social networks play an important role. Our empirical study appears in Section 4. Section 5 describes how our approach can be used for resource allocation by solving optimization models. In Section 6 we generalize our methodology to other applications. Section 7 gives a literature review with a conclusion in Section 8. --- II. SOCIAL NETWORK BACKGROUND AND NOTATIONS Figure 1 shows an example of a social network. The nodes represent individual workers or resources, and the edges show the distances between connected pairs of nodes. An edge from user A to user B indicates that A and B know each other or have had interactions with each other. Note the edges are not directed. This means that each edge represents a symmetric relationship, i.e. if A knows B, then B knows A. There are a number of metrics used to analyze a social network [22], [24]. We distinguish two types of metrics: network-based metrics that measure features of the network or its sub-networks; and local metrics which measure features of a particular node or edge in a network. Network-based metrics include size, density, cohesion etc. Typical local metrics include a number of node centrality measures, e.g. node degree, betweenness of a node etc. Some metrics, e.g. closeness, can be defined either at the network level or at the local level. In this paper, since we are interested in how resource team structure affects process performance, we will focus on network-based features. Next we define some of these metrics more formally and illustrate them using Figure 1. Given a network G=(V, E, W) where V is the set of |V| vertices (or nodes), E is the set of |E| edges between a pair of vertices, and W is the weight of each edge. • Network Size is the total number of vertices in the network. The size of the network shown in Figure 1 is 10. • Density is defined as the ratio of the number of edges and the number of possible edges in the network, i.e.!!! *!!!. The density of the network in Figure 1 is 0.38. • Closeness: This metric indicates how near a node is to all the others. We first compute all shortest paths between all node pairs in the network and compute their average, i.e. Avg!,! <unk>! (d!" ), where d!" is the shortest path between vertices i and j. In Figure 1, the closeness of the network is 2.93. • Strength: the strength of a vertex is defined as the sum of the edge weights of adjacent edges of this vertex. Also, network strength is defined as the total weights of edges in the network, i.e, w! ---!<unk>!, where w! is the weight of edge e. In our example the vertex strength for Node A is 4. The strength of the entire network is 25. However, in this paper, to make results more interpretable, we use weight to represent the distance between resources and inverse of the weight as the strength. In this way, the lesser the weight, the shorter is the distance between two resources, and the larger the strength between them. • Cohesion of network: this metric is defined as the minimum number of nodes needed to be removed from the network such that the network gets disconnected. The cohesion in Figure 1 is 1, as removing node F or G will disconnect the network. • Clustering Coefficient (CC) measures the degree to which nodes in a graph tend to cluster together. If node v has k! neighbors, then at most k! (k! -1)/2 edges can exist between them. If CC! denotes the fraction of these edges that actually exist, then CC is the average of CC! over all nodes. In Figure 1, CC! = 1, CC = 0.58. • Hierarchy measures the degree to which the network approaches a perfect hierarchy, i.e. degree to which all relations are unidirectional. For local metrics, for example, degree of a node specifies with how many other nodes a certain node is connected; thus, node E would be the most privileged, with 5 connections. Betweenness of a node measures the number of shortest paths that pass through the vertex. In Figure 1, despite E having more direct connections, node F is a bridge, without which the other nodes would be disconnected. Hence, it is a single point of failure, and plays the role of information broker. It helps to keep the network unit intact. The normalized betweenness of node F is 0.42, and of node E is 0.32. --- III. BUSINESS PROCESS SCENARIOS AND SOCIAL NETWORK FEATURES In this section, we introduce examples from medical treatment and software development to illustrate the I A C D B E H G F J 1 1 1 1 1 2 1 1 1 2 2 2 3 2 2 1 1 importance of alignment between social networks and business processes and discuss a few significant social network features in each example that have been explored in literature. We also describe the incident management process on which our empirical study is based. --- A. Medical Treatment Process Health care is fundamentally about social interactions. Figure 2 shows a network of interactions among healthcare professionals in a hospital environment. In this network there are professionals like surgeons, anesthetists, nurses, operating department practitioners (ODPs), home care assistants (HCAs), and administrators. The directed edges show the flow of communications among the individuals involved. Thus, Nurse manager 1 receives a large number of communications as indicated by her large in-degree. Research has shown that patterns of communication among the operating team members can affect decision making patterns, lead to communication breakdowns and drastically affect patient outcomes [15], [23], [25], [30]. In particular the effect of density, clustering and hierarchy metrics on performance is discussed in [23], [25]. --- B. Software Development Process Previous work [5], [6], [21], [28] has shown resource social networks have great impact on the quality of software. In most cases, resource social networks are an important factor to be considered during task assignment. Figure 3 shows development tasks in a software development project. These tasks are associated with each other due to shared artifacts (e.g. data tables, code, user interface etc.). To reduce dependencies and also improve efficiency, a common practice is to group tasks into clusters and try to assign a cluster of tasks to one resource or resources with strong ties. For the example shown in Figure 3 (a), these tasks may be grouped into three clusters, as highlighted in different colors. Moreover, some tasks, e.g. T6 and T7, may be identified as critical ones as they have high degree associations with others. We can leverage social network features to optimize resource allocation. A community is a group of resources that is densely interconnected. Usually, resources within a community share common interests, e.g. database, BPM communities etc. Figure 3(b) shows the corresponding social network of resources in the software development project. These resources can be clustered into three communities as highlighted in different colors. For example, tasks T1, T2, T5, and T6, all related to data model development, can be assigned to resources in a database community consisting of resource nodes 1-5. In addition, previous studies [6], [21] showed that resource closeness, betweeness, and degree metrics are significantly correlated with the proneness to software failures. These metrics need to be considered during task assignment. Further, a critical task can be given to critical positions such as a key contributor, coordinator, or influencer [20] in the community, so they can use their social position effectively to improve process performance. Thus, task T6 may be assigned to node 4 as this resource is the center of the community and also bridges communication between communities. A formal method for resource allocation by utilizing social networks is discussed later. --- C. IT Incident Management Process Many enterprises outsource the management of their IT systems to professional IT service providers. When facing issues with the systems, business users of the IT systems submit incidents to a service provider. Appropriate resources from the service provider are then assigned to work on the incidents and resolve the issues within timelines as defined in a service level agreement. Typically, resources are organized into different support teams by their specialty, e.g., database, operation system, etc. claiming the ownership of the incident (task "Update Incident Ownership") and then diagnoses it (task "Analyze Incident"). If needed, the resource may communicate with the customer for additional information. However, the incident may be transferred to another resource in the same support team or to another support team (task "Transfer Incident") for a number of reasons: (1) the incident may be misrouted due to insufficient information recorded; (2) the assigned resource may become unavailable; and/or (3) for the complexity nature of issues involved in the incident, multiple resources with different specialties may be required in the diagnosis and resolution. In such a case, the resource recommends an alternate resource to work on the incident or simply returns it to a dispatcher to determine the next appropriate resource. An incident is always assigned to one resource at a time. Once an incident is resolved, the resource restores the functionality of the system as required by the business user (task "Restore Service"). The business user validates and confirms the service provided by the resource (task "Confirm Service"). Once confirmed by the user, the incident is closed (task "Close Incident"). In general, for the resolution of complex issues multiple resources with different specialties may be required in diagnosis and resolution. Incidents, particularly ones regarding application outages, are often complicated as their root causes may lie in multiple layers in the IT environments, including the application, middleware, hardware, network infrastructure, etc. Thus, it may be better if a system can identify in advance a series of resources that would work on different aspects of a ticket. This is more like assigning a team of resources to a ticket. Of course, some of them may not be available when they are needed. In such a case, the team is dynamically reassigned based on the state of the ticket. It is clear from the above discussion that the various resources that resolve a severe incident work in a collaborative manner since a resource uses inputs provided by a previous resource in performing its task and may also consult with it to seek clarifications. Previous research has demonstrated the existence of such ties through social network analysis [20]. Furthermore, through empirical data, [20] shows that there are communities and workgroups in such a network and resources serve social roles such as contributor, influencer, coordinator, etc. These roles are identified based on local metrics such as degree and betweenness. Roles such as influencer and coordinator are authorities and hubs that provide advice and connect other resources together. They serve as the glue in incident resolution. A follow-up question is what networkbased metrics can be utilized during task assignment to improve performance. We continue with a formal analysis to answer this question. --- IV. EXPERIMENT: DISCOVERING SIGNIFICANT SOCIAL NETWORK FEATURES In the IT incident management process described above, each incident is handled by a group of resources including help desk agents and different support teams. We studied the process logs of 1562 incidents. The number of resources (excluding customers) involved in each incident ranges from 1 to 13, with an average of 3. IT incident management, as a knowledge intensive process, requires close collaboration between resources. The interactions between these resources, particularly, the interactions between support teams determines how efficiently a process instance can be executed. Next we use real case data to discover what specific features of resource social networks have significant impact on the process performance if they exist. For IT incident management, the most critical key performance index (KPI) is resolution time (or Time to Repair, throughput time), i.e. the elapsed time between close time and open time of an incident. In our case, the elapse time also contains the waiting time for pending customer response (i.e. execution time of tasks "Provide More Information" and "Confirm Service" as shown in Figure 4). In practice, the waiting time is not counted in measuring the performance of incident processing since it is largely out of the control of support teams. Therefore, resolution time is calculated as the elapse time deducted by the waiting time. It is well known that many factors contribute to this KPI, including incident priority level, technology domain, incident complexity, performer skills etc. Typically, a service level agreement defines the timeline for resolving incidents at each priority level. Besides these well-known factors, here we conjecture that the resource social networks can be another contributing factor. We collected 1562 closed incidents with detailed execution logs for each task and formulated a social network for resources (Help Desk agents and support teams) based on task execution logs. In this social network, each resource is represented as a node and a handoff between two resources as an edge. The weight associated with each edge is the inverse of the total number of handoffs between the two nodes. For example, if there are in total 5 incidents handed across two resources (in either direction) along the process flow, then the weight between them is 1/5. We use the weight to measure the distance between two nodes. When more handoffs occur between two nodes it implies they are closer. Other factors like execution time of consecutive tasks may also be considered in computing closeness, but we ignore them here. With this setup, we get a social network with 479 nodes and 1918 edges. The density of this social network is 0.017. In this experiment, we try to determine features of the resource social network that significantly affect the performance of this incident management process, i.e. resolution time. For this experiment, ideally, we need to first estimate the effect of those well-known factors on this KPI, e.g. priority level, technology domain difference, and problem complexity. Unfortunately, information regarding these factors is not disclosed in this data set. Thus, in order to isolate these factors as much as possible, we only consider a subset of incidents all processed by a key resource. For example, node 51 in Figure 5 worked on 158 incidents in total. We use the 158 incidents as our testing data set. By selecting all incidents that involved a single resource we can reasonably assume that these incidents were within the performer's expert domain and they had similar technical issues. Also, resource 51 serves as a technical lead in the support team and it is also identified as an influencer and a coordinator within the resource social network. As a general practice in this process, a technical lead always gets involved in high priority or complex incidents. Thus, his participation may indicate these 158 incidents have relatively high priority and complexity. Thus, with this data selection process, we can assume that these incidents are similar in terms of complexity and priority, and it is reasonable to use them for our experiment. The network constructed based on the 158 incidents is shown below in Figure 5. This network has 131 nodes and 602 edges with density 0.07. The degree of nodes ranges from 1 to 139 with an average of 9. Clearly, a few high degree nodes (i.e. with more edges) are placed in the center of the graph, surrounded by low degree nodes. In our case study, each incident is handled by a group of resources. For example, in Figure 5, we highlight 7 nodes (or resources) that collaborated on an incident. This incident was created by node 205, assigned by node 121 to a support team, picked by node 119 in the assigned support team, and transferred through a few resources from node 8 to 143 to 51, and then closed by node 274. This collaborative team forms a sub network. We are interested in exploring what features of the team, i.e. the sub network, significantly contribute to the performance of the incident resolution. Hence, we calculate social network features for the sub network of resources that collaborate on each incident, not for the entire network (or a single node) of Figure 5. Previous work suggested that network size, density, centrality, cohesion and other metrics are correlated with resource performance [4]. Consequently, we choose sub network size, density, strength, cohesion and closeness as our candidate features and use linear regression to test if any of these features is significant. The definition of these features is given in Section 2. The summary statistics of these metrics for the sub networks corresponding to these 158 incidents are shown in Table I. Table I shows that the mean resource count is 4.15, so four resources are involved in an incident on average. It takes averagely 103 hours or approximately 4 days to resolve each case. The mean density is 0.57 suggesting the typical sub network is moderately dense and 1.69 nodes must be removed to break it. The mean closeness indicates that the average shortest path length between any pair of nodes in a sub network is 0.41, no matter whether they are directly or only transitively connected. This short distance indicates resources in the sub network in general have a close relationship. The mean strength of 86.85 means that, on average there are about 87 handoffs among the resources in the sub network. We first built a linear regression model with resolution time as the dependent variable and all candidate features as features, the number of resources (res_count) and closeness are significant at the 1% and 5% confidence levels respectively, while the other features are not. In the second round, we removed the insignificant variables and only kept the two significant ones to build the linear regression model again. The results shown in Figure 7 indicate that these two variables are still significant at confidence level 0.001. This experiment provides good evidence that the number of resources involved in a case and the closeness among them have strong impact on resolution time. It is intuitive that the number of resources matters because having more resources implies more handoffs and thus a longer throughput time. Also, due to the high complexity of the incidents in this data set, an incident owner frequently uses group chats with all the people that have been involved in the case to resolve the incident. The effectiveness of such team interactions largely depends on the social distance among the resources, which can be measured by closeness of the sub network. This finding can also be demonstrated through a scatter plot as shown in Figure 8. Clearly, a large portion of points follow the plane defined by the regression model, but some points are far from it. This can be explained by the fact that resolution time is determined by many factors, not just social network features. Certainly, we cannot predict the resolution time of each case solely by social network features. In this experiment, our intention is to find social features that significantly affect process performance. Note that although we have discovered two significant social network features through the experiment, this finding only applies to the IT incident management process described in this paper. It may not always be the case for other business processes, but similar experiments can be performed to discover such features. Next, we discuss how to leverage these significant features in task assignment. --- V. MODELING OPTIMAL RESOURCE ASSIGNMENT A general approach to modeling resource assignment is to optimize with respect to the key social network features that are found by analyzing actual data from the network. If there are several metrics, the objective function is a weighted sum of each metric. We illustrate this approach in the context of incident management and software development. A social network is a graph (see Figures 1 and2) where each node represents a resource. Previous research [20] has shown that this graph can be organized into communities, and communities in turn have work groups and roles. Thus, we can label a resource node r! as: ( Tech_Role!, Com!, Grp!, Soc_Role! ) where i is the node id Tech_Role is the technical role the resource plays Com! is the community node i belongs to GRP! is the workgroup within the community i belongs to Soc_Role! is the social role of i within Com! or Grp!. Table II shows a few resources along with communities, workgroups and social roles identified in the previous work from real incident management data [20]. For example, resource 51 has "Support Team" as her technical role, she belongs to "Procure to Pay" workgroup in the community of "Purchasing & Payment", and plays a number of social roles as an influencer and a coordinator of this community and a key contributor of "Procure to Pay" workgroup. For the IT incident management process, based on our experiment, in assignment of a resource team, the goal is to find a subset of resources R!"# as a team from the available resource pool R that has the smallest closeness (denoted as C(R!"# )) among all feasible teams. This is expressed as: Min!!"# <unk>! C(R!"# ), such that for each task t, there is an assigned resource r! = ( Tech_Role!, Com!, Grp!, Soc_Role! ) that matches the task technical role, community, workgroup and social role requirements. We did not include resource count in the objective function because it is implied by closeness. Moreover, the constraints will specify the minimum number of resources required. Typically, technical role constraints are hard, while the social role constraints are soft and can be relaxed. Thus, we may relax the social constraints if the only suitable technical role that is available does satisfy them, or is from a different community or group than the one desired. Next we use a hypothetical example to illustrate this model. Assume that when a customer reports an issue, only the resources listed in Table II are available. This process shown in Figure 4 requires two technical roles, Help Desk and Support Team. Hence resource 205 is assigned to execute "Open Incident" and "Validate and Assign" tasks. Also, based on the social network shown in Figure 5, among these four support team resources, resource 121 is the nearest one to resource 205. So resource 121 is assigned to execute all the support team tasks in the process. When resource 121 has to transfer the incident, a new support team resource needs to be added. By the optimization model, this new node should be closer to both nodes 121 and 205 than any other node. In other words, among all triad sub networks consisting of nodes 121 and 205, the one with smallest closeness is optimal. In this case, resource 51 is recommended. Note this illustrative example is different than the actual case highlighted in Figure 5. This is a NP-complete assignment problem. A heuristic would find an initial pair of nodes with the minimum shortest distance between them. In subsequent steps it would enlarge this subset by adding nodes to it such that additional nodes have the shortest distance to the current subset of nodes until all the technical resource requirements are satisfied. Similar formal optimization models can also be used to assign resources in software development process. For example, based on previous work [6], [21], closeness, betweenness and degree metrics are significant social network features that can be used to predict software failures. Thus, we can formulate a task assignment model with the goal as finding a sub network (i.e. a group of resources) that has minimal weighted sum of these metrics, subject to the constraint that it matches the role requirement of each task in a process instance. Thus: Min!!"# <unk>! <unk>! C R!"# -<unk>! Bet!<unk>!!"# r - <unk>! Deg!<unk>!!"# r where C(R!"# ) is the closeness of the sub network R!"# of resources assigned to tasks, Bet r is the betweenness of resource r in the sub network Deg r is the degree of resource r, and <unk>!, <unk>!, <unk>! are weights of each metric These models can be solved using optimization software like CPLEX [16]. Initially, we can analyze historical process execution logs of the resource social network. If such data does not exist, we start with default values which may be revised periodically until the system reaches a steady state. When allocating resources for a particular process instance, the resources are first filtered based on skill, permission, availability and other hard constraints. Then the proposed models are solved to get a subset of resources that form a team for this instance. Often, the team may change dynamically based on resource availability, and the model can be solved again at an intermediate stage as needed to find new solutions. --- VI. DISCUSSION We presented an approach for identifying social network metrics that affect throughput time of a process instance and for modeling an optimization problem that can be solved to determine an appropriate resource assignment in a given environment. While there is agreement that process performance improves when the resources that participate in a collaborative process or modules of a software package are close to each other in a social network it is not clear what specific metrics of the social network play a role in delivering the improved performance. In the incident resolution process we found from our empirical study that resource count and closeness were two significant factors that had an impact on throughput time. On comparing results of social network analyses in other domains such as medical treatment and software development, it appears that these factors are not the same across all applications. Each application has a different degree and pattern of team interactivity, and a different mix of social network metrics is needed to capture it. The incident resolution process has some unique aspects that are different from those of other team processes. Here the interactions tend to be predominantly between successive resources that work on a case. Hence, the closeness among successive resources or a subgroup of resources is more important than the interactions among all resources on a team that works on an incident. This is reflected in the closeness metric. Further, the notion of a team is different from that in, say, a medical surgery process where interactions among all the team members are denser. Another difference between these two applications is that in incident resolution although an initial assignment of resources for an incident is made upon incident arrival, these assignments can change dynamically as the case progresses based on the availability of a certain resource at the time it is required and also the assessment of a resource working on the case as to who the next best resource would be. In contrast, in a surgery all or most resources have to be available for the entire duration of the procedure from start to finish. Clearly, it is hard to draw generalizations from a single empirical study and that is not our intention here. Rather it is to recognize and emphasize that different social network metrics influence the performance of cases in different application settings. By recognizing this observation, we have tried to associate different metrics with various applications where these metrics would be appropriate based on our understanding of the nature of interactions required in different applications and the results reported in literature. In the incident resolution case, group interactions between a pair or among a subset of the resources involved in a case are often an effective way to resolve complex incidents; hence, it is helpful to minimize the closeness of the sub-network. On the other hand, in surgery all surgical team members interact with one another. Hence, the density becomes an important metric [23], [25]. In software development resources who work on related modules must interact very closely. Their collaboration is tighter than that among resources in incident resolution. Hence, the closeness, betweenness, degree properties are critical [6], [21]. We also conjecture that centralization and hierarchy properties are more important where a single or a small group of resources are involved repeatedly in completion of a process, as in a strategic management process where, say, a CEO or some key people are intimately involved throughout various stages. This conjecture is supported in part by [3]. Finally, some processes are modular and consist of subprocesses. Within each sub-process the interaction is strong, but across sub-processes it is weaker. An example would be the order-to-cash process where different departments like sales, production, logistics and finance are involved in that order, and each performs its own subprocess. In this case, our hypothesis is that clustering coefficient may have significant impact on the process performance. Some limitations of our work are as follows. First, our data for the regression experiment assumed that all incidents in which a key influential resource was involved would be of similar complexity. Incident complexity varies widely and it is very hard to predict in advance how complex an incident may turn out to be. However, the fact that it passed through a certain important individual suggests that it exceeded a threshold of complexity. Second, there is a large number of social network metrics. We considered several social network metrics that we perceived to be important in the context of incident resolution in our experiment but we could not consider all of them. Thirdly, our results about significant social network metrics apply to only one domain and the extensions to other domains are conjectures that need to be verified further. --- VII. RELATED WORK The evolution of socio-technical systems dates back to the work of Trist [29] and refers to the interrelatedness of social and technical aspects of an organization. Sociotechnical theory focuses on joint optimization [8] that is, designing the social system and technical system in tandem so that they work smoothly together. Inspired by this early work, there is a new wave of interest in leveraging social computing in business process management, and particularly for allocating tasks of a process instance to resources that are mutually compatible [19]. There have been some initiatives in this line of research. First, researchers have analyzed process execution logs to understand resource social behaviors [1], [12], [20], [27]. [1] introduces a systematic approach and also a tool called MiSoN that creates social networks based on task execution logs. [12] gives rules and algorithms for finding resource interaction patterns in adhoc processes. A social network analysis for IT Incident management process was conducted in [20]. The idea of extracting social networks through organizational mining of process logs is introduced in [27] and algorithms for doing so are discussed. Second, social BPM studies how to fuse BPM practices with social networking applications to improve process performance. Studies in this area [7], [10], [13], [17] propose the integration of different social tools, e.g., instant messaging, Wikis, discussion forums, to engage stakeholders either inside or outside an organization in designing process models or collaborating on task execution. [18] proposed a system architecture and modeling notations in support of social BPM. Another concept in Social BPM is the Social Compute Unit (SCU) [11], [26], a collaborative resource unit formed dynamically during task execution. Our work aligns well with this concept. However, in contrast to the previous approach [ [26]] that identifies SCU based on task technical dependencies alone, here we proposed to leverage significant social features also to identify such optimal teams with regard to process performance. In addition, in the literature, there is a large body of work on the role of socio-technical networks in software development. Inspired by the original Conway's law [9], many studies continued to understand social network effects on resource performance [5], [6], [28]. By first building a dependency network among software modules and a contribution network among resources that contribute to modules, [6] showed by means of a predictive model that socio-technical features influence the failure-proneness, an important KPI, of the software. In [5], the authors give evidence to suggest that better communication among developers led to faster bug resolution. The collaboration patterns among developers are also studied elsewhere [28] suggesting they are important, although performance implications have not been evaluated. Socio-technical networks play a major role in healthcare. In [23], it is shown that distinct patterns and structures of communications emerge among the various medical teams' leadership based on their hierarchy and density metrics, and they affect the efficiency and effectiveness of team function. In [15] it is shown that, "serious communication breakdowns occur across the continuum of care, typically resulting from a failure in verbal communication between a surgical attending and another caregiver..." A

There is a growing interest in socio-technical networks that encompass people, process and technology. They capture connections among technical artifacts and human resources. Existing studies have shown that social networks among resources that collaborate to work on a business process instance have an impact on the performance of the instance. Hence, the social network among the resources should be aligned with the technical work structure. Thus, we propose to identify empirically the specific features of social networks (e.g. size, closeness, density, etc.) that have the most significant impact on the performance of a process. Then, we can leverage these significant social network features to optimize dynamic assignment of tasks to resources for better performance. In general, organizations can use this two-part approach to evaluate their resource social networks and shape their resource networks to achieve socio-technical network alignment. We use real data from business processes in the IT incident management domain for this study to demonstrate our proposed methodology.

a large body of work on the role of socio-technical networks in software development. Inspired by the original Conway's law [9], many studies continued to understand social network effects on resource performance [5], [6], [28]. By first building a dependency network among software modules and a contribution network among resources that contribute to modules, [6] showed by means of a predictive model that socio-technical features influence the failure-proneness, an important KPI, of the software. In [5], the authors give evidence to suggest that better communication among developers led to faster bug resolution. The collaboration patterns among developers are also studied elsewhere [28] suggesting they are important, although performance implications have not been evaluated. Socio-technical networks play a major role in healthcare. In [23], it is shown that distinct patterns and structures of communications emerge among the various medical teams' leadership based on their hierarchy and density metrics, and they affect the efficiency and effectiveness of team function. In [15] it is shown that, "serious communication breakdowns occur across the continuum of care, typically resulting from a failure in verbal communication between a surgical attending and another caregiver..." A social network analysis of two primary care practices [25] found widely differing decision making patterns as reflected in their respective network density, clustering coefficient, and centralization metrics. They hypothesize that network density would correlate positively with practice adherence to treatment guidelines, and network hierarchy would correlate negatively. The work in [30] showed that simply use of a checklist (which is similar to a process) by the senior surgeon during the conduct of a surgical operation resulted in significantly higher team communication scores in the surgical group. --- VIII. CONCLUSION While there is evidence to show that patterns of communication in a social network affect decision-making behavior and performance of a team process, it is not clear what specific metrics of the social network are important in different domains. Our empirical study using data from an incident resolution process confirmed that social network features do affect the performance in terms of resolution time, and identified the number of resources and closeness as the two metrics that affect performance in a statistically significant way. We also found in our study that other factors such as strength, density and cohesion were not significantly related to throughput. It should be noted that the influence of these factors on performance is not the same across all applications and depends upon the specific nature of the processes involved and the patterns of interactions among the collaborating resources required to complete them. Thus, we have developed models for optimizing resource assignment to tasks in a team process taking into account the nature of their interactions as reflected in the social network. Our methodology consists of two parts, where in the first part empirical data is used to determine important social network metrics and in the second part a model is applied to enable optimal resource allocation. In this paper, we tested the first part of the methodology through an empirical study to show that some social network metrics are more important than others. However, we leave the testing of the models described in the second part of the methodology for future work. Suffice it to say that the models need to be application domain and process specific, and the same models cannot be used in all situations. In future work it would be helpful to implement these models and evaluate their performance. We would also like to collect more data from processes in the incident resolution and other domains to validate our view that the relative importance of different metrics varies in different domains.

There is a growing interest in socio-technical networks that encompass people, process and technology. They capture connections among technical artifacts and human resources. Existing studies have shown that social networks among resources that collaborate to work on a business process instance have an impact on the performance of the instance. Hence, the social network among the resources should be aligned with the technical work structure. Thus, we propose to identify empirically the specific features of social networks (e.g. size, closeness, density, etc.) that have the most significant impact on the performance of a process. Then, we can leverage these significant social network features to optimize dynamic assignment of tasks to resources for better performance. In general, organizations can use this two-part approach to evaluate their resource social networks and shape their resource networks to achieve socio-technical network alignment. We use real data from business processes in the IT incident management domain for this study to demonstrate our proposed methodology.

Background The challenges faced by government, policy-makers and healthcare providers in designing and managing healthcare systems in response to an ageing population are acutely felt by rural practitioners. Demographic projections estimate a rise in the global population of older adults by 116% equating to 2.1 billion people by 2050 [1]. In Australia, the number of people aged 85 years and over is projected to double by 2042 [2]. While 28% of the general Australian population lives in rural and remote areas, 34% of older Australians (aged 65 and over) live in rural and remote areas [3]. A number of studies have already shown that the number of older adults hospitalised for trauma is growing rapidly in high-income countries [4][5][6]. In Australia, the number of major trauma patients aged 65 years and older doubled between 2007 and 2016, representing an increase from 25 to 37% of the major trauma population [4]. This demographic trend is at odds with the framework of current trauma systems, which were designed to manage younger patients with major trauma as a result of high-energy mechanisms [5]. Older patients experience greater morbidity than expected for the severity of their injuries, compared to younger patients with equivalent injuries [7][8][9]. Current protocols dictate that older patients with major trauma are transferred appropriately. Older patients who do not meet the criteria for transfers based on injury severity and mechanism, and haemodynamic instability, are typically managed locally [10]. Therefore, the responsibility of managing older trauma patients, some of whom may be at greater risk of adverse outcomes, rests with rural clinicians [11]. For older trauma patients who sustain trauma in rural areas, the risk of adverse outcomes associated with advancing age is compounded by the challenges encountered in rural healthcare such as geographic isolation, lack of resources, and accessibility [11]. Older trauma patients have a greater likelihood of death than younger trauma patients in a rural environment [7]. The need for specialised trauma care tailored to older adults is increasingly recognised. There are calls from stakeholders in the fields of surgery, nursing, geriatric medicine and trauma research, for the implementation of specialised guidelines and services for older trauma management [12,13]. New models of care are emerging, including specialised geriatric trauma centres integrating trauma surgery and geriatric medicine, as well as, proactive geriatric consultation services [14][15][16][17][18]. However, it is not certain that these models are adapted to trauma care provision in the rural setting. New care models are designed to target major trauma patients (commonly defined as an Injury Severity Score (ISS) > 12) [19]. Moreover, these are resource-intensive models, requiring specialists in fields such as geriatric medicine, and a critical mass of patients to justify the expenditure and infrastructure for these specialised services [20]. The direct translation of these models into rural areas is problematic due to the scarcity of health professionals, and specialised services such as geriatricians [21]. Thus, specific and novel care models for managing older trauma that are adapted to the rural context are needed. Little is known about the experience and challenges faced by rural clinicians who manage trauma in older adults [22]. An understanding of stakeholders' views is paramount to the effective development and implementation of a trauma system inclusive of rural communities. --- Aim The aim of this study was to explore the experiences and perspectives of clinicians who provide care to older trauma patients in rural settings. Specifically, we aimed to describe what factors healthcare professionals in rural areas perceive to enable the delivery of older trauma care; the barriers they encounter, and finally what changes may support their practice. --- Methods --- Study design We conducted a qualitative descriptive study to understand the experience of health professionals who provide care to older trauma patients in rural Queensland, Australia. Specifically, we sought to explore the enablers and barriers clinicians encountered, and what changes they want to see to support their practice. The qualitative descriptive study design was chosen as it is suited to obtaining rich descriptions from those experiencing the phenomenon under investigation which was critical to the aims of study [23][24][25]. In the trauma research literature, the definition of older adults is commonly set at age 60 years and over [26], thus this threshold was also used. With respect to rurality, we used the modified Monash model classification (MMM) of rurality, based on the Australian statistical geography standard -remoteness areas framework [27]. Participants were categorised as primarily practicing in a regional or rural area for MMM categories of three to seven, with three corresponding to towns of 15,000 to 50,000 residents, and seven to very remote communities [27]. Semi-structured interviews were conducted with individual clinicians via videoconference (Zoom<unk> video communications). The study's methods and results are reported according to 'Consolidated Criteria for Reporting Qualitative Research' (COREQ) [28] (See supplementary material, Appendix 1). The study was approved by the Monash University Human Research Ethics Committee (32479). --- Participants Participants comprised health care professionals (medical doctors, nurses, paramedics, and allied health professionals) with their primary location of practice in rural Queensland, or those practising in a metropolitan setting but providing care to patients transferred from rural areas. In 2019 in Queensland, of 21,735 medical practitioners, 28.4% practised in a rural area, and of 77,258 nurses, 31.5% practised in a rural area [29]. Participants were eligible if they had provided healthcare services within the past 24 months to trauma (defined as physical injury to the body [30]) patients aged 65 years or more. Purposive sampling was used to attain a breadth of backgrounds and experiences among participants. Recruitment to the study was facilitated by advertisements through research and professional organisations including the Jamieson Trauma Institute, the State wide Trauma Clinical Network, and the Australian Paramedics Association Queensland. Participants were invited to express their interest in taking part in the study by responding to the relevant organisation by email, the primary author (NF) then contacted the participant to confirm eligibility, provide further information on the study, and schedule an interview. --- Data collection The interview guide was developed based on the relevant contemporary literature on the research topic and the practice knowledge and expertise of the authors. Questions for the semi-structured interview were developed to elicit views of health professionals on their experience of older trauma care in rural settings (Table 1). The questions were structured around three concepts: 1. experience of rural clinicians of a typical case (as defined by the participant) of an older trauma patient; 2. experience of rural clinicians of a memorable case of an older trauma patient and; 3. views of rural clinicians on changes and/or recommendations to improve current practice. The interview guide comprised open-ended questions, and probing questions to elicit more detailed information (Table 1). Participants were asked to recount a typical, and a memorable case of patients they had managed, in order to evoke factors they would subsequently be asked about, including enablers, barriers and recommendations for trauma care. Interviews were conducted by the primary author (NF), who tested the interview guide with three pilot interviews. These were also attended by a trained qualitative researcher (CP), for the purpose of mentoring the primary author, and for quality assurance. All subsequent interviews were audio-recorded with participants' informed consent for verbatim transcription. Informed consent was obtained verbally at the start of each interview, documented, and recorded. Information on the participants' training and professional background with respect to older trauma and rural healthcare, were obtained at the start of the interview. --- Data analysis In line with the exploratory nature of the study, and breadth of participants' experience, thematic analysis was used to identify and develop themes from interviews. All interview transcripts were imported into NVivo 12 (1.6.2), a qualitative data analysis software to support data analysis. Data analysis was performed independently by two researchers (NF and CP). Themes were discussed for discordance; no further adjudication was required. Data analysis proceeded alongside data collection to --- Table 1 Interview guide for semi-structured interviews --- Biography -demographic and professional characteristics Question 1 -A typical case • Can you describe a typical case of a patient over 65 years with trauma that have you cared for? • What things do you remember were done well in that case? • From your perspective, what things could have been done better? --- Question 2 -A memorable case • I would like you to try and describe a case of an older patient with trauma that stood out for you? • If we take a step back from the two cases you just described, and consider in general, trauma care of older patients in rural areas. What do you think are some of the contributing factors to providing good care? • Again, if we consider trauma care of older patients in rural areas more broadly, what do you think are some of the contributing factors that prevent good care? --- Question 3-Changes • Tell me about some of the changes you would like to see to support health professionals delivering older trauma care in rural areas? enable the themes identified in the earlier interviews to be included in the interview guide and explored in subsequent interviews. Data collection concluded when data saturation was reached, which was defined as when no new information was forthcoming [31]. Both an inductive coding approach using open and axial coding, and a deductive coding approach, were employed to identify key themes in the results. After reading the transcripts, open codes were generated, and the first interviews were inductively re-analysed at the cessation of data collection to ensure no coding was missed in earlier transcripts. --- Results Of 24 clinicians who expressed their interest in participating in the study, fifteen took part in interviews; 10 medical practitioners, two registered nurses, two occupational therapists, and one paramedic. Eleven participants primarily practiced in a regional or rural area, whilst four participants had their primary location in a metropolitan area; three of whom had a current role in air retrieval of older trauma patients from rural sites. Twelve of 15 clinicians had over ten years of experience since acquiring their professional qualifications (Table 2). Several themes were identified within the formulated framework: 1) enablers of trauma care for older people in rural areas; 2) barriers to trauma care of older people in rural areas and; 3) changes to improve trauma care of older people recommended by rural clinicians. In addition,to facilitate guidance for trauma care improvement, the various themes were reported at patient-, clinician-and system-level (Table 3). --- Enablers of trauma care The resilience of people in rural areas Clinicians described patients from rural areas as tough and robust, "the classic bushy" (Table 4). Patients were frequently observed to still be working into older age and therefore sustained significant injuries related to Table 2 Demographic characteristics of participants a R regional/rural, M metropolitan b Years of experience since participants obtained professional qualifications. J: junior (less than 10 years); Senior (10 years and over) --- Profession --- Number of participants --- Primary location of practice a --- Level of experience b Medical doctor 10 R = 7; M = 3 S = 9; J = 1 Registered nurse 2 R = 1; M = 1 S = 2 Occupational therapist 2 R = 2 J = 2 Paramedic 1 R = 1 S = 1 --- Table 3 Summary of identified themes and subthemes --- Factor-level Themes --- Enablers Patient The resilience of people in rural areas --- Staff --- The experience and training of rural clinicians --- System --- Relationship between clinicians and the rural community Delivering better care through a multidisciplinary approach A robust system for major trauma Because of perceived stoicism, injuries may be missed or underplayed. Clinicians perceived patients as being attached to their land and community; "people here really would rather get their care close to home. The people who are living up here, choose to live up here. They don't like cities. " --- Barriers [DR2]. --- The experience and training of rural clinicians Participants described clinicians as often being the sole practitioner, and thus needed to be generalists with a broad skill set, and the ability to make clinical decisions and provide care with very few diagnostic tools and equipment "they can do a lot with nothing" [DR5]. --- Relationship between clinicians and the rural community Knowledge of patients from previous clinical and nonclinical encounters was also viewed as a strength; "You know them, you've met before, you've had a chat and I think that rapport that already exists is protective" [DR10]. --- A robust system for major trauma Participants viewed the trauma system as wellrehearsed and oriented towards the management of major trauma cases, with early prehospital notification, protocols with clear identification criteria for critical patients, and fast retrieval to metropolitan trauma centres, "our system has evolved so that we are much better at the both the pointy end of the resuscitation, access to theatre and the ICU side" [DR1]. --- Barriers to trauma care A life-changing event Clinicians reported that older rural patients may often live alone and remotely, and lose their independence after sustaining an injury; --- Being alone When transferred to another hospital, patients were also isolated, away from their community, thus unable to draw on their usual support. Patients may also die after sustaining trauma, away from their community. --- Not speaking up Clinicians viewed older patients, and patients from the country as having different expectations from the health system from patients living in urban areas, "the people up here don't complain. They will get most horrendous care and they will never complain" [DR2]. --- Rural clinicians feel unsupported Participants emphasised a widespread lack of staffing both in rural hospitals and community services, which was not specific to older trauma care. They identified a lack of staff to manage complex and urgent cases. Par --- Older adult trauma is hard and not sexy Participants reported that mechanisms of injuries were often not as dramatic as in younger patients, which may contribute to under-triage of older trauma patients "they've been on the ground... because the way ambulance triages elderly falls" [PM1]. Moreover, trauma in older patients could be perceived as comparatively dull "something a bit more exciting about the younger person with the motorbike accident and the gunshot wounds" [DR8]. --- The tyranny of distance/ systemic lack of resources A major barrier raised by participants was that distances from the injury site could be very large "distance is the biggest issue we have because... that always delays transfer. And that always delays timely treatment" [DR5], as well as a lack of hospital beds impeding patient flow, and rehabilitation facilities and community services in rural areas. --- A fragmented health system Participants reported that the different health services worked in isolation, a lack of streamlined referral pathways, which could contribute to delays of care when awaiting an opinion from specialised centres. Due to lack of integration in the health system, there was no incentive in providing care to a patient out-of-catchment "they have to be in the hospital for the hospital to take responsibility for that discharge plan" [OT1]. As a result, patients spent an excessive amount of time in acute beds awaiting transfers, solely to be able to access rehabilitation facilities, outpatient and community services. --- Changes to improve trauma care of older people Supporting patients in their hospital journey When possible, participants thought that an escort for older patients during retrieval would be beneficial, as is already the standard for children and Indigenous Australian patients, "if you weren't able to get a lot of history from the patient, having family (present) can really help" [DR8] (Table 6). --- Enabling discussions on advanced care planning Participants identified advanced care planning as a major area for improvement. The likely trajectory of the patient in light of their injury and current health state should be clearly explained to them, and should ideally involve senior clinicians and be given sufficient time. Additional recommendations put forward by participants included a documented advanced care plan as part of the retrieval checklist. --- Multidisciplinary and coordinated care as standard of care Participants advocated for a holistic approach, with early review by a general physician or a geriatrician "a lot of those trauma patients don't need to be in a big tertiary hospital, but they need good multidisciplinary, holistic care" [DR6]. Management should be based on an integrated or shared model of care between physicians and surgeons. --- A coordinator for inpatient and post-acute care Participants advocated for a dedicated position for care coordination in and out of hospital "a care coordinator who fights for the patient so that we have one bus that brings them in, they have their whole day of appointments and then go home again" [RN1], as well as to provide a link back to the patient's relatives and primary care physician. --- Enabling the training of rural clinicians in older trauma care Participants overwhelmingly advocated for more education on older trauma care. Effective education programs require engagement of participants "if people don't understand why they're doing something... it gives no benefit" [DR1], should be tailored to rural practice, and ideally be taught in rural centres. However, participants identified the lack of staff cover to take professional leave as a major barrier to accessing education, and proposed this may be remediated through covering rosters within individual professional organizations. --- Improving integration within the health system Participants thought that streamlining patient care would be facilitated by integrating the various health services into a trauma system "a robust system in which we have clearly identified a feeder system... where...there are actual people responsible for the support throughout that network" [DR1].This would require an accurate understanding of the level of care that can be provided at various locations and the designation of different levels of trauma services with capabilities to manage various aspects of geriatric trauma, e.g. integrated physician/surgeon care model, regional anaesthesia and observation in ICU. This would require governance with "a state-wide trauma coordinator... who is the first port of call as a senior clinician to assist with decision making" [DR3], and linking medical records. --- Care protocols and standardized referral pathways Participants also recommended a system that would assist in the early identification "a better trigger at triage that says this person is a risk", that is simple and easily applied in triage. Yet triggers should not be oversensitive "and rushing everybody to the higher priority. You don't want to be the boy that cries wolf with everybody" [PM1]. Protocols for common injury patterns (e.g. chest trauma, anticoagulation, head injury) that were easy to follow were thought to be particularly useful in rural settings. Streamlined referral pathways were also highly valued by participants. --- Supporting and expanding the use of telehealth Participants identified telehealth as a useful tool. Telehealth was sometimes used in resuscitation bays for the management of trauma patients, and could be expanded to other inpatient settings. For instance, for remote assessment of patients by specialized services in metropolitan centres "having some sort of administrative agreements, saying that we (can) make decisions based on the information that we see over an electronic platform" [DR10]. --- Discussion In this study, we gained insight into the experience of clinicians who provide care to older trauma patients in rural settings. To the best of our knowledge, this study is the first to explore the perspective of rural clinicians. Participants identified strengths both amongst older patients and clinicians, such as resilience and resourcefulness, which contributed to good care, despite the systemic lack of resources and integration within the healthcare system. They put forward relevant and practical recommendations, such as a streamlined referral pathway for older trauma patients and a state-wide care coordinator. These recommendations should be assessed against the current evidence, and tested in future studies. A recurring theme amongst enablers of trauma care was the breadth of experience of rural clinicians, who may be accustomed to practicing in relatively resourcelimited environment. Another enabler was the presence of a robust pre-hospital system to manage cases of major trauma. These strengths should be further enhanced in future interventions aiming at optimising trauma care to older patients in rural areas. The most common barrier was the perceived systemic lack of resources, both material and in the workforce. The dearth of staff in general, and of specialists, and the ensuing strain on rural clinicians, was thought to contribute to sub-optimal care and lack of staff retention of staff. Disproportionate attrition of healthcare workers in rural compared to urban areas has been described in the literature, as were strategies to enhance staff retention [32]. With the exception of pre-hospital care, participants viewed the health system in their state as highly fragmented. The fragmentation of healthcare, and poorer outcomes and inefficiencies that ensue are well known [33], and form the rationale for the creation of trauma systems [34]. Participants felt that the management of older trauma patients from rural areas transcends the boundaries of designated health services. Participants thus proposed that a centralised system of governance and funding, akin to pre-hospital care, should be designed to streamline the management of older trauma patients coming from rural regions. Another recommendation consisted of a case coordinator to facilitate transitions of episodes of care, from in-hospital to outpatient and primary care. An equivalent role in cancer care is thought to be of particular benefit to rural patients [35]. Whilst there is a dearth of studies on the perspectives of rural clinicians on older adult trauma care, one study examined the views of rural clinicians on care of the general adult trauma population. This study also identified that the lack of support for rural clinicians was compounded by the lack of resources, putting extra pressure on healthcare staff [22]. Participants in this study also identified the fragmentation of the healthcare system as a major barrier to providing trauma care [22]. --- Implications for trauma systems Rural clinicians face substantial challenges in providing care to order trauma patients. The perceived lack of support and ensuing attrition is likely to be exacerbated by the growing number of older adults presenting with traumatic injuries to rural centres. This study suggests that possible interventions exist. There is a call from rural clinicians for a centralised system for the management of trauma patients that would integrate rural centres. We should also recognise that some interventions should come from the bottom-up, as they can only be organised locally. --- Future directions Rural clinicians and older patients are important stakeholders, and should be included in discussions on adapting trauma guidelines in the rural setting. Future studies should assess the validity of these recommendations against the current evidence. For instance, the validity and benefit of a trauma-specific frailty index in aiding rural clinicians in early identification of patients at risk of adverse outcomes, may be tested [36]. The challenges faced by rural older trauma patients have previously been investigated post-discharge [37,38], but not during their hospital journey. Future studies should examine enablers, barriers and need for change from the perspective of patients during their hospital admission. --- Strengths and limitations To the best of our knowledge, this is the first study to examine the experiences of rural clinicians involved in the management of older trauma. Using a qualitative approach enabled us to gain in-depth and rich information on several aspects of older trauma care in the rural context. Moreover, the evidence base for trauma care of older adults in rural areas is sparse, in part due to the lower volumes of patients and participants, and the lack of research opportunities [21]. Although, our findings are specific to the local context of the study, they may prompt consideration of rural trauma care provision in countries with equivalent healthcare and trauma systems, such as North America and Scandinavia [22]. There are also important limitations to consider in this study. As participants were identified through voluntary professional organizations, there may be a selection bias. Participants who were made aware of the study, and who agreed to be interviewed may have perspectives that may be different from non-participants who were either not aware of the study or did not agree to participate. The perspective of participants interviewed here may therefore not be generalizable to all clinicians involved in managing older trauma patients in rural areas. Another important limitation is that perspectives gained here were solely from care providers, rather than consumers. --- Conclusions Rural clinicians are important stakeholders who should be included in discussions on adapting trauma guidelines to the rural setting. In this study, participants formulated pertinent and concrete recommendations that should be weighed against the current evidence, and tested in rural centres. --- Availability of data and materials The datasets generated and/or analyzed during the current study are not publicly available as this would compromise the privacy of individual participants who partook in the interviews, but are available from the corresponding author on reasonable request. --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12913-023-09545-x. --- Additional file 1. Authors' contributions NF contributed to the study conception and design, interviews and data collection, data extraction, analysis and interpretation, drafting of manuscript, tables, and referencing. CP contributed to the study design, data analysis and interpretation, and major revisions to the manuscript. JI, BG and PC contributed to the study conception and design, and major revisions to the manuscript. The author(s) read and approved the final manuscript. --- Declarations Ethics approval and consent to participate The study was approved by the Monash University Human Research Ethics Committee (32479). All methods in this study were carried out in accordance with the guidelines and regulations in the declaration of Helsinki. All interviews conducted in this study were audio-recorded with participants' informed consent for verbatim transcription. Informed consent was obtained verbally at the start of each interview, documented, and recorded. --- Consent for publication Not applicable. --- Competing interests The authors declare no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field --- • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background For older trauma patients who sustain trauma in rural areas, the risk of adverse outcomes associated with advancing age, is compounded by the challenges encountered in rural healthcare such as geographic isolation, lack of resources, and accessibility. Little is known of the experience and challenges faced by rural clinicians who manage trauma in older adults. An understanding of stakeholders' views is paramount to the effective development and implementation of a trauma system inclusive of rural communities. The aim of this descriptive qualitative study was to explore the perspectives of clinicians who provide care to older trauma patients in rural settings.We conducted semi-structured interviews of health professionals (medical doctors, nurses, paramedics, and allied health professionals) who provide care to older trauma patients in rural Queensland, Australia. A thematic analysis consisting of both inductive and deductive coding approaches, was used to identify and develop themes from interviews.Fifteen participants took part in the interviews. Three key themes were identified: enablers of trauma care, barriers, and changes to improve trauma care of older people. The resilience of rural residents, and breadth of experience of rural clinicians were strengths identified by participants. The perceived systemic lack of resources, both material and in the workforce, and fragmentation of the health system across the state were barriers to the provision of trauma care to older rural patients. Some changes proposed by participants included tailored education programs that would be taught in rural centres, a dedicated case coordinator for older trauma patients from rural areas, and a centralised system designed to streamline the management of older trauma patients coming from rural regions. Conclusions Rural clinicians are important stakeholders who should be included in discussions on adapting trauma guidelines to the rural setting. In this study, participants formulated pertinent and concrete recommendations that should be weighed against the current evidence, and tested in rural centres.

non-communicable diseases (NCDs), accidental and violent injuries, and effects of ecological disaster'. 3 The organisation and design of urban spaces, such as isolated and dark streets, poor traffic management, lack of green spaces, high-density urban areas, or city sanitation practices, have an impact on daily conditions that shape our health and contribute to health inequalities. However, in practice, legal frameworks for decision-making in local government may pay little attention to health outcomes or may prioritise other considerations that may appear, prima facie, contradictory-such as promoting a strong economy. This raises a series of important, connected questions. How can the decision-making process be improved, in England, for Local Planning Authorities to promote health and reduce health inequalities? What is the role of the law in this? Can human rights contribute? In this article, I explore the conceptual and practical links between urban development, health, law, and human rights. I bring together literature from global health, public health ethics, and human rights that seek to contribute to the shared goal of promoting human well-being. First, in Section II, I focus on the conceptual links, and I explain how research on the social determinants of health reveals the critical importance of the law in urban development and health. Then I discuss how law, conceptualised in a broad sense, may have an effect on the social determinants of health. In particular, I explore the law as a mechanism to shape structural power in our society, as a tool to regulate multiple areas interacting with urban development, and as a form of implementation of governance of public and private actors in urban planning in England. I argue that human rights cannot be discounted when researching the social determinants of health. 4 In Section III, I make innovative use of global health scholarship to explore domestic questions on the application of legal mechanisms and tools to promote health and reduce health inequalities in urban development in England. I suggest applying a human rights framework, as defined in international human rights law, and thus partially implemented and enforceable through the UK's domestic legal system. Finally, I conclude by reflecting on the added value and limits of this approach. --- I I. URB AN D EVEL OPM EN T, LA W AS A STR UCT UR AL D ETE RM IN AN T, A ND H UM AN R IG HT S A. The social determinants of health in the context of urban development Research on the social determinants of health shows how the daily conditions in which we are born, grow, live, work, and age, impact our health. 5 These daily conditions are themselves determined by the unequal distribution of power, resources, and money in our society. 6 The law, conceived in its basic meaning as 'a rule [...] [or set of rules] that is used to order the way in which a society behaves', plays a key role in organising the distribution of power and resources in a society, and, therefore, as shaping the social determinants of health. 7 In this section, I briefly introduce the social determinants of health and their conceptual links with the law, before applying this thinking to the context of urban environments in England. 1. The social determinants of health and the law By way of background, I use a definition of health that extends beyond the biomedical model-the absence of ill health-to encompass positive states of health and well-being. 8 In terms of practice, policy, and law, I am interested in the creation or perpetuation of the social determinants of health. 9 The findings from the World Health Organisation Commission on the Social Determinants of Health (WHO Commission), as well as the review of these findings 10 years later, show that the conditions of life, such as the level and quality of education, the availability of safe sanitation and potable water, the quality of the air one breathes, and the conditions of work, have a cumulative impact on the distribution of ill health across groups in the population. 10 The WHO Commission observed a'social gradient in health', meaning that health varies at each level of income, with worse health for each lower socioeconomic level. 11 These conditions of daily life are themselves impacted by the unequal distribution of power, money, and resources in our society. 12 As a result, any research aiming at informing policies to reduce health inequalities and promote justice in public health should look beyond healthcare policy and view health as a product of our social system. 13 Law plays a key role in the social determinants of health because it has a significant effect on the organisation and administration of our social system. In this article, I explore the role of law to improve the causes of the unequal distribution of ill health in the context of urban development. I am interested in health inequalities and NCDs; that is, diseases that are not transmissible and last for life, or a significant period in life. In the subsections below, I first posit that urban environments constitute an opportunity to observe how the law interacts with the social determinants of health. Then, I show how urban environments are, themselves, a social determinant of health. --- Observing the effect of law on the social determinants of health Researching the causes of ill health at population levels is challenging on several counts, one of them being the multiplicity of variables-economic, political, social, biomedical, and behavioural-that interact with each other. For instance, we may imagine that an individual's chances of getting lung cancer would depend on individual behaviour, such as smoking and alcohol use, genetic predispositions to developing this disease, and individual exposure to harmful agents, such as asbestos. Regular exposure to asbestos may be due to poor working conditions, lack of information on the dangers of exposure to such an agent, or lack of choice regarding one's source of remuneration. Previous lung infections also have a cumulative effect on the chances of that individual to develop, one day, lung cancer. These previous infections may or may not have been treated fully depending on the availability and quality of health care received, which itself could depend on an individual's financial resources to enrol into a private health insurance scheme, and the quality of hospital care of the geographical area that individual lives in. Finally, additional exposure to air pollution may contribute to the development of cancerous cells in the lung, and to accelerating the spread of the disease. In order to understand what laws, taken in a broad sense-primary and secondary legislation, jurisprudence, as well as the implementation of these rules (including through policy)-have an impact on health outcomes, it is necessary to take into account the time between the moment a law, a regulation, or a policy or programme is implemented, and the visible effect, if any, it has on health outcomes. 14 For instance, experts estimate that a national cancer plan may take 8-10 years before its effect may be observable on national estimates of cancer survival. 15 In the context of NCDs, there is an additional layer of complexity: the period of time between exposure to a harmful agent (air pollution or asbestos, for example) and the outcome (say, lung cancer). Indeed, NCDs are characterised by a plurality of causal factors, as well as the long time needed-sometimes years-for the first symptoms to manifest. This contrasts with infectious diseases, for which the incubation period is usually much shorter (for example, a few days), exposure to the virus can be located with more precision, and determining the cause(s) of a disease can thus be more accurate. With NCDs, the multiplicity of causalor rather, contributing-factors, coupled with the lengthy period of time between exposure to a given factor and manifestation of the first symptoms, mean that social environments as a whole must be considered in order to understand and implement effective prevention. In 2019 in England, 56.3 million people lived in urban areas (82.9% of the population). 16 These areas are managed by 333 local authorities with devolved powers and funding to make decisions on the social determinants of health, such as housing, work, local economy, and health policies. 17 In particular, the organisation and management of housing and neighbourhoods play a critical role in distributing health (in)equity among population groups. 18 As such, urban environments represent a valuable opportunity to observe the interaction of political, economic, social, and cultural determinants of NCDs and their unequal distribution in the population. Urban environments represent what Venkatapuram calls'social environments' in social epidemiology; that is, 'economic, political, and social policies and processes driven by social and cultural values that create and distribute the daily conditions of life'. 19 --- Urban planning (law) as a social determinant of health In addition to using urban environments as a laboratory, it is important to recognise that urban planning itself is a social determinant of health: 'The planning and design of urban environments has a major effect on health equity through its influence on behaviour and safety'. 20 The organisation, use, and management of land in urban areas have an effect on a myriad of health determinants that contribute to causing NCDs and deepening health inequalities. The WHO estimates that lack of physical activity, alcohol use, tobacco use, and unhealthy diets are the main causes of the four NCDs responsible for over 80% of premature NCDs deaths worldwide: diabetes, cancer, heart diseases, and respiratory diseases. 21 Urban planning determines the quantity and quality of green space, the type and frequency of transport, including the design of cycle lanes, the quality and quantity of housing, including social housing, and the frequency and location of fast-food and off-licence shops. These conditions have a knock-on effect on the causes of NCDs, regarding, for example, physical activity, car use and cycling, air quality, sanitation, and indoor air quality, as well as dampness. Planning law plays a critical role in this context. Planning law is the body of rules relating to the development; that is, 'any building, mining, engineering or other operations which are carried out in, on, over or under land, or the making of material change of use of buildings or other land'. 22 With that in mind, it is expected that planning law influences the social determinants of health. However, it is not the only field of law that does so. Urban development is concerned with a multitude of law fields, each having an effect on the social determinants of health. Before delving into these considerations, I will first explain how the law plays a role in shaping the unequal distribution of money, power, and resources. This is what the WHO Commission calls the 'causes of causes'. 23 --- Inequities of power, money, and resources in urban development The social determinants of health are themselves determined by structural issues of power, money, and resources. 24 In England, the way planning works creates a'system of systems' which is the result of a multitude of stakeholders having political and/or financial power to shape urban development. 25 The power conferred to Local Planning Authorities is in tension with the informal power of private parties, such as developers. This is because planning development operates in a 'plan-led' system, which means that granting or refusing planning applications is decided on a case-by-case basis, depending on whether they fit with a given local development plan unless material considerations indicate otherwise. 26 In essence, if a proposal 'fits' with policy criteria, the relevant planning application should be granted. The issue thus becomes a question of how well health is embedded into planning policy, and to what extent decision makers have discretionary power to interpret planning policy to promote well-being. Other priorities may trump health promotion, such as contributing to a strong economy. 27 This approach to planning differs from a zonal system, where local authorities decide in advance what use different areas of land should be allocated; for example, residential or commercial. In this latter planning system, developers have more certainty, but Local Planning Authorities lose flexibility in deciding what is built on a particular piece of land. The proposed changes from the 2020 White Paper to planning law in England would have moved the system towards more zoning, but several parties voiced their concerns to this approach in a consultation. 28 Until such changes are enacted, the planning system in England remains 'plan-led'. The (unequal) distribution of power, therefore, affects planning decisions which, in turn, have an effect on the determinants of NCDs and health inequalities. Researching the root causes of unhealthy urban development means looking at the role of law in shaping the structure of the social system, setting the rules within which urban development operates, as well as regulating powerful actors. I explore these three aspects below. --- B. Law as a structural determinant A primary function of law is to regulate, -including by providing and limiting-power and setting the rules for an orderly, cohesive society. Law is the tool that materialises the 22 social contract according to which citizens agree to give up a proportion of their autonomy to the State so that, in return, the State can regulate social, economic, and political interactions between its citizens. 29 Law, and within it multiple areas of law, determine structural societal rules that shape the distribution of power, money, and resources. It also sets the rules for national policies, such as the National Planning Policy Framework (NPPF), the core policy informing all local development plans and decisions from Local Planning Authorities in England. A spectrum of influence of the law can thus be seen, ranging from the overarching structure that implements the political ideology in which we live-a capitalist market economy-through to the norms that regulate a plurality of areas and having an impact on urban health, including the daily rules that organise decision-making in urban development. At one end of the spectrum, regulation (or lack thereof) pertaining to the market economy sets the background within which urban planning operates. In England, reliance on private actors to fund and deliver a public function, such as housing provision or road management, means that planning outcomes are influenced by the way these private actors operate. For instance, real estate investors are looking for high and/or regular financial returns, which means that they invest in projects with a potential to generate income. In this context, social projects, or projects that prioritise new outcomes, such as social equality and health, may represent financial uncertainty, which arguably is a risk that investors are less likely to take. In turn, developers are dependent on investors to fund the design and construction of new projects. In this context, Local Planning Authorities appear to have limited power over what gets built and how; a power which may be even more limited in the 'planled' English urban planning system. 30 Beyond the law as implementing the politico-economic ideology in which society, and within it, planning operates, the law regulates a multiplicity of fields having an impact on urban development and health. Indeed, several areas of law operating in, and in parallel to planning, have an impact on urban development, and, in turn, on the prevention, protection, and promotion of health in urban areas. 31 By way of illustration, the recent inclusion in England of a 10% social value in procurement rules contributes to determining the quality of large development schemes. 32 Similarly, building regulations are known to have a significant impact on health, as the 2017 fire at the Grenfell Tower tragically testifies. 33 Other areas include administrative law and its impact on the good governance of Local Planning Authorities, highways regulations and their impact on traffic, laws governing taxation, uses of buildings, and so on. 34 All these areas of law influence urban planning-what gets built, where, and howand have a significant impact on shaping the environment we live in, with the corresponding consequences on our health. By way of illustrating the variety of law fields that impact health in urban development, I will now discuss some concrete policy examples which have had direct and indirect effects on health. 29 --- The direct effect of laws on health in urban development The NPPF pays particular attention to environmental concerns by listing sustainable development as the objective of urban planning in England. 35 In this context, environmental law influences public and private actors' decisions by imposing standards on the built environment. For instance, the 'net zero' policy in the UK, which aims to transition cities and regions to zero carbon emissions by 2050, is expected to have an impact on air quality, which in turn has an effect on respiratory diseases, some cancers, cardiovascular diseases, and diabetes. 36 In 2019, the WHO published a report suggesting that air quality is the second leading cause of death from NCDs. 37 A national policy aiming at reducing greenhouse gas emissions, therefore, has a direct impact on the prevention of NCDs (if successfully implemented). Another example of the direct impact of laws on health is within the planning law itself; for example, a national policy aiming to increase access to green spaces for all communities in urban areas. Public Health England, an executive agency of the Department of Health and Social Care that has now been replaced by the Office for Health Improvement and Disparities and the UK Health Security Agency, collected evidence showing how access to green spaces has an impact on physical health, notably levels of cholesterols, cortisol, variability of heart rate, and blood pressure; as well as mental health outcomes such as depression, levels of anxiety and stress, as well as resilience, 38 which is in line with WHO findings and policy recommendations. 39 Evidence also suggests that access to green spaces benefits more deprived communities and may, therefore, help in reducing health inequalities. 40 Finally, transport and, in particular, the promotion of cycle lanes provide another example of the direct impact of laws on health in urban development. Evidence suggests that walking and cycling infrastructure in urban development is associated with increased walking and cycling, which, in turn, increases physical activity. 41 These examples show how laws may have a direct impact on health in urban planning. These laws are by no means limited to those studied within the siloed fields of healthcare law or planning law. They extend to fields beyond health such as environmental law, administrative law, and transport. In addition, laws may have an indirect effect on health outcomes in urban development. --- The indirect effect of laws on health in urban development Laws may indirectly impact health in two ways: by regulating outcomes and by regulating behaviours. By way of example, consider a national policy aiming at increasing the number of housing accommodations; for instance, the UK's target to achieve 300,000 new homes a year. 42 Such a policy may have indirect positive and negative effects on health. If the policy leads to an emphasis on the number of new homes over the quality of housing and affordability, it may result in decisions favouring expensive and/or poorly built houses. Standards 35 NPPF (n 26) paras 7-14. 36 necessary to promote good physical and mental health may not be respected, such as the Decent Home standards in England or the WHO housing standards having an impact on health. 43 A national policy emphasising housing targets may also overlook the accessibility of housing to persons with physical disabilities and older people. 44 In addition to the indirect impact of laws on health by regulating outcomes, law-understood as encompassing issues of governance-plays a role in regulating the behaviour of powerful actors. 45 It regulates government and non-government institutions; that is Local Planning Authorities, the Planning Inspectorate, and branches of government responsible for, or having an impact on, urban development, such as the Department for Levelling Up, Housing and Communities, the Department for Transport, or the Department for Environment, Food and Rural Affairs. Different types of specific laws govern the good functioning of powerful institutions and industries. In England, private actors such as developers, real estate investors, and development corporations, as well as non-governmental bodies such as the Environment Agency and Natural England, exercise power over what gets built and how. 46 As a result, any law governing these entities has the potential of limiting, or at least controlling, their power to protect the social determinants of health. This brings us back to the idea that the law regulates, in part, the distribution of power, money, and resources in our society, including powerful private and public actors in urban development. The direct and indirect effect of laws on health in urban development, together with the multiplicity of government and non-government actors, as well as law taken as structural rules organising our society in a capitalist market economy, are all different faces of the law acting as a key determinant of health. In this section, I have demonstrated how law, both as a concept of structural organisation, daily rules, and as a concept of governance of actors, shapes the daily conditions in which we are born, grow, work, live, and age; that is, the social determinants of health. I will now turn to human rights theory and discuss why human rights are relevant to researching the causes of ill health in urban development. --- C. Why human rights are relevant I argue in this section that fundamental human rights principles of universalism, interrelatedness, and indivisibility mean that when we talk about the social determinants of health, we are talking about human rights. A contrario, human rights and health cannot be researched without consideration of the social determinants of health because public health and human rights share a common goal of promoting human well-being. I turn to these two points below, before discussing how the right to health helps understand the links between public health and human rights principles in urban development. human rights are universally held does not show that such rights exist universally. Donnelly shows that human rights, although conceptually universal, are not enforced universally, nor are they understood universally at historical and anthropological levels. However, even for those who do not accept that human rights are fundamentally universal, Donnelly provides an account of what he calls'relative universalism'. He grounds human rights into a functional universality, to provide a practical answer to modern threats to human dignity: Human rights today remain the only proven effective means to assure human dignity in societies dominated by markets and states. Although historically contingent and relative, this functional universality fully merits the label universal -for us, today. 48 Universalism of human rights, whether understood conceptually as 'inalienability' or functionally to protect human dignity in our globalised world, suggests that human rights are interrelated and so taking one single human right in isolation makes little sense. To extend Donnelly's reasoning, conceptually, dignity stemming from 'being human' cannot mean that only a few human rights are relevant to being human. All human rights apply to us by virtue of being human. In a similar fashion, functionally, human rights would lose a great deal of their power if they were conceived independent from one another. Universalism, or even'relative universalism' as argued by Donnelly, requires us to think of human rights as interrelated. The so-called 'interrelatedness' of human rights is a fundamental principle reaffirmed in paragraph I.5 of the Vienna Declaration and Programme of Action. 49 From a practical perspective, as posited by the United Nations, this means that violation of one right results in conditions that lead to the violation of other rights. 50 For instance, violation of the right to housing, such as living in a poor-quality home with dampness, creates conditions in which the right to health cannot be fulfilled. Similarly, violation of the right to health, because of unaffordable health care or unreasonably long distances of travel to access care, creates conditions where a person's employment is put at risk, thereby potentially leading to violation of the right to work. On a more general level, violation of economic and social rights, such as housing, work, health, and education, creates conditions in which civil and political rights, including freedom of assembly, freedom of speech and thought, and right to participate in political life, cannot be realised fully, and vice versa. 51 The interrelatedness of human rights echoes research on the social determinants of health which shows that the daily conditions in which we live have an effect on the distribution of health in the population. These daily conditions are, in fact, our economic and social rights to housing, education, work, water, and health, as well as our civil and political right to freedom of thought and to participate in political life. 52 The WHO Commission's report discusses 12 areas as key drivers of the unequal distribution of health, of which seven at least can be linked to human rights (Table 1). The principle of interrelatedness of all human rights means that every human right is equally important. One right cannot and should not be more valued than another, and it would make little sense, conceptually and practically, to compare one right to another. This is what is referred to as the principle of 'indivisibility' of human rights. 53 In summary, when the social determinants of health are examined, human rights cannot be ignored because, in fact as well as in principle, the social determinants of health are about human rights. 54 Equally, it is not possible to research the right to health whilst ignoring social epidemiological evidence on the social determinants of health. 55 --- The common goal of public health and human rights In addition to being intricately linked to the social determinants of health, human rights share a common goal with public health. In 1994, Jonathan Mann and colleagues identified three conceptual links between health and human rights, which remain valid and relevant today: the impact of health policies on human rights; the health impacts resulting from human rights violations; and an inextricable linkage between health and human rights due to, inter alia, the 'central problem of defining and advancing human well-being'. 56 The common goal shared by public health and human rights, as well as the complementarity of approaches that they offer to protecting and promoting well-being, are a solid basis for including human rights in research on urban development and health. By way of illustrating how human rights are relevant to this issue, I will now consider the right to health itself. --- The right to health Health is a human right enshrined in Article 25 of the Universal Declaration of Human Rights, and in binding international human rights treaties, principally Article 12 of the International Covenant on Economic, Social, and Cultural Rights (ICESCR). 57 The right to health is also integrated into regional human rights instruments, such as Article 11 of the European Social Charter, to which the UK is a signatory. 58 In 2000, the content of the right to health was clarified with the publication of General Comment 14 by the Committee on Economic, Social and Cultural Rights (CESCR), the body responsible for overseeing the implementation of the ICESCR. 59 General Comment 14 interprets the right to health as extending to 'the underlying determinants of health, such as food and nutrition, housing, access to safe and potable water and adequate sanitation, safe and healthy working conditions, and a healthy environment'. 60 The underlying determinants of health share considerable resemblance with the social determinants of health. The First Special Rapporteur on the right to health, Paul Hunt, recommended that the right to health should be understood as part of a national health system, which itself is part of a broader social structure influenced by political and economic considerations. 61 This conceptualisation of the right to health as being influenced by, or even the result of, our 53 economic, social, and political system, shows similarities with the 'causes of causes'; that is, issues of power, money, and resources as formulated in social epidemiology. With this in mind, tackling the causes of unhealthy urban development necessitates not only consideration of the social determinants of health and the role of the law in shaping them, but should also draw from human rights to inform both health research itself and practical solutions to preventing NCDs and reducing health inequalities. Having defined the scope of the research, in the second part of this article I will now apply the tools that the law, including international human rights law, offers, as well as discussing the value and limits of using a human rights framework in urban development and health. --- I I I. APP LI CAT I ON O F T HE LA W AN D HU M AN RI G HT S T O UR BAN D EVEL OP MEN T AN D HE ALTH In Section II, I demonstrated how the social determinants of health provide a solid basis to the study of health in the context of urban development. The rich and longstanding research on the social determinants of health shows that the daily conditions in which we live are a determining factor of the unequal distribution of health in society. In turn, these conditions are shaped by the unequal distribution of power, money, and resources, factors for which the law is a key determinant. By conceptualising the law in a broad sense, encompassing the governance of powerful institutions and by considering it as part of wider socio-political phenomena, the study of law may shed light on the root causes of social inequalities and so on the causes of unequal distribution of health. Furthermore, the laws shaping urban planning belong to a multiplicity of areas within the law itself, including administrative law, procurement, environmental law, planning law, and building regulations. Given this, I now turn to the question of what mechanisms of law ought to be studied in this context. The Lancet-O'Neill Commission on Global Health and the Law reported on the impact of the law on health outcomes, looking at subnational, national, and inter-and supranational levels in four areas: (i) translating vision into action; (ii) strengthening the governance of national and global health institutions; (iii) implementing fair, evidence-based health interventions; and (iv) building legal capacities for health.62 I rely on these four legal determinants to illustrate what mechanisms of the law ought to be deployed in urban development to protect and promote health. I then apply a human rights framework to these mechanisms, before discussing the value and limits of this approach. A. Mechanisms of the law to promote health in urban development --- Translating vision into action Setting clear objectives and corresponding principles is needed to guide decisions within a sector-such as urban development-but also across sectors; for example, planning, public health, and the economy. Although the NPPF mentions health as a component of its core objective to achieving sustainable development, key stakeholders, such as planning officers and developers, lack guidance on how to implement this vision. 63 For instance, does promoting health provide grounds to refuse a planning application of a new housing development without any community garden, or with small spaces, or with poor quality housing? What can be considered, legally, poor quality housing? And how does it compare with the need to provide large numbers of accommodations? Does promoting health provide enough grounds to refuse a planning application for a car park, and in what circumstances? How does health weigh against material considerations that, legally, must inform decisions? 64 The NPPF specifies that its objective of achieving sustainable development and, within that, to'support communities' health, social and cultural well-being', does not constitute a criterion to base decisions on. 65 Rather, this objective should only inform the design of local development plans, their implementation, and planning authorisations, whilst taking into account other factors such as local circumstances. 66 As the Lancet-O'Neill Commission indicates, 'law provides both the mandate for [a goal] and the tools to achieve it'. 67 Providing the mandate for promoting health and reducing health inequalities could become a clear goal of the NPPF and of local development plans informing planning decisions in England. At first, explicitly stating this goal may appear at odds with the complex decision-making environment that defines urban planning in England. Health may be seen as one of many goals of urban planning, such as promoting a strong economy and delivering a targeted number of houses each year. 68 Nevertheless, health is a central value to society, and given the critical determining action of urban development on health, it appears as a sufficient reason to uphold it as (one of) the key goal(s) of urban development. This vision conceptualises health as the core value for our society. 69 Stating health as a main and explicit goal of urban development would be in line with the 'health-in-all-policies' agenda of the WHO. 70 Including health in all policies and beyond strictly the health sector is necessary to ensure cohesion in all areas to protect and promote health. Venkatapuram argues that this concern for health in all sectors is imperative to tackling the unequal distribution of economic and social conditions. 71 To some extent, the proposed Healthy Homes Bill-being read in the House of Lords at the time of writing-would implement a vision promoting health in new homes in the UK through the implementation of the 'healthy homes principles'. 72 Nevertheless, making health promotion, or even the reduction of health inequalities, a clear goal of urban development is not sufficient. In addition to providing a mandate, the law has the capacity to provide the tools to achieve a goal. 73 Taking the case study of universal health coverage, the Lancet-O'Neill Commission shows how the law may be of value in assisting governments making tough policy choices: Law and regulation are integral to making difficult choices because they establish processes and institutions to guide transparent decision-making. States use legal processes to express national health priorities, and legal frameworks define and delineate that which is possible for a state to achieve. 74 The use of the law as a guide to making policy choices is directly applicable to the context of urban development. In fact, once health priorities have been set out clearly at the national level, including in the NPPF, local development plans constitute the legal framework that define and delineate what is possible to achieve in any given urban context. By establishing clear standards that inform new local development plans, neighbourhood plans, and planning authorisations, Local Planning Authorities will be equipped to make decisions that are in line with the goal of promoting health and reducing health inequalities in urban contexts, and that are consistent between regions and over time. These standards must be discussed by experts in several disciplines, including urban development, health, and economics, in a transparent and participatory fashion. Welsh legislation is one example of how vision can be implemented into action and provide guidance on public policy. The Well-being of Future Generations (Wales) Act 2015 implements core principles of the United Nations Sustainable Development Goals, including promoting physical and mental health, building places resilient to climate change, and improving equality in society. 75 Under this law, public bodies in Wales are subject to a wellbeing duty, meaning that they must set clear objectives to implement the seven principles of sustainable development and take action to meet these objectives. The Act set five core ways of working, including collaboration, involvement of communities, and long-term impact. It also created implementation mechanisms. For instance, Public Services Boards foster collaboration between local authorities and other public bodies in a given area. A Future Generations Commissioner conducts research, assesses, and advises public bodies on their well-being objectives and progress towards them. The Auditor General evaluates public bodies' actions against the objectives set. The effectiveness of these compliance mechanisms has been questioned, notably due to the use of non-mandatory terms in the Act, as well as dependency on how the powers of the Future Generations Commissioner are exercised. 76 Nevertheless, the Act remains the first legislation in the world to impose a legal duty on public bodies to respect the well-being of future generations. By enunciating clear sustainable goals, creating implementation and monitoring mechanisms, and setting ways of working, the Act contributes to implementing the core vision of sustainable development in practice. --- Strengthening governance of national institutions, local planning authorities, and powerful private organisations The Lancet-Oslo Commission on Global Governance for Health, which preceded and informed the Lancet-O'Neill Commission, defined global governance as: The complex of formal and informal institutions, mechanisms, relationships, and processes between and among states, markets, citizens, and organisations, both intergovernmental and non-governmental, through which collective interests on the global plane are articulated, rights and obligations are established, and differences are mediated. 77 Although this definition was written in the context of global governance, its characteristics apply to national governance of urban development and health; that is, as a diffuse concept encompassing a broad range of actors, processes, and forces. Good governance is critical to a well-functioning State and the implementation of the law. 78 In contrast, issues of corruption, whether at national or local levels, undermine transparent decision-making and divert taxpayers' money away from social functions. 79 The 75 Well-being of Future Generations (Wales) Act 2015 (anaw 2). 76 The scattered and multilevel governance of these public and private actors creates unfavourable conditions to implement a shared vision of health, let alone the right to health. Law can be used as a tool to clarify functions in relation to population

The COVID-19 pandemic exacerbated profound inequalities in the conditions in which people live, work, and age. Law plays a critical role in shaping these structural health inequalities, which have existed for decades. This dynamic can be observed at the local level, with cities operating as environments unequally distributing the risks of non-communicable diseases between population groups. This article first focuses on urban development to explore the conceptual links between health inequalities and the role of law. I expand this observation and I posit that the social determinants of health are about human rights. With that in mind, I argue that human rights are necessary to address the issue of unequally unhealthy urban environments, hence recognising that people are entitled to a minimum essential level of the conditions in which they live, work, and age, which the State is responsible to fulfil. By way of strengthening my argument, I lay out how a human rights framework can improve these conditions and ameliorate unfair inequalities. Finally, I recognise and respond to the limits of a human rights approach. K E Y W O R D S : Health inequalities, Human rights, Right to health, Social determinants of health, Urban development, Urban planning, Legal determinants of healthIn 2019, 83% of the population of England lived in urban areas. 1 This proportion is expected to grow, with a faster growth rate in urban than in rural areas. 2 Infectious diseases spread quickly in cities-as the recent COVID-19 pandemic showed-but 'urbanisation is reshaping population health problems, particularly among poor people in urban areas, towards

on Global Governance for Health, which preceded and informed the Lancet-O'Neill Commission, defined global governance as: The complex of formal and informal institutions, mechanisms, relationships, and processes between and among states, markets, citizens, and organisations, both intergovernmental and non-governmental, through which collective interests on the global plane are articulated, rights and obligations are established, and differences are mediated. 77 Although this definition was written in the context of global governance, its characteristics apply to national governance of urban development and health; that is, as a diffuse concept encompassing a broad range of actors, processes, and forces. Good governance is critical to a well-functioning State and the implementation of the law. 78 In contrast, issues of corruption, whether at national or local levels, undermine transparent decision-making and divert taxpayers' money away from social functions. 79 The 75 Well-being of Future Generations (Wales) Act 2015 (anaw 2). 76 The scattered and multilevel governance of these public and private actors creates unfavourable conditions to implement a shared vision of health, let alone the right to health. Law can be used as a tool to clarify functions in relation to population health and encourage cooperation to enable coherent practices that, ultimately, promote health and reduce health inequalities. 83 The second challenge identified by the Lancet-O'Neill Commission relates to 'weak monitoring, compliance, and enforcement'. 84 To take one example, weak enforcement in building safety affects health and may lead to potentially disastrous consequences in urban development. The Grenfell Tower disaster is one extreme example of this, but, more generally, health consequences of poor building conditions can be observed, 85 such as the association between dampness or mould and increased risks of asthma. Law has the capacity to establish strong enforcement mechanisms, including by creating and governing dispute resolution bodies such as the New Homes Ombudsman, as well as mandating independent agents to check compliance during and post-construction. --- Implementation of evidence-based interventions Law is a means to implement interventions that we know scientifically work in any given context. In England, urban planners and developers may not be fully aware of the harmful impacts of planning measures on population health, 86 but it is important to first assess whether evidence of these impacts exists, or whether, whilst published in scientific journals, it is not known of by practitioners. If that is the case, efforts should be directed at bridging the gap between science and practice. 79 Gostin (n 45). 80 The third challenge relates to the multiplicity of international agencies in global health governance, Gostin (n 45) 1880. 81 Gostin (n 45) 1877. 82 Department for Communities and Local Government, Devolution: A Mayor for the West of England. What does it Mean? (Department for Communities and Local Government 2017) <unk>https://assets.publishing.service.gov.uk/government/uploads/ system/uploads/attachment_data/file/608525/Plain_English_Guides_to_Devolution_West_of_England.pdf> accessed on 4 August 2022. 83 Burris and Lin (n 45). 84 Gostin (n 45) 1878. 85 Grenfell report (n 33). 86 UPSTREAM report (n 38). Law may be used as a powerful means by policy-makers at national and local levels to implement 'what works' and guard against measures that are harmful to population health and deepen health inequalities. 87 For example, if sound scientific studies show that car parks located near schools worsen asthma in children, or that 'no traffic zones' increase healthy and green modes of transport such as walking and cycling, then such measures could be implemented by law. Evidence may show 'what works' in a specific urban context; for example, cycle lanes may not be appropriate in neighbourhood with high levels of street crime. In this case, it is important that local policy-makers look at what evidence is available in the context of their own city/region, or conduct impact assessments of measures that are suggested in another context to assess whether they would have the desired impact in their own local area. Lack of evidence of positive impact on health should not be an excuse to delay or deny the implementation of measures that may be beneficial to health. 88 This is especially true in the context of urban planning and health research, where causation at the population level is quasi-impossible to establish due to the number of variables confounding the outcome, the time between implementation of any given policy or programme and its impact, and the long latent period in NCDs between exposure to a harmful agent and observation of the first symptoms. The Lancet-O'Neill Commission pre-empted this potential policy inertia: While using the best available (albeit incomplete) evidence at the time, evaluating the law's effect to ensure continuous quality improvement in health legislation is imperative. Even then, though, law makers might need to accept some uncertainty, such as where there is clear evidence that a suite of legal measures is effective at addressing a complex health threat (e.g., obesity, or gun violence), but it is difficult to determine the specific contribution of any single intervention. 89 With regard to those health policies that we know work, the WHO has published a list of 'best-buys' for public health; that is, cost-effective and evidence-based. 90 Such laws and regulations are found in the WHO's Global Action Plan for the prevention and control of NCDs 2013-2020. 91 Some of the measures mentioned can be implemented using urban planning tools. For instance, 'eliminating exposure to second-hand tobacco smoke in all indoor workplaces, public places, public transport, and in all outdoor mass-gathering areas' may be implemented by Local Planning Authorities via the designation of non-smoking zones in the city, such as near schools and shopping centres. 92 Furthermore, zoning and the designation of building uses may contribute at the local level to'restrictions on physical availability of retailed alcohol'. 93 Two measures are directly implementable via Local Planning Authorities: 'Ensuring that macro-level urban design incorporates the core elements of residential density, such as connected street networks that include sidewalks, easy access to a diversity of destinations, and access to public transport', and 'Providing convenient and safe access to quality public open space and adequate infrastructure to support walking and cycling'. 94 Such measures would promote cities that are'socially cohesive' and 'ensure access to basic goods' in England and elsewhere, and therefore ultimately promote health equity. 95 87 Burris and Lin (n 45). 88 Gostin (n 45). 89 ibid. 90 ibid. 91 World Health Organisation, Global Action Plan for the Prevention and Control of Noncommunicable Diseases 2013-2020 (World Health Organisation 2013). 92 Gostin (n 45) 1875. 93 ibid. 94 ibid. Social determinants of health, human rights, law, and urban development 695 --- Building and strengthening legal capacities for health The final legal determinant discussed in the Lancet-O'Neill Commission report is legal capacity building, with action in three areas: Legal capacity for health refers to three interlinked dimensions: effective legal environments (which include the infrastructure for drafting, implementing, and enforcing laws that promote health with justice, as well as fairly resolving grievances that arise); a strong and growing evidence base, built on the rigorous monitoring and evaluation of existing laws; and an empowered transdisciplinary health law workforce. The latter includes connected networks of well-trained professionals -legal and non-legal -who share information and strategies, and who provide technical legal assistance. 96 The first two areas are directly relevant to urban planning. As for the third, the argument may be extended to legal capacity building and awareness of health and the law in urban planning. Legal capacity could be strengthened to empower decision makers in Local Planning Authorities with sound legal knowledge so that they have the confidence to make decisions in favour of population health. Some institutions are already in place to deliver courses and training to practitioners, such as the RTPI. This function may be used to train Local Planning Authorities, developers, real estate investors, and the Planning Inspectorate to consistently promote health in planning decisions, and to recognise the circumstances under which a new development project would harm health and/or deepen inequalities. In summary, the law may be deployed in (at least) four ways to promote health and reduce health inequalities in urban development in England (and elsewhere). I have focused on the legal determinants discussed in the Lancet-O'Neill report as a basis for my reflection, and I will now extend this to human rights. --- B. Application of a human rights framework to urban development and health As mentioned in Section II above, human rights theory and practice complement public health approaches. Here, I explore how a human rights framework may add analytical power to the legal determinants of health and fill some gaps. In particular, I discuss the first two legal determinants mentioned above and argue that human rights principles provide guidance in the implementation of policy coherence and good governance in urban planning. I will then show how a human rights framework might promote a change in the way health is conceived and interpreted by powerful actors in urban planning in England. --- Legal determinant 1 and the human rights analytical framework Human rights may complement the role of the law in making policy choices. Whilst a human rights framework is certainly not the only tool to facilitate choices in public policy, it may contribute to making such choices. A human rights framework, or rights-based approach, sets out clear principles that must guide policy objectives and decision-making at all levels. This analytical framework originated with General Comment 4 on the right to housing and has been applied to other human rights with slight variations, for instance, the right to food, the right to education, and the right to health. 97 95 Marmot (n 3). 96 Gostin (n 45). 97 In the context of urban planning, the framework as applied to the right to housing could constitute a solid basis to ground policy objectives into human rights principles, and to assist officers in Local Planning Authorities to grant or refuse planning permissions. In General Comment 4, the CESCR recognises the principles of legal security and tenure; availability of services, materials, facilities, and infrastructure; affordability; habitability; accessibility of housing to disadvantaged groups; location; and cultural adequacy of housing. 98 Whilst it is not possible to discuss all these principles in detail, I consider three of them below and explain how they could inform decisions in urban development in England: affordability, habitability, and locaton. First, the principle of 'affordability' of housing is defined by the CESCR as: Personal or household financial costs associated with housing should be at such a level that the attainment and satisfaction of other basic needs are not threatened or compromised. States parties should take steps to ensure that the percentage of housing-related costs is, in general, commensurate with income levels. States parties should establish housing subsidies for those unable to obtain affordable housing, as well as forms and levels of housing finance which adequately reflect housing needs. In accordance with the principle of affordability, tenants should be protected by appropriate means against unreasonable rent levels or rent increases. 99 This interpretation of 'affordability' might guide policy in what constitutes 'affordable housing' in England. For instance, under such a principle it would not be tolerable for developers to fulfil the condition of affordability by offering shared ownership of above market-priced new-built accommodations. In addition, General Comment 4 explains the principle of 'habitability' as: Adequate housing must be habitable, in terms of providing the inhabitants with adequate space and protecting them from cold, damp, heat, rain, wind or other threats to health, structural hazards, and disease vectors. The physical safety of occupants must be guaranteed as well. 100 The CESCR refers to the WHO Health Principles of Housing as a guide to promote adequate conditions of habitability. 101 Since General Comment 4 was written, other international guidance may have been published on what constitutes adequate standards of habitability, especially under the mandate of UN Habitat, which could inform decisions in urban planning. For instance, Local Planning Authorities may need to refurbish, or in a last resort, demolish and replace, current social housing in poor condition before accepting planning applications for a new luxury housing project. Interestingly, General Comment 4 adds to the framework the principle of 'location': Adequate housing must be in a location which allows access to employment options, health-care services, schools, childcare centres and other social facilities. This is true both in large cities and in rural areas where the temporal and financial costs of getting to and from the place of work can place excessive demands upon the budgets of poor 98 General Comment 4 (n 97) para 8. 99 ibid. 100 ibid para 8(d). 101 World Health Organisation (n 43). Social determinants of health, human rights, law, and urban development 697 households. Similarly, housing should not be built on polluted sites nor in immediate proximity to pollution sources that threaten the right to health of the inhabitants. 102 Whilst the connectivity between communities and between new projects and town centres is already informing decision-making in urban planning, the level of pollution may not be taken into account systematically. 103 The principle of location would mean that a newly built scheme in a busy area of a city with high levels of traffic should not be accepted until pollution levels are lowered. It would also mean that Local Planning Authorities have a duty to reduce traffic levels near residential areas. In this regard, the 2020 decision issued by the Coroner for London Inner South linking an individual death to air pollution is fully in line with a human rights-based approach to housing and urban planning, and may constitute a first step towards a new legal conceptualisation of accountability of the State with regard to air pollution and health. 104 Some principles of a human rights-based approach to housing are incorporated in the new Healthy Homes Bill for England and Wales, such as a minimum liveable space (principle b), access to natural light (principle c), inclusive, accessible and adaptable places for all, taking into account protected characteristics under the Equality Act (principle d), access to sustainable transport and walkable service (principle e), or reduction in carbon emissions and resilience to climate change (principles f and g). At the time of writing though, these principles would apply to new homes only. Taken together with the view that law provides the tools to translate vision into action, these principles may constitute a solid basis to guide the NPPF, local development plans, and to inform decisions to grant or refuse planning permissions by Local Planning Authorities. --- Legal determinant 2 and the right to a process Human rights can contribute to using the law as a tool of good governance to regulate behaviours and relationships of powerful governmental and non-governmental institutions having an impact on health in urban development. The principles of accountability, transparency of decision-making, participation of affected communities, and non-discrimination are well defined in human rights, with guidance on how to implement them. These principles form a right to a process applicable to all economic and social rights, including the right to health. 105 Non-discrimination (Article 2.2 ICESCR) and equality between men and women (Article 3 ICESCR) are overlapping principles that apply to all the rights protected by the ICESCR, including the right to housing and the right to health. 106 The application of human rights on a non-discriminatory manner is reiterated in many General Comments, including General Comment 4 on the right to housing and General Comment 14 on the right to health. In practice, a human rights framework would include non-discrimination as a core principle enshrined in the NPPF to guide decisions of Local Planning Authorities. The principle may even extend to private actors such as developers and real estate investors when it is recognised that such actors perform a public function; for example, building housing necessary to reach the housing target of a local area, or building affordable housing. In England, Scotland, and Wales, the Equality Act 2010 implements non-discrimination in national law. 107 The Act prohibits discrimination on 'protected characteristics' which are similar in scope to grounds of discrimination in international human rights law. 108 The equality duty applies to public bodies as well as private organisations carrying out public functions, which provides a solid basis, in national law, to implement equality and non-discrimination as principles to guide decisions in urban development. Likewise, Article 14 of the European Convention on Human Rights, enforceable in the UK under the Human Rights Act 1998, protects against discrimination in the enjoyment of Convention rights. 109 Of particular interest to planning decisions, potential indirect discriminatory effects of a policy or decision by a public authority are also protected under section 19 of the Equality Act. 110 Section 6(1) of the Human Rights Act adds that a public authority cannot act in a way that is incompatible with a right protected by the Act, including non-discrimination. 111 Despite these domestic protections, it is not clear how much the equality duty and the provision against discrimination weigh in decisions taken by Local Planning Authorities to grant or refuse planning permission, compared to other priorities such as the financial viability of a project. Integrating the equality duty and provisions against direct and indirect discrimination in decision frameworks of Local Planning Authorities would offer additional guidance, in particular with regard to the actual effects of new development projects on increasing or decreasing social inequalities and their related impacts on health. In parallel to non-discrimination, accountability of institutional actors in urban development may equip such institutions with processes to scrutinise decisions and means to redress inequalities. Accountability means that States, including their organisations and agents, are primary duty-bearers, and citizens are rights-holders. 112 Meaningful accountability is not about shaming;it is about the ability of citizens to participate in, and scrutinise, decisions that affect them, as well as to obtain redress when such decisions result in violations of their rights. As such, real accountability means implementing the principles of transparency, participation, and access to remedies. The Human Rights Act 1998 offers judicial remedies when an act of a public authority breaches the rights protected by the Act, including both monetary and non-monetary compensation. 113 This reflects international human rights law where remedies do not necessarily equate to financial compensation, but also to satisfaction, restitution, and guarantee of non-repetition to alleviate a situation where rights are violated. 114 However, judicial remedies-although necessary-are insufficient alone because they require the affected person(s) to bring legal action, something that people who lack time, knowledge and money may not be able to do. With regard to the right to health, Alicia Ely Yamin has argued that accountability of duty-bearers is about what the State is doing; how much effort it makes towards the realisation of the right to health (progress); and how it implements the right ( process). 115 I posit that this conceptualisation of accountability may be extended to the context of preventing NCDs and reducing health inequalities in urban development. Accountability would mean scrutinising planning decisions on the basis of whether they advance health and reduce social inequalities, for example, as is now the case in Wales with the Future Generations Commissioner. 116 Such concerns would be weighed against other priorities, in particular the financial viability of a project. In order to assess meaningfully health (dis)benefits against, say, financial viability, data would need to inform the projected impacts on health and social inequalities, in the same way that financial data is included to determine the viability of a project. Similarly, a human rights process would require active and meaningful participation of affected communities. Public consultation already forms an integral part of urban planning projects; for example, with nationally significant infrastructure projects. 117 However, it may be reduced to a simple box-ticking exercise when not conducted meaningfully because affected communities may not have the time to participate in such consultations, or they may not even realise that a public consultation is underway for a specific area in their town. In addition, a disproportionate number of people coming from disadvantaged communities may not be equipped to understand the legal jargon used in planning applications. Sometimes, the consultation may be conducted by a consultant whose priority is to foster good business relations with the developer. It may be conducted too late, after the project is already decided, which effectively strips communities from their voice. A meaningful consultation should be conducted by an independent third party, well ahead of a project, with information communicated in a clear and concise format, with data showing potential impacts of the project, including on health and inequalities, and shared openly. If a consultation requires affected communities to attend discussions such as workshops, offering financial compensation when people are obliged to take time off work would encourage, and may increase, participation of populations who would not usually be involved in such consultations. These examples illustrate how human rights principles of non-discrimination, accountability, transparency, participation, and redress can be implemented to ensure good governance of public and private actors and guide their actions and decisions in urban development. This right to a process would ensure better health promotion and reduction of health inequalities. It may also contribute to debates on making choices in public policy by stating which values ought to be prioritised and through which principles they should be implemented, something that the Well-being of Future Generations Act aims to assure in Wales. Beyond the application of human rights principles to the legal determinants identified in the Lancet-O'Neill report, using human rights may empower communities and generate systemic change. --- Social mobilisation Yamin and Andre <unk>s Constantin have noted the power of human rights as both legal claims and discourses for social mobilisation. 118 In the context of urban development and health, empowered citizens would be more likely to act upon harmful decisions, using either their voice to show discontent through the courts challenging a planning permission of a new project, or claiming protection of their human right to health, right to housing, right to life, or right to a healthy environment. The State may contest these types of action. For example, the reduction in England of the time limit from six to three months to file a judicial review claim substantially reduced the financial and material means of civil society organisations to contest the planning of large new infrastructure projects. 119 Nevertheless, applying human rights has multiple advantages to furthering health in urban development, in particular by analysing laws and policies, implementing good governance, as well as empowering communities and mobilising social movement. Yet, human rights theory and practice face their own limits, and will not solve all issues in urban development and health. --- C. The added value and limits of a human rights approach Before discussing the limits of a human rights framework, it is worth stating the benefits that such an approach conveys. Using human rights language turns moral values (for example, health) into legal entitlements (the right to health as protected by, inter alia, Article 12 ICESCR). Philip Alston expressed this idea in a compelling manner: Anyone who has spoken with development economists, government officials or experts from international agencies about economic and social rights will almost certainly have been told: 'What does it matter if we call something a right, a need, a goal, an objective, or a target? It makes no difference, since we all want the same result!' The reality, of course, is that it makes a world of difference. Needs can be deferred until those in power think it might be timely to address them. Needs can be defined and formulated by experts; they are usually seen to be eminently flexible and relative.... Rights, on the other hand, belong to individuals, who can and will assert them and strive to give them meaning and substance. They can be neither expropriated, nor defined, nor arbitrarily put on the back burner, by officials. 120 Referring to claims as human rights gives them legal and moral legitimacy, enhances coherence in policy-making and practice across different sectors, and, therefore, improves effectiveness of the policy-making and implementation.121 This is particularly relevant in the context of urban development, where competing objectives, short political cycles, and limited resources convolute decisions in Local Planning Authorities. However, using human rights may face important conceptual-and arguably practical-limits. Audrey Chapman discussed the application of human rights theory to the social determinants of health in an article published in 2010 that is still relevant today. 122 First, the conceptualisation of equality differs between human rights and social epidemiology. Chapman argued that whilst human rights are concerned with equality of everyone before the law (formal equality), the social determinants of health are concerned with substantive equality in society. The principle of non-discrimination in human rights could require both formal and substantive non-discrimination, in the letter of the law as well as in its effectintended and unintended, 123 and this broad understanding of discrimination is embedded into UK law in section 19 of the Equality Act 2010. 124 However, this is quite different from substantive equality at all levels of society as a foundational principle of research on the social determinants of health. 125 In other words, the social determinants of health aim to achieve equal societies, whereas human rights are targeted to prevent and remedy discriminatory use of the law and State power. The reach of measures or social action recommended by both approaches also differs. In human rights law, action should focus on the most vulnerable groups of society as a matter of priority, whilst the social determinants of health recommend action to remedy unequal societies in order to reduce social inequalities at a general level and in all health gradients for everyone. 126 In British urban development, using the concept of equality as defined in social epidemiology might help to better target health inequalities at all levels of society and prevent risks of NCDs in the future, and have an impact on the population as a whole. However, resource constraints in Local Planning Authorities mean that difficult choices must be made, and that vulnerable communities should be prioritised. In this case, a human rights approach to non-discrimination, although not perfect, may be better suited to guide public policy decisions. In addition, Chapman found that the scope of action in human rights is narrower than the social determinants of health. She referred to the 'underlying determinants of health' as enunciated in General Comment 14 and reports of the Special Rapporteur on the right to health, and showed that this term refers principally to the conditions necessary to be in good health, such as clean water and sanitation, adequate housing, adequate supply of nutritious food, or health-related education and information. 127 In contrast, the social determinants of health reach to daily conditions extending to the availability and quality of education, fair employment conditions, political empowerment, urban planning, as well as issues of distribution of power, money, and resources. 128 In that regard, the underlying determinants of health, as defined in international human rights law documents, would have a narrower impact than the social determinants of health in urban planning. To some extent, the underlying determinants already inform urban planning decisions; that is, provision of clean water and sanitation, or adequate supply of food in a local area. Public bodies may, though, benefit from implementing a vision of the social determinants of health to guide their objectives and decisions. For example, such a vision may encourage a fairer distribution of power and resources between central government and local authorities. This last remark reveals the limits of a human rights framework to address the prevention of ill health and amelioration of health inequalities in urban development. As demonstrated above, and with rich and compelling research on the social determinants of health, the unequal distribution of power, money, and resources lies at the heart of the interrelated problems of poor health outcomes and health inequalities. However, it is unclear whether human rights aim to, and indeed do, address systemic issues of power in market-driven societies. Whilst I admit that human rights cannot be apolitical, the process of their implementation and monitoring seems to avoid, as much as possible, political disputes. 129 This may be due, in part, to the fact that the implementation and monitoring of human rights are largely constrained by State sovereignty. Some may even argue that the implementation of human rights is dependent on States' goodwill. 130 Despite these limits, there are important benefits of a human rights framework, as I have sought to demonstrate here, and may contribute to debates on integrating values, such as health equity, in public policy and decision-making processes. For this reason, I concur with the thinking of human rights scholars who argue that a human rights approach, or, more specifically, the right to health, should be more concerned with addressing structural issues of power distribution in market capitalist economies. 131 As Yamin and Constantin said in their review of the health rights from the 1990s to 2017: [I]n order to fulfil their promise to change the systems that perpetuate inequality, human rights-based approaches to health'must adopt a holistic and integrated approach to health 126 Chapman (n 54); Marmot (n 3). 127 Chapman (n 54); General Comment 14 (n 50). 128 Marmot (n 3). 129 For instance, the UN treaty-based monitoring system prioritises dialogue between States parties, human rights bodies, and civil society, aiming to improve the implementation of human rights rather than shaming. 130 Donnelly (n 47). 131 systems' and examine 'the dominant assumptions underlying the structural determinants of... health... [including] strategies to address those factors, to reshape the possibility frontier for advancing... health.' In other words, they challenge power structures and the status quo. A human rights-based approach therefore changes decision-making processes and the issues and actors included in those processes, as well as outcomes. 132 --- IV. CO NC LU S IO N In this article, I have explored the conceptual and practical links between urban development, health, law, and human rights. I grounded the analysis in literatures across the disciplines of global health, law, and human rights. Using research on the social determinants of health as a basis, I have shown how law, conceptualised in a broad sense, is a key determinant of health. I argued that human rights-as enacted in international human rights law and partially implemented in UK domestic legislation-are relevant to research in health and urban environments. Principles of universalism, interrelatedness, and indivisibility essentially mean that the social determinants of health are about human rights. Furthermore, human rights share a common goal with public health, ie promoting human well-being. I relied on the Lancet-O'Neill report to show how the power of the law may be deployed to promote health and reduce health inequalities in urban development in England. I supplemented this analysis by explaining how a human rights framework may provide additional guidance, in particular by implementing non-discrimination, equality, accountability of the State and its agents, participation of affected communities, transparency of decisions, and redress. Although such principles are found in international human rights law, the principles of non-discrimination and equality have the added value of being justiciable in England through the Equality Act 2010 (in England, Scotland, and Wales) and the Human Rights Act 1998. Finally, I discussed the limits of a human rights framework, especially in relation to the consideration that they have for structural inequalities of power, and I argued that more attention should be paid to the 'causes of causes'. This article adds to the literature on two levels. First, conceptually, it explores the links between interrelated questions in public health, law, human rights, and urban planning to achieve the common goal of human well-being. In doing so, I have applied the global health concept of the legal determinants of health to the domestic context of English urban development. Secondly, this research offers practical suggestions on the use of a human rights framework to determine health as a core value of urban development to achieve public policy goals. Whilst contributing to debates on competing priorities in public policy, it highlights the need for further research in the problem of empowering regulators and decision makers effectively to balance competing public policy goals in urban development. This type of research should be informed by several disciplines, including law, urban planning, public policy, public health, and human rights, and be complemented, as far as possible, by empirical evidence. --- A CKN O WL EDG EM ENT S I would like to thank Professor John Coggon and Dr Edward Kirton-Darling for valuable comments on the draft of this article, as well as two anonymous reviewers at the Medical Law Review.

The COVID-19 pandemic exacerbated profound inequalities in the conditions in which people live, work, and age. Law plays a critical role in shaping these structural health inequalities, which have existed for decades. This dynamic can be observed at the local level, with cities operating as environments unequally distributing the risks of non-communicable diseases between population groups. This article first focuses on urban development to explore the conceptual links between health inequalities and the role of law. I expand this observation and I posit that the social determinants of health are about human rights. With that in mind, I argue that human rights are necessary to address the issue of unequally unhealthy urban environments, hence recognising that people are entitled to a minimum essential level of the conditions in which they live, work, and age, which the State is responsible to fulfil. By way of strengthening my argument, I lay out how a human rights framework can improve these conditions and ameliorate unfair inequalities. Finally, I recognise and respond to the limits of a human rights approach. K E Y W O R D S : Health inequalities, Human rights, Right to health, Social determinants of health, Urban development, Urban planning, Legal determinants of healthIn 2019, 83% of the population of England lived in urban areas. 1 This proportion is expected to grow, with a faster growth rate in urban than in rural areas. 2 Infectious diseases spread quickly in cities-as the recent COVID-19 pandemic showed-but 'urbanisation is reshaping population health problems, particularly among poor people in urban areas, towards

Introduction Educational content expressing values in contexts of social and cultural diversity refers to an assemblage of knowledge and wisdom, the transmission of which to children is desirable and promotes the upbringing of human beings in the framework of each individual's schooling and education [1]. This content is a part of education in the family context, which relies heavily on the role of families and wise adults who orient children's behavior [2]. Educational content expressing values allows the development of character and integrity as expected by indigenous societies; hence, there is a need to identify such content and include it in school education from the earliest levels of school, i.e., Initial Education. Under the Chilean General Law of Education (LGE), Initial Education refers to the education imparted to children in the country from birth up to the age of six years [3]. This level of education is the gateway to school education; therefore, the Education Ministry (MINEDUC) considers that the integration of both indigenous and non-indigenous educational knowledge is crucial for an individual's life [4]. The proposal for intercultural education at this level lacks guidelines to orient the work of teaching teams; thus, experiences in intercultural education are limited to particular scenarios, such as intercultural educational programs and diplomas [5,6], as in the case of educational content expressing Mapuche values. The Mapuche people ("people of the earth") are Chile's largest indigenous group. In cultural and identity terms, they are a binational people, scattered over both Chilean and Argentinean territory. In Chile, the Mapuche account for 79.84% of the indigenous population [7]). They live in the area included in the Valpara<unk>so to Los Lagos regions, within which area the following territorialities can be identified: pikunche, people who live in the territory of the North; nagche, people who live in the lowlands; wenteche, people who live in the plains; pewenche, people who live in the Andes mountains; lafkenche, people who live along the Pacific coast; and wijiche, people who live in the territory of the South. A review reveals that proposals for incorporating educational content expressing indigenous values into Initial Education have been developed in several countries, such as the following. In New Zealand, the incorporation of values allowed the implementation of the teaching resource entitled He M<unk>puna te Tamaiti ("every child is precious and unique"), which promotes culturally receptive educational practices and the integrated welfare of Maori children and families [8,9]. In Canada, Entewate'nikonri, or "sake" teaching (creation of artistic spaces following Mohawk practices), coordinates lines of the school curriculum with educational content expressing Mohawk values. This program is aimed to educate children based on educational objects that will promote the spiritual interconnection of the individual with the land, the community, and the history of creation [10]. In Mexico, the Milpas Educativas have incorporated educational content expressing values based on respect, encouraging children to value both the elders (sages) of their clans and members of their own families [11]. In Colombia, the Basctemingbe Yebna children's home and bilingual school draw on the culture, traditions, and values of the indigenous group to educate indigenous persons to form part of the people, stressing their identity and language, the earth, and the animals, the ancestors, and gods [12]. In Chile, educational content expressing Mapuche values is incorporated into Initial Education through intercultural educational programs developed by institutions such as the Integra Foundation and the National Association of Kindergarten Schools (JUNJI). These programs are intended to revive and incorporate Mapuche educational knowledge and wisdom, incorporating Indigenous Language and Culture Educators (ELCI) who support the work of Initial Education teachers and assistants [13]. Furthermore, some Initial Education institutions apply to have their Institutional Education Project (PEI) certified with the Intercultural Seal, autonomously imparting intercultural education [14]. In the Araucan<unk>a Region of Chile, educational guidelines have been drawn up to orient intercultural education in Initial Education; for example, a book entitled Kimkantuai<unk> brings a bilingual, intercultural approach to Initial Education [15]. This material offers educational proposals organized according to the ambits and focuses of the Curricular Bases for Initial Education (BCEP), based on Mapuche educational content implemented by kindergarten teachers and traditional educators. In the Mapuche context, educational content expressing values refers to knowledge and wisdom framed as ideas and reflections on respect for people, animals, and nature. It includes reflections based on the welfare of people and the different beings that make up the territory [5,16,17]. These reflections are associated with principles such as (a) respect for the family, the community, and the material and spiritual environment; (b) obedience to the adults of the family and community; (c) solidarity with the different members of the community; (d) hospitality to visitors; (e) patience, a requirement of Mapuche social rules; and (f) kindness, showing a friendly attitude toward people [17][18][19]. The Mapuche educational content presents similarities to those identified in other contexts; nevertheless, it is unique in its socio-spiritual basis, where a particular worldview emerges. This proposes a direct relation of the individual with the material and immaterial beings that make up the territory, a fundamental element of Mapuche education [16,20]. This review of background information from Chile and other countries shows the feasibility of implementing educational proposals incorporating educational content expressing indigenous values. Nevertheless, although intercultural educational programs to incorporate educational content expressing values have been implemented in Initial Education in Chile, they are inserted in a Westernized curriculum that involves largely inappropriate teaching methods. By the same token, they are reduced to particular scenarios, and their benefits are not extended to all children, both indigenous and non-indigenous [6,21,22]. Given the above, there is an urgent need to implement intercultural educational models that meet the needs and expectations of families and communities in indigenous contexts [23,24]. This implies continuing to develop education policies that incorporate Mapuche's wisdom and knowledge into the concept of education in the school education system. This is especially true of Initial Education, which is a vital stage in the construction of sociocultural identity [4,21] if we consider the cultural variations that exist in emotional socialization processes [25][26][27]. The problem is therefore found in the scarce incorporation of educational content expressing Mapuche values into the teaching of Initial Education professionals who work in kindergartens located in indigenous communities. This, in turn, may be attributed to ignorance of Mapuche educational wisdom and knowledge [6,21], resulting in teaching based on a Western structure, with a curriculum and with teaching and evaluation methods that follow a homogenizing logic. --- Mapuche Family Education: Kimeltuwün Mapuche family education is a process that guides the upbringing of children both in the context of their family and community and their immersion in Chilean society, to bring up future generations based on kimün (Mapuche knowledge) [28]. Kimün comprises knowing, learning, and feeling; it is linked with wisdom, teaching-learning, and awareness, and with knowledge about families, including the values considered desirable in the teaching of new generations [28]. Mapuche family education promotes cultural, social, and historical elements that have been handed down from generation to generation by word of mouth and social memory [29][30][31]. An important element of this is kimeltuwün, the action of educating, which encompasses the processes of teaching, contextualization, and the linking of what is learned with the natural and social environment [6]. Kimeltuwün is an educational model that guides the upbringing of the che (person), the final object of Mapuche education. The upbringing of the che involves developing autonomy, self-efficacy, and self-control, which are achieved through the facilities provided by adults for the activities performed by children [26] in the company of family and community members [32]. Thus, they learn about their culture and values through profound observation [18]. Some of the more important cultural patterns are spontaneous cooperation to achieve common objects as well as the motivation and initiative to participate in community activities [33]. Kimeltuwün is systematized in educational principles, methodologies, and objects [6]. The methods include (a) gübam, the structuring of learning; (b) wewpin, a synthesis drawn up from different ideas discussed by various people; (c) nüxam, learning based on conversation focusing on family relationships and social memory; and (d) pentukuwün, exchange of knowledge with members of the community or people from other territorialities [6]. Thus the object of family education is to bring up children to match the ideal of an educated Mapuche person (che); this requires bringing up both boys and girls on an equal footing, based on four objects of education. While it has been observed that differences exist between the roles performed by boys and girls [18,33] and that the value of respect when applied to Mapuche girls transcends the human and embraces the natural world of non-human living beings [17], no gender-based differences are observed when it comes to value content in their upbringing. The four objects are as follows: (a) kimche (wise person): it is expected that children will be brought up to know and understand their culture, nature, and relations with the spiritual beings belonging to the territory. In this context, they learn about history, the origin of the people, and their ancestry [6,18]. (b) norche (responsible person): it is expected that children will be brought up to act responsibly towards other people and to respect the members of the community [6]. To do this they are taught control, which means learning to identify corrective actions to avoid becoming a welu zuam (degenerate) [18]. (c) kümeche (good person): this refers to a person who does what is good and who has achieved stability by effort [5]. Children learn to be good people by participating in ceremonies like nguillatun and community activities like rukantun (collective building of a house), visiting families with sick members, and rendering thanks and offerings to the spirits of nature [18]. (d) newenche (strong person), refers to people with mental and physical strength [5,19]. Girls and boys are taught to be resilient, which is linked with a sense of strength. This is why emphasis is placed on teaching children to be proud of their origin [18]. For Mapuche society, education has a spiritual basis that includes a subset of their practices, languages, and cultural beliefs, connected with a network in which all beings are nested in their worldview [34]. Thus, it is necessary to provide a well-rounded education to children in the early years of their lives [18]. Recent investigations into Mapuche childhood have shown how this is possible [5,18,19]; the children's activities reported in these works reveal the ideal form of Mapuche education, based on the educational objects listed above (kümeche (good person), newenche (strong person), norche (responsible person) and kimche (wise person)). The Mapuche spirituality and worldview give this education a unique, individual stamp, by which this ideal can be distinguished from other models of indigenous education [5,6,18,19]. --- Materials and Methods The methodology used was qualitative educational research, describing and interpreting phenomena through evaluations of educational practice, from the perceptions and interpretations of social actors [35][36][37]. An interpretive paradigm was used, allowing us to study reality through the interpretations and significations of the participants, to obtain a deep understanding of the cases from a cultural and historical perspective [38]. The approach adopted was indigenous investigation, since this includes recognition of the episteme and the specific ways of seeing the world-from the indigenous angle-and implies following sociocultural research protocols [39]. Thus, we used Mapuche cultural protocols such as nütram (conversation) and wintrankontuwün (paying a visit). Our research approach therefore involved working in parallel with the participants to achieve recognition of the indigenous episteme [40]. --- Context and Participants The context was the district of Saavedra in the Araucan<unk>a Region of Chile, located in the territory of the lafkenche (people who live beside the sea). It has a population of 12,450 inhabitants, of whom 99.1% self-identify as Mapuche [7,41]. The selection of the participants was intentional The inclusion criteria were: (a) boys and girls enrolled in the middle kindergarten level; (b) parents who self-identified as Mapuche and who were responsible for a child aged under 4 years belonging to the kindergarten catchment areas; (c) kimches who lived in the communities; and (d) Initial Education professionals who had worked in the kindergartens for more than one semester. All others were excluded. --- Analysis Procedure Data were collected by semi-structured interviews, a method in which topics are interwoven and a comprehensive knowledge of reality is constructed [38,42]. The interviews followed the Mapuche oral tradition nütram (conversation). In the case of the kimches (sages), the interview followed Mapuche formalities, such as (a) pentukuwün, discourse, and sociocultural practice incorporating formal greetings between people [6,28]; and (b) witrankontuwün, the action of paying a visit, which includes rules for verbal and behavioral interaction between hosts and visitors [33]. The instrument was a scripted semi-structured interview, developed by the authors. Data analysis was based on Grounded Theory, an inductive method for generating theoretical proposals through analysis and comparison of the data collected [43]. The analysis procedure followed the coding paradigm, carrying out (a) open coding, to identify emerging categories; (b) axial coding, to refine and relate the categories defined; and (c) selective coding, selecting a central category and relating the other categories to it [43]. The ATLAS.ti program [44] was used to support qualitative analysis by coding. --- Ethical Aspects The project underlying the present investigation was approved by the Research Ethics Committee of Universidad Católica de Temuco. The study complied fully with the deontological guidelines laid down in international agreements, such as the Singapore Statement [45], and current Chilean legislation (Law 20.120) [46]. All the adult participants received an informed consent and confidentiality statement, while minors received an informed assent. They thus had the opportunity to decide whether or not to participate, being informed of the voluntary and anonymous nature of participation. Likewise, they were informed that participation did not involve any kind of health risk and would bring no kind of economic benefit. The data were stored confidentially, and both the interviews and the identity of the participants were coded for processing. The records will be eliminated after five years. Mapuche social and cultural protocols and formalities were observed, such as ma<unk>umtu (thanking the participants), pentukuwün (formal greetings and introduction), inatuzugu (investigating existing knowledge), and kimeltuwun (wisdom, knowledge, and methods of educational action) [47]. --- Results The results were organized into three categories: (1) educational content expressing Mapuche values; (2) teaching of educational content expressing values in the family and the community; and (3) teaching of educational content expressing values in Initial Education. --- Educational Content Expressing Mapuche Values Educational content that expresses Mapuche values comprises subjects important to teach in the upbringing of girls and boys, in the opinion of parents, kimches, and Initial Education professionals. It also implies attributing significance to these values from the perspective of the Mapuche worldview. This category may be divided into two subcategories: --- Educational Content Expressing Values and Ideal Character of a Person This category comprises educational values forming part of the ideal upbringing of a Mapuche person. The participants stress that children should be brought up to be (a) norche (responsible person), with special emphasis on speaking the truth in any situation, so that when they become adults, these values will be part of their lives, guide their actions, and allow them to play a role within the community; (b) newenche (strong person), learning the value of sacrifice, placing value on what one has-both material and immaterial-as a Mapuche person must be serious and hardworking; (c) kümeche (good person), which implies bringing up children to help other people, to be cheerful and affectionate, to respect adults, and to be grateful for what they have; and (d) kimche (wise person): wisdom implies knowing about a topic in-depth and sharing this knowledge. These types of knowledge are associated with roles or trades in Mapuche society, such as the role of lawentuchefe (a person who knows about healing plants) and the trade of woodworking. The above is reflected in the following testimony: "Er... I think that this is the value of being human, having the value of respect, and affection, and having the will, and the spirit of sacrifice. Our people, the Mapuche people, have always had this spirit of sacrifice, all for one and one for all; there are no greatly ambitious people as there are in other peoples and races". (Interview with kimche, Saavedra) The ideal upbringing for Mapuche society is linked to spiritual education, which resides in their way of seeing and understanding the world; it is therefore essential to bring up children with the virtues described above. This is apparent in the following testimony: "... It doesn't do you any good to work so much and have so much wealth, and afterwards you die and leave it all behind on earth... It's no good being like that, ambitious. You should be... good people, I always tell them [his children] that: don't fight with the neighbors or anything... ". (Interview with kimche, Saavedra) The above testimony shows the importance of bringing a person up to be aware of their surroundings and have the ability to live and relate harmoniously with all the beings that form part of this world, be they people, spiritual beings, or natural elements. --- Nature The second subcategory refers to nature, as an axis that runs through the educational content expressing essential values. This concept implies that to ensure the ideal upbringing of a Mapuche person, it is necessary to know, respect, and value nature and all that it provides, such as food. The concept of nature emerges as mapu (earth), encompassing everything on the planet. The value of nature, seen as an entity with which one is inseparably related, is emphasized in children's upbringing. Within this relationship, children are taught to value all the elements that make up nature, both material and spiritual. Thus, nature is to be respected and valued as the provider of food. The participants stress the value of the food that the earth brings forth, which is natural food, not processed like modern food, and therefore healthier. An example of this is found in the following testimony: "One-human being and two-Nature, immediately after. The human being and nature are alive, so they need: one-earth; two-water; three-air; and four-the sun or fire, for a normal existence" (Interview with kimche, Saavedra) (translator's note: Mapuche culture makes use of numerology, particularly duality and to a lesser extent the number four. This explains the enumeration of elements in this citation). The participants expressed concern because, nowadays, little value is attached to the earth, as nature is not valued or respected. They stressed the importance of educating children based on values to ensure that they take care of <unk>uke mapu (Mother Earth) and preserve natural resources, such as lawen (medicinal herbs). The concept of punishment by nature emerges in the narratives, referring to the consequences of undervaluing nature. The participants say that when there is a lack of respect and care, nature punishes those who attack her. This conception can be understood from the following testimony: "The earth is sacred, because the earth gives us food. It gives us everything: water, grass, animals; everything we see comes from Mother Earth, nature, the <unk>uke [mother], from chaw dios [God, the father]. They own all that, nature, and we have to take care of it while we live here". (Interview with kimche, Saavedra) To sum up, the first category allows us to understand in ontological terms which elements of knowledge form the basis of the teaching of educational content expressing Mapuche values to children; this content arises from family education, which will be described below. --- Teaching of Educational Content Expressing Values in the Family and the Community The second category refers to the teaching of educational content expressing values, from the perspective of Mapuche families and sages. The participants highlight the importance of teaching based on traditional Mapuche education, to educate children to achieve to the ideal upbringing described in the first category. This category is divided into two subcategories: --- Educational Content Expressing Values in the Home This refers to educational practices used by families to teach educational content expressing values. These practices have been handed down from one generation to the next, and according to the parents, are replicated in upbringing today. The teaching of educational content expressing values relies, in particular, on practices such as (a) observation-the children learn by observing the actions of family members, principally older siblings and parents. For this reason, importance is given to good behavior, since adults are a model to be followed; (b) prayer-children are encouraged to form a relationship with spiritual beings that are part of the territory, who provide food, health, etc.; (c) stories-educational content expressing values is conveyed by the telling of stories that have been relayed over the generations, known as epew. These stories teach children about what is or is not socially expected. Thus, stories about actions that occurred in the past are used to teach practices and behaviors that fit in with Mapuche family education; (d) advice-this is given in conversations to explain how people should behave socially. The person giving the advice reflects on how children should behave and why this behavior is expected of them. Examples of the practices described are given in the following testimonies: "'Look', I said to him... 'you know that's dangerous, don't do it... if your classmates are smoking drugs, don't do it. Because they have their own mothers and fathers, and you are the son of a different woman, the daughter of (... ), that's what you two are (... ). He never pointed the finger at anybody, and now that his son is pointing, I am going to feel terrible, to lose my self-esteem' I said. 'So no, not you. You can't be angry or refuse to have anything to do with him". (Interview with kimche, Saavedra) "In this case, yes; because I have a daughter. So I have seen attitudes-for example, my daughter may, I don't know... suddenly show a bad attitude, and if you say something to her... And then the same day, I don't know, in the evening or the next day, well Carlos will do the same, he does the same. He really does. It sometimes happens that my daughter loses her temper and slams the door. And what happens? When the next day, or another day, Carlos loses his temper and slams the door, then you say 'Ah-ha!'". (Interview with a mother) --- Teaching of Educational Content Expressing Values in the Family and the Community This refers to the significance assigned by the Mapuche family and community to the teaching of educational values. It also refers to the importance assigned to educating the new generations on values, as was done traditionally, as it is evident that these have changed over time. In this respect, the participants stress education focusing on values that were traditionally important for Mapuche society, such as respect, which according to the interviews has gradually lost its significance, as is seen in the following narration: "If we followed all those values inculcated by our forebears, I believe that we would be a very respectful generation because now it has been emphasized heavily. Our grandparents themselves say-my grandfather speaks so much sense, and the latest thing he said was that the children, the kids [small children], when they come by, practically beat their parents, their teachers (... )". (Interview with a mother) Another concept that emerges is the importance of educating children to be proud of being Mapuche. By teaching educational content expressing values, rooted in Mapuche knowledge, in their earliest years, they will be educated with knowledge and appreciation of their culture. Below is an example of this: "I always disagree with my partner, because I say 'Hey, they are descended from their ancestors'. I say, 'your grandmother was a machi' [spiritual leader of a community]. Because now they deny that they come from there, so that's how they cut themselves off. Because if you belong to a Mapuche indigenous people, you should go along, take part". (Interview with a mother) Finally, within family and community knowledge, the idea arises of educating children on duties, in other words, an upbringing that includes responsibilities, and at the same time, one that ensures that children value and respect adults, as they used to. In this respect, the participants stress that children should grow up with responsibilities; in this way, they will learn educational values such as helping, respect, and sacrifice, associated with the educational objectives described in the first category. An example of the above is found in the following testimony: "I taught my son... well, the responsibility of going to bed early so as to be able to get up in the morning, to be able to go to kindergarten the next day; he is taught respect, he is taught-well-that he has to respect adults". (Interview with a mother) This can also be seen in examples from everyday life, such as household chores, which include helping to keep the house tidy and looking after animals, as the following example shows: "Yes, to make a home for it [the mother goose], make a nest for it, because if you don't make it a nest then you aren't bringing the animals up. You don't breed birds and then abandon them. So you tell them that they must make and prepare a bed for it, so the gosling can be born, and then support them, give the goslings water and food; they eat grass, anything, but you have to look after them". (Interview with a mother) Based on the narratives presented in this category, we can see the importance assigned by family and community members to practices and knowledge lodged in the social memory; the narratives emphasize appreciation for traditional upbringing and certain positive aspects that should be maintained in raising children today. As we see from the testimonies, some Mapuche educational values-such as respect-have changed over time. This presents difficulties for society, since respect is one of the pillars of the ideal upbringing of the che (Mapuche person). Although changes are observed at the level of society-both Chilean and Mapuche-in how children are brought up under the influence of modernity, retaining the essence of these values is fundamental for the preservation of kimün (Mapuche knowledge). For this to endure, and for the teaching of values based on a dialogue of knowledge between the Mapuche and Western worldviews to continue, the incorporation of these versions of knowledge into Initial Education is fundamental, since this level of education is attended by children in their earliest years, a key age for acquiring educational content regarding values. --- Teaching of Educational Content Expressing Values in Initial Education The third category refers to teaching practices implemented in kindergartens for teaching educational content expressing values. This category also refers to the outstanding challenges for the incorporation of educational content expressing Mapuche values. The participants commented that strategies for incorporating Mapuche knowledge are implemented in the kindergartens, as every kindergarten has an Indigenous Language and Culture Educator (ELCI). Both parents/guardians and teaching teams recognize that progress is now being made towards a type of education that integrates the knowledge and wisdom of the Mapuche community. This change is considered positive, as exemplified below: "Yes, once a week, we do... a llellipun (rogation), on Mondays, when we are going to start the week, so things go right for us, so the children are all right; for everyone who works here in general, the children, everyone, we mark Mondays as the day of the llellipun (rogation) and during the other days too". (Kindergarten Director) Parents and guardians are pleased regarding the incorporation of Mapuche knowledge during everyday activities in kindergarten, because they value the fact that now Mapuche cultural practices are taught from early childhood. They also value the fact that the children are taught words in Mapuzungun (Mapuche language), which they then repeat at home, as the following testimony shows: "Our children are proud, especially here in the kindergarten, that this is happening and that from an early age they can greet people with mari mari [hello], er... 'go and drink some water' 'Mummy, it's ok', I don't know, 'go and eat a potato' 'No, Mummy, it's po<unk>i'". (Interview with a mother) This context brings out the role of the ELCI, who plays an active part in teaching educational content expressing values and, in general, in the incorporation of Mapuche knowledge (kimün) in the kindergartens. Although Mapuche knowledge is incorporated into everyday activities, either once a week or daily, the narrations of the participants show that this role is assumed mainly by the ELCI; when they are asked about the teaching of Mapuche educational values, this practice is immediately associated with the ELCI. This result is striking, as it indicates that the teaching of Mapuche educational values in the classroom is led by just one person. This is a source of concern regarding the integration of knowledge and wisdom associated with educational content expressing values, since the teaching of this content should be generalized in everyday activities in kindergarten. This idea is reflected in the following statements: "Now, we have an ELCI... the family has said... for example they are delighted with our ELCI, and the activities he has carried out; so, from an early age we are inculcating respect and appreciation for the Mapuche culture". (Interview with Kindergarten Director) "It's very good, what the kindergarten here is doing by having an ELCI; that the children are already picking up and absorbing the culture. And when they come home, I don't know, sometimes they say 'hello' without any encouragement. There they are, when they arrive,'mari mari' [hello], or they are eating, and instead of saying 'bread' they say 'kofke'. So seeing what they are picking up in the kindergarten is a great basis, it is a great basis they have". (Interview with a mother) Kindergartens therefore need two things in order to teach educational content expressing values: (a) Community participation in the kindergartens, in other words, community demand for participation in the kindergartens to incorporate Mapuche knowledge into teaching. The participants highlight the importance of including family and community members, as they know about family education and thus about educational content expressing Mapuche values. If a dialogue could be generated between school education and family education, Mapuche's knowledge could be incorporated by all of the members of the teaching teams, not just by the ELCI. This is expressed in the following statement: "What is very important is that we should be included, that the school should recover how we teach the children values at home and follow those values, inculcating the values that we were taught by our grandparents, and go on inculcating them". (Interview with a mother) (b) The need to incorporate Mapuche knowledge into the training of Initial Education teachers, in other words, the needs and demands of the teaching teams to acquire knowledge and tools for teaching educational content expressing values based on Mapuche knowledge. We see in the kindergartens that work has started to progress in the incorporation of Mapuche family education; however, the teaching teams need greater knowledge and competence to do this, as described below: "I think it should be taught as part of the curriculum, whether or not, er... a university is located in a Mapuche region; it should not be conditioned by that, it should be in the national curriculum". (Interview with Director of Kindergarten No. 1) "No, and I feel that this is also linked with the question of the Curricular Bases because, with the existing Curricular Bases, they should adjust to the people who are studying in the university to become Initial Education teachers. For our future colleagues, there should be a part, a section [of the curriculum], that teaches them some intercultural content". (Interview with Director of Kindergarten No. 2) What is suggested in these testimonies would be a key element to ensure that the teaching of educational content expressing Mapuche values is not the sole responsibility of one member of the team, the ELCI. If all the members of the teaching team knew about Mapuche family education in the early years of a child's life, the role of teaching educational values would be shared between all agents (directors, educators, education assistants, and ELCIs). To summarize, the results reflect that educational content expressing Mapuche values is a fundamental axis in children's education during the early years of their lives; this is supported by the families, the community sages, and the teaching teams in the kindergartens. Educational values play a fundamental role in educating children following the ideal upbringing of a person (che), as they are the qualities that a person must possess to self-identify as Mapuche. This implies a challenge for kindergartens, which educate children aged six and under, a challenge that can only be addressed through cooperative work between kindergartens, families, and the community. --- Discussion Our investigation enabled us to reveal, from the words of parents, Mapuche sages, and Initial Education professionals, the educational content expressing Mapuche values that would be desirable to teach children in the early years of their lives. This content is an essential axis of upbringing in early childhood, since teaching them fosters the expectation of bringing these children up according to the ideals of the Mapuche community [6,19]. This places emphasis on the role of childhood, since for the Mapuche people, it is a time window for learning kimün, which can thus be perpetuated from one generation to the next [18,21]. This is particularly important, as it can be connected with investigations that have shown the vital importance of incorporating components of Mapuche education systems into formal education [16,48]. Furthermore, the findings are related to the importance of highlighting cultural differences in the processes of emotional socialization, which occurs in the child's first years of school [25,26]. The teaching of Mapuche educational values occurs mainly within the family, which thus can be seen as the primary educator. The educational action called kimeltuwün is recognized in Mapuche family education. In this process, children participate actively in family and community activities, learning through profound observation and participation [6,22]. Some practices for the teaching of educational values are highlighted

There is a lack of knowledge about Mapuche educational values among teachers in kindergartens located in indigenous territories. The object of the present work was to identify educational content expressing Mapuche values, to contribute to an epistemic base for the education of boys and girls, both Mapuche and non-Mapuche, with an intercultural perspective. The methodology used was qualitative educational research, with a focus on indigenous research, following the interpretive paradigm, through a collective study of kindergartens attended by Mapuche children in the district of Saavedra. Sixty-seven people participated: children, parents, kimches (Mapuche sages), and Initial Education professionals. The preliminary results are organized into three categories of value content: (a) educational; (b) family and community education; and (c) teaching in Initial Education. It is concluded that the teaching of Mapuche educational values in early childhood should be coordinated between kindergartens, families, and the community; some ideas for teaching these values are suggested.

be desirable to teach children in the early years of their lives. This content is an essential axis of upbringing in early childhood, since teaching them fosters the expectation of bringing these children up according to the ideals of the Mapuche community [6,19]. This places emphasis on the role of childhood, since for the Mapuche people, it is a time window for learning kimün, which can thus be perpetuated from one generation to the next [18,21]. This is particularly important, as it can be connected with investigations that have shown the vital importance of incorporating components of Mapuche education systems into formal education [16,48]. Furthermore, the findings are related to the importance of highlighting cultural differences in the processes of emotional socialization, which occurs in the child's first years of school [25,26]. The teaching of Mapuche educational values occurs mainly within the family, which thus can be seen as the primary educator. The educational action called kimeltuwün is recognized in Mapuche family education. In this process, children participate actively in family and community activities, learning through profound observation and participation [6,22]. Some practices for the teaching of educational values are highlighted, such as teaching by example, giving advice, telling stories, and prayer, both of rogation and thanksgiving. These forms of teaching constitute an upbringing focused on spirituality, with the object of bringing children up to be wise (kimche), responsible (norche), strong (newenche), and good (kümeche) [18,19]. This is supported by recent findings in schools located in areas of social and cultural diversity in the Mapuche context. One study showed that the pupils who self-identified as Mapuche were described by their teachers as calm, respectful, and conciliatory, particularly in situations of conflict, presumably due to their upbringing and the transmission of knowledge in their family environment [49]. Another investigation showed that pupils who self-identified as Mapuche perceive more physical and verbal violence than non-Mapuche pupils, from which it may be presumed that Mapuche family upbringing stresses healthy coexistence and considers acts that other pupils do not see as violent to be unacceptable [50]. To raise a child according to this ideal requires the generation of a dialogue between the knowledge acquired by the children at home with their families and that which is taught in Initial Education, the first level of formal education in Chile. This work, therefore, reaffirms the importance of incorporating the knowledge that underpins educational content expressing Mapuche values, as these forms of knowledge can become lost when the children start to attend school [6]. The results of this investigation show that the teaching teams of the kindergartens recognize the need to for educational content expressing Mapuche values and that there is a willingness to integrate such content into formal education. For this to happen, it must be recognized that learning is constructed by interaction with the socio-cultural context [51]. Therefore achieving a dialogue between the knowledge acquired in family education and that stipulated in the curriculum requires coordinated work between schools (kindergartens), families, and the community [48]. Educational agents must therefore be mediators in the construction of learning [52]. For this to happen, they must mediate the incorporation of knowledge from the family and the community, to achieve the co-construction of knowledge. The term co-construction implies integrating a variety of forms of wisdom and knowledge from societies; forms that are recognized and valued in a framework of understanding and legitimacy, promoting new knowledge [53]. The school system, in spaces of social and cultural diversity-in the Mapuche context in particular-is the formal structure for the transmission of knowledge; its objective has been to teach children about the world from the worldview of the dominant other [54]. This has meant that the socialization of indigenous children occurring in the school system has been marked by expressions and experiences of structural violence [50], which has doubtless limited the personal growth of children belonging to a subordinated group [55,56]. There is thus a need to open spaces for change, not from a position of domination, but one of acceptance and cooperation [57]. The idea of an Initial Education with an intercultural perspective, therefore, needs to be rethought, to progress towards a curriculum that recognizes the value-related knowledge and wisdom of the children and their families and integrates this into educational practice [10,21]. --- Conclusions We conclude that there is a need to incorporate educational content expressing Mapuche values into Initial Education. To do this, local knowledge and wisdom need to be integrated into Initial Education, since currently teaching is restricted to a single line, i.e., Western knowledge. To reverse this situation, we stress the importance of achieving coordination among the teaching teams in kindergartens, the family, and the community/territory where the kindergarten is located. To this end, we propose five ideas: (a) Talking with the families whose children are in kindergarten in order to learn what educational values they teach their children, how they teach them, and the contents that underlie them; (b) Starting working with the community/territory where the kindergartens are located. To do this, we suggest communicating with the families to learn about their sociocultural protocols, for example, by communicating with the traditional authorities and asking their permission to come into the community; (c) recognizing the value of the kimches (Mapuche sages) in the communities, since they are repositories of the ancestral knowledge that has been handed down from generation to generation; (d) learning about the methods used by the Mapuche family and community to teach educational values by speaking with the teaching agents in the family and the community as mentioned above; and (e) cooperating with families and communities to teach educational values that combine school knowledge (from the curriculum) and Mapuche knowledge (kimün). Coordinated work between kindergartens, families, and the community will improve the transmission of kimün (Mapuche knowledge) from generation to generation since, if children are brought up to be proud of their origins, this knowledge will be transmitted to future generations. The above is a way of incorporating the sociocultural context in which the pupils live, benefiting both indigenous and non-indigenous children, to create an intercultural educational model for the local, regional, national, and world levels in contexts of colonization. The present work, therefore, adds to the few investigations that have been carried out on Mapuche children in Initial Education in Chile. It also helps to reveal the knowledge lodged in the social memory about the teaching of educational values in the context of the Mapuche family and community; this knowledge could be incorporated into the school context through an intercultural approach to education. The principal limitation of this study was the context delimited by the COVID-19 pandemic; the children's attendance at the kindergartens during the winter months was poor, which delayed the fieldwork by several months. Future work is therefore aimed to continue analyzing the remaining interviews and the field notes of observations in the classroom and the trawün (meeting) held for the systematization and closure of the study. We also plan to continue working to establish ideas for the teaching of educational content expressing values based on the Initial Education curriculum and the knowledge of the Mapuche families and community. This will enable us to contribute to educational models, such as kimeltuwün (educational action), to assist in the construction of intercultural education focused on educational values, to the benefit of all the children in Initial Education in the Araucan<unk>a Region. Author Contributions: K.B.C., E.R.M., G.F.V. and F.M.-T. were responsible for all tasks related to the design and development of the article, as well as data collection and analysis. All authors have read and agreed to the published version of the manuscript. --- Funding: The authors are grateful for the financial support received from the Chilean National Research and Development Agency (ANID), through the National Doctorate grant; Fondecyt Regular Project 1191956 'Family and school education: Emotional socialization in contexts of social diversity and cultural'; Fondecyt 1231178 "Sociocultural and educational ambivalence in the Mapuche context: epistemic tension between teachers with parents and students" and Fondecyt 1211213 "How and why do children read in contexts of social and cultural diversity in La Araucan<unk>a?: An approach to teaching-learning practices of reading comprehension, from various disciplines in Basic Education". --- Institutional Review Board Statement: The study was carried out in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of the Universidad Católica de Temuco, Exempt Resolution No. JI007872 on 27 June 2019. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. --- Data Availability Statement: The data presented in this study are available upon request from the corresponding author. The data is not publicly available as it is stored in a non-virtual format at this time. --- Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

There is a lack of knowledge about Mapuche educational values among teachers in kindergartens located in indigenous territories. The object of the present work was to identify educational content expressing Mapuche values, to contribute to an epistemic base for the education of boys and girls, both Mapuche and non-Mapuche, with an intercultural perspective. The methodology used was qualitative educational research, with a focus on indigenous research, following the interpretive paradigm, through a collective study of kindergartens attended by Mapuche children in the district of Saavedra. Sixty-seven people participated: children, parents, kimches (Mapuche sages), and Initial Education professionals. The preliminary results are organized into three categories of value content: (a) educational; (b) family and community education; and (c) teaching in Initial Education. It is concluded that the teaching of Mapuche educational values in early childhood should be coordinated between kindergartens, families, and the community; some ideas for teaching these values are suggested.

Background The tenacity towards achieving universal health coverage (UHC) is central to the post-2015 global agenda which commits "leaving no one behind" and is incorporated as a target in the Sustainable Development Goals (SDGs) [1]. A global framework has been developed to track progress in SDG indicators disaggregated by socioeconomic and demographic strata in order to allow assessment of the equitable distribution and financial risk protection [2]. The unbiased measurement of socio-economic status (SES) is crucial for such benefit-incidence analysis in health, population, and nutrition. Literature suggests that SES has diverse definitions and multiple ways to measure [3]. Conventionally, income is a core SES indicator and some SES measures are solely based on per capita income such as "Prasad's scale" [4]. Considering high level of unreliability [3], including the unwillingness of people to discuss about income, social scientists consider "consumption" or "expenditure" as better markers of SES than income [5]. Composite SES indices are used that usually incorporate education and occupation along with income to reflect three distinct and interrelated dimensions of class, status, and power of social hierarchy [6]. Others preferred "education" or "occupation" as proxy for SES. The problems with such proxy measures are that they divide population into unequal-sized groups making causal interpretations difficult [5]. However, all SES indices commonly used in epidemiological studies have their own strengths and weaknesses [3]. Researchers are working hard to identify suitable SES indices to measure inequality in different contexts. Several tools are now available with multiple combinations of component indices to assess SES in different contexts. For example, Kuppuswami et al. [7] has combined material possession, education, occupation, and income in his composite SES scale; Pareekh et al. [8] added caste and family type and created a new scale with a total of nine indicators; and Tiwari et al. [9] used seven profiles (housing, material possession, education, occupation, economic profile, cultivated land, and social profile) in his scale. In Gour's classification (2013), expenditure is combined with income, education, occupation, and living standard [10]. A similar SES scale has been proposed by Bhuiya et al. [11] for rural Bangladeshi population where social involvement, food, clothing, education, shelter, and health were incorporated in the composite SES indicator. Other indices such as Multidimensional Poverty Index (MPI) and unsatisfied basic needs (UBN), which are based on different economic theories, are capable of identifying non-income factors associated with social inequalities [12]. A newer and more objective way of measuring SES is wealth index (WI) where construction materials of dwelling houses and household assets are combined together through data reduction using statistical procedure of principal component and factor analysis (PCA & FA) methods to come up with a summary WI (usually in quintiles). Related asset information is usually extracted from household survey or census data. Results from validation study [13] showed almost the same interpretation as the SES index constructed from income, consumption, or expenditure [13]. The WI [5] is thus a composite and relative measure of households' SES [3]. The WI has been constructed from national household surveys such as Demographic Health Surveys (DHSs) in 56 countries and the National Family and Health Survey (NFHS) in India [12]. Although the method of choosing component variables is not well defined [14], experts opined that context-specific WI is a useful tool for measuring inequalities and widely used in low and middle income settings [12]. In this scooping review, we attempt to identify a range of indices used to measure SES in epidemiological studies in South Asian urban countries covering urban population. --- Methods This is a scoping review to identify different indices used to measure SES in South Asian urban health studies. --- Types of studies All epidemiological studies with use of any socioeconomic indices as an explanatory, outcome, or confounding variable were included in this scooping review. --- Population All eligible studies conducted in South Asian countries (as defined by the World Bank) [15] covering urban population were included. --- Types of interventions No specific intervention was targeted; rather, all studies including observational studies using different SES indices were considered. --- Outcome measures All health-related studies using SES indices were targeted. --- Inclusion and exclusion criteria All South Asian urban health studies, using any socioeconomic indices, and published in English language between January 2000 and June 2016 were included. Studies covering rural population in addition to urban dwellers were also considered. Any research design, without any restrictions on sample size, was allowed. Studies without the use of SES indices, conducted outside South Asian region, without urban population, published in other language (than English), beyond human health, and conducted beyond the mentioned time period (Jan 2000-June 2016) were excluded. Gray literature and unpublished works were excluded. --- Data sources and literature search We searched six electronic databases: Medline (through PubMed), the Cochrane Databases, Science Direct, the Web of Science, LILACS, and the Lancet Series to retrieve relevant articles. We used the following key search terms for population, intervention, comparison, and outcome (PICO) (Table 1). We developed a comprehensive and contextualized search strategy for each of the databases using key terms and database-specific index terms (see Additional file 1). Endnote software (version 7.0) was used for database management including duplication checking while EPPI reviewer software (version 4.6.0.1) was used for screening purposes. --- Screening process Two reviewers (KMSUR and MH) independently screened the title and the abstract of each included article based on a set of codes for inclusion and exclusion criteria. After screening titles and abstracts, full texts of included articles were screened using the same set of codes for inclusion and exclusion. Any disagreements between the two reviewers were resolved by the third reviewer (SH). --- Data extraction and analysis We developed a standard template to capture relevant aspects of the review objective. The template comprised of descriptive characteristics of the included studies such as author(s), year of publication, study design, analysis type, data source, geographic location, study theme, indices used to measure socioeconomic status, and types of population targeted (urban and/or rural). Extracted data were analyzed to address the review objectives. The obtained SES indices were categorized according to the similarity of component variables, formulation process, and their combinations (Table 2). --- Results Initial search yielded 3529 results of which 224 articles were discarded for duplication. The titles and abstracts of the remaining 3305 articles were screened applying the inclusion and exclusion criteria, and through this process, more 2924 articles were excluded, as they failed to meet the inclusion criteria, and 381 articles were selected for full-text review. Reviewing the full texts of these 381 articles, we identified 256 articles for the final analysis. A detailed description of the selection process is given in the PRISMA flow diagram (Fig. 1). The scoping review identified the urban health studies in South Asian region between January 2000 and June 2016 with the use of 25 different types of SES indices (Table 2). Detailed descriptions of these indices are provided in Additional file 2. For better understanding, we further divided these 25 SES indices in 5 major categories based on underlying approaches, ingredients, and their different combinations. Asset-based wealth index was the most frequently used SES indicator irrespective of year of publication, country of origin, or thematic area of the study. Uses of other SES indices were minimum and some with even single frequency (Table 2). Fewer number of studies using SES indices were available during the first 5 years of the study period (n = 29, 11%), number of studies using SES indices increased gradually over time, and the majority of studies included in the review were published after 2010 (n = 175, 68%). The highest number of articles with SES indices was published during year 2014 (n = 33, 13%) (Fig. 2). Table 2 shows the distribution of studies by types of SES indices used with their ingredients (or their combinations) and methods used. The largest proportion of articles used asset-based wealth index (n = 140, 54%) followed by indices based on income and expenditure (n = 80, 30%). A number of studies used wealth index combining education (n = 21, 8%), while fewer studies used indices based on occupation and education (n = 16, 6%). It was not possible to classify six articles (2%) to any of the abovementioned groups due to ambiguous description of use of indices and was classified as "indices without any description." Majority of the studies included in the review were from India (n = 145, 57%), followed by Bangladesh (n = 42, 16%) and Pakistan (n = 27, 11%) (Table 3). Fewer number of studies were from Nepal (n = 14, 5%) and Sri Lanka (n = 10, 4%), and only one was from Afghanistan. There was no published article from Maldives and Bhutan satisfying inclusion and exclusion criteria. Majority of included studies were descriptive in nature (n = 228, 89%). Fifty-five percent of included studies were from primary data while the rest 45% were secondary data analysis studies. Source of secondary data was mostly national-level household surveys such as Demographic and Health Surveys in Bangladesh (n = 15, 13%) and Nepal (n = 7, 6%) and National Family and Health Survey in India (n = 47 or 41%). Thematic area wise, majority of included studies were on "maternal, neonatal, and child health (MNCH)" (n = 98, 38%) followed by studies on "non-communicable disease (NCD)" (n = 84, 33%), "adolescent and women's health" (n = 32, 13%), "health systems" (n = 24, 9%), and studies on "TB, HIV, and other communicable disease" (n = 18, 7%). Table 4 shows the distribution of SES indices used by study design, country of origin, and thematic area of studies. Majority of reviewed studies used asset-based wealth index as SES markers irrespective of study design, country of origin, or thematic area of the study. Among the cross-sectional studies, asset-based wealth index was the most commonly used SES indicator (n = 125, 54%), followed by SES indices based on income and expenditure (n = 68, 29%), indices based on occupation and education (n = 20, 9%), and wealth index combining education (n = 15, 6%). However, among the 6 case-control studies, 4 (2/3rd) used SES indices based on income and expenditure. Other study designs also mostly used asset-based wealth index. All countries used asset-based wealth index in majority of cases except Sri Lanka where majority (58%) of the studies uses SES indices based on Thematic area-wise, MNCH-related studies used assetbased wealth index mostly (n = 67, 66%), followed by SES based on income and expenditure (n = 23, 23%) and wealth index combining education (n = 6, 6%). NCD-related studies used asset-based wealth index mostly (n = 37, 42%), Fig. 1 Scoping review flow diagram using PRISMA followed by income and expenditure (n = 33, 37%), education and occupation (n = 11, 12%), and wealth index combing education (n = 8, 9%). Adolescent health studies mostly used asset-based wealth index (n = 20, 63%) and income and expenditure (n = 9, 28%) as measure of SES. Health systems-related studies also used asset-based wealth index (n = 11, 46%) mostly followed by indices based on income and expenditure (n = 6, 25%) as SES marker. Income and expenditure was the most frequently used SES indicator in TB and other communicable disease-related studies (n = 9, 47%) followed by asset-based wealth index (n = 7, 37%). --- Discussion This scoping review is an attempt to explore the types and patterns of SES indices used in epidemiological studies conducted among South Asian urban population. The review revealed 25 different types of SES indices which can be categorized into 5 major groups. Asset-based wealth index was the mostly used SES indices in South Asian urban health studies. Uses of other SES indices were less frequent. Asset-based wealth index has been debated as the component variables are artificially constructed [14], and the method is criticized as arbitrary due to poorly defined concept of choosing variables. At the same time, its discriminating power depends on the nature and relationship of the included variables [14] which may differ in different contexts. Almost all studies considered in this scoping review were quantitative in nature and followed cross-sectional research design mostly. There is paucity of published literature on SES measurement using data from longitudinal studies, randomized controlled trials, and qualitative and mixed-method studies. Hence, there is need for more studies of these types using SES markers. Although asset-based wealth index [5] was the most frequently used measure in describing the socioeconomic status of the target population, a number of studies used asset-based wealth index after contextualizing the study theme and study setup. Almost half of the studies were secondary analysis where different national, international, and regional survey data were used. NFHS of India [16][17][18] and Bangladesh Demographic and Health Survey (BDHS) [19,20] were important among the national survey Equity and health systems 24 9 Adolescent and women health 32 13 Maternal, neonatal, and child health 98 38 Non-communicable diseases 84 33 data. Indian studies used a large variety of indices where the researchers contextualized the indicator of SES with the highest frequency of asset-based wealth index use. Income-related indicators like income itself or BG Prasad's classification [4,21] based on income (modified several times) were as prominent as asset-based wealth index. At the same time, we observe different income category in different countries probably because of divergent currency values in India [22], Pakistan [23], Sri Lanka [24], Nepal [25], and Bangladesh [26]. Standard of living index (SLI) is calculated by adding scores on material possession following the theory of summing values of measurable quantity [27]. Total index scores ranging from 0 to 14 was considered as a low SLI, a score between 15 and 24 as a medium SLI, and a score between 25 and 67 as a high SLI [28]. Many authors used asset-based wealth index in individual context, mostly in primary studies. These contextualized indices resembles Pareekh [8], Tiwari [9], or Gour's [10] classification. SES is estimated in Kuppuswami classification considering indicators like material possessions, highest education, highest occupation, and type of house [7,29]. The modified Kuppuswami classification is based on occupation, education, and income which were modified in 2007 [30,31]. Type of schooling was one of the important indices used by Indian authors in several studies [32][33][34] where school fees, medium of education, and type of school (public or private) were factored in the composite SES indicator. Most of the Bangladeshi studies used asset-based wealth index as the measures of SES; income was the second most frequently used indicator. Pakistani studies mostly used asset-based wealth index and "income and expenditure" based marker as SES indicator. Nepali studies used asset-based wealth index most frequently, and income was the second most used indicator. Sri Lankan studies mostly used income [24,35] as measures of SES. Asset-based wealth index [36] and education-based indices had been applied as well. We identified only one article conducted in Afghanistan which used asset-based wealth index as measure of SES [37]. However, we did not find any study from Maldives and Bhutan in this review. Independent of the development of new indices [38][39][40][41], most of the studies reviewed used traditional indicators such as asset-based wealth index [5] and indices based on income and expenditure. Further explorations are needed whether these common indices are capable of capturing the urban inequality properly as many urban inhabitants are transitory particularly in growing urban slums. Majority of the studies measured health outcomes of targeted population and focused prevalence for specific diseases in population groups. Further in depth review can be considered to explore whether the researchers could satisfactorily fulfill their purpose of measuring socioeconomic status by the indices which they used. It would be useful to undertake a content analysis of the methodological and policy papers. One of the main limitations of this review is that our analysis was limited to published scientific articles only and excluded the gray literature. This review also prioritized peer-reviewed published articles and did not include policy and institutional reports. This review provides an idea of mapping of the available indices but do not give any clarification regarding the validity and acceptability of the indices in different context, especially for the urban poor. This review provides a detailed description of different indices used in the South Asian region. Though each of the indices has its own acceptability and limitations [27], it has been observed that some authors tried to use contextualized indices based on the population. We have found asset-based wealth index as the most frequently used index, but its acceptance is debated in the literature [3,14,27]. In urban setting, PCA-based approaches to designate SES is challenging due to difficulty in identifying and allocating assets [27]. Though some authors tried to use a combination of different indices, to overcome these debates, we propose further reviews of these indices against the backdrop of ever-changing nature of material wealth situation in South Asian countries. --- Conclusion This scoping review aimed to identify the indices used to measure inequalities in health-related studies among South Asian urban population and found asset-based wealth index as the most used index. At the same time, other indices were identified which have been used in this region in different context. This review provides a distribution of all the available indices of socioeconomic status measurement. Further attempts should be made to explore the suitability of available indices to measure the socioeconomic status for the rapidly growing urban population with a transitory nature of relative wealth scenario of this region. --- Availability of data and materials Additional files are added. --- Additional files Authors' contributions SI, IA, KMSUR, SH, and MH conceptualized the review in consultation with the co-reviewers. KMSUR wrote the first draft of this manuscript with substantial inputs from all authors. SH, KMSUR and MH contributed to the literature search. Screening, collection, and analysis of data for all the included interventions was conducted by KMSUR and MH with close consultation from SH, SS, AR, MK, FH, SI, and IA. All authors provided input and reviewed and finalized the paper before dissemination. The corresponding author is the guarantor of this review. All authors read and approved the final manuscript. --- Authors' information --- Not applicable Ethics approval and consent to participate Ethical approval has been taken from the Institutional Review Board of Bangabandhu Sheikh Mujib Medical University (BSMMU). No additional formal ethical assessment and informed consent were required. --- Consent for publication --- Not applicable --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Universal health coverage (UHC) is a key area in post-2015 global agenda which has been incorporated as target for achieving health-related Sustainable Development Goals (SDGs). A global framework has been developed to monitor SDG indicators disaggregated by socioeconomic and demographic markers. This review identifies the indices used to measure socio-economic status (SES) in South Asian urban health studies. Methods: Two reviewers searched six databases including Cochran Library, Medline, LILACS, Web of Science, Science Direct, and Lancet journals independently. All South Asian health studies covering urban population, with any research-designs, written in English language, and published between January 2000 and June 2016 were included. Two reviewers independently screened and assessed for selection of eligible articles for inclusion. Any conflict between the reviewers was resolved by a third reviewer.We retrieved 3529 studies through initial search. Through screening and applying inclusion and exclusion criteria, this review finally included 256 articles for full-text review. A total of 25 different SES indices were identified. SES indices were further categorized into 5 major groups, e.g., (1) asset-based wealth index, (2) wealth index combining education, (3) indices based on income and expenditure, (4) indices based on education and occupation, and (5) "indices without description." The largest proportion of studies, irrespective of country of origin, thematic area, and study design, used asset-based wealth index (n = 142, 54%) as inequality markers followed by the index based on income and expenditure (n = 80, 30%). Sri Lankan studies used income-and expenditure-based indices more than asset-based wealth index. Majority of the reviewed studies were on "maternal, neonatal, and child health" (n = 98, 38%) or on "non-communicable diseases" (n = 84, 33%). Reviewed studies were mostly from India (n = 145, 57%), Bangladesh (n = 42, 16%), and Pakistan (n = 27, 11%). Among the reviewed articles, 55% (n = 140) used primary data while the rest 45% studies used secondary data. Conclusion: This scoping review identifies asset-based wealth index as the most frequently used indices for measuring socioeconomic status in South Asian urban health studies. This review also provides a clear idea about the use of other indices for the measurement SES in the region.

Introduction It is a commonplace that 'data' represents the future of not just humankind, but of all life and things on earth: this is the discourse of Big Data. From this standpoint, 'datafication' -the process whereby life-processes must be converted into streams of data inputs for computer-based processing -is deemed a natural stage of development, even it has already been deconstructed within a critique of the ideology of 'dataism' (Van Dijck, 2014). But the momentum of datafication still goes on growing through the general Internet of Things and developments in specific sectors (e.g. the Internet of Medical Things: Topol, 2015: 1). Meanwhile, debate is growing about the uses of data with ever less attention being given to whether data collection, in particular the collection from persons of data relevant to them as persons and the collection from things of data of potential relevance to persons (we encompass both in the term 'personal data'), itself raises problems. This article starts from the premise that the very collection of personal data continues to raise ethical issues to which we must pay attention (Nissenbaum, 2017). This divergence between two lines of data critique (use versus collection) matters, because it is unclear at present whether emerging regulation of personal data uses, in particular through the EU's General Data Protection Regulation (GDPR), will be sufficient to protect people against the harms caused by continuous data collection. The GDPR is based in fundamental human rights concerns, particularly the right to the 'free development of the personality' (Article 2 (1) of the German Constitution). This right protects the value and dignity of the person based on free self-determination, requiring the individual's capability to control the data and information produced about her/himself (Rouvroy and Poullet, 2009: 49-51). Certainly, the GDPR represents the most radical challenge so far to datafication and is already being incorporated into business practice in the EU and North America, and its significance has been widely acknowledged by Facebook and many other parties in the worldwide debate following the Cambridge Analytica revelations. 1 But the GDPR's preventive force is limited in cases where datafication serves the 'public interest', where consent has been given to data collection or where data processing is necessary for performing a contract to which a data subject is party (Articles 8, 9, 6(1)(b)). But in increasingly many areas of life (health insurance, employment contracts, education contracts with parents) data collection is incorporated as a basic requirement, leaving unposed the question of whether data should be collected. Indeed, it is now claimed in some quarters that data collection is so pervasive that it should be exempted from legal regulation, with regulation focussing only on data use (a position Nissenbaum (2017) calls 'Big Data Exceptionalism'). We argue in this article against this position, however, and address the potentially negative implications for individual autonomy of naturalising personal data collection. Our attempt to deconstruct discourse about data practices begins with this recognition that the act of data collection today operates on a much more general level, sometimes in arenas outside the remit of the GDPR. It is thus not enough to rely only on legal interventions such as the GDPR, and vital to challenge the general discourse that makes personal data collection seem natural. In foregrounding discursive deconstruction as a tool, this article recalls Van Dijck and Nieborg's (2009) article nearly a decade ago for this journal on the discourses of 'Web 2.0' whose growing mainstream status, they argued, 'urgently begs for deconstruction' (Van Dijck and Nieborg, 2009: 855). We are not of course the first to offer a wider critique of regimes of automated data extraction (see for example Andrejevic, 2013;Cohen, 2017;Fuchs, 2013;Gandy, 1993;Zuboff, 2015); less usual, however, has been to emphasise the 'denaturalisation' of datafication discourse as a tool to open up debate. As is well-established in many fields, static and dynamic aspects of the world become closed off from normative or epistemological debate when we treat them as if they were 'natural', part of 'nature'. There are important variants of this principle, whether in anthropology (Douglas, 1986), or STS (Bowker and Star, 1999;Espeland and Sauder, 2007;Porter, 1996), but the core idea remains the same. Also, naturalisation is more than reification (Bewes, 2002;Honneth, 2008). In naturalisation, a process is not (or is not only or necessarily) referred to as, or assumed to be, a 'thing'; it is referred to as, or assumed to be, a part of nature, that is, part of the totality of things, objects and processes that constitute 'nature'. Nature in this discursive sense is not necessarily an aspect of physical (for example, astronomical, biological, or zoological) nature, but rather refers to the totality that is always already 'there' in our lifeworld, providing the basis of all specific normative or epistemological questions. Nature is what can be assumed to be 'given' so that arguments about anything in particular can start. As such, nature is distinctively immune from critique; only by a process of de-naturalisation that unmasks how particular things or processes were first constructed as natural, can they be opened up once more to critique. How could personal data collection have come to seem neutral, even natural and necessary? For this, we need to look critically at European but also North American and business discourse about data. As we will see, such discourse (whether general or specific to sectors such as health and education) barely discusses the fact of data collection at all, making any ethical challenge difficult. By deconstructing dominant discourses about datafication, we seek to hold open a space for future normative debate about one of the most important practical shifts in contemporary everyday life: the routinisation of surveillance and the collection of personal data. In the conclusion, we then turn briefly to the normative implications of not naturalising personal data collection and instead treating it as potentially problematic from the start. We conducted between January 2016 and March 2017 an analysis of general discourses about data and specific (health and education) discourses, materials that, together, illustrate a range of perspectives on the collection and processing of data. The first dataset comprised two types of source: 20 authoritative publications and reports from the world's leading economic institutions and consultancy firms, including Deloitte, McKinsey, OECD, PricewaterhouseCoopers, United Nations, World Bank, and World Economic Forum ('WEF' hereafter), alongside (in addition to the GDPR) European and US government reports on data issues (France, Holland, the European Union, plus two US White House reports). The second dataset comprised selections from the public discourse (mission statement, annual reports, marketing language, blog posts) of 19 major corporations and research institutions in the health and education industries, including 23andMe, Fitbit, Google DeepMind Health, IBM Watson Health, PatientsLikeMe, and Wellcome Trust (health), and Blackboard, IBM Watson Education, Impero Education, and Pearson Education (education). 2 The rationale for this selection was as follows. Global consultancies, IGOs and INGOs, in addition to developing blueprints for global business and persuading stakeholders of the value of personal data, also play a role in rationalising the often contradictory perspectives about data for the general public, and so play a key role in stabilising practice about data. Government reports (from both Europe and the USA) provide examples of potentially critical policy discourse on data, reflecting the different traditions of thinking about data in Europe and the USA (for example, as a component of fundamental freedom versus an individual market asset), but also the influence of global business discourse. In the health and education sectors, the majority of digital platforms' approaches are grounded in that market-based understanding of data and privacy. We, however, give special attention to corporations in those sectors, because of their important, if contrasting, practical applications of data norms. By teasing out the similarities and differences across the general and specific discourses, we hope to offer richer insights into a wider process: the naturalisation of personal data collection. --- The naturalisation of data itself The first, and most fundamental, move in naturalising datafication is to state that today's 'world of near-ubiquitous data collection' (White House, 2014a: 4) is natural. This is achieved through the common metaphor of data as a raw material with value: personal data will be the new "oil" -a valuable resource of the 21st century... becoming a new type of raw material that's on par with capital and labour (WEF, 2011: 5, 7). Through this move, data's basis in a prior process of data collection is obscured. Sometimes, however, the ownership of this 'natural' asset is asserted to be blurred, and so not referable to personal ownership:... in contrast to the concept of ownership of physical goods, where the owner typically has exclusive rights and control over the good... this is not the case for intangibles such as data... The digital divide isn't about who owns data -it's about who can put that data to work (OECD, 2015: 195-197). At times this makes an interesting counterpoint to a more European-driven thinking that rejects the ownership of data and information for a different reason, in part because recognising such ownership:... makes it the responsibility of the individual to manage and protect their data, reinforces individualism and ignores the power relationship between consumers and businesses (CNNum, 2014: 30). A blurring of data ownership is also achieved metaphorically through the common idea that data are'merely' the 'data exhaust' exuded by people's lives, and so not ownable by anyone, until, that is, they are appropriated by corporations for value (UN, 2012: 9). On this view, the raw material of data needs a further set of processes to establish its value, merging data into larger pools of data assets. This aggregative logic is understood as applying to corporations in all sectors, but in the health sector (OECD, 2015: 336) in particular, the possibility of amassing data on a large scale generates, potentially, an obligation on individuals to submit to the monitoring that supports this. Very little is said here about the rights (or wrongs) of collecting this data from individuals. The term 'hyperconnectivity' 3 creates a reified explanation of these processes. The term 'hyper' implies moving inexorably to a higher level of organisation. The further effects of hyperconnectivity become then already determined: 'the exponential growth of mobile devices, big data, and social media are all drivers of this process of hyperconnectivity. Consequently, we are beginning to see fundamental transformations in society' (WEF, 2012b: xi). Concerns (such as privacy) are occasionally noted (WEF, 2012b: xi), but not discussed in detail; much more common is to expand on the benefits of 'the connectivity and functionality made possible by converged next-generation networks' (WEF, 2012b: 47). On this view, large-scale data collection facilitates a data-driven social transformation without reference to people at all. What then of the infrastructures built upon processes of datafication? Naturalising the infrastructure for data use The key features and infrastructures of data use are further naturalised by association with certain large-scale outcomes that seem to offer unquestionable benefits. The first step to naturalising the various infrastructures for using data is the claim that, like raw materials such as water and oil, data has no value unless used: 'data have no intrinsic value; their value depends on the context of their use' (OECD, 2015: 197;italics added). Although the very idea of 'raw data' is fundamentally problematic (Gitelman, 2013), the notion that data can be refined (UN, 2012: 13;Weigend, 2017) seems to authorise the notion that there is something prior which is 'raw' (at least unrefined). The features that make even unrefined data very different from a 'raw', let alone 'natural', substance are thereby completely obscured (Alaimo and Kallinikos, 2017). The unobjectionable notion that data, to be useful, must be put to use, when combined with other principles, can generate the much more contentious claim that only data use, not data collection, has problematic consequences: 'Policy attention should focus more on the actual uses of big data and less on its collection and analysis... it is the use of data that is the locus where consequences are produced' (White House, 2014b: xii-xiii). This view suits well the common idea that data, like technology, is in itself neutral (WEF, 2013: 3). But we would argue the availability of data is not natural (only constructed as such). Nor are data 'technologies' neutral, for they are already deep applications of underlying computing technologies whose 'neutrality' (or otherwise) depends precisely on how technologies of data processing are applied to aggregate data according to specific designs (boyd and Crawford, 2013;Gitelman, 2013). Talking primarily about data use covers this over. The next key move towards naturalisation involves thinking about all forms of data collection and usage together as something with important scaled-up effects: that is, an 'ecology'. This builds on underlying metaphors which naturalise the status of data as such, but also the idea that using data means connecting it up with other data in ever larger datasets: 'Data... is not consumed when used; it can be reused to generate value. Data grows ever more connected and valuable with use' (WEF, 2012a: 7). The power of this metaphor is enhanced by assuming that the ecological feedback loops are virtuous, not vicious, whether in the area of innovation or politics: 'All factors interact and co-evolve within an ICT ecosystem... a virtuous circle starts where improvements in one area affect and drive improvements in other areas' (WEF, 2012b: 6). Other It is easy to be swept away by this rhetoric and reach the conclusion that any obstruction to the flow of data in space and time is ipso facto bad: 'Because future, yet-to-bediscovered uses of data cannot be fully anticipated, a default policy of deleting data in all contexts can be harmful' (WEF, 2013: 12). The 'price' of using free internet services (data collection) can then be presented as not a cost at all, but as simply a 'fuel' necessary for broader benefits. Ignoring underlying issues with data collection, it is argued that policy should focus on optimising the free flow and use of data in this new ecological totality:'shift the collective mind-set about patient data to "share, with protections," rather than "protect"... data sharing could be made the default' (McKinsey, 2013: 13). One European government report offers a challenge to this perspective, urging that individuals get back 'full control over the data concerning their online activities and over the implications of the use of this data' (CNNum, 2014: 9), by, for instance, introducing expiry dates for consent given for data retrieval and the use of specific data over time. But this reservation is not reflected in the mainstream of global business discourse which is likely to have much wider influence. --- Macro-arguments for datafication Larger models for thinking about datafication and its long-term benefits move even further from any consideration of data collection. The most obvious macro-benefit is economic benefit, that is, the production of profits and growth: 'big data can play a significant economic role to the benefit not only of private commerce but also of national economies and their citizens' (McKinsey, 2011: 1-2). A second macro-benefit is improved knowledge and potentially the enhancement of life itself through improved human understanding (WEF, 2011: 5). Here, 'pro-social' arguments, on the benefits of'sharing... communicating and hyperconnectivity' (WEF, 2012b: 118), further naturalise both the general inevitability (as goal) and the necessity (as means) of automated data collection and processing. The claimed outcome is individual and social empowerment, but there is little reference to the potential negative implications for empowerment of data collection that datafication requires: some of the most profound insights are coming from understanding how individuals themselves are creating, sharing and using personal data... The impact of this "empowered individual" is just beginning to be felt (WEF, 2011: 7). online communities not only provide a place for members to support each other, but also contain knowledge that can be mined for public health research, monitoring, and other health-related activities (OECD, 2015: 351). These forms of 'empowerment' then allow for a further macro-benefit: a datafied 'common good', to which we return below. A 2014 White House Report is rare in at least seeing some light and shade in the debate, yet it still too slides quickly past data collection and onto issues of use: These collections of data are benign, in the sense that they are necessary for products and services that consumers will knowingly demand. Their challenges to privacy arise both from the fact that their analog sensors necessarily collect more information than is minimally necessary for their function... and also because their data practically cry out for secondary uses ranging from innovative new products to marketing bonanzas to criminal exploits (White House, 2014b: 16). Occasionally in the literature there surfaces the crucial point that datafication matters not only for social benefits, but also for power and government, that is, by producing 'actionable' intelligence (Amoore, 2013): 'today's analytics must go beyond data input and output and maintain relevance to the real world... transforming the data into actionable information' (WEF, 2012b: 90; italics added). But this is where the long-term social implications of datafication -as a large-scale process of organising the world through continuous surveillance -come into view. --- New data subjects, new privacy At this point, two deeper forms of naturalising argument are needed: one that rethinks the individual subject in general and the other (next subsection) that rethinks the structure of a datafied social world. Given the potentially negative implications for datafication for the subject's autonomy, it is unsurprising that contemporary discourse on data seeks to reshape our understanding of the subject itself. Reconceptualising the individual as newly 'connected' to multiple information and data gathering systems, involves not just empowerment, but also data-related mutual responsibilities: Individually, we are all limited in what we can know, but together hyperconnectivity makes it possible to overcome those individual limitations and mine different types of data to find insights (WEF, 2012b: 102). In so far as these'shared' rights and responsibilities depend on all parties' use of underlying infrastructures for data collection and sharing, the implications of such infrastructures for personal autonomy are hidden from discussion, since they are what must already be in place if the 'new deal' on data is to proceed. This potential shift in how the human subject is understood is reflected, indirectly, in a US and European discussions of changes in the nature of privacy: the physical sanctity of the home's papers and effects is rapidly becoming an empty legal vessel. The home is also the central locus of Brandeis' "right to be left alone." This right is also increasingly fragile... --- Achieving the 'Data Commons' There are few more resonant terms to capture the collective freedoms associated with the internet than 'the commons' (Lessig, 1999). The notion of the 'data commons' seems to capture a state of social well-being achieved through datafication that, according to some writers (Rifkin, 2013), makes historical notions of individual privacy obsolescent. But insofar as discussion of this 'data commons' operates without reference to the possible negative implications of data collection for individual autonomy, its use serves only to naturalise further such processes. The 'data commons' debate has been advanced particularly by the WEF. It is a domain 'in which... information [about health, education, and financial services collected by mobile devices and online platforms] benefits society as a whole' (WEF, 2012c: 3-4; compare UN, 2012: 17). However, when the costs and benefits of a data commons are set out more fully, the costs of data collection are nowhere mentioned: the fear that digital medical data will be used by employers or insurance companies to discriminate against individuals is a serious and valid concern. However, this needs to be balanced with the value that the data creates for individuals in terms of better treatment, the value for society in terms of better research and cures, and the value for governments and other healthcare providers in terms of reduced costs (WEF, 2012a: 20; italics added). Similar ideas, without the use of the term 'data commons', are also reflected in such concepts as 'information commons' (CNNum, 2014: 31) or 'data ecosystem' (UN, 2012: 17). A later WEF report, however, goes further, arguing that restrictions on data use must be limited if the data commons is to grow (WEF, 2013: 8). From this perspective, the basic trade-off of the internet's development -free services in return for data collection -appears a good deal to gain a larger benefit (the data commons). The OECD makes this trade-off explicit: Promoting open data and data commons:... Advocates for greater openness and transparency link the availability of government data and information to more socially inclusive service delivery; to participatory democracy; and to economic stimulation from the development of new products and services (OECD, 2015: 359; emphasis in original). Yet none of this acknowledges the possible shifts in the purposes of data use that the data commons involves, let alone the underlying social costs of data collection, and the possible rights of data subjects to object to changes of use. The notion of 'data commons' remains in any case metaphorical: it is not proposed that data actually be owned incommon by everyone, only that it be gathered together under wider corporate control. --- Some counter-arguments against unfettered data collection There are rare places in the general reports analysed where the costs of data collection are acknowledged, even if not developed fully: Integrating Internet connectivity into devices and things opens up new risks that information will be unintentionally put into the hands of people who should not have access to it... The outcome for our hyperconnected world might not necessarily be Big Brother, but it might not be far off either (WEF, 2012b: 50, 55). More broadly, a more critical White House report notes how big data analytics may have 'an immediate effect on a person's surrounding environment or decisions being made about his or her life' (White House, 2014a: 5, linking to resources of the US Constitution (both First and Fourth Amendment)) which points in a very different direction from the 'data commons': Flowing from this protection of physical spaces and tangible assets [by The Fourth Amendment] is a broader sense of respect for security and dignity that is indispensable both to personal well-being and to the functioning of democratic society... "Privacy"... addresses a range of concerns reflecting different types of intrusion into a person's sense of self (White House, 2014a: 11; italics added). A broader critique, however, is made in the European reports. For instance, the French Digital Council calls for a review of the emerging economic and social landscape, rejecting the market-based proposal (Lanier, 2014;Lessig, 1999) of tradable individual property rights in 'own' data: --- Domains of natural datafication: the Health and Education sectors So far, we have established how a general hegemonic discourse, with few exceptions, covers over the fact of collecting personal data and displaces attention to specific uses, indeed supposed benefits, of personal data. If we are right about the influence of this discourse, we would expect to see similar moves repeated in specific sectors, such as health and education, and this is indeed what we find for North America and the UK. We plan to analyse the distinctive features of these sectors more fully in separate articles, but there are reasons for highlighting these sectors briefly here: education, because it is where we expect mature human subjects to be formed, and so might expect the GDPR's concern with the right of self-development of the personality to be most vigorously protected; and health, because health data is given special treatment in the GDPR as a sector where public interest can be expected to override individual rights (see especially Article 9(2)(h), (i) and (j)). Discussion of the collection of personal data remains, however, limited in crucial ways, as we note below, developing concerns in specialist critical literature. First of all, both health and education sectors' discourse about big data follows the pattern already seen of shifting the focus from the problems related to data collection towards those associated with data use: With new data becoming [naturally] available, innovators have taken the opportunity to build applications that make it easier to share and analyze information. As discussed later in this paper, these advances are starting to improve healthcare quality and reduce costs (McKinsey, 2013: 5). The proliferation of technology has created more data and at the same time has made it more accessible; educators just need the tools to put it to work to shape more personalized learning (IBM Watson Education). Individuals' involvement in the health and education systems is presented here as naturally generating vast quantities of meaningful data that can be garnered continuously. Consequently, expanded forms of data collection are ready to be treated as authorised in both health and education cases, legitimating commercial use of the collected data. The argument goes that the automated mining and processing of data (as natural resource) will enhance the service and care offered to individual data subjects. In a recent study, for instance, Lupton (2016a) provided a useful account of how different modes of'selftracking' (more or less voluntary) are advocated as a means to achieve better health. Examples from our analysis, in line with those findings, include: data'support medical professionals as they make decisions' (IBM Watson Health), or help teachers'respond in real-time to each individual's performance and activity on the system' (Knewton, Adaptive Learning White Paper). Yet, no limit is discussed on the type and amount of data that can be collected, as long as the data is anonymised, de-identified, or otherwise 'does not reveal your identity' (according to clothing and fitness apparel retailer Under Armour). These discourses help us imagine a new environment, in which the logic of data-sharing and unrestricted data flow become the starting-point for advancing health and education domains and, potentially, society as a whole. Elsewhere, health platforms talk more broadly of 'deliver[ing] a complete, connected, and fun experience that's 24/7' (Fitbit; italics added). This deep expansion of data generation and data collection is presented as a natural unfolding, fulfilling the dream of hyperconnectivity. Meanwhile, any discussion of the potential costs of datafication, for example privacy and human autonomy, is sidestepped. Let us now see how this general discourse plays out under the specific circumstances of each domain. --- The health sector The use of data in the health and health-related sectors has been relatively well discussed (for example, Lupton, 2016aLupton,, 2016b;;Neff, 2013;Neff and Nafus, 2016). With the historic sensitivity regarding health data's confidentiality, health data actors tend to rely on consensual data sharing aimed at social benefits on a large scale (but note Lupton, 2016: 7-10 on 'pushed self-tracking' and 'communal self-tracking'). For that reason, alongside data companies' assurances of protecting data through anonymisation and deidentification are claims throughout the health sector that data sharing is a means to make common life better. Combining two prominent cases: 'When people share their [health and medical] experiences, they help each other live better' (PatientsLikeMe), and so'make a real difference to people's lives across the world' (Google DeepMind Health). At the same time, encouragements to share data address individuals in terms of their assumed affinity with other individuals doing, or being, the same: 'A male in his 40's will see that there are 4.5 MILLION other people in the database today -that are JUST LIKE YOU' (IBM Newsroom). Such discourse arguably takes the most radical form when genetic data companies like 23andMe seek to mobilise the general public to rethink their individual rights, by drawing on the human race's genetic commonality as the anchor for a new kind of 'community': 'Coming together as a global genetic family can only help all of us understand how we are genetically related to the world around us. (Remember we are 99.9 percent genetically the same!)' (23andMe). Just as we saw the general discourse of a 'data commons' supplementing the idea of new data subjects, here we find a vision of a newly empowered and connected human community mobilised to encourage a cumulative practice of data sharing that extends into more general aspects of life and, arguably, produces a more nuanced picture of how health care is enacted and experienced today (Ruckenstein and Schüll, 2017). In such vision, sometimes data and the infrastructures installed to support its collection and use generate new forms of value and reorder relationships between the agents involved in the practices of health care, by blurring the previously existing distinction between different types of medical data and clinical practices (Hogle, 2016). But underneath this discourse are signs of a more fundamental reorganisation of health care itself, with datafication becoming part of a wider neoliberal repositioning of the individual in relation to health care (Clarke et al., 2003): people acquire an ethical responsibility to contribute to the common good by caring for themselves, including, quite possibly, by submitting to continuous health tracking. Rights to privacy and freedom, where they restrict information flow to specific contexts, should, some argue, be overridden, since 'privacy has hindered the effective development of new treatments and shared understanding of how to manage disease' (PatientsLikeMe). This move however, in the context of a health sector which treats patient confidentiality as a basic principle, puts heavy reliance on the idea that personal health data can be securely anonymised. That goes against the grain of recent work in health and genetics law which casts doubt on the sustainability of anonymisation of individual data (Evans, 2016: 5;Kaplan, 2015). To the extent that the anonymisation of health data cannot be trusted, we are required to return to this article's more basic question: should personal health data be collected and under what conditions should it be made available for wider use? --- The education sector In the education sector, the principle of confidentiality is less heavily ingrained, and the logic of automated data collection can operate with fewer initial constraints. This is not to suggest that the education domain entirely ignores privacy concerns or individual confidentiality (see for instance Pearson, 2014b: 55). But it matters that datafication in the education sector has a background in pedagogic monitoring and individual measurement (compare school league tables that from the start exempted some level of individualised data collection from ethical scrutiny). Few roadblocks exist in education to intensified surveillance through datafication, which 'allow[s] individual students to be monitored and tracked through their production of digital data' (Williamson, 2016: 56). The result, potentially, is to transform the very nature and goal of education, and reconfigure ideas of 'good schooling' (Breiter and Jarke, 2016) and good relationships between teachers and learners (Selwyn, 2014). In the UK and USA, education actors that implement 'adaptive learning' to optimise learning and learning environments through 'the measurement, collection, analysis and reporting of data about learners and their contexts' (Siemens and Gasevic, 2012: 1) are acquiring a new centrality. Adaptive learning is claimed to be 'absolutely data-driven' (Knewton) and, as such, inevitably relies on the transformation of all elements of the educational process into (ideally continuous) data flows about every student. At the same time, adaptive learning is characterised as a superior alternative to the educational model of the pre-datafication era -now described as an irresponsible 'factory-model', in which all children had to receive the same content in the same manner (IBM Education), impeding 'a deliberate and continuous approach to the improvement of learning and teaching' (Pearson). As in the health case, we find a clear application of neoliberal logic: individuals appear to have an ethical responsibility to submit to adaptive learning and so to data collection. The result is a deep normalisation of datafication and an emerging education environment of dataveillance, to which schools, students and parents are expected to commit in the name of a 'better education'. In a recent report, Pearson Education employed the metaphor of 'digital ocean' to capture such environment, which can record every single 'fleeting experience' of individuals (Pearson, 2014a). Not only does this analogy give corporations the right to appropriate data in this 'digital ocean', but it also implies that they must do so: otherwise, it is implied, important educational experiences will be lost forever. Furthermore, in education discourse, the costs of continuous surveillance are not even acknowledged as a risk in their own right. Unlike in the health case, personalised surveillance is presented as inherent to the educational process, indeed necessary for its further personalisation! It is not that privacy is exactly forgotten; rather, personalised data collection and storage are not presented as surveillance at all, but the expansion of a new notion of 'digital citizenship': Real-time monitoring is not about policing kids. Rather, it's about providing opportunities for mentorship, teaching and learning... This allows students to be responsible, safe and good digital citizens -both in school and out in the world (Impero Education, 2016). The links between continuous surveillance and the education of children are so deeply naturalised here that the panopticon of teachers monitoring pupils in real-time appears neither chilling nor threatening. On the contrary, such possibilities are emphasised as a selling-point, giving teachers 'a full bird's eye view of the entire classroom' (Impero Education), which enables them to 'immediately intervene a highly personalized way' (Blackboard). Going further, data collection is imagined to 'create a virtuous circle of real-time data that solves issues relating to student leavers lacking necessary skills' (IBM, 2016), and 'allow[s]' children to become good digital citizens. What is the consequence of this failure even to pose the question of whether personal educational data should continuously be collected, stored and used? It is to naturalise something rather shocking: how the free space where students develop and grow as adults and emerge into responsible, educated citizens -a space always in modernity assumed to be integral to what education is (Dewey, 1938) -has installed within it today an apparatus of continuous surveillance and behaviour modulation (Cohen, 2017). In the brave new world

As World Economic Forum's definition of personal data as 'the new "oil" -a valuable resource of the 21st century' (2011: 5) shows, large-scale data processing is increasingly considered the defining feature of contemporary economy and society. Commercial and governmental discourse on data frequently argues its benefits, and so legitimates its continuous and large-scale extraction and processing as the starting-point for developments in specific industries, and potentially as the basis for societies as a whole. Against the background of the GDPR, this article unravels how general discourse on data covers over the social practices enabling collection of data, through the analysis of highprofile business reports and case studies of health and education sectors. We show how conceptualisation of data as having a natural basis in the everyday world protects data collection from ethical questioning while endorsing the use and free flow of data within corporate control, at the expense of its potentially negative impacts on personal autonomy and human freedom.

of children are so deeply naturalised here that the panopticon of teachers monitoring pupils in real-time appears neither chilling nor threatening. On the contrary, such possibilities are emphasised as a selling-point, giving teachers 'a full bird's eye view of the entire classroom' (Impero Education), which enables them to 'immediately intervene a highly personalized way' (Blackboard). Going further, data collection is imagined to 'create a virtuous circle of real-time data that solves issues relating to student leavers lacking necessary skills' (IBM, 2016), and 'allow[s]' children to become good digital citizens. What is the consequence of this failure even to pose the question of whether personal educational data should continuously be collected, stored and used? It is to naturalise something rather shocking: how the free space where students develop and grow as adults and emerge into responsible, educated citizens -a space always in modernity assumed to be integral to what education is (Dewey, 1938) -has installed within it today an apparatus of continuous surveillance and behaviour modulation (Cohen, 2017). In the brave new world of datafied education, surveillance becomes a paradoxical condition of educational freedom: the idea [of digital monitoring] is to allow students the online freedom they need to grow, learn and survive in a digital world, with the safety net of keyword monitoring to protect against the risks (Impero Education, 2016). The main drive behind this combination of adaptive learning and continuous dataveillance is encapsulated in Pearson Education's mission: 'to help people make measurable progress in their lives through learning' (Pearson; italics added). The issue here is not so much measurement itself as the implicit rethinking of the young human subject of education as an entity that needs to be continuously tracked and measured in order to become 'free'. This paradoxical notion of freedom, and how freedom can be learned, has major implications for the teacher's role. As Selwyn notes (2014: 52), knowledge about whether a student is 'effective' or 'deviant' may now be obtained primarily through observation of data, relegating teachers to the role of moderators of system evaluations. This changes fundamentally the relationship between teachers and taught, with the former coming to see pupils less through the'mutual experience of interaction' in the classroom (Dewey, 1938) and more through the metrics of collected data. --- Conclusion: Reopening the ethics of datafication In this article we have shown how deconstructing those discourses, both general and specific, that serve to naturalise not only particular uses of personal data, but the underlying collection of data from persons reconnects contemporary debates about Big Data with the fundamental insight that 'the principal business model of the internet is based in mass surveillance' (Schneier, 2013). This is the truth that Big Data Exceptionalism tries to move past. Continuous surveillance (more precisely, for imageand text-based modalities, 'dataveillance' : Clarke, 1988;Van Dijck, 2014) is not prima facie compatible with the practices of democracy, or indeed with principles such as autonomy which are generally assumed to underlie democracy and a good life. This conflict is not therefore something which can properly be covered over, without introducing a fundamental contradiction into everyday life. The European Community's GDPR (in force from late May 2018) acknowledges this fact far more decisively than legal provision in North America, starting out from the recognition that data processing (if not necessarily data collection) affects people's 'fundamental rights and freedoms' (Recitals ( 1)-( 2)), and linking also, if only in passing, to the principle of 'human dignity' (Article 88; see Floridi, 2016). These normative values give the GDPR a different character, as a discourse, from those market-driven business discourses analysed in this article which in most cases start, unproblematically, from the 'facts' of a changed everyday practice. But, as noted earlier, the GDPR will not intervene in arrangements for data collection and processing that have been consented to or contractually agreed, and is framed only in terms of the protection of the data subject from arrangements which process their data, rather than protecting individuals from the impacts of a generalised environment of data collection (Davenport, 2014). There is good reason, therefore, to insist on posing questions about the appropriateness of data collection more generally. If 'privacy' (as traditionally protected, for example, through informed consent) is insufficient to address an environment of continuous, multidirectional data collection (Barocas and Nissenbaum, 2014), we must dig down to the ethical principles which underpin privacy as a broader norm (Hildebrandt, 2015). The beginnings of a consensus may be emerging around the value of autonomy, as the basis for challenging assumptions about the naturalness of the continuous collection of personal data. Autonomy can be understood in terms of the individual subject's space of freedom, the 'breathing room' it needs for an ethical life (Cohen, 2012: 149). Or it can be expressed in terms of the fear that continuous data collection may have a chilling effect on citizens' 'independent critical faculty' (Cohen, 2000(Cohen, : 1424)), a point also found in German legal thinking (Rouvroy and Poullet, 2009). Alternatively, a fundamental objection to data collection can be expressed in terms of a deepened notion of privacy as 'the right to reasonable control over the construction of one's identity' (Hildebrandt, 2015: 80, drawing on Agre, 2001: 7). The US constitutional lawyer Neil Richards argues that redefining 'privacy' to include 'intellectual privacy' would protect against'surveillance or interference' when a subject is in the process of 'generating ideas' (Richards, 2015: 5), for example, while reading. 'Privacy', understood merely as the right to block interference within a particular physical space, is clearly insufficient to counter continuous online tracking (White House, 2014b). 4 Whatever the details of these legal proposals, datafication's nature as a set of social relations involving countless institutions and individuals requires us to formulate any intervention in terms of a social value which recognises, as for example did the philosophy of Hegel, the deep social grounding of individual freedom (and autonomy) in mutual recognition. For Hegel, freedom was a freedom 'to be with oneself in the other'. 5 Such freedom is not simply a private good, but a value of social life, necessary for the quality of human beings' life together. How, one might ask, can one be 'with oneself' if one is being continuously surveilled by external forces, whether state or corporation? 6 If it is at all plausible that the continuous collection of data raises questions for autonomy and freedom in this sense, then it is vital that, rather than naturalising the collection of data as a fact of life, we hold open the question of whether such data collection is ethically appropriate or not. That requires deconstructing, as we have done, those discourses that perform just that naturalisation: general discourses about big data's economic value and applied discourses about the collection and use of data in health and education. Debates about the normative implications of collecting and processing personal data will no doubt continue, since they underpin new forms of social knowledge, indeed an emerging social order. They are at the heart of the worldwide debate that the Facebook/Cambridge Analytica scandal launched. Deconstruction will never by itself be --- Is it [i.e. data] an unsaleable asset, a common asset, private transferable property, or a right of use or usage? There are... issues concerning the enforcement of fundamental freedoms (CNNum, 2014: 9, 30). The French report not only problematises the idea that data is a naturally existing entity, but makes a bridge to thinking about the consequences of data collection from the perspective of 'fundamental freedoms' (CNNum, 2014: 29). A recent Dutch government report similarly claims that automated and ubiquitous collection and extraction of data undermines the'social space between the individual and others' that helps keep the individual apart from 'institutions that want to observe and direct our behaviour' (WRR, 2017: 7). Big data, it argues, must be used 'in a way that serves to protect both personal and social freedom' (WRR, 2017: 7; italics added). Both the Dutch and French Councils therefore challenge the discourse that data and its collection/use are a natural part of human life in the big data era, in ways that parallel the principles underlying the GDPR. The gulf, however, between these cases of European policy thinking and the general data discourse of world business organisations remains. Let us now explore how that general discourse that naturalises data collection is applied in two contrasting sectors, health and education, where US and UK corporations, operating with a more market-oriented notion of data, are dominant. enough, but we hope to have shown that, without attention to the task of deconstructing increasingly standardised discourses about data, we lack a basic tool in the much needed debate about datafication's consequences for social freedom. --- Funding The authors gratefully acknowledge the financial support of the Enhancing Life programme (http://enhancinglife.uchicago.edu/), funded by the John Templeton Foundation and administered by the University of Chicago, which funded this research under the project title 'The Price of Connection'.

As World Economic Forum's definition of personal data as 'the new "oil" -a valuable resource of the 21st century' (2011: 5) shows, large-scale data processing is increasingly considered the defining feature of contemporary economy and society. Commercial and governmental discourse on data frequently argues its benefits, and so legitimates its continuous and large-scale extraction and processing as the starting-point for developments in specific industries, and potentially as the basis for societies as a whole. Against the background of the GDPR, this article unravels how general discourse on data covers over the social practices enabling collection of data, through the analysis of highprofile business reports and case studies of health and education sectors. We show how conceptualisation of data as having a natural basis in the everyday world protects data collection from ethical questioning while endorsing the use and free flow of data within corporate control, at the expense of its potentially negative impacts on personal autonomy and human freedom.

Introduction Black sexual and gender minority individuals are disproportionately impacted by HIV in the United States (US) [1][2][3][4][5][6][7][8][9]. In 2017, Black men who have sex with men (BMSM) nationally accounted for more than a quarter of all new HIV diagnoses in the US, with a majority of new HIV diagnoses occurring among younger BMSM aged 13-34 [1]. Among transgender individuals, an estimated 14% of transgender women were living with HIV in 2017, with the highest prevalence estimate (i.e., 44%) being among Black transgender women (BTW) [10]. High incarceration rates have also been identified among BMSM and BTW in the US [11,12]. A landmark longitudinal study, the HIV Prevention Trials Network (HPTN) 061 study conducted in six urban US cities (i.e., District of Columbia, San Francisco, Los Angeles, Boston, Atlanta, and New York City), highlighted the disproportionate burden of incarceration among sexual and gender minority populations. Consisting of primarily cisgender BMSM, the HPTN 061 study showed an elevated lifetime history of incarceration (60%) and high incarceration incidence (35%, 95% CI = [31%, 38%]) among sexual and gender minority participants. Furthermore, BTW had a greater odds (3.26) of incarceration history compared with cisgender BMSM [12]. However, to date, there is not a comprehensive understanding of the impact of incarceration among these priority populations and there remains a dearth of information about the connection between criminal justice involvement and HIV among BMSM and BTW who sit at the intersection of multiple marginalized identities (i.e., Black race, gender identity, sexual orientation) [3,5]. Thus, in the current systematic review, we sought to expand understanding of the burden of incarceration as well as the relationship between HIV and criminal justice involvement among Black sexual and gender minority populations in the US [11,12]. Previous studies have shown that network characteristics (e.g., high sexually transmitted infection [STI] rates among sexual partners), psychosocial characteristics (e.g., mental health and substance use/misuse), and socio-structural factors (e.g., unstable housing, racism/stigma, unemployment) contribute to HIV vulnerability and poor access to HIV services for Black sexual and gender minority populations [1,[13][14][15][16][17][18][19][20][21][22][23][24][25]. What is less understood are the reasons surrounding why these characteristics may or may not be more pronounced and/or exacerbated among criminal justice-involved (CJI) BMSM and BTW. To address this gap, we focus our systematic review on quantitative studies that have examined the various intersections between criminal justice involvement, HIV, and key characteristics (e.g., network, psychosocial, socio-structural) known to be associated with HIV vulnerability and poor access to HIV services (i.e., HIV care and treatment). We use the term CJI to include those who are currently detained or under criminal justice supervision (probation or parole) or who have a history of arrest, incarceration, or supervision. We also highlight epidemiological studies focused on YBMSM given that they account for the majority of new HIV diagnoses among all men who have sex with men (MSM) and the need to fully engage this population in order to reach HIV elimination [1,14]. Pre-detention, detention, and post-release incarceration periods have been identified as important intervention opportunities for evidence-based and informed HIV interventions [15,16]. A recent systematic review found that the most common interventions within criminal justice settings consisted of HIV/STI screenings as well as educational interventions [17]. In addition, relatively few interventions were focused on MSM and TW, perhaps, because this approach may expose these populations to unwarranted attention and violence within criminal justice settings [17]. Of the 58 studies identified in this review, only one was focused specifically on CJI among BMSM [17]. The information gathered from the current systematic review can be used to inform gaps in existing research focused on CJI BMSM and BTW in the US and guide the development of future research and interventions. --- Methods The final systematic review search was completed on September 6, 2018, and included all articles up to that date. Relevant keywords and Medical Subject Headings (MeSH) terms based on a priori knowledge and manuscript goals were used to search the PubMed, PsychINFO, and SocINDEX databases. The finalized terms inputted into these databases are available in the appendix. The search yielded a total of 393 articles that were then exported into an Endnote library [18]. Seventy-four (74) duplicate articles were removed and the remaining articles were assessed for eligibility. --- Article Eligibility The systematic review was focused on quantitative epidemiological studies conducted in the US that provided data for individuals 18 and older. Thus, qualitative and non-US studies were excluded. Studies had to (1) report criminal justice involvement estimates (e.g., previous incarceration history or incidence) that were specific to BMSM and/or BTW participants and/or (2) contain CJI estimates specific to BMSM and/ or BTW within the context of at least one of the following conditions: (a) HIV, (b) STIs, (c) sexual risk behaviors, (d) mental health outcomes, (e) substance use/misuse, and (f) socio-structural experiences (including stigma/discrimination, housing status, education levels, and employment history). Articles were reviewed for inclusion or exclusion in an Endnote library. Detailed information on article identification and selection, including exclusion criteria, is available in Fig. 1 (PRISMA diagram). Each ineligible article was categorized by a single exclusion reason. An initial review of titles, abstracts, and full texts of articles was conducted by a member of the research team, and uncertainties related to eligibility were resolved with input from the primary author. In total, 47 articles were identified for inclusion. --- Data Abstraction General article characteristics (e.g., location, type of study, primary outcomes) are described in Table 1. All relevant point estimates from the eligible articles were abstracted and described in Tables 234. Within Tables 234, each article with qualifying data is presented along with article-specific estimates and organized according to variables of interest. Table 2 includes articles with HIV/STI estimates. Table 3 includes articles involving sexual behavior, substance use, and mental health. Table 4 includes all articles that assessed and provided information about socio-structural characteristics to include stigma/discrimination, education, employment, and housing. --- Quality Assessment Eligible studies were assessed for quality using a ten-item risk of bias tool developed by Hoy et al. [19] for population-based prevalence studies (range: 0, worst-10, best). The risk of bias tool included domains such as the method of sample selection, non-response bias, and reliability and validity of the study instrument. Some limitations with applicability of the scale existed due to its intended use to rate observational studies of clinical disease prevalence. The scale was modified to more appropriately assess the quality of included articles based on the senior investigator's knowledge of epidemiological studies of sexual and gender minorities and based on previous work [26]. Initially, 5 articles were scored independently to assess interrater reliability. A member of the research team generated a quality score for each article using the scale items with input from senior investigators. We organize our findings in the following order: (1) a summary of the studies included in the systematic review; (2) studies describing criminal justice involvement estimates among BMSM and BTW; (3) studies focused on the intersection of HIV and criminal justice involvement to include the HIV care continuum; (4) studies focused on criminal justice involvement and STIs; (5) studies focused on criminal justice involvement and sexual risk behaviors; (6) studies focused on criminal justice involvement and substance use/misuse; (7) studies focused on criminal justice involvement and mental health outcomes; and finally (8) studies focused on criminal justice involvement and socio-structural factors. --- Results --- Summary of Included Articles A total of 47 eligible articles published over a span of 18 years were identified for inclusion. Aggregate characteristics of the included articles are summarized in Table 1 to include studies by region, design, and primary outcomes (e.g., HIV, STIs, mental health, and substance use). Tables 2-4 containing all abstracted article data are available as part of the supplementary materials. The mean rating for article quality based on the risk of bias tool [19] was 7.04 (SD = 2.00), with scores ranging from 4 to 9. Notably, 25 published articles (53.2%) were analyzed utilizing datasets similar to at least one other article (i.e., based on the same local cohort or from the same multi-site study such as the HPTN 061 study and the National HIV Behavioral Surveillance survey). Information on usage of similar datasets is detailed in a separate column for each applicable article in Table 4. Nearly all studies were implemented in urban settings, with just one article providing specific estimates for a rural location [27]. All studies in the Midwest (n = 5) were conducted in Chicago, IL. More than a quarter (n=13) of all studies were conducted in the West Coast, primarily in Los Angeles County, CA. Criminal Justice Involvement Estimates Among BMSM and BTW All included articles described criminal justice involvement estimates for BMSM and BTW. However, the type (e.g., prior history, arrest, currently detained) and measures used varied substantially across studies. The most common measure, reported by 28 studies (60%), was prevalence of incarceration history. Of these, 19 articles reported lifetime incarceration history estimates [11,12,[28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44]. Eleven of these articles had estimates focused on BMSM of all ages (i.e., samples of all or primarily BMSM) [11, 28-31, 34-36, 40, 42, 43], five particularly among YBMSM [33,[37][38][39]44], one among BTW only [41], and two among both BMSM and BTW [12,32]. Incarceration estimates focused on BMSM had values ranging from 25.5 to 84%, while those on YBMSM ranged from 2.4 to 53.5% [32,33,[37][38][39]44]. The three estimates among BTW of all ages had relatively small sample sizes, but estimates were uniformly high in contrast to estimates among BMSM of all ages, ranging from 46.9 to 80% [12]. Seven studies (15% of total) assessed more recent estimates of incarceration prevalence ranging from the past 2 months to 2 years [22,35,[45][46][47][48][49]. Among these, two articles contained estimates specific to BTW [45,46] and one among YBMSM in particular [49]. Six articles described prevalence estimates for history of arrest [25,39,[50][51][52][53], varying from lifetime history to the past 6 or 12 months, though one had an unspecified window [50], Two analyses based on the same sample of YBMSM broadly assessed lifetime criminal justice involvement (inclusive of both arrest and incarceration) [13,54]. Incarceration frequency and duration were relatively uncommon variables of interest with only three articles providing estimates of incarceration duration [29,55,56], four on incarceration frequency [11,25,32,55], one on arrest frequency [51], and one on length of time since release [57]. Two estimates among these were specific to BTW [32,55]. Only one analysis explored incarceration incidence, reporting an annualized incidence rate of 35% [31,38] among a sample of 1278 BMSM [11]. Twelve articles (25.5%) also contained comparative incarceration estimates by race and/or ethnicity [22,27,30,31,33,39,41,46,47,52,53,58]. Seven (14.9%) studies were conducted among detainees in a jail setting [56,[59][60][61][62][63][64], and an additional four reported on a post-incarcerated population in which all participants had been recently released from jail or prison [55,57,65,66]. Key Findings Related to the Intersection Between HIV and Criminal Justice Involvement Among BMSM and BTW Seventeen (17) studies examined the intersection between HIV and criminal justice involvement among BMSM and BTW. These articles are described in Table 2. Among these, 14 were focused primarily on BMSM [11,13,22,28,34,37,40,44,54,56,57,62,64,67], two primarily on BTW [45,65], and one on both BMSM and BTW [46]. We describe the specific studies in detail below by four categories of interest. Incarceration History Among BMSM and BTW Living with HIV Four studies reported a high prevalence of incarceration among BMSM living with HIV [28,34,44,46]. Three of these analyses reported lifetime incarceration history estimates ranging from 28 to 84% [28,34,44]. The fourth article examined the prevalence of recent incarceration among BMSM by HIV serostatus, reporting a prior 12-month incarceration prevalence of 21% among BMSM living with HIV compared to 14% among BMSM who were not living with HIV (n = 760) [46]. Only one article, by Bukowski et al. [45], provided estimates on incarceration prevalence specific to BTW living with HIV, finding that 42.3% (n = 111) of BTW with previously undiagnosed HIV were incarcerated in the past 2 years, compared to 31.5% of BTW who were not living with HIV (n = 232), aOR = 1.92 [1.15, 3.22], and 39.7% of BTW with diagnosed HIV (n =70). In summary, a high percentage of BMSM and BTW living with HIV have a high frequency of contact with the criminal justice system. --- Independent Associations Between Criminal Justice Involvement and HIV Four studies explicitly examined the independent associations between HIV and criminal justice involvement among BMSM and BTW [11,22,40,46]. Three of those analyses utilized cross-sectional data [22,40,46] and one utilized longitudinal data [11]. Anderson-Carpenter et al. [46] found that recent incarceration history was associated with HIV-positive status (aOR = 1.69, CI = [1.31, 2.16], p <unk>.001). Oster et al. [22] found that 12% of newly diagnosed BMSM living with HIV were incarcerated in the past 12 months but did not find an independent association between incarceration and HIV infection. Similarly, Magnus et al. [40] did not find an independent association between HIV infection and incarceration history among 178 BMSM. The study by Brewer et al. [11] was the only analysis to longitudinally explore the relationship between incident incarceration and HIV acquisition. Brewer et al. [11] did not find an association between incarceration (last 6 months) and HIV acquisition (adjusted hazard ratio 1.69, CI = [0.64-4.44]) among a sample of 1278 BMSM followed for 12 months after controlling for demographics and unprotected receptive anal intercourse (URAI). In summary, findings related to the independent association between HIV and criminal justice involvement are mixed, but the current evidence favors no association between incarceration and HIV. Criminal Justice Involvement and the HIV Care Continuum Components Among BMSM and BTW Five studies examined the relationship between criminal justice involvement and the HIV care continuum, with four focused on BMSM and one on BTW [37,45,54,56,67]. Koblin et al. [37] found that intention to test for HIV was negatively associated with lifetime incarceration history (aOR = 0.37, CI = [0.16, 0.89], p =.008) in a sample of 169 consisting of 91.7% YBMSM and 8.3% YBTW. Yet Mimiaga et al. [67] found that in a sample of 197 BMSM, those who had not received an HIV test in the past 2 years were less likely to report a history of incarceration. This is likely due to an increased likelihood of receiving an HIV test during incarceration, as those who had ever obtained HIV testing in jail had a lower odds of not testing for HIV in the 2 years prior to study enrollment relative to those who were never tested in jail (OR = 0.53, CI = [0.13-0.99], p =.049) [67]. The sole article focused exclusively on BTW showed that incarceration was significantly and positively associated with undiagnosed HIV and negatively associated with viral suppression in multivariable models such that BTW who were virally suppressed had significantly lower odds of incarceration compared to those who were virally unsuppressed (aOR = 0.15, CI = [0.03, 0.71], p <unk>.05). 45 Thus, criminal justice settings may serve as an important venue for the first step (i.e., HIV testing/diagnosis) of the HIV care continuum and not subsequent stages (e.g., viral suppression) among BTW. The findings from Schneider et al.'s [54] examination of the associations between criminal justice involvement and the HIV care continuum broadly aligned with those of Mimiaga et al. [67]. Such that, any history of criminal justice involvement was associated with engagement in the overall care continuum (aOR = 2.35, CI = [1.13-4.88]), as well as linkage and retention to care, adherence to anti-retroviral treatment, and viral suppression. Although longer incarceration duration events were associated with improved engagement, repeated incarceration events were associated with poorer HIV care engagement [54]. In contrast, Vagenas et al. [56] highlighted low rates of access to HIV treatment, retention in HIV care, and viral suppression among jail detainees including YBMSM. Thus, the relationship between criminal justice involvement and the HIV care continuum may be highly dependent upon the specific CJI measure assessed among BMSM. However, criminal justice settings may serve as a critical venue for HIV testing services for both BMSM and BTW. Key Findings Related to Criminal Justice Involvement and STIs Among BMSM and BTW Only 7 studies (15%) examined the relationship between criminal justice involvement and STIs among the populations of interest. Three studies were among BMSM [25,29,67], one specifically among YBMSM [13], and three on the same jail population that consisted primarily of MSM [62][63][64]. None of the studies reported specific relationships between criminal justice involvement and STIs among BTW. However, three of 7 studies specifically examined the independent association between one or more STIs and criminal justice involvement among BMSM [13,25,29]. One article found a significant relationship between incarceration history and syphilis seropositivity, and another between incarceration and self-reported lifetime history of STIs (including syphilis, chlamydia, gonorrhea, or herpes) [13,29]. Nelson et al. [25] reported a significant relationship between a recent conviction and any STI in the past 6 months (aOR = 3.97, CI = [1.58, 9.94], p <unk> 0.05). Contrary to the mixed findings between HIV and criminal justice involvement, these studies show that incarceration history is independently associated with a greater odds of STIs among BMSM with limited information particularly among BTW. Key Findings Related to Criminal Justice Involvement and Sexual Risk Behaviors Among BMSM and BTW Eleven of 47 articles (23%) assessed sexual risk behaviors [11-13, 25, 29, 36, 49, 57, 65, 66, 68]. A range of sexual risk behaviors were assessed to include the following: exchange sex, three articles [57,65,68]; jail sex, two articles [57,68]; history of condomless or unprotected sexual acts (e.g., any condomless sex and condomless insertive and/or receptive anal intercourse), six articles [12,25,29,49,57,65]; frequency of condomless or unprotected sex, four articles [36,57,66,68]; age of first intercourse, one article [11]; number of sexual partners, eight articles [12,13,25,29,36,57,65,68]; and gender identity of sexual partners (i.e., male, female, transgender), seven articles [12,25,29,36,57,66,68] with three transinclusive articles [36,57,65]. Overall, the analysis by Harawa et al. [57] was the most comprehensive assessment of sexual risk behaviors prior to entering a jail setting, but also surveyed the prevalence of sex within jail by partner gender. The analysis by Wohl et al. [68] was the most comprehensive assessment of sexual risk behaviors within a jail setting. Only one study assessed sexual behaviors among transgender women and reported high rates (<unk> 65%) of condomless sex, exchange sex, and multiple partners in the past 90 days among small CJI samples of primarily BTW [65]. None of the included studies specifically examined the direct relationship between incarceration and sexual risk behaviors among BTW. However, findings on the relationship between incarceration and sexual risk behaviors among BMSM were mixed. Two articles did not find any significant relationship between lifetime incarceration and having multiple partners or unprotected and/or condomless sex in the past 6 months among BMSM [29,36]. Nelson et al. [25] actually reported a protective effect of lifetime incarceration for condomless receptive anal sex (aOR = 0.73, CI = [0.58, 0.91], p <unk> 0.05) among BMSM. However, two articles found a relationship between more recent incarceration and unprotected sex among BMSM [29,49], and the same article by Nelson et al. [25] reported a significant relationship between recent conviction and multiple male partners in the past 6 months (aOR = 0.56, CI = [0.33, 0.93], p <unk> 0.01). Schneider et al. [13] also found a significant relationship between a history of criminal justice involvement and having a greater number of sexual partners among YBMSM (50.8% among YBMSM with a history of criminal justice involvement vs. 66.1% with no history of criminal justice involvement; p = 0.0001). Interestingly, Schneider et al. [13] found that CJI YBMSM were more likely to have CJI homophily in their confidant and sexual networks such that 59% of CJI YBMSM reported at least one sexual partner who had been in jail during the past 6 months compared to 26.5% of YBMSM without a history of criminal justice involvement (p <unk> 0.0001). Furthermore, network CJI history (i.e., confidants and sexual partners) was associated with greater encouragement of HIV/STI transmissible sexual behaviors [13]. Key Findings Related to Criminal Justice Involvement and Substance Use/Misuse Among BMSM and BTW Nine studies (19.1%) assessed substance use/misuse behaviors among study participants. These ranged from three studies (6.4%) on general alcohol use [12,56,65], one (2.1%) on binge drinking (n=1) [65], and nine (19.1%) assessing substances such as marijuana, methamphetamine, crack/cocaine, opioids, stimulants, and multiple substances [11-13, 29, 46, 56, 57, 61, 65]. Two studies (4.3%) also assessed whether participants received an alcohol/drug treatment program during custody [29,57]. One article utilized the Addiction Severity Index (n=1); the others did not address severity of substance use disorders [56]. Only one article had estimates focused primarily on BTW [65], finding high usage rates across a range of substances including crack/cocaine and binge alcohol, but in limited sample sizes of 20 or less. The remaining studies reported findings based on samples of all or nearly all BMSM and did not provide separate estimates for BTW. Overall, high levels of substance use were identified among BMSM and BTW. Several studies found an independent association between previous criminal justice involvement and substance use among BMSM with no studies examining this relationship among BTW [11-13, 29, 41]. Bland et al. [29] found that incarceration history was significantly associated with injection drug use, lifetime history of substance use, and prior substance abuse treatment in a sample of 197 BMSM, with a higher odds among those incarcerated for <unk> 90 days. Interestingly, BMSM with longer incarceration trajectories (i.e., <unk> 90 days) were more likely to use crack cocaine during sex than those with shorter incarceration histories [24]. In Brewer et al.'s [12] two papers reporting on BMSM enrolled in the HIV Prevention Trials Network (HPTN) 061 study, any alcohol or drug in the past 6 months was significantly associated with a higher odds of lifetime incarceration history, while stimulant and marijuana use was linked to a higher odds of incident incarceration [11]. Vagenas et al. [56] examined age differences between younger and older BMSM, finding no significant differences between these groups in terms of drug type and addiction severity, except for cocaine usage, which was more frequent among older BMSM. Among YBMSM, network CJI history (i.e., confidants and sexual partners) was associated with a higher likelihood of drug use [13]. Key Findings Related to Criminal Justice Involvement and Mental Health Outcomes Among BMSM and BTW Five articles (11%) assessed mental health outcomes with no reported estimates specific to BTW even though they were enrolled in the larger cohort studies [11-13, 29, 56]. The most common condition measured were depressive symptoms using the Center for Epidemiologic Studies Depression Scale (CES-D). Two of the five papers were based on subsets of participants from the larger cohort of BMSM enrolled in the HPTN 061 study, with final analytic samples of 1278 and 1521 participants [11,12]. In these papers, the prevalence of depressive symptoms among BMSM reporting any incarceration during study follow-up or in their lifetime was 27% and 46%, respectively [11,12]. One article was focused on a statewide sample of 197 BMSM and found a similarly high prevalence of depression (i.e., 46%) among those incarcerated for <unk> 90 days [29]. In short, relatively high levels of depressive symptoms among BMSM were reported. Three of the five articles examined the relationship between a history of criminal justice involvement and depression among BMSM, with only one of the studies showing a statistically significant relationship between incarceration history (<unk> 90 days) and depressive symptoms (AOR = 2.76, CI = [1.22, 6.24], p =.01) [29]. In terms of differences by age group, no significant differences between younger and older CJI BMSM were observed among included studies (50% for younger vs. 58.2% for older, p=0.54) [56]. However, interesting findings were observed for the relationship between network history of criminal justice involvement and depression, such that network stability was associated with a reduction in respondent depression [13]. In summary, the findings on the relationship between criminal justice involvement and depression among BMSM remain mixed. For other mental health outcomes, among YBMSM specifically, a higher prevalence of criminal justice involvement history among network members (i.e., confidants and sexual partners) was associated with higher levels of anxiety and distress [13]. Schneider et al. [13] concluded that network criminal justice involvement history may limit access to the types of relationships that support and maintain mental health [69] and/or create situations of instability that further exacerbate mental health problems among non-institutionalized YBMSM [70]. Key Findings Related to Criminal Justice Involvement and Other Socio-Structural Characteristics Among BMSM and BTW Only six studies (12.8%) assessed one or more of the four other socio-structural characteristics of interest within the context of criminal justice involvement (i.e., stigma/discrimination, housing, employment, and education) [11,13,56,57,65,66]. In terms of stigma/discrimination, among BMSM enrolled in the HPTN 061 study, incarceration during study follow-up was associated with high levels of perceived racism (aOR 1.82; 95% CI = 1.02-3.27) [11]. All six studies (12.8%) documented high instances of housing instability among CJI BMSM ranging from 28 to 62% of samples [11,13,56,57,65,66]. Of these studies, two included estimates focused on YBMSM [13,56]. Schneider et al. [13] found that CJI YBMSM were more likely to report unstable housing in the past 12 months compared with non-CJI YBMSM (28.9% vs 20.7%, p=0.063), though the difference was not significant. Among YBMSM, a history of criminal justice involvement was associated with housing instability. In addition, network criminal justice involvement history (i.e., confidants and sexual partners) was associated with a higher likelihood of housing instability [13]. Vagenas et al. [56] compared housing instability at baseline by age group among detained BMSM and did not find any significant differences (38.9% among younger vs. 32.7% among older, p = 0.63). Harawa et al. [57] and Li et al. [66] reported a high rate of housing instability of 62.7% in the past 12 months among post-incarcerated BMSM. Among BTW specifically, only one study reported a similarly high rate of housing instability (44%) among a small sample of 16 predominantly HIV-positive BTW with a history of criminal justice involvement [65]. Five articles (10.6%) provided estimates of educational attainment among CJI participants, including one focused on BTW [65], two on BMSM [11,57], one on YBMSM [13], and one comparative analysis including estimates for older BMSM and YBMSM [56]. All five articles across these various groups reported over 10% (range = 10.4-38.9%) of their samples not having completed a high school education [11,13,56,57,65]. The lowest estimate of 10.4% was in a sample of 285 CJI YBMSM, but all other estimates across groups (YBMSM, BMSM, and BTW), were <unk> 25%. Of these articles, Harawa et al. [57] specified that having a GED, high school diploma, or some college were categorized as having completed a high school-level education, while the other four articles did not delineate a classification method. Two articles (4.3%) examined the relationship between criminal justice involvement and education level among BMSM. The study by Brewer et al. [11] showed that BMSM with less education were more likely to be incarcerated during study follow-up (aOR=1.83, CI = [1.27, 2.62], p <unk> 0.01). Schneider et al. [13], however, found no significant difference in education between CJI and non-CJI YBMSM at baseline. No studies examined the relationship between criminal justice involvement and education level among BTW. Three studies (6.4%) provided employment estimates for CJI BMSM but none were identified specifically for BTW [11,13,57]. All were quantitative studies with city/county (i.e., Los Angeles County and Chicago) and multi-state (i.e., six HPTN sites) samples of BMSM ranging from 212 to 1278 participants. The Chicago-specific study by Schneider et al. [13] did not find any significant differences in employment between CJI and non-CJI YBMSM at baseline which may be point to high level of need regardless of incarceration status. The multi-city HPTN study showed a greater odds of incarceration during study follow-up among men who were unemployed compared with those who were employed at baseline (OR=2.22, 1.63-3.03) [35]. Thus, unemployment is a risk factor for future incarceration among BMSM. No studies examined the relationship between criminal justice involvement and employment among BTW. --- Discussion As of the date of this review, there was a paucity of research focused on the intersection of criminal justice involvement, HIV, and associated factors among BMSM and BTW. The objective of this systematic review was to provide a comprehensive overview of the intersections between criminal justice involvement, HIV, and HIV-related characteristics among BMSM and BTW. Included studies examined a wide range of criminal justice experiences, but the most common measure of criminal justice involvement was lifetime history of incarceration with a robust non-representative national sample composed primarily of BMSM enrolled in the HPTN 061 study reporting a high (i.e., 60%) lifetime history of incarceration among BMSM with a greater likelihood of incarceration history among BTW [12]. The dearth of studies up to 2018 specifically focused on CJI BTW across all of the categories of interest is quite evident. Even though several studies enrolled BTW, specific analyses were not conducted and/or reported. A limited number of trans-inclusive published studies at the date of this review showed that criminal justice involvement may be more pronounced for BTW compared with BMSM. The geographic diversity of studies was additionally limited with most studies occurring in Los Angeles, CA and Chicago, IL. Geographic limitations may partly be explained by investigator interest and presence in those locales as well as the challenges associated with building and maintaining academic/criminal justice setting research collaborations, which may be even more pronounced during the COVID-19 pandemic [71]. Consistent with national data documenting the high prevalence of HIV among CJI persons, BMSM and BTW living with HIV had a high frequency of contact with the criminal justice system [72]. Even though we observed mixed findings as it relates to the independent association between criminal justice involvement and HIV infection, the current evidence favors no association between criminal justice involvement and HIV among BMSM with the largest cohort of BMSM to date not finding an association between recent incarceration and HIV acquisition. Findings related to the independent association between criminal justice involvement and sexual risk behaviors among BMSM were mixed. Criminal justice involvement was positively associated with substance use among BMSM with a strong network influence particularly among YBMSM such that network CJI history (i.e., confidants and sexual partners) was associated with a higher likelihood of drug use. Thus, network interventions may be appropriate for YBMSM. In terms of STIs, we observed an independent relationship between criminal justice involvement and STI prevalence among BMSM, but inability to establish a temporal relationship precludes assessment of evidence for causality. Regardless, it is plausible that criminal justice settings provide an opportunity for increased STI testing services which are coupled with opt-out HIV testing within these settings [17]. Inc

Black men who have sex with men (BMSM) and Black transgender women (BTW) are impacted by dual epidemics of HIV and incarceration. We advanced understanding of the relationship between criminal justice involvement, HIV, and other key HIVrelated characteristics among these key populations in the US. We conducted a systematic review up to 2018 and 47 articles met the inclusion criteria of scientific publications involving quantitative findings of US-based HIV-related studies focused on criminal justice-involved (CJI) BMSM and BTW. Overall, there was a dearth of studies focused specifically on BTW. Criminal justice involvement was relatively high among BMSM and BTW and more pronounced among BTW. The current evidence favors no association between incarceration and HIV acquisition among BMSM with limited information about BTW. Criminal justice involvement was associated with a greater likelihood of STIs among BMSM with mixed results for sexual risk behaviors. Criminal justice settings served as an important venue for HIV testing/diagnosis for both BMSM and BTW. However, these settings were not conducive for subsequent stages of the HIV care continuum. Studies pointed to an independent association between criminal justice involvement, substance use, housing instability, and greater odds of incarceration among BMSM who were unemployed and had limited education. Future incarceration was associated with high levels of perceived racism among BMSM. Among young BMSM, high network criminal justice prevalence was also associated with sexual risk behaviors, poorer mental health outcomes, drug use, and housing instability. CJI BMSM and BTW represent a critical subpopulation to end the HIV epidemic in the US. Keywords HIV . Criminal justice-involved . Corrections . Black men who have sex with men . Black transgender women * Russell Brewer

BMSM and BTW living with HIV had a high frequency of contact with the criminal justice system [72]. Even though we observed mixed findings as it relates to the independent association between criminal justice involvement and HIV infection, the current evidence favors no association between criminal justice involvement and HIV among BMSM with the largest cohort of BMSM to date not finding an association between recent incarceration and HIV acquisition. Findings related to the independent association between criminal justice involvement and sexual risk behaviors among BMSM were mixed. Criminal justice involvement was positively associated with substance use among BMSM with a strong network influence particularly among YBMSM such that network CJI history (i.e., confidants and sexual partners) was associated with a higher likelihood of drug use. Thus, network interventions may be appropriate for YBMSM. In terms of STIs, we observed an independent relationship between criminal justice involvement and STI prevalence among BMSM, but inability to establish a temporal relationship precludes assessment of evidence for causality. Regardless, it is plausible that criminal justice settings provide an opportunity for increased STI testing services which are coupled with opt-out HIV testing within these settings [17]. Incarceration may also disrupt sexual relationships leading to an increase in casual sexual partnerships and associated increases in the likelihood of being exposed to STIs [25,73]. As of the date of this review, there remains a dearth of information about the relationship between criminal justice involvement and STIs among BTW. A limited number of studies explored the relationship between criminal justice involvement and the HIV care continuum. Criminal justice-involved settings served as an important venue for HIV testing/diagnosis for both BMSM and BTW. The relationship between criminal justice involvement and the HIV care continuum particularly among BMSM seems highly dependent upon the specific CJI measure assessed. For example, any CJI history was associated with improved HIV care outcomes across the HIV care continuum. However incident jail stays and/or more frequent incarceration events may negatively impact HIV care. Interestingly, in one study of BTW living with HIV, those who were virally suppressed had significantly lower odds of lifetime incarceration compared to those who were virally unsuppressed. This may point to the syndemic nature of health disparities production among BTW [45] and existing barriers to viral suppression within criminal justice settings for this population [74][75][76][77]. For example, a qualitative study consisting primarily of transgender women of color described a non-affirming correctional culture in which their feminine identity was not recognized and a combination of provider bias, limited provider knowledge or inexperience caring for transgender patients, and restrictive correctional policies (e.g., sex-segregated environments based on genitalia and strict rules surrounding access to transition-related medical care) limited access to needed medical care [74]. In terms of the socio-structural characteristics, only one study examined the relationship between stigma/ discrimination and criminal justice involvement showing that incarceration during study follow-up was associated with perceived racism among BMSM. This is not a surprising finding given the pervasive nature of racial inequality and systemic racism in the US [78]. African Americans are more likely to be stopped by police and feel somewhat or very unsafe during their interactions with law enforcement, more likely to experience threats or use of physical force by law enforcement, and twice as likely to die from a fatal shooting by an on-duty police officer than their White counterparts [79][80][81]. Furthermore, BMSM's positioning at the intersection of multiple identities expose them to other forms of discrimination including homophobia [5,82,83]. Few studies comprehensively assessed mental health outcomes among the CJI priority populations. Depressive symptoms were the most commonly assessed outcome, with high estimates of depressive symptoms among BMSM, but inconsistent associations with criminal justice involvement among BMSM. Network stability was identified as a protective factor in terms of mental health particularly for YBMSM. This finding also highlights the importance of network factors among YBMSM. High rates of housing instability were found among CJI BMSM and BTW, with no statistically observed differences between CJI vs. non-CJI participants. However, criminal justice history was strongly connected to housing instability particularly among YBMSM. Findings indicated that more than 10% of BMSM and BTW with CJI have not completed a high diploma or GED. However, we note that, nationally, 15% of Black men ages 25 and higher have not completed high school [84]. While we did not observe differences in education level and employment for CJI YBMSM vs. non-CJI YBMSM [13], the largest multi-site cohort of BMSM did find an independent association between criminal justice involvement and education as well as criminal justice involvement and employment, particularly as it relates to incarceration incidence [11,12]. We note that research indicates that higher levels of both education and income offer relatively less protection against criminal justice involvement for African Americans than they do for Whites [85]. --- Conclusion This systematic review provides some of the first comprehensive examinations of the burden of incarceration among Black sexual and gender minority populations as well as the intersection of criminal justice involvement, HIV, and associated factors among BMSM and BTW. Most studies were crosssectional in nature, limiting the ability to make causal inferences between criminal justice involvement, HIV, and associated factors. Our review is limited to published scientific studies up to 2018 in the three scientific databases used. Additional studies have been published after the cutoff date used for this systematic review. For this review, we did not include studies focused on HIV pre-exposure prophylaxis (PrEP). Thus, a future research area may include examining the relationship between criminal justice involvement and PrEP outcomes among BMSM and BTW. In spite of these limitations, this analysis provides a comprehensive overview of the state of the science as it relates to criminal justice involvement, HIV, and HIV-related characteristics among Black sexual and gender minority populations in the US. It highlights the need for more focused studies on BTW to include the relationship between CJI and STIs among this population as well as the importance of CJI network dynamics particularly among YBMSM. It also underscores the importance of the intersection between criminal justice involvement and HIV in any response to Ending the HIV Epidemic in the US [86,87]. work. The opinions and ideas expressed here are not necessarily those of the sponsor. Author Contributions RB, SLR, AK, KF, JAS, AH, LW, NTH contributed to manuscript creation, design, analysis, and interpretation. RB, SLR, TE, JAS, and NTH contributed to data collection. All authors significantly revised the article. RB and SLR reviewed and completed the final version for publication. All authors have read and approved the final manuscript. --- Declarations Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent Informed consent was not required for this systematic review. --- Conflict of Interest The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Black men who have sex with men (BMSM) and Black transgender women (BTW) are impacted by dual epidemics of HIV and incarceration. We advanced understanding of the relationship between criminal justice involvement, HIV, and other key HIVrelated characteristics among these key populations in the US. We conducted a systematic review up to 2018 and 47 articles met the inclusion criteria of scientific publications involving quantitative findings of US-based HIV-related studies focused on criminal justice-involved (CJI) BMSM and BTW. Overall, there was a dearth of studies focused specifically on BTW. Criminal justice involvement was relatively high among BMSM and BTW and more pronounced among BTW. The current evidence favors no association between incarceration and HIV acquisition among BMSM with limited information about BTW. Criminal justice involvement was associated with a greater likelihood of STIs among BMSM with mixed results for sexual risk behaviors. Criminal justice settings served as an important venue for HIV testing/diagnosis for both BMSM and BTW. However, these settings were not conducive for subsequent stages of the HIV care continuum. Studies pointed to an independent association between criminal justice involvement, substance use, housing instability, and greater odds of incarceration among BMSM who were unemployed and had limited education. Future incarceration was associated with high levels of perceived racism among BMSM. Among young BMSM, high network criminal justice prevalence was also associated with sexual risk behaviors, poorer mental health outcomes, drug use, and housing instability. CJI BMSM and BTW represent a critical subpopulation to end the HIV epidemic in the US. Keywords HIV . Criminal justice-involved . Corrections . Black men who have sex with men . Black transgender women * Russell Brewer

INTRODUCTION The current global economic crisis caused by the SARS-CoV-2, or COVID-19, is leading toward the possibility of adapting the economy to make it more focused on the person, aligned with the general idea of the B (or Benefit) Corporations (B Corps). The B Corps are hybrid companies defined for-profit. Still, they are socially responsible beyond corporate social responsibility (CSR), as they make profits but not at the expense of stakeholders (Romi et al., 2018). B Corps use an alternative model of business to link for-profit and not-for-profit models (Castellani et al., 2016). They have incorporated a clear societal purpose into their missions, intending to achieve a positive social impact (Perkins, 2019), as they internalize their social and environmental effects (Stubbs, 2017). The existing economic literature on the B Corps is very scarce. Studies on B Corps have so far focused on solving business organization problems (Hiller, 2013;Rawhouser et al., 2015;Wilburn and Wilburn, 2015;Stubbs, 2017;Bianchi et al., 2020), studying consumers' perceptions on the B Corps (Marquis et al., 2011;Jin, 2018), and analyzing the combination of economic, social, and ecological objectives to go beyond a CSR defined by purely economic factors (Vargas-Balaguer, 2014;Vargas-Balaguer and Caillet, 2016). However, the triple combination of economic, social, and ecological elements of the B Corps in Latin America has not been studied in detail, which constitutes our primary contribution to our work. B Corps provide a shared collective identity for internal and external validation; they are focused on societal impact rather than maximizing profits, and they attempt to legitimate this form of sustainable entrepreneurship by influencing the business community and government officials (Stubbs, 2016). As a result, there is a strong public-private collaboration given the externalities generated during the establishment and operation of this type of company. The public administration can issue B Corp's certifications in exchange for drawing up Annual reports, or firms can obtain private certificates (Benefit Impact Assessment, BIA) issued by B labs. These B labs measure firm's externalities on purpose (pc01), certification areas (pc02), stakeholder groups (pc03), and social contribution (pc04), as will be shown in our empirical model. Externalities can also be certified (and re-certified every 2 years) by applying third-party independent standards (Alcorn and Alcorn, 2012;Hiller, 2013;Castellani et al., 2016;Nicholas and Sacco, 2017). B Corps aim to be certified to attract consumers, as certified B Corps are defined by having a strong social/environmental responsibility, as they are agents of change and customers' care (Bianchi et al., 2020). Audited and certified B Corps are a third-party signal of achieving a social purpose business model innovation to help organizations to capture value above economic gains. As a result, B Corps participate in activities endowed with ethical, sustainable, or moral goals guided by five B Corp's certification paths: brand wagoner, reprioritizer, evangelist, inertial benchmarker, and reconfigurer (Moroz and Gamble, 2020). Regarding the methodology, our study is based on grounded theory and its methods of systematization and knowledge emergence (Glaser and Strauss, 1967;Strauss and Corbin, 1990;Glaser, 1992). Grounded theory, in contrast to the approach obtained by logico-deductive methods, is theory grounded in data that has been systematically collected through social research (Goulding, 2002). We have chosen grounded theory "to conceptualize what is going on in people's lives-from their perspectives-and to propose theories that can explain and predict processes" (Nathaniel et al., 2019, p. 17). As a result, our research question deals with knowing if the B Corps satisfy social demands or whether they prefer to maximize their profit in a competitive market. Therefore, the objective of this paper is to investigate the main motivations that lead consumers to buy products and services offered by the B Corps in Latin America (Marquis and Matthew, 2015;B Lab, 2017;Winkler et al., 2019) and how this relationship creates (or not) positive impacts in the community (Cao et al., 2017). --- BACKGROUND B Corps are a different type of companies, based on the common good (Groppa and Sluga, 2015), that operate under a model created in 2007 by the American NGO (non-governmental organization) B Lab, which developed a certification to be included under the company's logo and name. Established in several countries, System B is the organization in Latin America that coordinates B Corps and oversees by organizing corporates' communication and visibility strategies, as well as implementing training policies to expand these businesses. B Corps show steady growth from their creation (B Lab, 2017). We analyze B Corps for Latin America only, given its heterogeneity and diversity, because Latin American B Corps have traits linked to familyrelated issues that are distinct from other B Corps located in different continents (Table 1). B Corps have points in common with the Economy of Communion (Caravaggio, 2018). Both fight against social and/or environmental problems, trade products, and services with the common good in mind and organize critical labor and ecological practices to benefit firms (Cea-Valencia et al., 2016). The origin of the Economy of Communion lies in the implementation of Christian values in the organization (Linard, 2003) by targeting the common good to achieve humanistic management of the firm (Frémeaux and Michelson, 2017). As a result, organizations work for the common good through a lucrative economic activity (Bianchi et al., 2020) but endowed of an open-minded social to benefit society. As a result, the creation of a three-fold social, economic, and ecological benefit transforms organizations into B Corps guided by high standards of transparency and accountability (Vega-Mu<unk>oz et al., 2018). Besides, B Corps perform as public-owned firms (Sharma et al., 2018), because they focus on reducing socioeconomic and environmental distortions (Gueneau, 2015). Consequently, to be classified as a B Corp, organizations must meet high social, environmental, and transparency standards (Hsu and Chen, 2020). In turn, they must commit a shared decisionmaking procedure by considering the long-term goal of the group (Sanchis-Palacio and Campos-Climent, 2019), as B Corps guided by product newness, low competition, recent technology, and export orientation are more prone of achieving entrepreneurial growth (Carreón and Saiz-Alvarez, 2019). B Corps work for the common good to benefit society, and the firm (Groppa and Sluga, 2015), where the economic, social, and ecological benefits generated are maximized (Bonilla-Garc<unk>a and López-Suárez, 2016). Among these types of companies, the certified B Corps stand out. These accredited organizations are companies that have accepted voluntary third-party social participation and environmental audits conducted by B Lab, a non-profit company (Moroz et al., 2018) focused on suggesting ideas rooted in business opportunities and social work (Xin, 2005;De Smet et al., 2019). Consequently, B Corps must go through a certification process defined in four areas: environment, workers, communities, customers, and the business model (Castellani et al., 2016), and the organization is classified as a B Corp when it obtains a minimum of 80 points out of 200 (Zebryte and Jorquera, 2017). After certification, B Corps are encouraged for continuous improvement to achieving leading positions in their sectors. Globally, this certification process involves more than 500 national and transnational NGOs (Moroz et al., 2018). External auditors measure and evaluate NGOs' activities and impacts to analyze to what extent and how audited companies incorporate socially responsible business practices in their operations. Overcoming the certification process involves new business opportunities and better access to resources (Reiser, 2012;Rawhouser et al., 2015). Certified B Corps reach a strong business reputation. As new training and networking schemes are being designed continuously, certified B Corps improve their score every year (Moroz et al., 2018). They are in a continuous improvement process to increase job creation and the civic commitment of the firm (Masson, 2011). This organizational process creates positive externalities to benefit society, especially stakeholders. Although certified B Corps are encouraged to improve their positive impact on society and the environment endlessly, there is significant variability in how they do so (Conger et al., 2018). These hybrid organizations certify their behavior of fulfilling positive social and environmental actions, in addition to generating economic profitability to benefit their stakeholders (Abramovay et al., 2013). Directors and managers of these hybrid companies must balance the rights of their shareholders to receive dividends with the interests of their clients, external collaborators, and workers. As a result, both the B Corp's value chain and its working environment optimize, and the communities where firms operate achieve a higher standard of living (Della Mea, 2013;Ferraro et al., 2015). Consequently, companies are increasingly committed to continuous, stakeholder-driven change toward the implementation of sustained and socially responsible business practices (Delmas and Toffel, 2008;Shepherd and Patzelt, 2011). Some studies have found that consumers demonstrate a preference for companies that support social or environmental causes (Girling, 2012) and reward these companies by purchasing their products or services (Jin, 2018). As we know, B Corps are hybrid companies between organizations with social purposes and socially responsible firms that propose solutions to social or environmental problems (Masson, 2012). Among these problems to be solved, stand out the access to quality education and conscious consumption, and how to deal with the issues of garbage reduction, obesity and prison recidivism, access to credit, drinking water, energy, quality food, unemployment, ecosystems' regeneration, and the valuation of biodiversity (Abramovay et al., 2013). As a result, when social and environmental problems worsen, consumers increasingly prefer to buy products and services from responsible hybrid organizations (Bianchi et al., 2020). B Corps do not only operate with the logic of profit, as they increase both working and the living environments of their firms (Barton et al., 2018;Townsend, 2018). These corporate hybrids combine business attributes endowed with (or without) profit (Battilana and Lee, 2014;Rawhouser et al., 2015) to create a positive social or environmental impact to benefit their stakeholders (Chen and Kelly, 2015;Stubbs, 2017). Added to creating value for investors, workers, the community, and environment, innovative B Corps break traditional management paradigms and empower managers, as B Corps have the legal obligation to look upon other interests apart from their shareholders (Tapia and Zegers, 2014) by following a postpositivist perspective guided by subjective influences (Squires, 2009;Nurjannah et al., 2014). --- METHODOLOGY Many researchers work with grounded theory, as they relive the reality of the phenomenon to be studied (Strauss and Corbin, 1998;Birks and Mills, 2015) and review the direction and framework of their research in real time when both new findings and information emerge (Nübold et al., 2017). Grounded theory uses induction-related procedures that generate an explanatory theory of the phenomenon analyzed (Glaser and Strauss, 1967). In this study, we emphasize a conceptual and theoretical approach based on grounded theory (Boe and Torgersen, 2018), since concepts and data relationships are continuously produced and reexamined to be considered rigorous as scientific research (Strauss and Corbin, 1998). Linked to symbolic interactionism, grounded theory ensures to know what is happening and why in a social group (Strauss and Corbin, 1990) to formulate grounded theories with empirical analysis (Martin and Turner, 1986;Strauss and Corbin, 1990;Andrade-Rhor, 2019) on human behavior and the social world (Kendall, 1999). This theory is especially useful when analyzing different organizations and groups (Glaser, 1992) with relatively unstructured information (text data) and theoretical sampling (Hernández-Sampieri et al., 2014;Ilias et al., 2019). In this work, and given the heterogeneity and diversity of Latin America as a geographical area (López, 2016;Manzano, 2016;Dini and Stumpo, 2018;Bernasconi et al., 2019;Paolasso, 2020), we have used the B Corp Directory for Latin America to contact CEOs and managers working in B Corps. We have chosen Latin America because its diversity and heterogeneity make the conclusions obtained in this study applicable to other regions and continents of the planet. Based on grounded theory, we collected and analyzed the speeches of 147 B Corps representatives from 14 countries: Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, Guatemala, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela, before being codified (Cxxx, Product/Service, Country) by using four precodes, as identified in the B Corps: purpose (pc01), stakeholders and interest groups (pc02), social contribution, economic growth, and human development (pc03), and certification (pc04). We complemented these pre-codes with codes (axial coding, acxx) that were interpreted as emerging and relationship functions, such as "it is associated with, " "it is part of, " "it is the cause of, " "it contradicts, " "it is one, " and "it is the property of, " to shape a resulting proposition (rpxx). The results are shown and discussed in the next sections. --- RESULTS From the accounts analyzed later, B Corps combine profit (economic, social, and ecological) and social development to create a production model defined by the impact of the firm on four components (or factors): purposes, stakeholders, social contribution, and certification. Business impact is monitored by standards of transparency and accountability in management (Figure 1). --- Discourses on Purpose (pc01) The B Corps' Latin American model is moving toward the creation of a commercial and production model for the social, economic, and ecological benefit of stakeholders in contact with the firm. This relationship generates positive externalities, as shown in the following statement from a Brazilian company linked to financial services: "To mobilize capital for a positive socio-environmental impact by helping different actors in the financial market, such as banks, asset managers, and insurance companies, to incorporate the social, environmental, and government criteria in their decision-making investment process." (C123, S, BR). In general, the social and ecological purpose is mostly related to commercial firms, but the most significant economic impact is observed in the production process. This fact is found in the statement made by the following Colombian company: "The first B Corp came to Colombia in the year 2000 to revolutionize the coffee industry in the country. This new business model has caused producers to receive prices almost three times above the market, but more importantly, it helped to discover exemplary producers who are producing excellent coffee." (C027, S, CO). Along with these economic, social, and ecological purposes, the organization includes the perspective of the client to solve the service offered in the market. This goal is achieved with customers' participation ("learning community") in the design of the production process ("co-creation of solutions") and not only with the identification of business problems, as happens in the following Mexican service company: "We are a B Corp that offers consulting, advisory and training services, and designing processes to co-create solutions tailored to the challenge to be solved in a collaborative environment." (C140, S, MX). This triple combination of the economic, social, and ecological purposes in the production stage of the good or service offered to the market must be linked to human resources, as happens at a Peruvian company caring on its workers: "Our company helps our collaborators, as we respond to their social and environmental needs through a strategic alliance signed with Traperos de Ema<unk>s. An organization ruled by the Brothers of Charity." (C007, S, PE). From all of the above, the B Corps' goal is associated with the generation of social, economic, and ecological benefits, only if they are sustainable, viable, bearable, or equitable (ac01). --- Stakeholders and Interest Groups (pc02) When the economic dimension is taken into account only, B Corps' stakeholders in Latin America tend to privilege interest groups, as they are the primary beneficiaries of the productive aspect of the good (consumer) or service (user) offered in the market. However, when the ecological, social, and economic dimensions of production are combined, interest groups tend to be less powerful. This fact happens in this B Corp of Paraguay, where interest groups benefit from the ecological dimension only, and stakeholders take advantage of the social and economic dimensions of the business model: "Our company aims to offer healthy and inclusive food, to create opportunities in the world of work for people in vulnerable situations, and to solve problems related to the environmental impact generated by waste by recycling waste." (C017, P, PY). Likewise, stakeholders influence the State as explicitly as is recognized in the following statement made by another B Corp in Paraguay: "We have strengthened the mission of private, social, and government projects. When we created our company, we realized that all the projects with purpose needed computers and reliable communication services." (C063, S, PY). B Corps are a model for generating scientific and technological knowledge adjusted to the challenges and dilemmas of contemporary societies. In this case, stakeholders of this Chilean B Corp exceed the margins of being social agents to benefit the community: "The development of societies is closely linked to their ability to apply scientific knowledge to face the challenges of an increasingly complex and demanding world." (C041, S, CL). Thus, in the Latin American B Corps, stakeholders, as part of the company, are also transformed into interest groups (ac02). --- Social Contribution, Economic Growth, and Human Development (pc03) More significant economic development does not necessarily imply higher human development because the factors affecting human development do not equally impact on economic growth. Hence, increasing GDP per capita is not enough. Still, it is necessary to improve living conditions, respect for the environment, and increase social welfare (Amate and Guarnido, 2011). Therefore, the social contribution carried out by B Corps contributes to impelling economic growth and human development, as seen in the following Ecuadorian B Corp: "Our goal is to solve people's needs to generate well-being, good health, and quality nutrition, through true solutions by developing the agricultural and livestock sector to generate shared value through good practices and fair trade. We also apply a responsible and innovative approach to the generation of sustainable business and products to promote a healthy planet through productive practices with low environmental impact." (C049, P, EC). The same idea is seen in this Chilean B Corp, but now from other areas of social activity: "The company promotes gender equality within the organization, ensuring equal pay for men and women to promote gender parity through the incorporation of women at the managerial level. This B Corp maximizes the potential of each member of the organization and establishes clear regulations regarding discrimination in the workplace and sexual harassment. Statutes have protected the company's social mission since its creation. Our firm has been created, first, to do good to the world and, second, to be economically profitable, and not vice versa." (C042, S, CL). From all of the above, the social contribution of the B Corps contributes to economic growth and human development in Latin America (ac03). --- Certification (pc04) B Corps are generally interested in getting a certification in any of the three areas (economy, society, and environment) to improve their corporate image, strengthen their brand in the market, and attract new clients and partners. As an example, the following statements from two B Corps in Latin America are included, the first Chilean and the second Colombian: "Renewable/Clean Energies: Recognizes products/services that reduce GHG emissions by providing renewable or cleaner energy than fossil fuels; Workforce Development: Recognizes the provision of jobs with good quality standards and access to training for people in vulnerable situations." (C018, P/S, CL). "Companies face the challenge of remaining relevant in the new digital environment. In this process, they need to minimize their risks, optimize their opportunities, and, above all, fulfill their strategy leveraged on the appropriate technology, Choucair, through focused software testing in business. As a result of this business strategy, we have enabled and enhanced our competitive advantage in the digital transformation of firms." (C031, S, CO). Then, B Corps' certification focuses on achieving a triple balance (social, economic, and ecological) orientation to strengthen the business model of the firm (ac04). In short, the B Corp's business model in Latin America is oriented toward achieving a triple income statement (economic, social, and ecological) that is beneficial for both the organization and society (rp01). --- DISCUSSION According to our results, the B Corp model is based on social and economic factors that reduce the centrality of the market, especially when they are combined with ecological and social dimensions. Our idea complements Sharma et al. (2018), who conducted 24 interviews with B Corps' leaders to verify that B Corps generally change their practice settings while undergoing the evaluation and reevaluation processes to achieve certification. Exogenous factors, such as the firm's size and sector, and endogenous factors, such as the nature of the firm and their business strategies pursued to maximize their impact on the society, are affected when firms transform into B Corps. B Corps adapt to the market, and they fulfill an inductively derived theoretical framework based on three building blocks: affordability, interpretability, and social references. As a result, organizations change and have a positive impact on administrations (public and private), as they can make good practices happen. B Corps' components are not new, as they are globally recognized actions or efforts inserted into the cooperative model and the social and solidarity economy (Campos, 2016;Saiz-<unk>lvarez, 2016;Sanchis-Palacio and Campos-Climent, 2019). Our contribution is how they combine, articulate, and contextualize. The B Corps model renews the activity of the company and explores how these activities can have external effects that stimulate social well-being beyond the limits of the organization (Stephan et al., 2016). In this way, the creation of shared value (CSV) developed by Porter and Kramer (2011) is crystallized in the search for new capitalism that transcends CSR (Mu<unk>oz-Mart<unk>n, 2013). This fact opens the discussion on new forms of social resignification of business (Bocken et al., 2014;de Bakker et al., 2020). Based on the discourses related to the business practices carried out by B Corps, the combination of productive management skills, good business reputation, and CSR policies linked to the economic, social, and ecological benefits of the firm contributes to achieving a certificate that strengthens the corporate image of the organization. Certification related to corporate identity, rather than a document that classifies the type of production and social contribution carried out by the firm. In other words, being a certified B Corp is the recognition and integration into a learning community. Having in mind this recognition, organizations are struggling to be certified as B Corps, especially in the case of environmentalrelated certifications. This desire is especially intense in female entrepreneurs (Grimes et al., 2018) and supports the central theoretical argument of our research that responds to the efforts of participating in identity work by strengthening their sense of self-coherence and distinction through an authentication process to benefit B Corps. Unlike the efforts made by traditional companies to implement business strategies to stimulate corporate policy and management based on transparency and accountability, transparency and accountability at B Corps are transversal and associated with the production process of goods and services. In this way, B Corps are a pilot experience to advance a useful model to meet social goals demanded by civil society in a socioeconomic context defined by climate change, the regular surge of economic crises, and resource depletion. Consequently, social impact is getting increasingly important on a planet characterized by inequality and social imbalances. --- CONCLUSION AND PERSPECTIVES First, the findings provided by our research confirm, more broadly, that these firms use social-based market laws to respond to the economic, social, and environmental problems B Corps face. As a result, B Corps can develop a more inclusive and sustainable economy to benefit society. Our results are consistent with those recently identified for the Taiwan case in Huang et al. (2019). Second, B Corps go beyond the notion of CSR. While the CSR model is focused on compensating society for part of the damage generated by the organization, or for the company's desire to benefit the community where the firm locates, the B Corps model contains, as part of its operation, both the economic contribution and the social purpose. This social purpose is a fundamental part of the production structure and not a consequence of a successful company after profit and capital accumulation. Third, B Corps move away from traditional companies. Traditional firms continue focusing on maximizing their profits without taking care of market imperfections, social cohesion, and equity. We have shown in this paper that the main motivations that drive B Corps are based on the conviction that it is possible to combine the concepts of social development and economic growth. This study has expanded the scarce existing research on B Corps that are beyond organizational issues (Stubbs, 2017), consumer behavior, market convenience, and responsible consumption (Bianchi et al., 2020). B Corps are now facing an adverse scenario caused by the SARS-CoV-2 (COVID 19) health crisis. They are learning from the adaptations that firms are implementing to maintain business activity in contexts of social distancing, quarantine, and teleworking. Given its economic advantages and social impact, it is foreseeable that there will be a gradual increase in the creation of B Corps, once the economic effects of this pandemic have disappeared. As a result, these socioeconomic-related firms can be one of the main pillars of the COVID-19 post-crisis. --- DATA AVAILABILITY STATEMENT Publicly available datasets were analyzed in this study. This data can be found here: https://sistemab.org/empresas-b-americalatina/. --- AUTHOR CONTRIBUTIONS <unk>A-D, AV-M, and JS-<unk>: conceptualization. AV-M and DC: methodology design. <unk>A-D: formal analysis and writingoriginal draft preparation. DC: validation. AV-M and JS-<unk>: writing-review and editing. JS-<unk>: translate and supervision. All authors have read and agreed to the published version of the manuscript. --- Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The current global health and economic crisis caused by COVID-19 has opened the possibility to adopt the B Corp model and focus more on the person. Based on grounded theory, we have examined 147 organizations from 14 countries listed at the B Corp Directory for Latin America. Latin American B Corps have traits linked to familyrelated issues that are distinct from other B Corps located in different continents. Our main findings are that B Corps develop a more inclusive and sustainable economy to benefit society, go beyond the notion of CSR, and move away from traditional companies, as B Corps combine social development and economic growth.

in refugee settlement and integration, providing both emotional and instrumental support (Mayblin and James 2019). Previous work has highlighted the ongoing need to examine how TSOs mediate labor market transition for refugees (Lee et al. 2020), particularly as the fl ows and profi les of refugees and the policy dynamics of receiving countries are constantly evolving (Lugosi et al. 2022). Studies have also pointed to the importance of recognizing variations between refugees, considering how their unique profi les and experiences may shape their engagement with TSOs and, consequently, how TSO practices address their distinct needs (Garkisch et al. 2017). Research within the framework of street-level bureaucracy (Lipsky 2010) has focused on the intermediary role of service providers, including civil society actors and public offi cials (Siviş 2021;Üstübici 2022). However, the work of TSOs in delivering services for migrants and refugees within the UK remains under-researched (Mayblin and James 2019). Only a few studies have been conducted in the UK and they mainly focused on confl icts between welfare-to-work policy and the ethos of assistance toward refugees, issues of solidarity and the provision of legal services, and the promotion of voluntary work (Cal<unk> et al. 2022). Our study contributes to knowledge by exploring the role of TSOs in enabling refugee integration into the British labor market. More specifi cally, in response to the call by Eun Su Lee and colleagues (2020), it examines how "bottom-up" solidarity by TSOs manifests in social and organizational practices to address limitations in state-led service provision and how refugees have responded to the support off ered by these actors. In so doing, it fi lls gaps in scholarly literature on the roles and practices of TSOs as the conduits for refugee employability. Drawing on in-depth qualitative interview data from refugees and associated organizational stakeholders in the South of England, the article identifi es key practices of support provided by disparate networks, which includes civil society and private sector actors. Th e article conceptualizes the support provided by TSOs through the notion of bottom-up solidarity, which is enacted to fulfi l essential functions of labor market integration. We argue that bottom-up solidarity interventions by TSOs are tailored to refugees' needs and that their responsive and fl exible interventions help to create highly personalized pathways to labor market integration. More specifi cally, the data are used to identify four forms of TSO support practice that characterize the enactment of bottom-up solidarity facilitating refugee access to the labor market. We argue that these domains of practice seek to address the wider circumstances faced by migrants, thus removing additional stressors that interfere with their labor market transition. Th ese practices also involve personalized activities, including the creation of voluntary opportunities, which seek to overcome barriers caused by the lack of past work experience. TSOs attempt to bridge gaps in social capital through intermediation with external actors and engage in tailored, instrumental coaching interventions to improve employability. --- Contextualizing Integration: Th e Multilevel Dynamics of Migrant Integration Policies In the context of this study, integration is viewed as complex, ongoing, relational, and negotiated processes involving change among multiple actors, as opposed to one-way adaptation and simplistic category shift s from non-integrated to integrated (Klarenbeek 2021). Moreover, the focus in this article is more specifi cally on labor market integration processes associated with the capacity of refugees to access and remain engaged in decent work that supports their independence and wellbeing. Literature on integration oft en focuses on just one scale of analysis-European, national, or local-and does not explore the interactions between levels. However, studies on multilevel governance consider the relations between diff erent layers (Scholten 2013). Scholten proposed four confi gurations of relations between levels: centralist (top-down), localist (bottom-up), multilevel, and decoupled. Th e top-down relationship between the diff erent levels of government implies "a clear central codifi cation of the division of labor between levels and control mechanisms to ensure that policy implementation at the local level follows central rules and refl ects the central policy frame" (Scholten and Penninx 2016: 93). Th is approach embraces the idea of "national paradigms of migration or integration, " that is, nation-based approaches to integration (ibid.). In contrast, the bottom-up approach is a localist perspective where local government actors "formulate policies, respond to local policy agendas, and exchange knowledge and information horizontally with other local governments" (Scholten and Penninx 2016: 94). Th is approach frames integration policies in terms of specifi c local modes (Borkert and Caponio 2010;Penninx et al. 2004). Multilevel governance refers to "interaction and joint coordination of relations between the various levels of government without clear dominance of one level.... " It "is thought to be most eff ective" and "in terms of policy frames... " it "is likely to engender some convergence between policy frames at diff erent levels, produced and sustained by their mutual interaction" (Scholten and Penninx 2016: 94). Th e fourth type, decoupled, is "characterized by the absence of any meaningful policy coordination between levels" and it "can lead to policy confl icts between government levels" (Scholten and Penninx 2016: 94). National and local integration policies are seen to be diverging (J<unk>rgensen 2012;Poppelaars and Scholten 2008). Such a complex framework of migration governance policies across various levels and settings requires a "local turn" to foster integration and ensure both take-up and local implementation (Scholten and Penninx 2016;Siviş 2021). Literature has recognized the crucial role of non-governmental organizations (NGOs) and private organizations (Cal<unk> et al. 2022). Researchers have argued that a more polycentric, decentralized, and consensual approach is more responsive and eff ective than "top-down" regulatory methods (Gunningham 2009). Whilst part of the governance literature refl ects on the shift in the regulatory methods from a "command and control" approach to soft law mechanisms (Gunningham 2009) and the interactions between diff erent layers of government (Scholten and Penninx 2016), scholars have called for a "local turn" in migration and integration studies (e.g., Glick Schiller and <unk>a<unk>lar 2011). Research in this area has considered how state actors-so called "street level bureaucrats"-support refugees more generally through service provision (Hinger et al. 2016;Ulutaş 2021). Studies have also examined how these actors rationalize their actions in their selective enactment of support (Üstübici 2022). However, it is important to stress that "street level bureaucrats" refers to state actors (Lipsky 2010), while support for refugees, especially for labor market integration, relies on collaborative arrangements with actors from commercial and third sector organizations (<unk>çduygu and Diker 2017;Lugosi et al. 2022;Siviş 2021). Th ese actors enact interrelated practices and are, arguably, proxy actors of the state, but they do not have the same resources; nor are they bound to state institutional arrangements in the same way as government ones (Bagavos and Kourachanis 2022;Lugosi et al. 2022;Üstübici 2022). Th eir support service oft en goes beyond narrowly defi ned institutional roles or practices (Strokosch and Osborne 2016). Consequently, to better understand how localized support for labor market integration operates, it is necessary to examine how TSO actors enact "bottom-up" solidarity. Enacting "Bottom-up" Solidarity in Labor Market Integration for Refugees in England <unk> 73 --- Th e "Local Turn": Bottom-Up Solidarity and the Role of TSOs in Providing Services to Refugees Despite the abundance of studies on refugee access to the labor market, there have been calls for research exploring diff erent aspects of labor market integration, including cross-sector partnerships, successful forms of engagement, and collaborative eff orts between diff erent actors (Lee et al. 2020;Lugosi et al. 2022). Th e bottom-up solidarity concept adopted in this study responds to these calls, focusing on the role of the TSOs, and is consistent with the spirit of the local governance literature. Although scholars identify diffi culties in establishing a single unifying defi nition of "solidarity, " affi rming that there are diff erent forms (Bauder 2020), we argue that referring to solidarity in this context is appropriate. Hence, our categorization and the concept in governance literature are complementary but diff er in their focus. Foregrounding our interpretation of bottom-up solidarity stresses the importance of understanding the everyday practices through which it is enacted according to the specifi c needs of particular refugees. Within this setting, the bottom-up approach contributes to an understanding of practices of organizing the integration of refugees into the labor market. Solidarity captures "bonds, expression and experience in which constitutional (legal) and existential (sociological and political) dimensions predominate and are at the very least in a mutual (if not always directly causal) and dynamic relationship"; indeed such "a complex notion, with no clear-cut normative dimension or consensual conceptualization" off ers "diverse possibilities in terms of meaning, scope, expectations and implications" (Morano-Foadi 2017: 227). Importantly, when solidarity is employed in the context of integration and, more specifi cally, in labor market integration for refugees, it presents some peculiar organizational connotations as many dynamics and actors shape its processes. It is thus characterized by the capacity to tailor services, which enables TSO actors to address refugee experiences, capabilities, and needs. TSO activities potentially span a variety of roles that directly and indirectly mediate the transition of refugees into work; they facilitate interaction, the exchange of information, and mobilize resources, thus creating opportunities for refugees, which are conducive to employability (Mayblin and James 2019;Morano-Foadi et al. 2021;Perna 2019). TSOs thus act as conduits that interpret and enact state responsibility; however, their services are oft en sporadic and underfunded, and their enactment of solidarity can take diverse forms that need to be better understood (Bagavos and Kourachanis 2022;Morano-Foadi et al. 2021;Siviş 2021). Our study thus contributes to knowledge by examining how bottom-up solidarity is performed by UK TSOs. --- Th e Role of TSOs in Labor Market Integration for Refugees in England Th e UK government has devolved responsibility for integration from central or local authorities to regional or localized organizations and networks, encompassing TSOs and other non-state actors (Morano-Foadi et al. 2021). Refl ecting the "decoupling" concept adopted in multilevel settings (Scholten and Penninx 2016), coordination between national and local levels and civil society organizations has decreased (Morano-Foadi et al. 2021). Consequently, TSOs have a wider set of roles and range of responsibilities in facilitating labor market integration. In England, TSOs dealing with refugees are "oft en small, local, volunteer run organizations which oft en fi ll gaps of public services" (Cal<unk> et al. 2022: 874) and remain the only source of support (Mayblin and James 2019). Moreover, in the last decade, a noticeable reduction in the availability of tailored support for asylum seekers and refugees is more evident (Mayblin and James 2019). Funds allocated to local authorities and refugee community organizations have been cut signifi cantly (Mayblin and James 2019;Phillimore and Goodson 2010). Th e government has gradually moved to deliver services for refugees through outsourcing and multisector partnership arrangements (Mayblin and James 2019;Strategic Migration Partnership n.d.). Critics have argued that the state has increasingly functioned as an "enabler, " delegating some procedural functions to commercial service providers while also promoting the role of civil society organizations (Cal<unk> et al. 2022;Haugh and Kitson 2007). Th e state's aim was to encourage greater levels of voluntarism with the involvement of charities, private enterprises, and social enterprises in the running of public services (Cal<unk> et al. 2022: 873;Lugosi et al. 2022). TSOs were considered as being "locally embedded or better able to articulate the needs of local communities" (Haugh and Kitson 2007: 983) and to thus better understand specifi c societal needs than many state actors. TSOs provide migrants and refugees with basic services, especially at the time of their arrival in the host country (Garkisch et al. 2017). Yet, TSOs generally have limited resources; they depend on governmental or other forms of funding that are restricted in terms of timescales, objectives, and the activities they support; and they must oft en compete for funding (Garkisch et al. 2017). However, the ability of UK local authorities to support and engage in service coordination has diminished (Bales and Mayblin 2018;Darling 2016). Th e Refugee Integration and Employment Support (RIES), which provided refugees with "a personal development worker who facilitated access to key social welfare agencies such as Jobcentre Plus" (Phillimore 2012: 7), was dismantled. Th e MIF (Migration Impact Fund) was also scrapped (Th omas 2019: 13). Welfare and employability support across the whole welfare system and tailored support from employment agencies and Jobcentres diminished (Anderson 2013). In the context of these dynamics and contrasting forces, our study examines how TSOs enact bottom-up solidarity to support labor market integration for refugees. --- Methods Th is research adopted a pragmatist approach that sought to develop context-sensitive understanding of organizational processes and actionable knowledge (Kelly and Cordeiro 2020). Consequently, the study focused on TSO actors and their activities in a specifi c county in the South of England, considering the local dimensions of refugee labor market integration. Th e decision to examine this service ecosystem was driven by the potential relevance of the data to our objective to identify and assess inter and intra-organizational support mechanisms. Th is setting represented what Michael Patton (2015) referred to as a "data rich" empirical case, insofar as refugee labor support was provided by a network of TSOs working in combination with local authorities and social enterprises. Studying this context thus enabled us to analyze activities at the level of individual actors, organizations, and among networks of organizations. Moreover, we could examine individual experiences in their social and organizational context to understand the key domains of practices through which local TSOs supported refugee labor market integration. Th e project was approved by the university's ethics committee. Participants were invited directly by members of the research team; they were given information sheets and informed written and/or verbal consent was obtained. Twenty-nine people were interviewed. Six organizational participants were recruited based on their personal knowledge of and involvement in diff erent domains of refugee reception and processing. Th ey included representatives from the local authority responsible for managing the Syrian Vulnerable Person Resettlement Program for the city; two diff erent refugee charities based in the city; a local education and training provider; a local social enterprise supporting refugees; and the International Organization for Migration, which also collaborated with these actors. Twenty-three refugee participants were selected and recruited through their involvement with one of the refugee support charities and refugee networks. Th e charities disseminated information about the project among their clients, and the refugees were approached personally while at the charities, given information about the project and its ethical protocols, and invited to participate. Th e refugee cohort consisted of 13 females and 10 males, 12 of whom had come to the UK via the resettlement program, and 11 of whom were recognized refugees who had gone through the asylum process in the UK. Th eir ages ranged from 20 to 50, and their nationalities included Afghans, Algerians, Egyptians, Indonesians, Iraqis, Syrians, and Yemenis. Semi-structured interviews were conducted in English to explore experiences, perceptions, and attitudes of the refugee cohort regarding their labor market access. Diff erent interview protocols were created for the key stakeholder groups, but they all utilized an open questioning approach to elicit rich accounts of institutional, organizational, and other factors facilitating and inhibiting transition into and within the labor market. Interviews with state and TSO stakeholders explored their views regarding factors that supported or limited refugee access to work, and their activities and experiences in supporting refugees. Interviews with refugees began by exploring their backgrounds, education, and pre-migration work, before examining post-migration experiences of trying to access work, for those who had paid or unpaid jobs, and their career plans. Th e digitally recorded interviews were transcribed. Transcripts were anonymized with pseudonyms and adjustments to any data that could identify the participants. Data analysis was conducted manually and began during the fi eldwork. Th e data were analyzed thematically (Salda<unk>a 2016) over several cycles, with three researchers initially scrutinizing and coding the data independently. Th e multidisciplinary nature of the team, with backgrounds in law, social sciences, and organization studies, enabled us to identify diff erent issues at diff erent cycles of analysis. During the initial analysis cycles, the team adopted an inductive, data centric approach, using line-by-line "open" coding (Salda<unk>a 2016). However, we remained conscious that the refugees and organizational stakeholders had shared transformative ambitions, that is, the desire to facilitate labor market transition. Th is acted as a broad sensitizing concept for our initial analysis, and we focused primarily on examining a) what forms support took for diff erent organizations, andb) what types of practices individual actors engaged in. Th e analysis subsequently adopted "focused" coding (Salda<unk>a 2016) of individual experiences, where participants highlighted incidents, actors, organizations, processes, and actions that they saw as infl uential. Th rough this process, attention was placed on specifi c aspects, such as the consequences of diff erent practices, including the challenges and opportunities they created. Th e team met at interval periods to discuss similarities and diff erences in our interpretations of the data. We reviewed individual fi ndings to identify overarching subthemes refl ecting key dimensions of support activities and behaviors. At this stage, we agreed that the data pointed to the everyday enactments of support, as manifested in the social practices and experiences. Th is encompassed practices that addressed labor market entry directly and indirectly. It also considered practices that were aimed at refugees, as well as at external actors and organizations. --- Enacting bottom-up solidarity Previous studies have discussed how support is enacted in everyday practices of state and nonstate actors working with refugees and asylum seekers in relation to their wider reception and settlement (Ulutaş 2021;Üstübici 2022). However, research focusing on labor market interventions oft en point to generalized areas of activity, such as acting as informational hubs, rather than examining how support is enacted (Siviş 2021)-hence the importance of identifying how it manifests in practice and distinguishing its diff erent forms, based on evidence from frontline practitioners. More specifi cally, the data made it possible to distinguish between interventional practices that directly aimed at labor market interventions and those that were indirect forms of interventions. Th ese were key to facilitating access to work, addressing peripheral factors that oft en inhibited the refugees' ability to focus on employment. Th e data also enabled us to distinguish between interventions that were aimed specifi cally at refugees and those focusing on employers or other types of organizational stakeholders, who either employed or mediated the refugees' transition into work. Regardless of the practice, a distinct value of the solidarity enacted by TSOs was in their fl exibility. For example, charitable organizations providing support for refugees and asylum seekers could be very effi cient in off ering tailor-made services to refugees in assisting their job searches. Th eir services were adaptable and responsive to specifi c individual needs-taking into consideration varied backgrounds, gender, and language skills, as well as other factors like past traumatic experiences. While these represented individualized enactments of solidarity, it is also important to stress that the diverse actors involved had limited resources and specifi c domains of expertise, which was refl ected in the scope and focus of the support they off ered. However, their individual practices, when viewed holistically, addressed a wide range of needs, providing indirect and direct forms of intervention. --- Transforming Circumstances Transforming circumstances refl ects an explicit recognition of the multidimensionality of challenges for refugees and the corresponding need to address them holistically. Th is is advocated by contemporary researchers and practitioners who stress the need to adopt a multipronged strategy for refugee support service provision (Boenigk et al. 2021). In short, it was necessary for TSOs to overcome a range of inhibitors that compromised refugees' capacity to fi nd, obtain, and sustain employment. Bottom-up solidarity aimed at labor market integration should be viewed in the wider context of TSO support services, representing indirect interventions. Th is helps to appreciate how TSO expertise was both sensitive to and responsive to refugee needs. Moreover, it reinforces that labor market integration does not and cannot function as a standalone activity, disconnected from the wider personal and organizational challenges that refugees encounter. Many of the refugees noted the importance of TSO support in addressing basic psychological, social, and administrative hazards, which enabled them to concentrate on employment-related activities. For example, as Zara, a female refugee, stated: When I went [to the charity] for the fi rst time, it was the only hope for me. Th ey [the charity] helped me a lot during my immigration process for the fi rst time, like in documentation, when I applied for medical exemption to make a medical exemption certifi cate. Th ey helped me to fi ll a form... for the housing benefi t as well. And now they're still helping; they're Enacting "Bottom-up" Solidarity in Labor Market Integration for Refugees in England <unk> 77 trying to fi nd me a job. And the other thing, they give me other support: they organized me a teacher, a teacher for my English. Which is free. Zara was a graduate in her late twenties who had never worked in her country of origin and had been residing in the UK for approximately one year at the time. Her refl ections highlight how employability and support for labor market transition should be thought of as a bundle of practices that encompass building capacities that will eventually facilitate entry into work and career progression (Cal<unk> et al. 2022). TSO interviewees oft en discussed their labor market support in the context of their wider activities that sought to eliminate additional barriers. For example, Adele, who was part of the service delivery team at a local educational provider, observed: First of all, we start doing the bills, sorting out gas, electricity, and if the family or the people or the children had been put in the right benefi t, if not we need to change their benefi t.... And child tax credit, sometimes they send the letter: "you are not eligible for it" and this kind of stuff. All this process takes time. But anyway, bills, Internet, schools... the children will be already allocated to specifi c schools but doing their free school meals if the children need to, if they need to buy any bus pass, if they're eligible for it or not. And school uniforms as well, all these. In principle, the educational provider had a relatively narrow brief, focusing primarily on developing refugees' work-related capabilities, especially their language skills. However, as Kirsty Strokosch and Stephen Osborne (2016) have argued, TSOs regularly redefi ned their roles in practice, extending the scope of activities to encompass wider areas beyond narrow conceptions of training and education. Refugees encountered multiple challenges, which were unique to them, for example, concerning their immediate families. Family separation was an understandable source of stress, and many of the refugees without recourse to funds had to dedicate considerable time and eff ort to raising money to support legal claims and trying to negotiate the country's immigration system. Th is also had several notable consequences for their labor market transition. For example, less time could be dedicated to learning and attending English language classes, which could have improved their chances for gaining employment; and their family separation could have increased the risk of social isolation and psychological stress, inhibiting work. Bottom-up solidarity in relation to labor market integration should therefore be seen as embedded within a wider set of support practices. Th ese practices sought to remove potential risks and barriers, which then enabled the creation of pathways to employment. Importantly, the removal of inhibitors and mitigation of risks was based on the ability of TSOs to understand the specifi c needs of refugees. However, bottom-up solidarity oft en involved active facilitation practices that sought to help refugees progress along those pathways. A key set of practices focused on creating direct opportunities. --- Creating Direct Opportunities Creating direct opportunities refers to the capacities of TSOs to provide paid and/or voluntary work for their clients in their organizations to enhance their employability. De-recognition of qualifi cations, restricted network capital, lack of language skills, and limited work experience in the localized labor market are strong barriers to accessing work and career mobility for refugees (Lee et al. 2020). Recruiting refugees represents potential psychological and economic risks for organizations, which are consequently reluctant to hire them out of a fear that their integration into the workforce would require additional time and resources at the expense of organizational performance. Th is risks perpetually excluding refugees from labor market participation because of restricted opportunities to gain any relevant experience and, subsequently, to develop employability competencies. Addressing this substantial constraint was a central feature of bottom-up solidarity, which involved making direct interventions targeting refugees. In line with previous research (Cal<unk> et al. 2022;Siviş 2021), many of our participants pointed to the support of various actors and networks in facilitating their access to the labor market. However, our fi ndings also helped us understand the practices through which such support was enacted by TSO actors. Specifi cally, short, informal volunteering opportunities were leveraged to secure part-time work, which could be used to gain full-time employment and subsequently expedite transition across employment sectors (cf., Martin 2012; Tomlinson 2010). TSOs were again in a unique position to facilitate these developmental journeys, in part because they understood their clients' specifi c needs and exclusionary factors. Moreover, TSOs had the capacity to negate the associated risks and, being functioning organizational entities, their premises and operations could be mobilized as resources to support refugees. Th is was illustrated by one of our female refugees, Ola, a graduate in her early forties who had been residing in the UK for less than a year. Ola explained the constraints of seeking paid work: To be honest, I didn't try to ask for paid work, because it meant commitment and I'm looking aft er my mother at the moment, so I can't commit to part-time or full-time job. But I'm working as a volunteer with an organization... which helps and supports new Syrian families. Importantly, volunteering was viewed by the refugees and the organizations as part of longer processes that unfolded over time, through which refugees gradually built their confi dence, developed employability-related capabilities, widened their social networks, and constructed career profi les that could be appreciated by employers embedded in the locality. Th ese opportunities were also transformative insofar as they helped to build trust between diff erent actors. For example, volunteering with a charity acted as a facilitative space to expand Zeinab's employability, as she explained: I will say the fi rst [job] with [the charity] was because I volunteered, then the volunteering helped them and helped me. So the volunteering helped them to see me and see my skills and, you know, use me. And of course [the colleague] was saying-this is a good person; I'm now going to take her. Zeinab was a graduate in her thirties with a background in a specialist technical service fi eld who had been residing in the UK for approximately two years. Her observations pointed to her refl ective learning and development, but it also showed how these facilitated subsequent labor mobility. Depending on the organization and the skills of refugees, they could morph into longer term, paid positions, as Zeinab affi rmed: Th e data suggested that practices of bottom-up solidarity involved fl exibility and the mobilization of resources in direct response to the needs of their clients (see also Strokosch and Osborne 2016). Dedicated asylum and migration charities were in a unique position to perform or enact solidarity because they knew and appreciated their clients' distinct requirements. More importantly, they had the organizational resources and capabilities to create bespoke learning and developmental opportunities. Arguably, they also had a certain level of risk appetite, insofar as they were prepared to give refugees work opportunities (for example, in frontline service provision) that commercial organizations would have resisted because of the reputational or fi nancial risks presented by refugees' limited language skills. Solidarity thus involved the creation of pathways and supporting refugees to traverse them. Th e examples above point to the role of creating direct opportunities. However, bottom-up solidarity oft en involved intermediation, where TSOs sought to exert their infl uence beyond their organizations in support of refugees. --- Intermediation and Trust Brokering Intermediation and trust brokering refer to the ability of TSOs to create positive representations of refugees and to negotiate on their behalf with third parties, which were oft en central to the enactment of solidarity. Th ese forms of intervention were aimed at employers or employment agencies that facilitated access to work, rather than at refugees themselves. It typically took the form of "sponsorship, " which was "episodic and focused on creating opportunities" (Ayyala et al. 2019). For example, Zeinab explained how charities acted as intermediaries so that the experience, confi dence, and trust gained through voluntary employment at the TSO could then be leveraged to gain further work: "... they called me because some people went to [another charity], to ask for help and were given my number at [this charity]. Th ey called me to translate the form. " Th ese types of brokering utilized the resources and network capital of TSO actors in representing refugees, in eff ect mobilizing "bridging social capital" (Putnam 2000) across multiple organizations. Th e extent to which this bridging capital was operationalizable depended on the existing credibility of the recommending actor and the presence of an established relationship of trust between the organizational actors. More important was the purpose for which bridging capital was mobilized within the context of bottom-up solidarity. TSO actors sought to generate advantages for refugees, helping them to traverse social, cultural, and psychological obstacles inhibiting their transition into work. Th ese obstacles oft en stemmed from employers having preconceptions of refugees or poor understanding of the risks (and opportunities) involved in employing them. For example, Mary, a member of a local social enterprise, constantly searched for and sought to exploit opportunities to engage with diff erent employers: Well, for example, I met with the MD of a big construction company.... So I suddenly say, "have you thought of recruiting someone from the Syrian refugee population?... Have you thought of building a pathway for Syrian refugees into work? I' d like to talk to somebody who would be willing to start thinking about a pathway. " And he said to me he hadn't even given it a thought. Th ese were arguably longer-term, strategic forms of sponsorship, which sought to establish pathways for multiple refugees to obtain jobs. However, intermediation oft en took more tailored, tactical forms, insofar as TSO actors sought to protect refugee interests, as Mary explained:... there was much more liaison work with support agencies, you know, it wasn't just they've turned up here today and everything is fi ne. It was, "this problem's arisen, can you help?" And all of that. So there was more liaison, I think, than we had anticipated. We hadn't factored in the fact that their regular ESOL [English for Speakers of Other Languages] learning, if I can call it that, clashed with what we were doing and, yeah, and our tutor was just brilliant because as she did a lot of one-to-one, she did a lot of juggling of times and days. And because the pilot project ran over Easter, there were no ESOL classes over Easter, so in fact that in a 80 <unk> Sonia Morano-Foadi, Peter Lugosi, and Clara Della Croce way turned out to be a good thing, because then they could come and do the work experience. And do it that way round. Mary's refl ections highlight the challenges of intermediation. Her considerations help us to appreciate the dynamic, improvised, and adaptive character of support practices. Solidarity was constructed toward and performed according to the specifi c needs of individual refugees. Th is relied on TSO actors understanding the circumstances faced by refugees and being willing and able to respond to them in their enactment of solidarity. However, solidarity in the form of intermediation was not enacted by TSOs alone. It is also important to recognize the refugees' own agency, resources, and practices (see also Borkert and Caponio 2010;Strokosch and Osborne 2016). One of the respondents, Nabila, observed in her role as the leader of a selfinitiated "women's group" how important it was to create a supportive environment that could infuse trust and enact solidarity for refugees. She explained: When I fi rst arrived, I asked my friend [to fi nd me volunteer work], she wrote an email. I rent a room in the community council. I start inviting people. We start just for fun but then the project became bigger and bigger. And now off er a lot of things. We also do training... Th e city council gave me money for three months, through an English lady [who] helped me. She helped me to fi nd a consultant to write the constitution of the group.... In our group

Th is article examines the role that third sector organizations (TSOs) play in supporting refugees' access to the labor market in England. TSO practices are conceptualized through the notion of "bottom-up" solidarity. Data gathered through interviews with refugees and representatives from charities, social enterprises, and public authorities are used to identify how TSO actors enact bottom-up solidarity and, in turn, facilitate integration of refugees into the labor market. Th e fi ndings show how labor market transition is built on the transformation of the wider circumstances faced by refugees. Data also demonstrates how the creation of direct employment opportunities, coupled with intermediation and trust brokerage, and alongside episodic and extended coaching, is key to enacting "bottom-up" solidarity.

to respond to them in their enactment of solidarity. However, solidarity in the form of intermediation was not enacted by TSOs alone. It is also important to recognize the refugees' own agency, resources, and practices (see also Borkert and Caponio 2010;Strokosch and Osborne 2016). One of the respondents, Nabila, observed in her role as the leader of a selfinitiated "women's group" how important it was to create a supportive environment that could infuse trust and enact solidarity for refugees. She explained: When I fi rst arrived, I asked my friend [to fi nd me volunteer work], she wrote an email. I rent a room in the community council. I start inviting people. We start just for fun but then the project became bigger and bigger. And now off er a lot of things. We also do training... Th e city council gave me money for three months, through an English lady [who] helped me. She helped me to fi nd a consultant to write the constitution of the group.... In our group we have also organized a [language] class for women who have little children.... We run the class in partnership with [local educational provider] for women with children with a volunteer from [charity]. We have also started to help people to fi nd job experience, help with references and other stuff. Nabila was a graduate in her thirties who had worked as a social worker and had been residing in the UK for approximately one year. Her refl ections illustrate how solidarity networks actively assisted in promoting English language services, voluntary work, and other forms of support to respond to the immediate needs of newly arrived refugees. Solidarity was thus more than service provision carried out exclusively by TSOs; it was resource integration for a common set of goals, which operated on the principles of co-production as diverse stakeholders mobilized their resources in pursuit of mutually desired outcomes (Strokosch and Osborne 2016;Lugosi et al. 2022). Nevertheless, TSOs had key roles in these networks because of the reputational and social capital that they could mobilize. Th eir expertise and capacity to mobilize resources in support of refugees oft en took the form of targeted coaching. --- Coaching Coaching was a direct form of intervention aimed specifi cally at refugees. Following Tatiana Bachkirova and colleagues (2010), coaching is conceived here as a developmental process, involving focused interactions and the deployment of techniques and strategies to promote desirable goals, in this case concentrating on labor market access domains. In this context, coaching can be seen as a "person-centered approach, " insofar as it is "non-directive" (avoiding simple prescription) and seeking to facilitate self-determination (Joseph 2010). Stephen Joseph (2010) has argued that this type of coaching concentrates on goals and focuses on helping coachees to develop eff ective solutions to attain them. Within bottom-up solidarity, coaching can thus be viewed as "solution-focused, " underpinned by brief interactions that stress the role of coachees in self-directed learning (Cavanagh and Grant 2010). Coaching is a complex phenomenon that can assume multiple forms (Bachkirova et al., 2016), but we foreground these features (i.e., development orientation, narrow domain scope, person-centric, and solutions-focused) in relation to bottom-up solidarity for two main rea-Enacting "Bottom-up" Solidarity in Labor Market Integration for Refugees in England <unk> 81 sons: fi rst, to distinguish it from therapeutic and counselling activities that may have been provided by other actors in the same or diff erent organizations, which addressed refugee trauma and psychological problems; and second, to distinguish it from mentoring, which we view as a related but diff erent set of developmental practices involving longer and more substantial interactions aimed at refl ective transformation (De Cuyper et al. 2019). Th e refugee charities and social enterprises in this empirical context set up employment advice services and helped refugees to navigate administrative problems, for example, concerning driving licenses. Th ey oft en encompassed a range of employment-related activities as suggested by Zara, who we quoted previously, who had not worked previously in her country of origin: Th ey helped me with draft ing a CV and they also helped me to apply for a job. Because I have applied, I've submitted my CV throughout the country but I haven't received any response, but [the charity] helped me on how to apply using a specifi c form, how to fi ll a form, and how to deal with an interview, like these kind of things. Coaching was a dimension or manifestation of bottom-up solidarity that emerged as a surgical intervention in a narrow fi eld of practice for a specifi c goal. Coaching was constructed and performed around refugees' individual needs and circumstances. Th erefore, the coaching engagement of TSOs was built on sensitivity toward and understanding of the unique challenges that refugees face. For example, as Mary, the social enterprise representative, highlighted:... in practice, one of the refugees had much more caring responsibilities for his family than any of the support agencies were aware of. So he found it diffi cult to attend on the specifi c days that the program was planned in for. So he had one-to-one support and one-to-one coaching. Th at was his issue. One of the other refugees who is not in settled accommodation is moving from one place to another, so this made his attendance diffi cult. Th e third refugee, who was also known to other statutory services, as they were trying to address concerns with his family. So we arranged meetings with other organizations, with ESOL learning, and explained that they were doing things elsewhere, we were trying to juggle all of that and make them attend here [at the social enterprise]. Th e enactment of solidarity was based on TSO actors' capacity to adapt their individual and collective practices in close collaboration with their clients (Strokosch and Osborne 2016). In part, their responsiveness was driven by their client focus, and their desire to perform localized solidarity through fl exible, multi-dimensional interventions. However, it is important to stress that adaptability was also made possible because these actors operated outside of the institutional constraints faced by actors working in state-run or state organizations, who are usually subject to their governance regimes and institutional logics (cf. Hinger at al. 2016;Siviş 2021;Üstübici 2022). --- Conclusion Extant literature has identifi ed the general domains of practice through which TSOs may support labor market integration for refugees (Bagavos and Kourachanis 2022;Garkisch et al. 2017;Siviş 2021). Our research contributes to this body of knowledge by showing in detail how TSO actors develop the employability of refugees and facilitate their constructive engagement with work and employment. Our study is one of only a few conducted in the UK that identify and distinguish the forms of support enacted by TSOs in their service provision (Cal<unk> et al. 2022;Mayblin and James 2019). Specifi cally, our fi ndings show how the enactment of support could take direct and indirect forms of solidarity and focus on refugees and those who employ them or mediate their pathways to work. Furthermore, past work has suggested that, in England, these services are oft en delivered by small, local, volunteer-run organizations with limited resources, whose work compensates for the inability or unwillingness of the state to engage directly in supporting labor market integration for refugees (Morano-Foadi et al. 2021;Phillimore and McCabe 2010). Consequently, we have conceptualized the dynamics of this form of service provision through the notion of bottom-up solidarity. Th e invocation of bottom-up solidarity stresses that, in the context of fragmented state engagement, TSO actors take active responsibility in interpreting their roles and in mobilizing their resources to create bespoke interventions. However, it is important to emphasize that bottom-up solidarity refl ects a state of aff airs rather than a prescriptive model of intervention. Bottom-up forms of support may be understood as being components of a constructive strategy, as TSOs might be better in understanding and responding to refugees' distinct needs. Recent shift s in the scale and forms of global migration have raised new challenges that states and their agents have not been able to adequately comprehend and which, coupled with other economic and political pressures, have constrained their ability to develop comprehensive responses. By exploring how bottom-up solidarity manifests in practice in compensating for limitations in state-led service provisions, our study has refl ected on the potential role of TSOs as agents of change. Regardless of whether the notion of bottom-up solidarity is read as an opportunity or as a failure of state provision, the fi ndings of this study have highlighted some of its potential forms and dimensions. Moreover, it has recognized that enactments of solidarity intersect with, and directly address, needs associated with refugees' unique characteristics. Importantly, invoking the notion of bottom-up solidarity stresses the importance of several characteristics of TSO practices: fi rst, that they are underpinned by ambitions to support refugee empowerment and successful labor market transition; second, that they are built on specialist competencies, which are attuned to and enacted in relation to the distinct needs of specifi c migrants; and third, they are agile, insofar as they can be tailored to achieve distinct work-oriented goals, but they are embedded in a wider set of practices supporting refugees' settlement holistically. Based on this understanding of bottom-up solidarity, we note that further research is needed: fi rst, to better understand refugees' pathways to integration and, in particular, their engagement with the labor market beyond simply access to work; second, related to the previous point, to understand the institutions, actors, and the "touchpoints" when refugees engage with them to identify practices supporting and inhibiting labor market integration. Such mapping exercises, examining TSO practices and refugee agency and experiences, can help to identify what and how resources are mobilized by diff erent actors in their enactment of bottom-up solidarity in other organizational and national contexts. Moreover, future work can help to assess the eff ectiveness of diff erent practices, moving beyond cross-sectional analysis and subjective indicators of success to evaluate their impacts over longer periods, including their infl uences on incomes. However, such longitudinal research could continue to use subjective wellbeing or other psychological indicators to assess effi cacy. Assessing the impacts of bottom-up solidarity could also help to identify where greater levels of investment and coordination are required. Even if governments seem politically and ideologically driven to limit immigration, understanding the eff ectiveness of mechanisms helping integration activities and best practices among existing service providers could be valuable for the UK and beyond. Improved support may require the establishment of new governance structures, which can better coordinate the activities of individual actors and agencies within a holis-Enacting "Bottom-up" Solidarity in Labor Market Integration for Refugees in England <unk> 83 tic strategy. However, support may be most eff ectively delivered by valuing refugees' experiences and enabling them to realize their potential to contribute to the communities in which they settle. Focusing on improving targeted funding for specialist service provision and investing in knowledge transfer and knowledge management activities could be part of government strategy in the medium term. Specifi cally, charities and social enterprises may already have capacities to address specifi c needs, such as CV development or job matching, alongside the provision of other forms of support, including counselling as well as coaching and mentoring. Service development may, therefore, rely on more eff ective forms of knowledge management, where networks of organizational agents are better informed and where local or regional providers off er more targeted services for refugees, and bottom-up solidarity is enacted as an ecosystem of TSO actors. --- <unk> ACKNOWLEDGMENTS Th e authors wish to thank Ms. Casey Alves for her assistance during the draft ing of this article. Th e usual disclaimer applies. <unk> SONIA MORANO-FOADI (LLB and Mphil [equivalent], PhD) is Professor of Law at the Oxford Brookes University's School of Law. She is also the Director of Postgraduate Research Students. Her research focuses on the relationship between citizenship, migration, and fundamental rights in the context of an enlarging European Union. Her work is characterized by an innovative methodology based on a combination of empirical fi ndings and theoretical and doctrinal discourses on concepts of European citizenship, migration, fundamental rights, and integration. She has published on a wide range of subjects including economic migrants and refugees, asylum seekers, and victims of human traffi cking. She is on the editorial board of the Journal of Immigration, Asylum and Nationality Law. <unk> PETER LUGOSI (PhD) is Professor of Culture and Organisation at the Oxford Brookes Business School, Oxford Brookes University. His interdisciplinary research has examined migrant adaptation, labor market transition and work-related experiences, and migrant entrepreneurial practices, particularly in the hospitality, food, and creative sectors. His work has also explored linkages between migration and wider phenomena, including urban change and the evolving visitor economies of destinations. He has studied the practices of state, charitable, and commercial organizations supporting migrants and refugees in Europe, the UK, and Brazil, and has focused more recently on mentoring initiatives and the management of mentoring programs involving migrants. <unk> CLARA DELLA CROCE (LLM, PhD) is a Senior Lecturer in Law (Education) at the School of Law, School of Oriental and African Studies (SOAS), University of London and a Knowledge Exchange Fellow at the School of Law at Oxford Brookes University. Her areas of expertise are international migration law, refugee and immigration law, human rights law, European law, and development law. She worked for more than a decade at an Oxford-based charity supporting migrants, refugees, and foreign national prisoners. She also worked for the United Nations Conference on Trade and Development (UNCTAD) and for the Orga-nization of American States (OAS), spent time in Brussels at the European Commission, and previously worked in Brazil and the UK as a commercial lawyer.

Th is article examines the role that third sector organizations (TSOs) play in supporting refugees' access to the labor market in England. TSO practices are conceptualized through the notion of "bottom-up" solidarity. Data gathered through interviews with refugees and representatives from charities, social enterprises, and public authorities are used to identify how TSO actors enact bottom-up solidarity and, in turn, facilitate integration of refugees into the labor market. Th e fi ndings show how labor market transition is built on the transformation of the wider circumstances faced by refugees. Data also demonstrates how the creation of direct employment opportunities, coupled with intermediation and trust brokerage, and alongside episodic and extended coaching, is key to enacting "bottom-up" solidarity.

Introduction In India 2.4 million people are estimated to have the human immunodeficiency virus (HIV) [1] and there is a lack of reliable sexually transmitted infection (STI) incidence and prevalence data for India. However the Indian National AIDS Control Organisation (NACO) has estimated the annual STI incidence in India at 5%, equivalent to 40 million new cases each year [2]. Groups perceived to have higher risk sexual behaviour are targeted with HIV and STI prevention interventions under NACO's National AIDS Control Programme (NACP). Movement in search of work is common in India and other developing countries. The current NACP is focused on migrants, especially single male migrants, who are perceived to be at high risk of HIV and STI infection and are therefore an important bridging group for the transmission of these infections from core groups to the general population [3]. However the majority of data on male migrants are from sub-Saharan Africa [4,5,6,7,8] and there is a lack of data from within India to guide national prevention programmes. Here we present the first estimates of the prevalence and determinants of HIV infection among male migrants in India. The data are from a survey of migrant men aged 18 to 49 years conducted between 2005 and 2006 in the diamond and textile industries in the highly industrialised city of Surat in western India. We chose these industries since they are the two main industries in Surat and the majority of employees are migrant men. In 2005 the city had a high HIV sentinel surveillance prevalence (.5% among STI clinic attendees or high risk groups and.1% among antenatal clinic (ANC) attendees). It is situated in Gujarat, a medium HIV prevalence state (.5% among STI clinic attendees or high risk groups and,1% among ANC attendees). The high risk groups are comprised of female sex workers and men who have sex with men [9]. At the time of the study, local nongovernmental organisations (NGOs) carried out workplace based HIV/STI prevention interventions during which they provided information and education about HIV and STIs and their prevention and distributed condoms. --- Materials and Methods --- Study Design Multi-stage stratified probability sample survey. --- Sampling Frame In 2005, textiles were sold through 120 markets employing an estimated 130,000 men. The diamond industry employed approximately 500,000 men working in 20,000 diamond units housed in larger buildings known as complexes. We selected and mapped four diamond sub-areas and 19 textile markets which had received partial HIV intervention coverage including HIV and STI information, education and communication work and condom distribution by NGOs a year prior to the study start date. The sampling frame consisted of 226 diamond complexes and 19 textile markets. --- Sampling We randomly selected 41 diamond complexes and 11 textile markets from the sampling frame. Diamond complexes were stratified into four strata according to size. The largest stratum consisted of the single largest complex sampled with certainty. Complexes were randomly selected within the remaining three strata, to give a total of forty. Units were randomly selected with replacement from each complex with probability proportional to size. The number of workers sampled from each unit varied across strata; five for stratum one [smallest complexes], ten for stratum two, 20 for stratum three and 30 for stratum four [largest complex]. These numbers were selected taking into account the size and number of complexes from each stratum to ensure that the probability of selection for any diamond worker was roughly equal. The eight largest markets were sampled with certainty and three were randomly selected from the remaining markets with probability proportional to size. The total number of men sampled from the markets was proportional to market size amongst the eight largest markets and fixed in the smaller three sampled markets. Within each market workers were stratified by one of ten occupation groups and selected proportional to the size of the stratum. Within both industries workers were selected from workplaces by systematic sampling. Selection was without substitution if a migrant man refused to participate. If a nonmigrant man was selected, other men were systematically approached until a migrant worker was found. --- Eligibility Men aged 18 to 49 years were eligible to participate if they had not been born in Surat city. --- Sample Size A target effective sample size of 456 men was required to estimate an expected HIV-1 point prevalence of 5%, i.e. a 95% CI 3% to 7%, within 2%. We assumed a design effect of 1.45, based on an average cluster size of 10 workers and an intracluster correlation of 0.05, and a participation rate of 60%. We therefore aimed to approach 1100 men to recruit 661 participants. --- Questionnaires We developed three standardized interviewer-administered questionnaires in English and two local languages. These included a sexual behaviour and HIV/STI knowledge, attitudes and perceptions questionnaire, and clinical history and examination questionnaires. Sexual behaviour data included lifetime sexual behaviour, and three most recent female and male sexual partnerships and partner types in the past year. Clinical data included STI treatment history and symptoms and signs related to HIV and STIs. Data collected on non-responders included industry, age, place of birth, marital status including whether or not married men were living with their wives. --- Consent We obtained written informed consent from literate participants and witnessed informed verbal consent from illiterate participants in their language. --- Data Collection A unique study number was used to link questionnaires with laboratory samples. Six interviewers collected demographic and sexual behaviour data and six doctors collected clinical data. Participants were interviewed by study staff at government and private health centres out of hours. Laboratory staff collected the biological samples. Each participant was given a torch as a gift after participation. --- Biological Samples We collected blood samples to test for HIV-1, Herpes Simplex Virus types 2 (HSV-2), and Treponema pallidum (syphilis) and first catch urine samples to test for Chlamydia trachomatis, Neisseria gonorrhoea and Trichomonas vaginalis. Participants were given all the results except HIV and offered free HIV counselling and testing at local government facilities. --- Laboratory Testing We tested for HIV-1 with two methodologically different enzyme-linked immunosorbent assays (ELISAs) (J Mitra Microlisa HIV EIA (HIV subgroup O & subgroup C) and Genedia 1/2 Elisa 3.0) and used an enzyme immunoassay (EIA) for indeterminate or minimally reactive results (PBS Orgenics EIA Immunocomb II HIV 1 & 2 Bispot). We tested all serum samples for syphilis with rapid plasma reagin (RPR) and treponema pallidum haemagglutination (TPHA) (Omega Diagnostics Immutrep). All positive RPR samples were confirmed with a different RPR (Plasmatec). Western blots were used to confirm syphilis if a result was TPHA positive but RPR negative (Euroimmune TP IgM and IgG). Active syphilis (primary and secondary syphilis) was defined as RPR titre $1:8 [10]. We used IgG and IgM ELISAs to test for HSV-2 (Philco) and cultured T.vaginalis (TV In-pouch). Polymerase chain reaction (Roche amplicor) was used to detect chlamydia and gonorrhoea in urine sample aliquots pooled into fives. If a pool tested negative, all samples were reported as negative. If it tested positive, a second aliquot of each sample in the pool was tested individually. The validity of this approach is well established [11,12]. Processed samples were stored between 2 to 8uC at study sites and during transportation and at 220uC at the storage centre in Surat. They were transported either on the day of collection or the following day to Biocare Laboratories Ltd., Ahmedabad for HIV-1, HSV-2, syphilis and T.vaginalis testing. Urine sample aliquots were sent weekly to Auroprobe Laboratories, New Delhi for chlamydia and gonorrhoea testing. All samples were stored at A four level conceptual determinant model was developed for any prevalent STI diagnosis on the day of interview. The model is based on the relationship between the biological determinants of STI infection and their effect on STI outcomes. Ro = BcD summarises the relationship of the biological determinants of STI infection where Ro is the reproductive number of an infection defined as the average number of secondary cases which arise from a new case of infection, B is the efficiency of transmission per contact, c the 280uC except for T.vaginalis which was kept at room temperature up to 37uC until the final report at five days. Ten percent of each of the following samples, HIV-1, HSV-2 IgM and IgG, RPR and TPHA including all positive samples and randomly selected negative samples were sent for external quality control (EQC). All positive chlamydia and gonorrhoea samples and 2% of negative samples also had EQC. There was good agreement between EQC and study results (HIV-1, CT and NG 100%, RPR and TPHA 97.6%, HSV-2 IgM 94% and HSV-2 IgG 92.9%). T.vaginalis samples were not sent for EQC due to poor survival beyond five days. --- Statistical Methods Double-entry for data. Data were double-entered and verified using Epi Info 6 software. Outcomes. The outcomes were HIV-1 and 'any STI'. 'Any STI' was defined as chlamydia, gonorrhoea, T. vaginalis, HSV-2 or syphilis diagnosed positive on laboratory testing. Determinants. The determinants included socio-demographic factors and sexual behaviour for both HIV-1 and 'any STIs', and also reported and laboratory diagnosed STIs for HIV-1. Laboratory diagnosed STIs were grouped into likely recently acquired STIs (active syphilis (RPR$1:8), chlamydia, gonorrhoea, T.vaginalis), ulcerative STIs (serologically diagnosed herpes simplex virus 2 and syphilis) and non-ulcerative STIs (chlamydia, gonorrhoea and Trichomonas vaginalis). Effect measures. We calculated prevalence, unadjusted and adjusted odds ratios (OR). Only % and p values were calculated for rare determinants, i.e. less than 5% of participants, and the p values obtained using a survey command were confirmed with Fishers' exact test. Statistical analysis. We used Stata v.8 for data analysis. The survey commands in STATA were used to take stratification, clustering and weighting into account. For the diamond industry the first randomly selected sampling unit (primary sampling unit or PSU) was each person in the largest stratum and each complex in the remaining strata. For the eight largest textile markets, strata were defined by job type and PSU was defined as the participants recruited to the study. The three smallest markets together made up one stratum with PSU defined by each individual market. To take into account significant differences between diamond participants and non-responders in the length of time men had lived in Surat and whether or not married men were living with their wives in Surat and between textile participants and nonresponders in marital status we weighted the participant data prior to analysis to increase the generalisability of the results to the study population. Regression analysis for a diagnosis of 'any STI'. The outcome for regression analysis was 'any STI' since the prevalence of individual STIs was relatively low. So that the study was adequately powered to detect associations between the determinants and 'any STI' and since there were no significant interactions between industry and the determinants, data for both industries were combined prior to further analysis. Underlying determinants of STIs can be used to identify populations for HIV/STI prevention interventions. We developed a conceptual framework (Figure 1) similar to that for HIV [13]. The model is based on the relationship between the biological determinants of STI infection and their effect on STI outcomes. Underlying determinants act through proximate determinants to influence biological determinants which in turn determine the STI outcome. Two model selection procedures were performed based on logistic regression. A base model was built from a forward stepwise model selection procedure of proximate determinants, using a significance level of p,0.05. A full model resulted from selecting further underlying determinants significant at p,0.05 and added to the base model. Underlying determinants had a higher threshold for inclusion in the full model compared to the proximate determinants, as only those which could not be fully explained by proximate determinants were retained in the full model. Evidence for interaction between underlying and proximate determinants was explored separately for each model. --- Results --- Comparison of Survey Participants and Non-responders The overall participation rate was 77% (845/1099). Diamond workers had a higher response compared to textile workers (83.2%, 456/548 vs. 70.8%, 390/551). Diamond participants and non-responders were similar with respect to birthplace (Gujarat excluding Surat 78.2% vs. 69.7%, respectively), median age (both 26 years) and marital status (married: both 59%). However diamond participants reported a significantly longer stay in Surat (median 8 years vs. 5 years, p = 0.01) than non-responders and if married, were significantly more likely to report living with wives than non-responders (67.7% vs. 41.9%, p,0.01). Textile industry participants and non-responders were similar in age (median 26 years vs. 24 years, respectively), birthplace (Indian state other than Gujarat: 92.8% vs. 95.0%, respectively) length of stay in Surat (median 9 years vs. 8 years, respectively), and reporting they lived with wives (50.2% vs. 53.5%). Textile participants were significantly more likely to report they were married compared to nonresponders (70.5% vs. 61.5%, p = 0.04). --- Survey Completion, Missing Data and Inter-observer Variability Of 845 participants, 839 (99.3%) men completed the survey. There were few missing data (range 0.2% to 6.3% across items). Four men declined to provide biological samples. There was a high level of agreement between study doctors in their diagnoses of genital discharge, ulcers and circumcision and among study interviewers with respect to their documentation of participants' responses to seven sexual behaviour questions. We excluded data from one diamond worker who only gave samples, and report the findings for the remaining 455 diamond and 390 textile workers. --- Socio-demographic Characteristics, STI Symptoms, Sexual Behaviour, Self-reported Circumcision and Antibiotic Use Compared to diamond workers, textile workers were significantly more likely to report they were born outside Gujarat, had exposure of susceptible persons to infected persons and D the duration of infectivity. The model has four levels: underlying determinants, proximate determinants, biological determinants and the outcome, 'any STI'. The underlying determinants act through the proximate determinants to influence the biological determinants which in turn determine the outcome 'any STI'. The underlying determinants included socio-demographic, economic and lifestyle factors. The proximate determinants included sexual behaviour and reported STI symptom related factors including treatment, and circumcision. Figure 1 Sexual behaviour lived in Surat for longer, were not living with their wives, had lower incomes and were not Hindus (Table 1). Self-reported circumcision and recent alcohol consumption, illicit drug use, antibiotic use and STI symptoms on the day of interview were uncommon. Textile workers were significantly more likely to report they had ever been sexually active, had ever had vaginal intercourse and had more than one lifetime female partner compared to diamond workers. Among married men, textile workers were significantly more likely to report female partners apart from their wife in the past year. --- HIV-1 Infection Eight participants were diagnosed with HIV-1, (see Table 2). Age and any STI were significantly associated with HIV-1 (see Table 3). Men with any STI and those with serological syphilis or HSV-2 were far more likely to be diagnosed with HIV-1 compared to men without STIs. There were no infections among men aged less than 25 years, unmarried men, those who reported consistent condom use in the past year or a history of circumcision. Most HIV diagnoses were in men who were living with their wives and had lived in Surat for more than ten years. We did not analyse HIV-1 further since few men had the infection. --- STI Infection We diagnosed STIs in 9.5% of participants (see Table 2). Of these, 38.9% were likely recently acquired STIs. The two most common diagnoses, syphilis (3.8%) and HSV-2 (3.4%) accounted for 71% of all STIs although none of the participants were noted to have genital ulceration on examination. Chlamydia and gonorrhoea were both uncommon (0.7% and 0.9%, respectively). Diamond workers were almost twice as likely to be diagnosed with any STI and the most common STI, syphilis, compared to textile workers. Few syphilis diagnoses were due to active syphilis (16%). Of eighty-three men diagnosed with STIs, 21.7% (nine syphilis, six N a is the total unweighted denominator per determinant. It was 845, the total number of participants in the study, for all the following variables: socio-demographic characteristics, STI symptoms, antibiotic use, self-reported circumcision status and ever had sex with male of female partners. N was 654, the number of men who reported ever having had vaginal intercourse for the determinants age at first vaginal intercourse in years, the number of lifetime female partners with whom had vaginal intercourse and whether or not participants had vaginal intercourse in the past year. For the determinant consistent condom use for vaginal intercourse in past year, N was 590 which was the number of men who had vaginal intercourse in the past year. For the determinant type of female partner in the past year, N was 523, the number of married participants in the study who had vaginal intercourse in the past year. n b is the unweighted number of participants by variable category. n c is the total unweighted number of participants by industry. d is the weighted column % and 95% confidence interval for participants by industry for each category of each determinant. e Ever had sex with male and/or female partners includes vaginal, oral, or anal intercourse. * Other religion includes Islam, Buddhism and Jainism. doi:10.1371/journal.pone.0043576.t001 trichomonas, one chlamydia, one gonorrhoea and one HSV-2) reported they were virgins. STIs were diagnosed in similar proportions of men who reported ever having had sex and never having had sex (9.4% of 664 men and 9.8% of 177 men). Age, length of stay in Surat, industry and antibiotic use were significantly associated with having 'any STI' in unadjusted analysis (see Table 4). We also analysed our data using a lower cut-off of two years stay in Surat based on a recent survey among male migrants in India [14]. Living in Surat for more or less than two years was not associated with either having HIV-1 or any STI (1.0% vs. 0.8%, p = 0.87 and 9.0% vs. 12.1%, p = 0.21, respectively). We included all participants in the multivariate analysis since sexual behaviours were not significantly associated with having any STI. The base model consisted of antibiotic use in the two weeks prior to interview. The final model included antibiotic use, length of stay in Surat and industry. After adjusting for other factors, men who had used antibiotics in the two weeks prior to interview, had lived in Surat for at least ten years or were diamond workers had significantly increased odds of having any STI (aOR 2.57, aOR 1.98 and aOR 1.83, respectively) (see Table 5). --- Discussion The study participants had a much lower than expected HIV-1 prevalence of 1.0% and a STI prevalence of 9.5%, arising from a mixture of acute and non-acute infections, and reflecting both recent and lifetime risk. Diamond work, living in Surat for at least 10 years and antibiotic use in the two weeks prior to interview were each associated with higher adjusted odds of having an STI; antibiotic use had the strongest association (aOR 2.57). There were some socio-demographic differences between men from the two industries in that textile workers were more likely to report lifetime sexual activity, multiple female partners and among married men non-spousal partners in the past year whereas diamond workers were more likely to have an STI and report recent antibiotic use. However, as described in the methods, analysis of the determinants of 'any STI' was based on combined data from the two industries because interaction testing revealed no significant evidence that associations with determinants differ between industries. The low HIV-1 and STI prevalence may in part be explained by most participants reporting only one sexual partner in their lifetime (78.9%). In the majority of cases this was a spouse and men who were more than 25 years old were much more likely to be married than younger men. The association between longer stay in Surat and STIs, after adjusting for age, is consistent with an increase in the rate of STI acquisition on migration to Surat. However the lack of overlap of STIs in individuals suggests that many of the sexual encounters may be low risk events. It is unclear why having an STI was associated with antibiotic use in the two weeks prior to interview and may be due to chance. This is especially likely given that past syphilis and HSV-2 over a lifetime account for the highest prevalence of ''any STI''. In addition there was no association between STIs and self-reported STI symptoms on the day of interview. The strengths of our study include HIV-1 and STI results verified by EQC, a high participation rate, low item non-response and low HIV or STI testing refusals. The limitations include the cross-sectional study design which limited our ability to interpret causal associations, social desirability bias and STI diagnosis. STI diagnoses in self-reported virgins in the study are likely to reflect social desirability bias for self-reported sexual behaviour. This is supported by the data from focus groups we conducted with 37 migrant men in Surat's diamond and textile industries in 2007 in which they discussed the stigma faced by men who had pre-or extra-marital sex. The method by which we obtained the sexual behaviour data, face to face interviews (FTFI), may also have contributed to this bias. In our study, two Indian national general population surveys and a recent Indian sexual behaviour survey which used FTFIs, sex with female sex workers (FSWs) in the past year was reported by 2.4%,,1%, 3.4% and 2%, respectively [15,16,17]. However in the latter study when a more anonymous method, polling booth surveys, was used 11% of participants reported sex with FSWs. Some of the STIs, syphilis, HSV-2 and gonorrhoea, found in self-reported virgins may be due to false positive laboratory test results. This can be a result of cross-reactivity of N.gonorrhoeae with non-pathogenic Neisseria species and of HSV-2 IgG with HSV-1 IgG [18]. RPR biological false positives can occur with other conditions and RPR sensitivity can vary between 44% and 76% in primary syphilis and between 70% and 73% for late latent or late syphilis [19]. Therefore all positive RPRs were confirmed by a different manufacturer's RPR test and TPHA and Western Blot were used as described in the methods. Some of the syphilis may have been either acquired congenitally or through blood transfusions; HIV-1 may also have been acquired through blood transfusions. The study sample may have been at lower risk of HIV and STI acquisition than the study population since within the textile industry participants were more likely to be married than nonparticipants and diamond industry participants were more likely to live with their wives than non-participants. To try and take these differences into account and make our results more generalisable to the study population, we weighted on marital status for the whole study sample and on duration of stay in Surat for diamond industry participants since they differed in this from nonresponders. However this may not have corrected for systematic differences in sexual risk behaviour and the results should therefore be treated with a degree of circumspection. On comparison of our results with those of other studies of male migrants from less developed countries, HIV prevalence was higher among Nepalese migrants who had worked in India (10.3%) [20] and considerably higher among migrants within Africa (range 24.4% and 33.9%) [4,5,6,7,8]. Migrants within Pakistan and China also had low gonorrhoea and chlamydia prevalence similar to the study participants (Pakistan, 0.5% vs. 1% and China, 0% to 3.5% vs. 1.8% to 0.5%, respectively) [21,22,23]. Nepalese migrants had a higher syphilis prevalence (25%) compared to the study participants and Chinese migrants (1%) [20,22]. HSV-2 prevalence was higher among Chinese migrants (5.5%) [24] compared to the study participants. As in our study, longer length of stay at the migration destination was associated with STIs among the Nepalese and Chinese migrants [20,22] and older age was shown to be associated with STIs and HIV [22,25]. Some of the variations between studies will be due to differences in study design, sampling, laboratory tests and how migrants are defined. There are conflicting data for how either living with a spouse or length of stay at a migration destination are associated with higher risk of HIV and STI infection [5,20,22,26,27]. Other studies have assumed that men who migrated to a destination recently will have higher risk behaviour than men who migrated prior to this [14,28,29]. However we found that living with a wife or living in Surat for at least 10 years was associated with having HIV-1, although the numbers were small. Living in Surat for at least 10 years was also significantly associated with having an STI among our participants. In contrast, living in Surat for less than two years was not associated with either HIV-1 or STI infection. This suggests that migrant men who have resided in Surat for a longer duration may have more access to sex partners than men who have recently migrated there. We postulate that this may be due good employment prospects and income in Surat so that there is spare money available to spend on sexual partners. HIV/STI risks and behaviours may differ between migrants who have employment and migrants who do not. Although the participants' HIV prevalence was similar to ANC attendees in the 2005 sentinel surveillance (1.3%), a proxy group for Surat's general population [9], this is three times greater than the 2006 national Indian general population HIV prevalence estimate (0.36%) [30]. The HIV prevalence among study participants is therefore likely to represent an increase in risk after migration since the majority (698) of the 845 study participants were from low HIV prevalence Indian states. The sentinel surveillance HIV prevalence in Surat has remained above 5% among STI clinic attendees and high risk groups and above 1% among ANC attendees since 2005 [31,32,33]. Commercial sex work is likely to have played a significant role in this especially following the closure of the red light area (RLA) in 2003which greatly hampered the implementation of a HIV/STI prevention programme to the sex workers. Mathematical modelling based on Indian data suggests that male migration itself can lead to a change in sexual networks in their place of origin [34]. In a South African cohort study HIV acquisition in female partners of migrant men in the place of origin was independent of the men's HIV status [7]. Further prospective studies are required to clearly understand the association between migration and HIV in populations at risk [35]. A substantial proportion of STIs were likely recently acquired STIs (38.9%) among participants and the increased risk of HIV acquisition and transmission due to the presence of STIs is well established [36,37] Similar to other studies of male migration in Asia, we found a lower prevalence of HIV and STIs among male migrants in our study compared to studies from Africa. Our findings challenge the concept that migrants are always at higher risk of HIV infection compared to the general population in the area to which they migrate, though the percentage of possibly recently acquired STIs is relatively high, indicating the risk for future, continued disease spread. Even if migration causes an increase in HIV and STI risk for individual migrants, the prevalence of infections in the areas of previous residence, the patterns of mixing between migrants and others, and the proportion of the population that are migrants all influence whether migrants will have a higher prevalence of infection than the general population in the area to which they migrated. Our findings suggest the need for data specific to a setting which can be contextualised with other local information to guide HIV/STI interventions. --- Author Contributions

Background: Our objective was to estimate for the first time the prevalence and determinants of human immunodeficiency virus type 1 (HIV-1) and sexually transmitted infections (STIs) among male migrants in India. Methodology/Principal Findings: We conducted a multi-stage stratified probability sample survey of migrant (defined as not born in Surat city) men aged 18 to 49 years working in the diamond and textile industries in Surat city. Behavioural and biological data were collected. Biological data included laboratory diagnosed herpes simplex virus type 2 (HSV-2), syphilis, chlamydia, gonorrhoea, Trichomonas vaginalis (together defined as 'any STI') and HIV-1. Likely recently acquired STIs included chlamydia, gonorrhoea, T.vaginalis and syphilis with rapid plasma reagin $1:8. The response rate was 77% (845/ 1099). Among 841 participants, HIV-1 prevalence was 1.0%, 'any STI' prevalence was 9.5% and 38.9% of these STIs were likely to have been recently acquired. Being a diamond worker, Surat resident for 10+ years and recent antibiotic use were each associated with higher odds of 'any STI' (aORs 1.83 (95% CI 1.09-3.09), 1.98 (95% CI 1.22-3.22) and 2.57 (95% CI 1 .17-5.64), respectively) after adjusting for the other two factors and age. The main study limitation was social desirability bias for self-reported sexual behaviour; STIs were diagnosed in some self-reported virgins. Conclusions/Significance: HIV and STI prevalence were lower than expected, but prevention interventions remain necessary in Surat since almost 40% of STIs among participants were probably recently acquired and sentinel surveillance HIV prevalence remains high. The participants had a similar HIV prevalence to Surat antenatal clinic attendees, a proxy for the general population. This suggests migrants are not always at higher risk of HIV compared to the general population in their migration destination. Our findings highlight the need to contextualise research findings from a specific setting with other local information to guide HIV/STI prevention interventions.

Introduction In March 2020, policymakers worldwide took measures to slow down and limit the spread of SARS-CoV-2, including lockdowns, home-isolation, and the restriction of free movement. These measures taken to fight the spread of the virus had a significant impact at individual, social, and economic levels. A forced stay at home, combined with the uncertainty of the pandemic and its progression, unstable socio-economic conditions, financial constraints, and physical threats, among others, increased the tension at home for many --- Materials and Methods --- Study Design and Setting In the summer of 2020, an international consortium of more than 45 research institutes was formed under the coordination of Ghent University (Belgium) to set up the study to consider how primary care practices were organized during the COVID-19 pandemic (PRICOV- 19). This multi-country cross-sectional study focused on quality and safety in primary care practices during the COVID-19 pandemic. Data were collected in 38 countries by means of an online self-reported validated questionnaire among primary care practices. The questionnaire was developed at Ghent University in multiple phases, including a pilot study among 159 general practices in Flanders (Belgium). More details are described elsewhere [40,41]. The questionnaire was translated into 38 languages following a standard procedure. The Research Electronic Data Capture (REDCap) platform was used to host the questionnaire in all languages, send out invitations to the national samples of general/family practices, and securely store the answers from the participants [42]. --- Sampling and Recruitment The data reported here were collected between November 2020 and December 2021, except for Belgium, where data were partially collected from 20 May 2020 onwards. The timeframe for data collection varied between countries from 3 to 35 weeks. In each partner country, the consortium partner(s) recruited primary care practices following a predefined recruitment procedure [40], which is shown in the Supplementary Materials (Table S1). Sampling and recruitment are detailed elsewhere [40,43]. One questionnaire was completed per practice, at least one reminder was sent in all countries, and the overall response rate was 27.8%. --- Measurements 2.3.1. Characteristics of the Respondent and the Practice This paper only used the questionnaires completed by GPs and GP trainees. Questionnaires completed by other health professionals working in a GP practice were not included. In all instances in the analysis where we refer to GP, this includes GPs and GP trainees unless otherwise stated. The questionnaire contained questions regarding the respondent (position in practice and years of work experience) and the practice (number of GPs and GP trainees working in practice and location of the practice). Additionally, respondents were also asked to compare their practice's average patient population (including patients with a migration background with difficulty mastering the local language, with limited or low health literacy, with financial problems, psychiatric vulnerability, over the age of 70, chronic conditions, little social support or limited informal care) to other practices in their country through seven 3-point Likert-scale items, ranging from 'Below average (1)' to 'Above average (3)'. --- Patient Flow Patient flow related questions included questions on the appointment system (walk-in hours and video consultations), triage (protocol when answering phone calls, telephonic triage, available information on how to refer a patient to a triage station), and changes in the GP's role (involvement in actively reaching out to patients that might postpone healthcare, increased responsibilities, task shifting and further training). Additionally, respondents were asked about the occurrence of specific incidents since the COVID-19 pandemic due to the complexity of primary care and the high degree of uncertainty (e.g., 'A patient with a serious condition was seen late because he/she did not know how to call on a GP'). These five coded as 'yes (1)' or 'no (0)' items were summed into one scale ranging from 0 to 5, Cronbach's alpha = 0.752. --- (Pro-)Active Communication Specific initiatives to contact vulnerable patients (with a chronic disorder, a chronic condition that needed follow-up care, psychologically vulnerable patients and/or those with previous problems of domestic violence or with a problematic child-rearing situation) were reported through four items coded as 'yes (1)' or 'no (0)'. 'To what extent have patients talked to you about financial problems since the COVID-19 pandemic?' through four 5-point Likert-scale items, ranging from 'Not at all (0)' to 'Much more than before (5)'. These items were rescaled to 3-point Likert-scale items 'Not at all or less than before (0), 'As much as before (1)' and '(Much) more than before (2)' as there were insufficient units of observation in the extremities. Finally, the questionnaire also contained four 'yes (1)'/'no (0)' items concerning communication (practice website, leaflet, answering machine, and COVID-19 specific leaflet) with patients. These items were summed into one scale ranging from 0 to 4, Cronbach's alpha = 0.654. --- Wellbeing of the Respondent The wellbeing of the GPs was assessed using the expanded 9-item Mayo Clinic Wellbeing Index (eWBI) [41]. Seven items are responded to with a 'yes (1)' or 'no (0)', and the remaining two items are responded to on a 7-point or 5-point Likert scale, ranging from'strongly disagree' to'strongly agree'. The first seven items are summed into one scale ranging from 0 to 7, Cronbach's alpha = 0.791. If the respondents replied'strongly disagree' or 'disagree' one point was added to their score, those who reported 'agree' or'strongly agree' had one point subtracted from their score. No adjustments were made for those with middle/neutral responses. Being at risk of distress is defined as a score of <unk>2, as per previous studies [41]. --- Data Analysis Data was imported into SPSS27 for initial data cleaning and manipulation. All statistical analyses were conducted with R software version 4.0.3 on the database of 33 countries, available as of 3 November 2021. (a) To describe the sample, as well as the frequency of screening for DV by GPs and disclosure of DV by patients to the GP during the COVID-19 pandemic, simple descriptive statistics were analysed, and group differences were computed using a (post hoc) chi-square test as all assumptions of the chi-square test could be met. (b) To identify key elements that could potentially explain differences in screening for and disclosure of DV (between those who answered 'Not at all or less than before', 'As much as before' and '(Much) more than before' to the screening for and disclosure of DV questions) we chose an exploratory study. Two stepwise forward-clustered ordinal logistic regressions were therefore conducted to analyse the association between the multiple factors and the two outcome variables. To avoid multicollinearity, the correlations were checked between all variables. There were no strong correlations (<unk>0.60). To ascertain the power required for the multiple logistic regression analyses, we assumed a ratio of ten cases per predictor based on the simulation by Peduzzi et al. [44]. The respondents were clustered by country, leading to a multilevel analysis. Main terms with p <unk> 0.05 were included in the multivariate models, and variables that produced at least one beta estimate significantly different from zero were retained. It was also determined whether these added main effects significantly improved the prediction of both outcome variables using a likelihood ratio test. Furthermore, the Akaike information criterion (AIC) and Bayesian information criterion (BIC) were used to compare the relative quality of one model to another by balancing a model's goodness of fit against its complexity. By doing so, it takes into account the risk of overfitting as well as the risk of underfitting. Models were then ranked from best to worst, with the "best" model showing the smallest AIC and BIC. Ordinal logistic regression was chosen because the outcome variables consisted of three categories (not at all/less than before = 0, as much as before = 1, (much) more than before = 2). Logit is the most commonly used link, and it allows the production of odds ratios by exponentiation of the model estimates. Negative estimates of the independent variables show that one value of an independent variable compared to its following value is more likely to receive lower values on the ordinal dependent variable and vice versa for positive estimates. --- Ethical Approval The study was conducted according to the guidelines of the Declaration of Helsinki. The Research Ethics Committee of Ghent University Hospital approved the protocol of the PRICOV-19 study (BC-07617). Research ethics committees in the different partner countries gave additional approval if needed in that country. All participants gave informed consent on the first page of the online questionnaire. --- Results The analysis included 4295 GPs with a valid value for the variables concerning DV. A description of the main characteristics of the sample is shown in Table 1. Approximately onequarter of the respondents were in each 10-year age group. In terms of location, the majority (43.2%) of the respondents were working in practices based in cities/suburbs and 34.6% work in single-handed practices. Screening for and disclosure of DV during the COVID-19 pandemic was significantly more likely to have been undertaken by respondents working in practices that estimated the proportion of patients with limited health literacy, financial problems, psychiatric vulnerability, and/or little social support in their practice above average for their country. Additionally, disclosure of DV by the patient was significantly more likely reported by respondents working in practices that considered their proportion of patients with a migration background above average. Compared to before the pandemic, almost half of the respondents (47.9%) reported no disclosure or less disclosure by the patient, four out of ten (40.9%) reported as much disclosure as before and only 11.4% reported (much) more disclosure during the pandemic. These proportions varied per country, as shown in Figure 1, but there are, however, almost no countries reporting much more disclosure during compared to before the pandemic. Similar results were found for the screening for DV by the GP, with four out of ten respondents (38.7%) who did not screen or screened less for DV during compared to before the pandemic. Almost half of the respondents (49.2%) screened as much as before, and only 12.1% screened (much) more than before the pandemic (Figure 2). Supplementary Materials Table S2 (Disclosure of DV by the patient) and Table S3 (Screening for DV by the GP) present the results of the multilevel ordinal logistic regression analysis. Model I, showing the intercept-only model, has an ICC of, respectively, 10.5% and 9.9%, meaning that approximately 10% of the variance in both, disclosure of and screening for DV, is attributable to the country. Each subsequent stepwise model shows a better goodness-of-fit (based on a smaller AIC and/or BIC value). Variances at the group and individual levels reduce when adding variables, indicating that the large variance in disclosure of and screening for DV between countries is reduced by adding individual-level variables. Compared to before the pandemic, almost half of the respondents (47.9%) reported no disclosure or less disclosure by the patient, four out of ten (40.9%) reported as much disclosure as before and only 11.4% reported (much) more disclosure during the pandemic. These proportions varied per country, as shown in Figure 1, but there are, however, almost no countries reporting much more disclosure during compared to before the pandemic. Similar results were found for the screening for DV by the GP, with four out of ten respondents (38.7%) who did not screen or screened less for DV during compared to before the pandemic. Almost half of the respondents (49.2%) screened as much as before, and only 12.1% screened (much) more than before the pandemic (Figure 2). Supplementary Materials Table S2 (Disclosure of DV by the patient) and Table S3 (Screening for DV by the GP) present the results of the multilevel ordinal logistic regression analysis. Model I, showing the intercept-only model, has an ICC of, respectively, 10.5% and 9.9%, meaning that approximately 10% of the variance in both, disclosure of and screening for DV, is attributable to the country. Each subsequent stepwise model shows a better goodness-of-fit (based on a smaller AIC and/or BIC value). Variances at the group and individual levels reduce when adding variables, indicating that the large variance in disclosure of and screening for DV between countries is reduced by adding individual-level variables. Reporting a perceived higher or equal average number of patients with a psychiatric vulnerability in practice compared to the rest of the country, (much) more screening for or disclosure of financial problems during the COVID-19 pandemic, and actively reaching out to patients with a known history of DV or problematic child-rearing situations during the COVID-19 pandemic were all positively associated with (much) more disclosure of and screening for DV. On the other hand, a greater number of GPs in practice and a prevalence of five different types of incidents (where patients were seen late) were only associated with (much) more disclosure of DV by the patient, while the presence of print versions of the triage information in every consultation room, GPs who (strongly) agreed that they were more involved in actively reaching out to patients who might postpone healthcare during the COVID-19 pandemic, and the presence of at least three communication items (e.g., leaflets, website, answering machine) for the patients in the GP's practice, were associated with (much) more screening for DV by the GP (Table 2). --- Discussion Since the start of the COVID-19 pandemic, experts feared an increase in DV due to the isolation, movement-restricting measures, uncertainty, and impact of the pandemic on the socio-economic conditions, health threats, etc. As the crisis progressed, reports of DV did increase [6][7][8][9][10]. Hence, an expected need for the most accessible caregivers, namely GPs during the pandemic, to address this risk in their patients. Yet, in the 33 countries participating in this study, only 12% of the GPs reported having screened more or much more for DV than before the pandemic, and 11% of the GPs reported more or much more frequent disclosure of DV by the patients. On the other hand, half of the GPs reported no disclosure at all or less than before the pandemic, and four out of ten GPs did not screen or screened less than before the pandemic. This confirms previous research [29,[45][46][47] that found that most GPs are insufficiently aware of the general magnitude of DV and its impact on people directly and indirectly exposed to it, or inadequately trained to approach and manage it. Based on our study, screening for and disclosure of DV during the COVID-19 pandemic occurred significantly more often during compared to before the pandemic in practices that estimated the proportion of patients with limited health literacy, financial problems, a psychiatric vulnerability, and/or little social support in their practice above average for their country. Additionally, disclosure of DV by the patient also occurred significantly more often during compared to before the pandemic in practices that considered their proportion of patients with a migration background above average. Awareness of GPs, concerning risk factors of DV, is a major step to identify vulnerable people. However, these results could not be entirely confirmed with the regression analyses. Only a higher estimated proportion of patients with a psychiatric vulnerability in practice came out as a significant predictor of screening for, as well as disclosure of DV during the COVID-19 pandemic. This suggests that GPs who estimated a higher proportion of vulnerable patients within their practice, compared to the average in the country, might not necessarily have a higher proportion of vulnerable patients, but are rather more aware of the risk of DV that these groups of vulnerable patients might face. However, while it is important to have increased attention to certain vulnerable groups, it is also important to keep in mind that anyone can become a victim of DV, especially in stressful times such as a COVID-19 pandemic with isolation and movement-restricting measures. To achieve this awareness and knowledge, appropriate training to implement tools and integrated models of care are needed. However, despite the acknowledgement of the need for further professional education/training for GPs about DV, Silva et al. found that this does not occur in the majority of courses [48]. Specializing young GPs may include practical training periods among disadvantaged patient groups/living areas or patients with a migration background. In addition, community-oriented approaches that include pathways or interventions, which may provide appropriate care to DV victims after they are identified should also be in place. The availability of integrated intervention increased preparedness after training in managing the complexity of DV in clinical practice; however, it was insufficient to catalyse identification and specialist referral. The lack of guidance in the usage of information about DV also has been reported as a barrier in practice [49]. According to our study, most significant associations concern (pro)active communication. More disclosure of and screening for DV was associated with GPs who reported (much) more disclosure of and/or screening for patients' financial problems, and who have actively been reaching out to patients with a history of DV or problematic child-rearing situations during the pandemic. Additionally, GPs who reported to have been more involved in actively reaching out to patients, in general, during the pandemic, and those who had at least three communication items (e.g., leaflets, website, answering machine) for the patients in the GP's practice, were also more likely to screen for DV. (Pro)active communication that has been culturally adapted for the population seems to enhance trust and openness towards the patients and may help them disclose sensitive situations such as DV during consultations. However, victims generally find it difficult to raise the topic of DV themselves, and it has been evidenced that many of them prefer that the GP starts the conversation and asks directly about it. GPs, however, tend to wait for a specific clue or for the patient to address a possible history of DV, instead of routinely raising the topic without any suspicion or specific reason [30][31][32][33]. The researchers argue that the strong commitment of key individuals to addressing DV is vital for response sustainability, and would be bolstered by prioritizing DV response as a national health priority with dedicated resourcing [50]. Towards that direction, there is adequate evidence that available screening instruments can identify DV [51]. These instruments need to be translated and culturally adapted in different countries so that GPs can be trained in their use and interpretation. Previous studies supported that effective interventions generally included ongoing support services that focused on counselling and home visits, addressed multiple risk factors (not just IPV), or included parenting support for new mothers [51]. GPs need to be aware of the supportive services that are available in their community and refer victims who disclose DV to these services. The agreement on and dissemination of guidelines on screening for potential exposure to DV, the identification of barriers and facilitators for implementing such screening, and the training of GPs in interviewing victims of DV, providing these victims first with psychological help and a warm referral when needed, needs to be prioritized in the health policy agenda. Interestingly no clear evidence was found for practice-related factors. More or much more disclosure of DV by the patient to the GP was associated with practices with more than one GP. This association was, however, not be found for screening for DV. No significant associations were found in the task-shifting variables due to the COVID-19 pandemic or the mental wellbeing of the GPs. While evidence has accumulated of increased workload, overburden, and burnout of primary care physicians during the COVID-19 pandemic [43,52], it did not seem to affect screening habits of the GP, or stop/encourage patients to disclose during the COVID-19 pandemic. --- Strengths and Limitations The study was conducted in 33 different countries. Although the response rate was only 28%, the analysis included 4295 GPs with a valid value for the variables concerning DV. Respondent GPs represented well each 10-year age groups of GPs currently working in primary health care. Most GP practices were based in cities or suburbs, but rural GP practices were quite well represented. The countrywide spread of the COVID-19 pandemic and announced lockdowns varied greatly between the European countries during data collection. Given the voluntary nature of the participation of the GPs, a selection bias is possible. The study questions asked about DV compared with the situation before COVID-19 pandemic, so the GPs in the different groups (much more, as much as before, and not at all/less) may be very different from each other. The study is a cross-sectional study, thus causal relationships could not be established. Another limitation may be related to the question that was asked concerning DV. We asked about screening/disclosure compared to the time period before the COVID-19 pandemic. When GPs responded "much more than before" we do not know if they already screened a lot and now even more, or if they never screened before while now they have started screening the population. Furthermore, when they say "less than before", it might be that they screened absolutely everyone in practice, but due to the COVID-19 pandemic they have less time and now "only" screened 95% of the patients. We do not know all of this for sure, so the GPs in each response category group (much more, as much as before, and not at all/less) can be very different. --- Conclusions While reports of DV increased during the COVID-19 pandemic, only a fraction of the GPs in our study reported (much) more screening for or disclosure of DV. According to our results, most of the significant associations with screening for and disclosure of DV concerned (pro)active communication, such as screening for other issues such as financial problems, or actively reaching out to vulnerable patients with, for example, a history of DV or problematic child-rearing situations. This leads us to believe that GPs are first of all insufficiently aware of the general magnitude of DV and its impact in society, as confirmed in previous research. Further professional education/training for GPs about DV is therefore highly and urgently needed. Such training should be implemented as a basic academic course in the GP curriculum. Additionally, community-oriented approaches that include pathways or interventions should also be in place. Secondly, once the GPs are sufficiently aware of the supportive services that are available in their community, as well as sufficiently trained (e.g., in the identification of barriers and facilitators, in interviewing skills, in offering first psychological help, etc.) and own enough tools to help them support and refer victims of DV, guidelines on screening should be agreed. This should go hand in hand with the appropriate resourcing of PC to undertake this screening and ongoing support. --- Supplementary Materials: The following are available online at https://www.mdpi.com/article/10.3390/ijerph20043519/s1, Table S1: Responses in each country for the domestic violence variables, sampling method, and overall response rate. Table S2: Results of logistic mixed-model analysis of potential predictors for disclosure of DV by the patient to the GP during the COVID-19 pandemic., Table S3: Results of logistic mixed-model analysis of potential predictors for screening for DV by the GP during the COVID-19 pandemic. Author Contributions: S.W. led the conceptualization and design of the study. S.W. and E.V.P. led the overall coordination of the study. E.V.P., C.C., R.G.B., P.K., M.K.L., L.M., A.T. and S.W. coordinated the sampling and data collection in their institution/country. All authors agreed on the focus of this paper and discussed the analysis approach. E.F. and I.K. led the authorship group, drafted the first version of the paper, and submission. E.F. undertook data analysis for this paper. E.F., I.K. and R.G.B. contributed to the introduction. E.V.P., C.C., and E.F. wrote the methods section. E.F., A.T., S.W. and I.K. undertook data interpretation. P.K, M.K.L., L.M., A.T., E.F. and I.K. wrote the discussion. C.C. and S.W. undertook final editing of the manuscript. All authors have read and agreed to the published version of the manuscript. --- Funding: The PRICOV-19 data collection was undertaken without funding in individual countries. A small grant was received by Ghent University from the European General Practice Research Network (EGPRN) to cover data cleaning; no grant number applies. The APC for this paper was funded by the King Baudouin Foundation (Belgium). --- Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki and was approved by the Research Ethics Committee of Ghent University Hospital (project number BC-07617). Additional research ethical approval was obtained in participating countries if required in that country or by the participating institution's policies. --- Informed Consent Statement: The responses were anonymous; however, all respondents provided informed consent at the start of the questionnaire. --- Data Availability Statement: The anonymized data is held at Ghent University and is available to participating partners for further analysis upon signing an appropriate usage agreement. --- Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

The COVID-19 pandemic left no one untouched, and reports of domestic violence (DV) increased during the crisis. DV victims rarely seek professional help, yet when they do so, they often disclose it to their general practitioner (GP), with whom they have a trusting relationship. GPs rarely screen and hence rarely take the initiative to discuss DV with patients, although victims indicate that offering this opportunity would facilitate their disclosure. This paper aims to describe the frequency of screening for DV by GPs and disclosure of DV by patients to the GP during the COVID-19 pandemic, and to identify key elements that could potentially explain differences in screening for and disclosure of DV. The PRICOV-19 data of 4295 GP practices from 33 countries were included in the analyses, with practices nested in countries. Two stepwise forward clustered ordinal logistic regressions were performed. Only 11% of the GPs reported (much) more disclosure of DV by patients during COVID-19, and 12% reported having screened for DV (much). Most significant associations with screening for and disclosure of DV concerned general (pro)active communication. However, (pro)active communication was performed less frequently for DV than for health conditions, which might indicate that GPs are insufficiently aware of the general magnitude of DV and its impact on patients and society, and its approach/management. Thus, professional education and training for GPs about DV seems highly and urgently needed.

INTRODUCTION The mental well-being and physical fitness of medical doctors represent an important issue for health services worldwide (1,2). Some studies focused only on female physicians, while others surveyed recent medical graduates (3,4). General practitioners (GPs) represent the largest group of physicians (4). Other studies focused on specific issues such as smoking (in France, Greece, Switzerland, and Poland) or alcohol consumption (in Estonia, Hungary, Germany, USA, and India) (5)(6)(7)(8)(9)(10)(11). Mental health problems, particularly depression due to overwork and emotional pressure, are severe (12,13). Indeed, work satisfaction is known to be negative predictor of mental disorders, especially among women (14). Doctors often fail to follow the same current preventive health recommendations they advice their patients. Moreover, research suggests that a large proportion of doctors have never registered with a general practitioner (15). Quality of life is linked to a number of factors, to profession and working position (16). Studies focused mainly on psychological health, thus information concerning the physical health and relation to social circumstances remains limited. No previous study evaluating these differences between medical specialties was found. We aimed to evaluate the morbidity, demographic characteristics and lifestyle factors prevailing in the study cohort of Hungarian physicians and compare the data to those of the general population. --- MATERIALS AND METHODS --- Subjects The subjects were recruited from 423 physicians who graduated in 1979 at Semmelweis University of Medicine, Budapest, Hungary. Hundred and thirty-eight doctors who participated at the 30-year jubilee meeting (November 2009) were asked to complete a questionnaire. It was also sent to 88 colleagues who were not present but their addresses were available. --- Survey Design The questionnaire consisted of 42 multiple-choice questions and 11 sub-questions, in the following sections: • demographic data: gender, age, marital status, number of children, and place of residence; • medical career: pre-and post-graduation intentions, the number of years spent in different fields of medicine; place and position of their job; their overall job satisfaction; • anthropometric data: current and 1979 height and weight; • morbidity data; • self-reported health: rating own health status, comparison to patients of the same age, and participation in routine screening; • lifestyle factors: physical activity and leisure time activities, sleeping, smoking habit, and alcohol consumption. They were inquired about the recent problems of the Hungarian healthcare system. In some questions, there were overlaps between answers and there are lacking data in others, therefore numbers and percentages are often not identical. --- Statistical Analysis Continuous variables were compared using Student's unpaired t-test or ANOVA. Categorical variables were evaluated using a <unk> 2 or Fisher's test, as appropriate. --- RESULTS 208 questionnaires (83 men and 125 women) were accepted and analyzed. --- Demographic Findings The mean age (<unk> SD) of participants was 55.2 <unk> 2.1 (range: 54-67 years) with an average of 2.20 <unk> 1.09 children for males and 1.87 <unk> 1.03 children for females. Male primary care physicians have more children (2.43 <unk> 0.82) than women (1.87 <unk> 0.93); however, there was no significant difference between specialties (p = 0.34). Ninety percent of men and 73% of women were married, 6% and 15% divorced, respectively. Of all the women 3% lived with their spouse and 8% were widowed. The geographic distribution of place of residence and employment is presented in Table 1. Among those who had previously left Hungary, most immigrated to America or Western European countries. --- Type of specialty --- Men (N = 83) Women (N = 125) time --- Medical Careers Significantly (p <unk> 0.01) more men were specialists in surgery (Table 2). Most of the men working in the specialty had intended to work in this field at the time of graduation, whereas women were more likely to have deviated from their career paths (p = 0.04). Since graduation, 30% of men and 48% of women changed their working place and/or specialty (p <unk> 0.05), usually based on family (9% and 24%; p = 0.02) or professional reasons (18% and 12%), respectively. However, for both genders career changes were equally likely to be for financial reasons (5%) or because of changing place of residence (4%). Around 10% of women and 23 % of men had previously worked abroad. The participants were working in the whole range of healthcare: administrative leaders (including the Minister of Health), ten of them were university professors (in Hungary or abroad). Around 40% of men and 22% of women worked in senior positions, 36% and 41% as private practitioners, and 25% and 34% within the public healthcare system, respectively. The number of on-call duties (night shifts) decreased by decades from 6-8 monthly in the first decade of carrier to about 2, thirty years later, mainly due to appointment to a senior position. --- Anthropometric Findings A significant increase in weight and BMI was observed for both genders and across all specialty groups (Table 3). There was no significant difference between the genders or medical specialties. The largest increase in body weight and BMI (mean: 5.27; 95% CI 3.84-6.69) was observed among female primary care specialists. At the time of graduation, the majority of the respondents had normal BMI values, whereas at the time of the survey many of them, mostly men, were considered overweight or obese (Table 4). --- Morbidity Medications were used by 41% of men and 44% of women, respondents on regular medication mostly live in big cities (p = 0.041). The most frequent diagnoses can be found in Table 5. Few of the respondents mentioned psychological morbidities. Of those diagnosed with a lipid disorder, all were living outside of Hungary; women were more likely to use lipid-lowering therapy than men, mainly for secondary prevention. With respect to treatment adherence, 56% of men and 70% women always, 38% and 14% seldom, and 4% of both genders never follow medical recommendations, without any difference regarding specialty. Over the past 30 years, 73% of men and 77% of women had received an average of 1.4 hospitalizations (besides delivery). Of the respondents, 46% reported that as patients, they received preferential treatment from medical personnel. When treating patients, 28% of the doctors paid more attention to the patients suffering from the same conditions as the doctors themselves. Only 36% of the male and 20% of the female doctors had their own family physicians (55% in administrative terms for both genders). Female physicians were significantly less likely to be treated by family physicians than men (in 29%). --- Self-perceived Health Status Twelve percent of male and 22% of female compared their own health status and appearance to that of their patients, and 70% of men and 64% of women believed that their health status was better. Almost 21% of men and 15% of women believed that their profession had a negative impact on their health status. Of the males, 37% regularly attended screening examinations or diagnostic procedures in comparison with 57% of female doctors. --- Men (N = 83) Women --- Table 4. Moving to higher BMI categories since time of graduation --- Lifestyle Indicators Self-reported sleep patterns were generally good, reported to be longer at weekends (both genders) and in all specialist groups (Table 6). Female doctors in surgical specialties reported an average of 7.2 hours of sleep per night between workdays -a significantly longer rest time when compared to other medical specialties (p = 0.049). --- Physical Activity Twenty two percent of male and 12% of female reported sport activity on a daily basis, while 15% and 12% of women reported no physical activity. Men were significantly more active than women, although female physicians considered regular exercise to be more important than males. Male primary care physicians performed 4.4 exercise sessions weekly, while female primary care physicians only 1.2 sessions (p = 0.001). --- Smoking Only 5% of men and 6% of women were regular smokers at this time. For those who quit smoking, the mean age of quitting was 44.1 <unk> 12.1 years for men and 39.6 <unk> 10.6 years for women. --- Alcohol and Substance Abuse Based on self-reported data, 9% of men and 10% of women abstained from drinking alcohol, while 58% and 72% only imbibed occasionally, and 8% of both genders reported regular drinking. The amount of consumed alcohol was 4.8 <unk> 4.1 units per week for men and 1.65 <unk> 0.98 units per week for women. The mean age when regular drinking commenced was 27.6 <unk> 7.9 years in men and 32.8 <unk> 10.4 years in women. No differences between medical specialties were found, but male surgeons reported higher consumption (5.1 <unk> 10.6 unit per week; p = 0.048). Substance and narcotic use was not inquired. --- Men (N = 83) Women (N = 125) --- Leisure Time Women were significantly more likely to read novels (40% versus 16%) while men undertake physical activities (53% versus 35%). --- Burn-out Thirty years after graduation, 41% of male and 37% of female doctors were content, while 44% of male and 51% of female were often satisfied with their profession. --- DISCUSSION --- Demography Male doctors have more children than females, and the fact that primary care physicians tended to have larger families may reflect the different working conditions between specialties (13,17). Within the cohort, divorce rates were smaller than in the USA (18). The distribution of medical specialties follows a similar pattern to the overall Hungarian medical community (17). Budapest is geographically over-represented. In Hungary, there is limited mobility between cities and regions. --- Medical Careers In other countries, general practice was commonly chosen for lifestyle reasons; desire to have close contact with patients and shorter postgraduate training have also been cited as factors motivating graduates to enter primary care positions (19,20). Since primary care was recognised in Hungary as board specialty only 30 years ago, it was not a career option for the respondents. The major problems affecting healthcare system were rated as a lack of financial resources, followed by human resource constraints. The average age of doctors in Hungary is quite high, for primary care the average age is over 57 years. There is no incentive for junior doctors to specialise in primary care due to low salaries. The informal payment (tipping) system used to obtain better access or a higher quality service is still widespread in Hungary and that has a negative effect on the healthcare system. In the last years, many of experienced doctors have left Hungary in favour of other countries with better work condition and 6-8 times higher salaries. --- Anthropometric Changes in BMI were expected to match global trends in obesity and are similar to the findings of a study on US physicians of the same age groups (20)(21)(22). The higher proportion of weight gain of primary care specialists may be explained by the characteristics of the specialty and by individual lifestyle factors (2). --- Morbidity The incidence of hypertension among doctors was higher, but lower than in the USA, the prevalence of other illnesses had the same ratio than in the average Hungarian population (17,20). --- Self-reported Health Status Female physicians were found to be more conscious of their appearance and health status than males. This can be explained by the different gender attitudes. All of the respondents judged their health status to be better than that of the general population, which is similar to another survey (23). Male physicians rarely participate in regular health screens. About one third of the respondents do not have their own local family doctor. This may be because in Hungary, one can access secondary care or tertiary hospitals without attending the family physician, whereas in other countries cannot (24,25). --- Lifestyle Indicators Females demonstrated better scores for healthy eating, rate of weight gain, and alcohol consumption. The proportion of smokers among physicians was lower than in other countries (7,11,26). --- Limitations It was impossible to find more persons of this cohort. Data was largely obtained using self-reporting questionnaires, thus the possibility of recall bias cannot be ruled out. Other limitations may include missing data in the findings. The relationship between working conditions and drugs prescribed for medical reasons was not evaluated. To our knowledge no other study with such a wide range of questions has been published covering this geographical area and from this time period (27,28). --- CONCLUSIONS Attitudes, orientations, motivations and working circumstances of physicians have changed during the last decades. Working in the medicine field has never been a well-paid job in the Eastern European countries (27). We believe that members of this cohort have chosen medicine rather a profession than a mere job. Although physicians do not always follow their own professional advices, their lifestyle proved a little bit healthier than that of the population at large, especially in women, i.e., their health outcomes, except hypertension, were better. In general, respondents were not satisfied with the financial and working conditions of the recent Hungarian healthcare system. --- Authors <unk> Contributions IR -constructing questionnaire, corresponding with participants, text writing, literature search, final text editing, ESz -analyzing data, ZLmanaging questionnaires, literature search, text writing. --- Conflict of Interests None declared Ethics approval was previously obtained through the regional Ethical Committee (South-Budapest region). This study did not have any funding; only departmental resources were available.

There are few studies from East and Central European countries on health-status, lifestyle and social circumstances of medical professionals. We evaluated data of a cohort of physicians who had graduated 30 years ago in Hungary and compared the data of their professional carrier, life style, health outcomes, and medical specialties. Questionnaires compiled by an expert group and filled in by 208 physicians (83 men and 125 women) were analysed. Men mostly work as surgeons, women were mostly employed as primary care specialists. Women changed their specialty and/or place of work more often than men. Male primary care physicians had more children than women and others specialists. At graduation, most of them had a normal BMI. Since then, a significant increase in weight and BMI was observed in both genders and across all specialty groups. The largest increase in body weight and BMI (mean 5.27) was recorded among female primary care physicians. Recorded physical activity was low in general, with male primary care specialists being most active and female primary care physicians the least. Female doctors in surgical specialties had longer resting time. Male physicians rarely participated in regular health screenings. The incidence of hypertension was higher than the Hungarian national average for that age. About 5% of primary care physicians identified themselves as regular smokers. Abstinence and regular daily alcohol consumption were reported in equal ratio. Burn-out symptoms were rarely experienced. This generation had started its medical profession before the significant progressive changes in the medicine occurred in the last decades. While physicians do not always follow their own professional advices, their lifestyle proved a little bit healthier than that of the population at large, especially for women and their health outcomes, except hypertension, were also better. In general, they were not satisfied with the financial and working conditions of the recent Hungarian healthcare system.

but also in terms of family education provided. Free time available for families is also very necessary in order to fulfill good communication between families and maintain good relationships between parents and children. So that as parents are not only required to fulfill needs in the form of physical (goods and money) but also in terms of family education provided. Free time available for families is also very necessary in order to fulfill good communication between families and maintain good relationships between parents and children. So that as parents are not only required to fulfill needs in the form of physical (goods and money) but also in terms of family education provided. In line with developments in the tourism sector that are supported by gender equality, the table shows the participation rate of the female labor force in the Province of Bali by district/city for 2021-2022. Female TPAK or female labor force participation rate is the percentage of the labor force to the female population aged 15 years and over, where the population aged 15 years and over is part of the workforce. The LFPR aims to indicate the percentage of the working age population that is economically active in a country or region. The higher the TPAK figure, the higher the available labor supply (Bali Province Central Statistics Agency, 2022). Every ten jobs in the world are included in tourism activities (WTTC, 2018). In the 2021 female labor force participation rate, Bangli Regency achieved the highest, namely 79.16 percent, compared to the total TPAK in Bali Province which was only 67.61 percent. This very significant difference in numbers of 11.55 percent makes Bangli a friendly district with the participation of women workers. However, in 2022, Bangli Regency is ranked second highest with an enrollment rate of 79.42 percent, below Karangasem which recorded the highest rate of 82.15 percent, while the total Province of Bali is 69.62 percent. Judging from these data, Bangli Regency has not experienced a decrease in the number of female labor participation rates from 2021 to 2022, but experienced a significant increase of 0.26 percent. Labor force participation refers to the percentage of the working age population involved in economic activities, namely those who are currently working or looking for work. Meanwhile, education level refers to an individual's level of education, which can affect their chances of getting a productive job. It can be seen in the data on the population of Balinese women aged 15 years and over who are included in the workforce according to districts/cities with the highest education level: which may affect their chances of finding a productive job. It can be seen in the data on the population of Balinese women aged 15 years and over who are included in the workforce according to districts/cities with the highest education level: which may affect their chances of finding a productive job. It can be seen in the data on the population of Balinese women aged 15 years and over who are included in the workforce according to districts/cities with the highest education level: Bangli Regency is in the second lowest position with regard to the highest level of education completed. This is because women workers in the informal sector do not need a higher education, but what is prioritized is professionalism and skills in work. With regard to the highest education completed. It is supported that the percentage of women's involvement as professionals in Bangli Regency is in the low category. It can be seen in the data on the involvement of women as professionals in the Province of Bali: The percentage of women involved as professionals involved in several business sectors. Business fields are divided into 17 sectors, in this classification there is no specific mention of business fields in the tourism sector. Because tourism activities are very broad and interrelated with each other. However, tourism activities in general are closely related to activities in the category of business fields providing food and drink accommodation, so that tourism is often approached in this category. However, it is important to note that the category of business fields providing food and drink accommodation does not 100 percent describe tourism. There are other categories of business fields that are also related to tourism, for example trade, transportation, corporate services and other services. And conversely, it should also be noted that what is included in the category of food and drink accommodation provider business fields are not only tourism activities, but also other activities related to tourism. In the table the percentage of women's involvement as a professional workforce consisting of several sectors, namely agriculture/forestry/fishery, mining and quarrying, processing industry, electricity and gas procurement, water procurement/waste/waste management and recycling, construction, wholesale and retail trade ; repair and maintenance of cars and motorbikes, transportation and warehousing, provision of accommodation and food and drink, information and communication, financial and insurance services, real estate, company services, government administration/defense and mandatory social security, education services, health services and activities social, and other services. In the Bangli Regency table, the percentage of women involved as professionals is in a low position, in 2019 with a total of 43.46 percent, in 2020 with a total of 41.79 percent and in 2021 with a total of 42.93 percent. Compared to Bali Province in 2019 it was 48.05 percent, in 2020 it was 49.12 percent and in 2021 it was 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. In the Bangli Regency table, the percentage of women involved as professionals is in a low position, in 2019 with a total of 43.46 percent, in 2020 with a total of 41.79 percent and in 2021 with a total of 42.93 percent. Compared to Bali Province in 2019 it was 48.05 percent, in 2020 it was 49.12 percent and in 2021 it was 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. In the Bangli Regency table, the percentage of women involved as professionals is in a low position, in 2019 with a total of 43.46 percent, in 2020 with a total of 41.79 percent and in 2021 with a total of 42.93 percent. Compared to Bali Province in 2019 it was 48.05 percent, in 2020 it was 49.12 percent and in 2021 it was 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. in 2019 with a total of 43.46 percent in 2020 with a total of 41.79 percent and in 2021 with a total of 42.93 percent. Compared to Bali Province in 2019 it was 48.05 percent, in 2020 it was 49.12 percent and in 2021 it was 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. in 2019 with a total of 43.46 percent in 2020 with a total of 41.79 percent and in 2021 with a total of 42.93 percent. Compared to Bali Province in 2019 it was 48.05 percent, in 2020 it was 49.12 percent and in 2021 it was 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. in 2020 it will be 49.12 percent and in 2021 it will be 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. in 2020 it will be 49.12 percent and in 2021 it will be 50.89 percent. The position of Bangli Regency is considered low according to the data presented. In line with data on labor force participation, Bangli has the highest figure, while in terms of the percentage of women's involvement as professionals, it is in the lowest position. This is the problem in this research and examines women's involvement in the informal sector in tourism. This research is very interesting because it is based on the workforce of female actors in the tourism sector. One of the villages that really preserves culture as a cultural heritage and tourism business sector for future generations, namely Penglipuran Tourism Village in Kubu District, Bangli Regency, Bali Province with an area of 112 hectares, with an area division where fifty hectares of arable land is used as agriculture, forty-five hectares of bamboo forest, four hectares of timber forest, nine hectares of settlements, four hectares of shrines and public spaces. Geographically it is at an altitude of 600-650 meters above sea level, so the temperature is quite cool. The total population of Penglipuran Tourism Village in January 2021 is 1,111 people with a total of 277 families, whose livelihoods include craftsmen, souvenir sellers, agricultural food sellers, homestead managers, laborers, civil servants, tour guides and other tour providers. The Penglipuran Tourism Village community adheres to Hinduism, maintains customs, mutual cooperation values, kinship and local wisdom based on the concept of Tri Hitha Karana (Penglipuran Village, 2022). This village is located in a mountainous area with charming views, beautiful bamboo forests, full of strict customary regulations to ensure the continuity and preservation of customs and culture that have been preserved to this day. Penglipuran Tourism Village continues to make improvements by completing and maintaining infrastructure and facilities so that domestic and foreign tourists can feel comfortable and enjoy visiting the village, thereby supporting the capabilities and expertise of potential human resources to manage the village (Mareni et al., 2022). The advantage of this Penglipuran traditional village compared to other villages in Bali is that the front of the house is similar from the main end of the village to the end of the village. The village has been laid out since the time the village was formed, arranged in such a way that the main area is higher and then slopes down to the downstream area. In addition to the same front, the shape of the material used in the construction of the house is also uniform (Pertiwi, 2016). Penglipuran Traditional Village is a rural area with a special arrangement of traditional village structures to show a beautiful rural face. The physical arrangement of the village structure is inseparable from the popular culture that has been passed down from generation to generation. The beauty of the Penglipura traditional village can be felt from the moment you enter the village area, with the green grass and hedges along the road adding to the coolness of the village area. The icon of Penglipuran Village is a community house with the same yard, between one resident's house and one resident's house on the left or right side, there is no boundary in the form of a dividing wall. Every resident's house has a close relationship with each other because there are no boundaries, boundaries are not an obstacle for not having a good relationship even though they do not have bloodline relatives, residents are not reluctant to one another. So that the formation of social capital that has been invested since the formation of Penglipuran Village. Each house also has the same angkul and pewaregan bale and has not changed since Penglipuran Village was founded. Penglipuran Village customary regulations also regulate the pattern of building community houses which do not allow residents to build terraced pattern houses, this is because in order to preserve Penglipuran Village so that it remains beautiful. Penglipuran Village also has regulations to protect women from polygamy, people who practice polygamy are removed and accommodated in a special place called "Karang Memadu". This term only exists in Penglipuran Village, which is a special part of Penglipuran Village that protects women from polygamy. The development of community-based tourism village areas is in line with the government's principles of sustainable development. Development requires community participation, both men and women. The role of the local community must be the main player in village development (Hamid et al., 2020). The development of the Penglipuran tourist village is very rapid from year to year so that it has a significant impact on the welfare of the community (Yasa, 2017). In this regard, this research attempts to explore more deeply the supply of female workers, especially in the world-famous tourism sector. Penglipuran Village where most of the people use the tourism sector as a livelihood. The main occupation of the people of Penglipuran Village is as fruit traders, traders of lolo cemcem or turmeric, traders of cassava klepon, bamboo handicrafts, souvenir traders and Balinese clothes. The research was conducted in Penglipuran Village because it is a village whose ancestral heritage is still intact and has not changed over time. In this context, it can be said that Penglipuran Village has often even repeatedly faced changes and has been able to maintain its cultural heritage without following the changing times and not leaving cultural values behind. In this regard, this research attempts to explore more deeply the supply of female workers, especially in the world-famous tourism sector. Penglipuran Village where most of the people use the tourism sector as a livelihood. The main occupation of the people of Penglipuran Village is as fruit traders, traders of lolo cem-cem or turmeric, traders of cassava klepon, bamboo handicrafts, souvenir traders and Balinese clothes. The research was conducted in Penglipuran Village because it is a village whose ancestral heritage is still intact and has not changed over time. In this context, it can be said that Penglipuran Village has often even repeatedly faced changes and has been able to maintain its cultural heritage without following the changing times and not leaving cultural values behind. In this regard, this research attempts to explore more deeply the supply of female workers, especially in the world-famous tourism sector. Penglipuran Village where most of the people use the tourism sector as a livelihood. The main occupation of the people of Penglipuran Village is as fruit traders, traders of lolo cem-cem or turmeric, traders of cassava klepon, bamboo handicrafts, souvenir traders and Balinese clothes. The research was conducted in Penglipuran Village because it is a village whose ancestral heritage is still intact and has not changed over time. In this context, it can be said that Penglipuran Village has often even repeatedly faced changes and has been able to maintain its cultural heritage without following the changing times and not leaving cultural values behind. --- II. LITERATURE REVIEW A. Labor Offerings Labor supply is the relationship between wages and the amount of labor offered. Labor offers come from people who are of working age and who are directly involved in the world of work. In this regard, the supply of labor is also inseparable from changes in wage rates, therefore in order to know the effect on the number of hours worked, the concepts of substitution effect and income effect are needed. The substitution effect explains that when wages earned increase, the costs incurred for leisure are greater, or in other words, the opportunity cost of leisure time will increase. The substitution effect concludes that there is a positive relationship between wage rates and hours of work offered. Meanwhile the income effect explains when the wage rate increases, a person's income also increases. The concept of the income effect is negative, meaning that if the wage rate increases, the number of hours offered will decrease (Santoso, 2012). --- B. Tourism Worker Concept Tourism workers are individuals who work in various sectors of the tourism industry which includes travel, leisure and recreation (Tribe, 2020). The main characteristics of tourism workers include friendly demeanor, good communication skills, knowledge of tourist destinations, and flexibility in dealing with various situations. They play an important role in providing satisfactory service and maintaining the sustainability of the tourism industry, which contributes to economic growth and social development in various regions around the world (Page, 2019). --- C. Working Woman Concept Women workers refer to the views and understanding of the role and participation of women in the world of work. This concept covers various aspects, including rights and employment opportunities, challenges faced, traditional roles, issues of gender equality, and the impact of women's role in society and the economy (Bener<unk>a, 2020). Employment Rights and Opportunities: Women workers should have the same rights and opportunities as their male counterparts to get jobs, promotions and recognition based on competence and qualifications (Kabeer, 2019). --- III. RESEARCH METHODS This type of research is descriptive and associative research. This research is descriptive in nature and aims to explain the results of the supply of female workers as tourism actors in Penglipuran Village. This research is associative in nature to determine the effect of several variables on the female workforce actors in the tourism sector have a relationship between the independent variables and the dependent variable. The location of this research was conducted in Penglipuran Village, Bangli Regency. The choice of this location was based on the fact that the female labor force participation rate in Bangli Regency, from 2021-2022, the female labor force participation rate has increased significantly by 0.26%. One of the villages in Bangli Regency, and this village is well known by the world of tourism, namely Penglipuran Village. The attraction of Penglipuran Village is that it has a mixed reef to protect polygamy against women, for people who practice polygamy it is mandatory to live in a mixed reef. The part that is the privilege of women is to avoid polygamy. This makes Penglipuran Village a location with women's relevance. Pengliputan Village has neatly arranged buildings built on the same land area. Inside the house there is a main door which is commonly called angkul which has the same shape and is symmetrical. And synonymous with a unique and beautiful village atmosphere. Penglipuran Village has a concept that makes it look attractive and has a unique appearance, housing is lined up very neatly, the roads are clean and quiet without any motorbikes passing by. Penglipuran Village which is a traditional settlement that is thick with local wisdom. housing is lined up very neatly, the road is clean and quiet without any motorbikes passing by. Penglipuran Village which is a traditional settlement that is thick with local wisdom. housing is lined up very neatly, the road is clean and quiet without any motorbikes passing by. Penglipuran Village which is a traditional settlement that is thick with local wisdom. --- D. Variable Identification In this study, the dependent variable is working hours (Y). Working hours (Y), used to work as a tourism sector worker in Penglipuran Village, Bangli Regency. Working hours are measured in units of hours in one month The independent variables in the study are divided into interest variables and control variables. The interest variables are social capital (X1) and income (X2), while the control variables (Ki) are: education (K1), number of children <unk> 6 years (K2), family income (K3), husband's income (K4), age ( K5), and free time (K6). The social capital variable is measured by empirical data, which means making measurements or observations based on real data or observable facts. And income (X2), the total income the respondent earns during one month from the place of work, is expressed in units of rupiah. Education (K1), which has been taken in this case is expressed in the number of years of education. Number of children under six years (K2), number of dependents of children under six years of age. The number of dependents is expressed in units of person. Family income (K3), the total family income earned in one month from each place of work, expressed in rupiah units. Husband's income (K4), the total income the respondent's husband earns during one month from his place of work, expressed in rupiah units. Age (K5), the age of the respondent expressed in numbers with units of years according to the age of the respondent. Leisure time (K6), free time owned by respondents is measured in units of hours in one month. expressed in rupiah units. Age (K5), the age of the respondent expressed in numbers with units of years according to the age of the respondent. Leisure time (K6), free time owned by respondents is measured in units of hours in one month. expressed in rupiah units. Age (K5), the age of the respondent expressed in numbers with units of years according to the age of the respondent. Leisure time (K6), free time owned by respondents is measured in units of hours in one month. --- E. Population and Sample The population in this study is the population of married female tourism actors in Penglipuran Village. Data collection can be done on a sample or part of the population, but can be considered to represent the entire population. As previously explained, primary data collection was carried out through distributing questionnaires and interviews. The sample used in this study is the female population of tourism actors in Penglipuran Village. In this study used the calculation of the sample according to the Slovin formula. --- F. Analysis Techniques The multiple linear regression analysis technique was applied to this study after the selected data mining approach was carried out and analyzed using Orange. The Linear Regression method in Orange is used to predict the numerical value of a variable based on other variables. The goal of linear regression is to create a model that can be used to predict the value of the unknown variable based on the value of the known variable. The advantage of using Orange analysis is that it can analyze ordinal data and ratio data in one model. --- IV. DISCUSSION --- A. Distribution of Answers The average score on the social capital variable contained in Appendix 4 is 3.4. This figure means that the average respondent chooses neutral and agrees with the social capital variable. The average respondent is neutral and agrees with the statement that network, trust, norms which are social capital are sufficient to become part of the female workforce in the Penglipuran Village tourism sector. Even so, there are also people who disagree with the variable social capital as part of the supply of labor, seeing the changes that have occurred in society. There has been a reduction in klepon and loloh home-based businesses and traders who sell klepon and loloh. Changes occur, seeing as time goes by, there are more and more variations of traders with the aim of attracting tourists to visit. Most people rent clothes, sell souvenirs and preserved food. Changes in the form of exchanging goods for goods, to exchanging goods taken for money. Remembering that you often forget to borrow goods for sale on the other hand. "Back in 2009, early tourism here was good, I still sold Klepon and Loloh. In the past, if my Klepon ran out, I would definitely ask to go to the other side, so I borrowed his stuff first. Later, if there were no visitors, I would go to the place for Klepon or Loloh to pick it up. goods. But seeing the conditions after the pandemic, I stopped selling klepon and loloh because it was quiet and there was more and more competition for traders, so the klepon and loloh that I sold were stale." explained Nengah Cariasih, a snack and instant drink seller at Warung Bu Nengah in Penglipuran Village. (25/5/2023) This contradicts Mrs. Wayan Murni's opinion, the owner of the Pekak Bagus shop. "If I call, the one I sell runs out, I run to Mrs. Suyastini whose house is number 21, there I usually borrow the goods first, then when I buy the goods go to the one who usually makes calls and I'll return them again to Mrs. Suyastini, for example I take 20 packs of klepon and 10 bottles of lol, tomorrow I will return it according to what I took." (26/05/2023). Furthermore, the results of the analysis show that the income earned by the respondents each month is according to the characteristics of the respondents. Respondents' income consisted of several ranges, but the highest income range was Rp. 3,000,001 to Rp. 5,000,000 with 18 respondents. This shows that the income earned is in accordance with the effectiveness of working hours used by respondents. In working it contains working hours with the aim of knowing the time spent working. Working hours are calculated in a month by adjusting income, the respondents' working hours are at most 151-200 hours per month with 29 respondents. The rest, below 151 and above 200 hours per month. This shows that women's working hours vary, looking at the activities they take in their daily lives as housewives. In terms of recent education, it is known that the number of respondents with a high school level is higher than that of other levels of education, namely 32 out of 70 respondents. There are female workers whose final level of education is below that of a senior high school graduate. This means that female workers in Penglipuran Village have various levels of education. In terms of the work that is owned, there is a salary or wages earned, the range of the highest salary or wages obtained by the husband of the respondent is IDR 1,000,001 to IDR 2,000,000 as many as 35 respondents. This is adjusted to the work you do and the position you hold. Income is seen from two sides in terms of women as wives and men as husbands in a family. So if both women and men work, family income increases to cover living expenses. In this study, the highest respondent's family income was in the range of Rp. 5,000,001 to Rp. 10,000,000 for 38 respondents. This shows a high family income because the contribution of women influences family income and improves family finances On the other hand, there are a number of dependents of children under six years old, usually children under six years of age require a fair amount of nutrition to improve their children. In the family it is certain that they have had children under six years old, the respondents here do not all have children under six years. The number of respondents who have children under six years is 15 people with one child under six years and two respondents have two children under six years. Field results show that most of the respondents are aged 41-50 years, namely as many as 26 people or 37 percent. This age is the age group who are experts or professionals in working to do their main job in the tourism sector In addition to time for work, respondents also have free time to relax or take a break from activities that have been carried out during the day. The respondent's free time calculated in a month is adjusted to the work time spent by the respondent, the highest respondent's free time is in the range of 50-100 hours per month. This means that every worker has the right to also have time to rest from working time spent to make ends meet --- B. Model Building This study uses data mining analysis with the software used for this research is Orange version 3.33.0. Based on the results of the regression analysis using the orange tool, it can be seen the results of data visualization using the results of the questionnaire data as shown in Figure 1. --- Figure 1. Visualization of Regression Analysis with Orange The visualization of the regression analysis consists of 12 stages for the analysis of all dependent and independent variables. Consists of 12 widgets, namely widget file, select columns, data table results of rank, rank, select columns linear regression, data table results of linear regression, future statistics, test and score, distance, MDS, scatter plot --- Table 4.1 Rank Analysis Results In the regression analysis using the rank widget on Orange, the results show that the independent variables social capital with univar.reg of 262,469, and RRelief of 0.555 and income with univar.reg of 0.102 and RRelief of 0.277 rank in the top two in their effect on the variable hours of work. this shows that the two variables have a significant effect on working hours. These results indicate that social capital and income have a strong relationship with the number of hours worked. It can be assumed that the higher the level of social capital and income, the more likely a person will spend more time in working hours. Based on the results of the analysis, --- Figure 3 Stages of Regression Analysis Using Orange Figure 4.3 is part of the data analysis workflow process using the linear regression widget to find the relationship between the independent variables and the dependent variable. In the widget used, the algorithm model that will be used in predicting and classifying machine learning on Orange, namely linear regression, can be seen in the following figure the results of linear regression analysis: --- Table 2. Results of Linear Regression Analysis The results of the linear regression analysis after connecting the select columns to the linear regression widget, to see the regression results, you must connect the linear regression widget to the table data to produce a regression analysis. Figure 4.4 is the result of the regression analysis with the number of respondent data of 70 respondents, so that the results of the regression analysis can be obtained. Based on the results of the regression analysis conducted, it was found that there is a positive relationship between working hours and social capital. The regression coefficient of 0.810 indicates that each unit increase in hours worked is associated with an increase of 0.810 units in social capital. This positive relationship indicates that the higher the hours worked, the higher the level of social capital owned. On the income side of the respondents also found a positive relationship between working hours and the income of the respondents, based on the results of the regression analysis conducted. It was found that there was a positive relationship between working hours and respondents' income. A regression coefficient of 0.002 indicates that each unit increase in hours worked is associated with a 0.002 unit increase in the respondent's income. Based on the results of the analysis of the variable working hours with the husband's income, it was found that there was a negative relationship between working hours and the husband's income. The regression coefficient of -0.005 indicates that each unit increase in the hours worked by the respondent or the wife is associated with a 0.005 unit decrease in the husband's income. This negative relationship indicates that the more hours worked by the respondent or the wife, the husband's income tends to decrease. This is caused by several factors, such as the division of tasks and roles in the household, the increase in child care costs or other household responsibilities, or the influence of the dynamics of the work sector of each partner. The results of the regression analysis found that there was a negative relationship between working hours and the number of children under 6 years. A regression coefficient of -12,931 indicates that each unit increase in hours worked is associated with a -12,931 unit decrease in the number of children under 6 years of age. This negative relationship indicates that the more hours worked, the number of children under 6 years tends to decrease. This may happen because more time is spent working, so the time available to care for children is limited. Other factors such as the role of the spouse or child care support systems can also affect the relationship between hours worked and the number of children under 6 years. The results of the regression analysis between working hours and the age of the respondents have a negative effect with a coefficient of -0.571. Which means, if the respondent has long working hours it will have an impact on the age of the respondent, the working hours of many aged workers tend to be younger. This negative relationship indicates that the more hours worked by the respondent, the age tends to decrease. This may be due to a heavier workload or stress experienced by respondents due to long working hours, which in turn can affect the aging process. In the results of the regression analysis between working hours and family income has a negative effect with a coefficient of -0.004. The regression coefficient of -0.004 indicates that each unit increase in hours worked is associated with a 0.004 unit decrease in family income. This negative relationship indicates that the more hours worked, the family income tends to decrease. This happens because there is a time limit spent on the main job so that the respondent cannot work extra outside the main job. In the results of the regression analysis analysis between working hours and education, negative results were obtained and the coefficient result was -1.494. This is because the higher working hours owned by the respondent as a tourism actor does not have an impact on the education held by the respondent. This is because the tourism sector often requires special skills and practical expertise. Such as, mastery of foreign languages, communication skills, expertise in customer service, knowledge of culture and history, and the ability to manage diverse situations. Workers in the tourism sector can develop these skills through training, courses or hands-on work experience. The regression analysis between working hours and free time is negative with a coefficient of -0.106. This is because the higher the working hours the respondents have, the higher the working hours spent by tourism workers, the less free time they have. This may be due to working on holidays as well as maintaining 24-hour service availability in the tourism industry. As a result, tourism workers may have less time to rest, relax and engage in activities outside of work. --- Table 3. Test and Score Results from Linear Regression On the results of the analysis intest and scores widgetin data mining Orange shows several important evaluation metrics, namely MSE (Mean Squared Error), RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), and R2(Coefficient of Determination). The MSE value of 0.672 shows the average of the squared difference between the predicted value and the actual value in the prediction model. The lower the MSE value, the better the quality of the model's predictions. In this context, the relatively low MSE value (0.672) indicates that the prediction model has a low error rate. The RMSE value of 0.9820 is the square root of the MSE value. RMSE measures the degree of prediction error in the model on the same scale as the dependent variable. The lower the RMSE value, the better the quality of the model's predictions. In this context, a relatively low RMSE value (0. 820) shows that the predictive model has a low error rate in the same units as the dependent variable. The MAE value of 0.632 shows the average of the absolute difference between the predicted value and the actual value in the prediction model. MAE measures the level of prediction error in the model regardless of the direction of the error. The lower the MAE value, the better the quality of the model's predictions. In this context, the relatively low MAE value (0.632) indicates that the prediction model has an absolute low error rate. The R2 value of 0.771 can be concluded to be worth 0.717 which shows that 71.7 percent of the variation in working hours can be explained by the variables of social capital, income, education, number of children under 6 years old, husband's income, family income, age, and free time. Meanwhile, 28, The other 3 percent is influenced by other variables that are not included in the research model. With the results of this analysis, it can be concluded that the regression model used is relatively good at predicting the value of the dependent variable. Even though there are still prediction errors, the low MSE, RMSE, and MAE values indicate that the regression model has a low error rate. In addition, the relatively high R2 value indicates that the independent variable is able to explain most of the variation in the dependent variable and low MAE indicates that the regression model has a low error rate. In addition, the relatively high R2 value indicates that the independent variable is able to explain most of the variation in the dependent variable and low MAE indicates that the regression model has a low error rate. In addition, the relatively high R2 value indicates that the independent variable is able to explain most of the variation in the dependent variable --- C.

The location that became the object of research was tourism in Penglipuran Village, Bangli Regency. Based on the results of secondary data, data on the total participation rate of the female workforce, the female population in Bali aged 15 years and over who are included in the labor force, and the percentage of women involved as professionals in the province of Bali. Data collection was carried out using a questionnaire and using regression analysis with Orange. Based on the results of the regression analysis conducted, it was found that there is a positive relationship between working hours and social capital. It was also found that there was a positive relationship between working hours and respondents' income. Regression results with a negative relationship between working hours and husband's income, education, family income, number of children under 6 years old, age, and free time. Working hours can play a role in building social relationships, developing networks, and increasing interaction with other individuals or groups. This affects the accumulation of social capital owned by respondents. In the income variable the number of hours worked by respondents, the income earned also tends to increase. This shows that the additional effort of working more hours can contribute to an increase in the income of respondents.

concluded to be worth 0.717 which shows that 71.7 percent of the variation in working hours can be explained by the variables of social capital, income, education, number of children under 6 years old, husband's income, family income, age, and free time. Meanwhile, 28, The other 3 percent is influenced by other variables that are not included in the research model. With the results of this analysis, it can be concluded that the regression model used is relatively good at predicting the value of the dependent variable. Even though there are still prediction errors, the low MSE, RMSE, and MAE values indicate that the regression model has a low error rate. In addition, the relatively high R2 value indicates that the independent variable is able to explain most of the variation in the dependent variable and low MAE indicates that the regression model has a low error rate. In addition, the relatively high R2 value indicates that the independent variable is able to explain most of the variation in the dependent variable and low MAE indicates that the regression model has a low error rate. In addition, the relatively high R2 value indicates that the independent variable is able to explain most of the variation in the dependent variable --- C. Discussion In accordance with the visualization carried out in the regression analysis using the rank widget on Orange, the results show that the independent variables social capital and income occupy the top two ranks in their effect on the working hours variable, this indicates that these two variables have a significant influence on working hours. These results indicate that social capital and income have a strong relationship with the number of hours worked. It can be assumed that the higher the level of social capital and income, the more likely a person will spend more time in working hours. Based on the results of the analysis, it can be concluded that social capital and income have an important role in influencing the number of hours worked. From the linear regression widget side with the results, that there is a positive relationship between working hours and social capital. The results of the regression coefficients show that each unit increase in hours worked is associated with an increase in social capital. This positive relationship indicates that the higher the hours worked, the higher the level of social capital owned. In this case, working hours can play a role in building social relationships, developing networks, and increasing interaction with other individuals or groups. This affects the accumulation of social capital owned by respondents. This is in accordance with the opinion (Putnam, 2015) that social capital has an important role in increasing the efficiency and effectiveness of economic activities, especially working hours, where in this study as the dependent variable that can improve the quality of life and welfare of the community as well as increase the ability of the community to deal with change and crises. On the other hand, there are factors that influence the occurrence of social capital, namely the social and economic conditions of society, geographical conditions and the social environment, as well as cultural and historical factors (Woolcock, 2013). On the income side of the respondents also found a positive relationship between working hours and the income of the respondents, based on the results of the regression analysis conducted. It was found that there was a positive relationship between working hours and respondents' income. The results of the regression coefficients show that each unit increase in hours worked is related to an increase in the respondent's income. This positive relationship indicates that the more hours worked by the respondent, the income earned also tends to increase. This can be interpreted that the additional effort of working more hours can contribute to an increase in the income of respondents. The results of this analysis are in accordance with the opinion (Mankiw, 2019) that income is one of the main factors affecting workers' working hours. The higher the income, the greater a person's ability to increase working hours. However, there are also other factors that influence increasing working hours, such as family needs or other factors that influence a person's decision to increase working hours. Income has a positive effect on working hours according to the opinion of Todaro and Smith, income is a measure of a person's or a person's ability to produce goods and services, and is an important indicator in evaluating the workforce (Todaro, 2014). With the results of the analysis obtained, it has answered the formulation of the problem and explained the results of the research hypothesis --- CONCLUSIONS The conclusion of this study is that the variable social capital has a positive effect on labor supply and income has an effect on the supply of female labor in Penglipuran Village. Suggestions from this study are that the role of the community is needed in advancing tourism by socializing and educating the people of Penglipuran Village about the concept and benefits of social capital. The information conveyed should include how social capital can contribute to increased employment opportunities and economic development. Then by encouraging active participation in community activities. Communities in Penglipuran Village can be encouraged to actively participate in community activities that focus on developing social capital. This can include gotong royong activities, community meetings, and other activities that strengthen social relations among villagers. On the other hand by increasing access to training and education. Efforts need to be made to increase the access of the Penglipuran Village community to training and education relevant to the world of work. This could involve collaborating with existing educational institutions, job training, or skills development programs in the area. Communities in Penglipuran Village can be encouraged to develop diverse sources of income. In addition to the main job, they can look for side business opportunities or explore local economic potential such as tourism, handicrafts, agriculture or other local products. This can help increase the number of hours worked and income earned. Improvement of skills and education with efforts needs to be made to improve the skills of the people of Penglipuran Village through training and education that are relevant to the needs of the labor market. This can help them find higher paying jobs or increase their entrepreneurial skills. Education can also open access to jobs that require special qualifications or skills --- ACKNOWLEDGMENT On this occasion, the author would like to thank Prof. Dr. dr. Ida Ayu Nyoman Saskara, M.Sc as Thesis Supervisor for her time, guidance, input and motivation during the completion of this thesis thanks to Dr. Putu Ayu Pramitha Purwanti, SE, M.Sc as the Thesis Examiner Lecturer who has provided input, corrections and suggestions for the perfection of this thesis. Do not forget to thank Drs. Sudarsana Arka, MP as Academic Advisor who has provided motivation and encouragement to the writer while attending lectures at FEB Unud., and thanks to the people of Penglipuran Village as respondents in preparing the thesis.

The location that became the object of research was tourism in Penglipuran Village, Bangli Regency. Based on the results of secondary data, data on the total participation rate of the female workforce, the female population in Bali aged 15 years and over who are included in the labor force, and the percentage of women involved as professionals in the province of Bali. Data collection was carried out using a questionnaire and using regression analysis with Orange. Based on the results of the regression analysis conducted, it was found that there is a positive relationship between working hours and social capital. It was also found that there was a positive relationship between working hours and respondents' income. Regression results with a negative relationship between working hours and husband's income, education, family income, number of children under 6 years old, age, and free time. Working hours can play a role in building social relationships, developing networks, and increasing interaction with other individuals or groups. This affects the accumulation of social capital owned by respondents. In the income variable the number of hours worked by respondents, the income earned also tends to increase. This shows that the additional effort of working more hours can contribute to an increase in the income of respondents.

Introduction Freedom of religion in our country is not absolute, adherents may practice worship and beliefs that are freely chosen and respect adherents of other religions to live side by side for generations. This is for religious adherents in carrying out worship and beliefs the government regulates by regulating in the Legislation to maintain religious harmony. The advancement of technology makes it easy for everyone to access and express opinions anywhere and anytime. Without any restrictions, it results in a wide scope with an easy internet connection. The author is interested in discussing examples of blasphemy cases in Muhammad Kace's youtube content. Muhammad Kace's youtube content conveyed something from his words that caused hatred towards Islam. The video shows Muhammad Kace saying that Rasulullah is close to the jinn. Of course, this statement caused problems among the people of Islam. Because spreading information has the effect of causing hatred and hostility between groups of people belonging to various religions to cause disturbances between groups deliberately in public to cause hostility. The author formulates the problem in the research conducted and the problem formulation is as follows: a. How were the blasphemous activities carried out by Muhammad Kace? b. What is the legal regulation of blasphemy committed by Muhammad Kace? --- Research Method The author of the journal chose the normative method. The author chose to use the normative research method because the research used a statutory approach. According to (Marzuki, 2017) in his book explains the normative method, a statutory approach method is carried out to understand all laws and regulations related to the legal issues handled. According to (Marzuki, 2017), the normative method can use primary legal materials and secondary legal materials. The author collects research materials through literature studies starting with classifying legal materials, reading systematically the legal materials available as a guide to answering the formulation of the problem in analyzing the research the author uses the interpretation method, the method begins with the opinions that exist in these problems interpreted through the interpretation of language, laws or systematically about the problem to be discussed, by looking at legal concepts from a normative point of view (Sidharta, 2009). --- Results and Discussions --- How Were The Blasphemous Activities Carried Out by Muhammad Kace? The blasphemy activities carried out by Muhammad Kace include one of the acts carried out by broadcasting and distributing news to the wider community. In the activities carried out to broadcast and disseminate Electronic Information or Electronic Documents to the wider community through electronic media. What Muhammad Kace did cause a sense of hatred or hostility of individuals and or certain community groups based on ethnicity, religion, race, and intergroup in this activity the perpetrator wanted an effect that might result from his actions. This has the potential to cause a sense of hatred or hostility in the spoken word causing it to be an indicator of conveying the intended meaning. The dissemination of this information expresses hostility, causes hatred, and even humiliation of certain groups in society. Of course, the task of the police in law enforcement is very important to handle this example of blasphemy. Muhammad Kace made a video on his personal YouTube account containing his views on the teachings of one religion, namely Islam, by calling the Prophet close to the jinn and conveying translations of verses from the Koran, which was protested and condemned by the Islamic Ulema Council. In his broadcast, Muhammad Kace also made it clear that whomever Allah's people pray, the jinn swarm. Some experts argue that blasphemy committed by Muhammad Kace through YouTube social media is included in cybercrime by the opinions of several experts quoted by the author. According to (Wulandari, 2021), cybercrime is a threat to stability, making it difficult for the government to keep up with crime techniques that have been carried out with computer technology, especially internet and intranet networks. In addition, (Tuanany, 2021) argues that it is a result of the rapid development of information technology so that every development essentially has an effect like two sides of a coin, each of which is related and inseparable from the positive and negative sides. Another thing with (Pasedan, 2015) opinion is that the development of technology and media has caused all forms of insults that were once only possible manually, to now occur through electronic means. Forms of defamation can occur through electronic means by disseminating information through cyberspace networks social networks (Facebook, Twitter, email, and BBM), through sites and websites that can be accessed by anyone. Another case with (Satria, 2014) opinion defines cybercrime as a cybercrime that uses computer facilities or through electronic networks linked to the internet that is illegally prohibited by certain parties and violates a criminal provision. --- What is The Legal Regulation of Blasphemy Committed by Muhammad Kace? The Indonesian Ulema Council reported Muhammad Kace with his YouTube account containing blasphemy of Islam. Muhammad Kace's actions are under Article 45A paragraph (2) Jo Article 28 paragraph (2) of Law Number 19 of 2016 which contains information and electronic transactions contains every person intentionally and without the right to disseminate information aimed at causing hatred or hostility of individuals and/or certain community groups based on ethnicity, religion, race, and intergroup (SARA). Muhammad Kace was sentenced to a maximum of 4 years and a maximum of 6 years in prison (Adare, 2013). According to the author, from several sources read and then summarized by the author, the blasphemy committed by Muhammad Kace is an act included in the provisions of Article 28 paragraph (2). Quoted in Law Number 19 of 2016 concerning Amendments to Law Number 11 of 2008 concerning Information and Electronic Transactions contains: 1. Parties who are harmed by the actions of a person or group of people are associated with the element of SARA, 2. The act of displaying images of persons sanctified by a religion that is contrary to the original images 3. Making writings that demonize the contents of the holy book of a religion that is different from the teachings of that religion 4. Disseminating matters of a personal nature contrary to or violating the norms of decency and morality The main offense of Article 28 paragraph (2) of the ITE Law is the act of disseminating information that creates a sense of hatred or hostility towards individuals or community groups based on ethnicity, religion, race, and intergroup (SARA); Defamation of Religion (SARA) can be in the form of information that is disseminated in the form of images, videos, sounds, or writings that are meaningful and contain elements of SARA. If a person invites and broadcasts others to have a sense of hatred and hostility towards individuals or groups of people based on SARA issues Whatever the form if it is in the form of uploading on a social media account setting can be accessed by the public, or broadcasting something on a group conversation application with an open nature where anyone can join the group conversation, the traffic of content or information is not controlled, anyone can upload and share with other words without another group (Wahid & Labib, 2005). The actions prohibited in this article have the motive to arouse hatred based on SARA. Law Enforcement Officials must prove the motive for arousing which is characterized by the presence of content inviting, influencing, mobilizing the public, and inciting divisions to cause hatred, and or hostility. In addition, between groups of people outside of ethnicity, religion, and race the definition of intergroup refers to the Constitutional Court Decision Number 76/PUU-XV/2017. Meanwhile, on the other hand, the delivery of opinions, statements of disagreement or dislike for individuals or groups of people are not included in the prohibited actions, unless it can be proven that there is an invitation, influence, and mobilize the community, inciting to complain to cause hatred or hostility based on the issue of SARA differences sentiment. On the other hand, according to the author, from several references read, quoted under the rules contained in several explanations above, the acts studied by the author have a relationship that can be regarded as one continuous act as stipulated in Article 64 paragraph (1) of the Criminal Code regulates concurs which is quoted and reads that a person commits several acts, each of which constitutes an offense crime, between which there is such a relationship that it must be considered as one continuous act. Meanwhile, on the other hand, the acts committed in the continuing act have not yet been decided by the judge. According to the author, this article can be related to the consequences that must be accepted in the Muhammad Kace case. According to Article 64 of the Criminal Code, in principle, the absorption system applies, namely that only one criminal regulation is imposed, and if it is different, one criminal regulation is imposed, and if it is different, the provisions containing the heaviest basic criminal punishment are imposed. In this case, Article 45A paragraph (2) Jo Article 28 paragraph (2) of the ITE Law stipulates that the act is punishable by a maximum imprisonment of 6 (six) years and/or a maximum fine of Rp 1,000,000,000.00 (one billion rupiahs). Article 156a letter a of the Criminal Code carries a maximum penalty of 5 years and Article 156 of the Criminal Code carries a maximum penalty of 4 years, so in this case, the heaviest penalty is applied, namely a maximum imprisonment of 6 years. --- Conclusion From the author's writing above, it can be concluded that Muhammad Kace is proven to have committed blasphemy (SARA), this can be proven from several statements delivered from his personal YouTube account. In his youtube account that the author researched, Muhammad Kace said that if there are people of Allah praying, they are in league with the jinn. This was reported by the Indonesian Ulema Council. In this case, it can be related to one of the articles read by the author, namely article 45A paragraph (2) Jo article 28 paragraph (2) and if proven guilty under existing regulations Muhammad Kace must be willing to carry out all the consequences of the statement delivered.

The variety of cultures and religions in Indonesia has led to a variety of conflicts. The diversity of cultures and religions sometimes leads to conflicts among various communities. Since the advancement of globalization, one of which is in the field of technology, there have been many cases of deviance in society. Advances in globalization in the field of technology provide a way for people to access anything using the internet. One example is the emergence of deviant content such as religious content containing blasphemy (SARA). As technology advances, everyone can access and express their opinions easily. Recently, a case was shocked by a person named Muhammad Kace through his lecture regarding his views on his YouTube account on Islam assuming that the Prophet was close to the jinn. Muhammad Kace in his YouTube channel conveys content that certainly contains hate speech and contains information that can invite chaos among the public and contains blasphemy against Islam. From the description presented by the author, the author is interested in studying the case of blasphemy through social media Muhammad Kace. In writing the journal, the author uses the normative method. The purpose of writing a journal is that people are careful in using and conveying things in public because it can result in conflicts of different views from the people who receive the information conveyed. Especially conveying religious information is so sensitive because it results in offending many parties and results in violating the rule of law. https://badanpendapatan.riau.go.id/home/hukum/8495315769-doc-20170202-wa0015.pdf

Introduction This study aimed to investigate the association between non-suicidal self-injury (NSSI) and psychological symptoms, victimization, interpersonal problems and perceived social support in individuals with a diagnosis of transexualism. Transsexualism is a formal diagnosis of the International Classification of Disease (ICD-10) [1] which describes individuals who present with discomfort or distress caused by the discrepancy between a person's gender identity (their psychological sense of themselves as men or women) and the sex they were assigned at birth (with the accompanying primary/secondary sexual characteristics and/or expected social gender role). Sometimes, the distress is sufficiently intense that individuals undergo transition from one point on a notional gender continuum to another -most commonly from Male-to-Female (people known as trans women or trans female) or Female-to-Male (people known as trans men or trans male). This typically involves changes to social role and presentation, and may necessitate their taking cross-sex hormones and/or having sex reassignment surgery [2,3]. Cisgender people (as a complement to trans people) are individuals who have a match between the gender they were assigned at birth, their bodies, and their personal identity. The ICD-10 diagnosis for transsexualism is under-review. It is proposed that the ICD-11 recognizes individuals who do not identify as either male or female and are therefore not part of the gender binary (such as gender queer) [4]. This has been acknowledged in the DSM-5, where the diagnosis of gender dysphoria incorporates all individuals on the gender spectrum [5]. Non-suicidal self-injury refers to the direct and intentional injury of one's own body tissue without suicidal intent, such as cutting, burning, and hitting oneself [6] and has been found to function as a way to regulate emotions and self-punishment [7][8][9]. NSSI was recently included as a Non-Suicidal Self-Injury Syndrome in section III of the DSM-5 [5]. While in adolescents, the prevalence rates of NSSI range between 14-39% in the community and between 40-61% in psychiatric samples [10,11], for adults, the prevalence rates of NSSI are estimated to be around 4% in the community and around 21% in psychiatric units [11], indicating a decrease in NSSI behavior with age. In the general population, the prevalence of NSSI in women and men is rather similar; whereas in the psychiatric population NSSI is more prevalent in females compared to males [12,13]. Recent research suggests that lesbian, gay, bisexual, and transgender (LGBT) individuals are at greater risk for mental health problems, including suicidal and non-suicidal self-injury [14][15][16][17]. Mustanski and Liu [17] investigated a sample of 237 LGBT youths (age range 16-20 years; 8.8% self-identified as trans) and reported that 11 (52.4%) out of 21 trans individuals engaged in a suicide attempt during their life-time and 4 (19%) during the past year. Factors which were related to a life-time history of attempted suicide were: hopelessness, depression, conduct disorder symptoms, impulsivity, victimization, age of first same-sex attraction, and low family support [17]. Concerning NSSI, Walls et al. [18] investigated a sample of 265 LGBT youths (age range 13-22 years; 4.9% self-identified as trans) and reported that 6 (47.2%) out of 13 trans individuals in their study of 265 LGBT youths stated that they had engaged in self-cutting behavior during the past year, with younger participants more likely to have engaged in NSSI than older adults (56.5% vs. 40.8%). Additionally, NSSI was found to be more common in trans men than trans women, with cutting as the most common type of NSSI found, particularly in the arms and wrists [19]. Nickels et al. [20] investigated the functionality of self-cutting in lesbian, gay, bisexual and trans (LGBT) individuals and found that, similar to non-trans people, emotional release, stimulation, and self-punishment were the most common functions of NSSI in trans individuals. Factors that are known to increase the risk of engaging in NSSI in the general and psychiatric populations are childhood traumatic experiences, psychological symptoms (such as anxiety, depression), body dissatisfaction, identity problems, and interpersonal dysfunction [21][22][23]. Traumatic experiences and the subsequent psychological reaction can cause stress, causing emotions such as anxiety and depression, and behaviors like NSSI and suicidal behaviors (i.e., stress model) [24]. Protective factors that appear to prevent individuals from engaging in NSSI are social support from parents and friends (i.e., buffering hypothesis of social support) [24], as well as good coping skills [26,27]. Similar findings were reported in a sample of LGBT youth. For example, Walls et al. [18] showed that peer victimization, homelessness, and depression increased the risk of engaging in self-cutting, whereas higher age and having knowledge of a supportive adult decreased the risk. Additionally, House et al. [15] found that experiences of interpersonal trauma and sexual discrimination were associated with increased levels of both non-suicidal and suicidal behavior in a sample of LGTB adults (age range 18-80 years; 14.6% transgender). The aforementioned studies largely consisted of small numbers of self-identified trans individuals and did not control for stage of transition or gender (natal or desired gender). They are, however, the only studies investigating psychological and interpersonal symptoms in trans people with and without NSSI. --- Aims Therefore, the aim of the current study was to investigate the rates of NSSI in a large sample of trans individuals, who have received a diagnosis of Transsexualism according to the ICD-10 [1] and who were at the initial stage of their transitional process, therefore they had not been treated with cross-sex hormones by the clinic. The study will investigate known factors associated with NSSI behavior in the general population such as psychological symptoms, victimization, interpersonal functioning, and perceived social support and will compare those variables between trans individuals who do and do not engage in NSSI, taking into account gender status (trans men/trans women), and the interaction between the presence or absence of NSSI and gender status. Based on the literature regarding NSSI and transsexualism, it is hypothesized that NSSI will be more prevalent in trans men and younger participants; will be displayed primarily on hands and arms; will be triggered by negative affect [15,18]; and will serve affect-regulation and self-punishment functions [20]. Additionally, it is hypothesized that individuals with NSSI will show higher levels of psychopathology, victimization, interpersonal problems and less perceived support from others [21,23]. Comparing trans women with trans men, it is hypothesized that anxiety and depression will be higher in trans women [28][29][30][31][32]. However, with respect to victimization, interpersonal problems and perceived social support no clear hypotheses with respect to gender are made, nor is a hypothesis made for the interaction effect of NSSI and gender on the aforementioned measures. --- Methods --- Participants The sample consisted of all individuals who completed the assessment process at a national gender identity clinic in the United Kingdom between January and December 2013. All consecutively admitted individuals who fulfilled diagnostic criteria for transsexualism (F64.0) as defined by the ICD-10 [1] were included in this study. All individuals completed a battery of questionnaires prior to their clinical assessment. In order to receive a diagnosis, individuals were assessed by two independent senior clinicians with experience in the field of transgender health and who worked at the national gender identity clinic. Following independent assessment, both clinicians reach an ICD-10 diagnosis [1]. In case of disagreement regarding a diagnosis of transsexualism, a third assessment was carried out by another senior clinician. Only individuals who fulfilled diagnostic criteria for transsexualism made by two independent clinicians were included in this study. The study was approved by the NHS local research ethics committee. --- Main Outcome Variables Self-cutting and its characteristics were assessed by means of the Self-Injury Questionnaire (SIQ) [33]. Participants were asked whether they had ever deliberately cut themselves (yes/no) and if they had, how long ago they last did this (in the last week, month, several months ago, more than a year ago, or never). If they injured themselves during the last week or month, they were also asked to indicate which body parts were injured; how many days/month and times/day the cutting occured; and how often and how much pain they felt during the cutting. Additionally, they had to specify what kind of emotions they felt immediately before and after the self-cutting (e.g., relief, anger, sadness), as well as the functions of the cutting on a 5-point Likert scale ranging from 1 ('not at all') to 5 ('very much'). Finally, they also had to indicate whether they planned the behavior, whether they realized how it came about, if they took care of their wounds and if they hid the cutting from others [33]. The Symptom Checklist-90-Revised (SCL-90-R) [34] is a 90-item self-report symptom inventory to measure psychological symptoms and psychological distress. Items are rated on a 5point Likert scale ranging from 0 ('Not at all') till 4 ('Extremely'). The SCL-90-R assesses psychological distress in terms of nine primary symptom dimensions and three summary scores (global scores). The symptom dimensions are labeled Somatization, Obsessive-Compulsive, Interpersonal Sensitivity, Depression, Anxiety, Hostility, Phobic Anxiety, Paranoid Ideation, and Psychoticism. The global measures are referred to as Global Severity Index (GSI), the Positive Symptom Distress Index (PSDI), and the Positive Symptom Total (PST). Good construct validity and reliability has been established for the SCL-90-R [35] and it has been used frequently across the literature on trans individuals [36][37][38][39][40]. The Rosenberg Self-Esteem Scale [41] is a self-report measure of global self-esteem. Items are rated on a 4-point rating scale ranging from 0 ('Strongly disagree') to 3 ('Strongly agree'). Its total score is calculated by summing the item scores with higher scores indicating higher self-esteem. The Hamburg Body Drawing Scale (HBDS) [42] is a pictorial measure that asks participants to indicate how (dis)satisfied they are with specific parts of their body. Thirty-three different body characteristics (e.g., voice, skin, arms, chest or breasts) are identified on a body drawing and participants are asked to rate each on a 5-point Likert scale ranging from 1 ('very satisfied') to 5 ('very dissatisfied'). In addition, they are asked to give a rating for their overall whole body (dis)satisfaction and "other", if applicable. The Experiences of Transphobia Scale [43] assesses experiences of discrimination or victimization on the basis of gender identity or gender presentation. The questionnaire was based on the Transgender Violence Study and measured people's lifetime experiences of violence and harassment and experiences of any form of economic discrimination as a result of being trans (e.g., verbal abuse, physical abuse, fired from a job, problems getting a job, and problems getting health or medical services due to gender identity or presentation). All 5 items are to be rated on a 4-point Likert scale ranging from 0 ('Never') to 3 ('Several times'). The Inventory of Interpersonal Problems-32 (IIP-32) [44] consists of 32 items to be rated on a 5-point Likert scale ranging from 0 ('Not at all') to 4 ('Extremely'). It consists of eight subscales of interpersonal problems: Hard to be Assertive, Hard to be Sociable, Hard to be Supportive, Hard to be Involved, Too Dependent, Too Aggressive, Too Caring, and Too Open. Higher subscale scores indicate greater interpersonal difficulties. The IIP-32 is a shortened version of the original questionnaire, yet the psychometric properties are retained; a confirmatory factor analysis demonstrated high reliability with alpha coefficients of.70 to.88 [44]. The Multidimensional Scale of Perceived Social Support (MSPSS) [45] is a 12-item, self-report scale designed to tap social support from family, friends, and significant others. Items are rated on a 7-point Likert scale ranging from 1 ('very strongly agree) to 7 ('very strongly disagree'). The instrument includes three subscales to address these three types of support (family, friends, significant others). The mean total and subscale scores range from 1 to 7, and a higher score indicates greater perceived social support. --- Analyses All data analyses were performed by means of SPSS 22 [46]. To determine the prevalence of NSSI and its characteristics descriptive statistics were used. To investigate the association between NSSI and gender, the Chi-Square statistic was calculated. MANCOVAs were applied with psychological symptoms, victimization, interpersonal problems, and social support as dependent variables and the presence or absence of NSSI and gender (trans men/trans women) and their interaction as independent variables, controlled for age. Finally, a logistic regression analysis was performed to investigate which variables predicted the presence or absence of NSSI in trans participants when taking all variables into account. The level of significance used was p <unk> 0.05. --- Results During the recruitment period 272 individuals were referred to the clinic, of whom 155 (57%) completed their assessment and fulfilled the diagnostic criteria for transsexualism according to the ICD-10 [1]. One hundred and seventeen individuals were excluded, of whom 31 (11.4%) did not attend, 42 (15.4%) did not have their assessment completed during the study period, 15 (5.5%) had socially transitioned and were on cross-sex hormone treatment before referral; they were referred to the clinic for a second opinion regarding sex reassignment surgery (SRS), and 29 (10.6%) did not fulfill a diagnosis of transsexualism. Of the total sample, 52 (33.5%) participants were trans male and 103 (66.5%) were trans female. The socio-demographic variables of the trans men and trans women are displayed in Table 1. The mean age of the participants was 34.52 years (SD = 14.21). Trans men were significantly younger than trans women [F(1, 153) = 49.94, p <unk>.001]. With respect to civil status, the majority of participants were single (64.7%) or divorced (13.1%, TF>TM). Concerning employment level, the majority of the participants were employed (44.5%), followed by unemployed (22.6%, TF>TM) or student (13.5%, TM>TF). The mean age of coming out was 27.15 years (SD = 13.61), with trans men coming out at a younger age than trans women [F(1, 119) = 30.54, p <unk> 0.001]. --- NSSI in trans men and trans women Overall, 36.8% (N=57) of the trans participants had engaged in NSSI (i.e., self-cutting), whereas 63.2% (N=98) never had. Participants who engaged in NSSI (M age = 26.98, SD = 9.77) were significantly younger than participants who did not report engaging in NSSI [(M age = 38.91, SD = 14.58), F(1, 153) = 30.20, p<unk>.001]. Additionally, a significant association was found between gender (trans men versus trans women) and the presence or absence of NSSI. The results showed that 30 of the 52 trans men (57.7%) and 27 of the 103 trans women (26.2%) had a history of engaging in cutting. Based on these results, it can be concluded that the prevalence of life-time NSSI is significantly higher in trans men compared to trans women (<unk> 2 (1) = 14.73, p <unk>.001). Overall, 13 participants (10 trans men and 3 trans women) reported to have engaged in NSSI very recently (within the last week or month) and these participants completed the second part of the questionnaire which collects more detailed responses about their NSSI behaviour. --- Specific NSSI behaviors Of the 13 participants who had recently engaged in NSSI, most injured their arms, hands, fingers, and nails (61.5%), followed by their legs, feet, toes (15.4%) or a combination of different body parts (2.,1%). With respect to NSSI frequency, the majority of the participants engaged in self-cutting 1-5 days a month (75%) and 1 to 2 times a day (58.3%). Self-cutting was often preceded by feelings of sadness (M = 4.36, SD = 0.92), anger at oneself (M = 4.18, SD = 1.25), anxiety (M = 3.45, SD = 1.75), and anger at others (M = 3.36, SD = 1.63). After self-cutting, feelings of relief (M = 3.83, SD = 1.26), sadness (M = 3.50, SD = 1.88) and anger at oneself (M = 3.25, SD = 1.66) were most prevalent. The most important functions of NSSI were "To avoid/suppress painful negative feelings" (M = 3.67; SD = 1.67), "To avoid/suppress painful images/memories" (M = 3.64, SD = 1.63), "To avoid/suppress suicidal thoughts" (M =3.42; SD = 1.38) and "To punish myself" (M = 3.33, SD = 1.67). The majority of participants did not plan their self-cutting (75%) and were often aware how the self-cutting came about (so no dissociative state; 41.7%). Finally, 50% of the participants'seldom' or 'never' took care of their wounds and 83.3% hid their wounds from other people. --- Clinical Symptomatology With respect to clinical symptomatology, analyses of the 155 trans individuals showed a significant main effect for NSSI [Wilks' <unk>=.85, F(10,140) = 2.50, p <unk>.01] and for gender [Wilks' <unk> =.86, F(10,140) = 2.29, p <unk>.05], without a significant interaction effect [Wilks' <unk> =.97, F(10,140) = 0.37, ns]. Trans women reported significantly higher scores on paranoid ideation, interpersonal distrust, anxiety, depression and obsessive-compulsive complaints compared with trans men. Moreover, trans individuals who reported engaging in NSSI had significantly higher scores on all psychological symptoms, as measured by the SCL-90-R, compared to participants without NSSI (see Table 2). On the Rosenberg Self-Esteem Scale, there was only a significant main effect for gender, but not for NSSI, which indicates that trans women reported a significantly lower level of selfesteem compared to trans men. The analysis also showed that trans individuals with and without NSSI did not significantly differ with respect to self-esteem (see Table 2). Finally, with respect to the total body dissatisfaction score, the results showed main effects for NSSI and for gender, but no significant interaction effect (see Table 2). Trans women reported significantly more body dissatisfaction compared to trans men based on the global HBDS score. On the item level, trans women disliked their hair, armpit hair, pubic hair, feet, forehead, nose, chin and skin significantly more than trans men. Additionally, trans participants who engage in NSSI reported significantly more body dissatisfaction than those without NSSI. On the item level, participants who engage in NSSI disliked their voice, chin, and (upper) arm more than participants who did not engage in NSSI. --- Transphobia On the transphobia total score, there was no significant main effect of NSSI or gender, or a significant interaction. However, on the item level, there was a significant interaction between NSSI and gender with respect to physical abuse (see Figure 1). In trans men, the probability to have been a victim of physical abuse is similar in the NSSI and non-NSSI groups; however, in trans women the probability to have been a victim of physical abuse is much higher in the NSSI compared to the non-NSSI group [F(1,147) = 6.36, p <unk>.05] (Table 3). --- Interpersonal Problems With respect to interpersonal problems, there was a significant main effect for NSSI [Wilks' <unk> =.90, F(8,141) = 1.88, p <unk>.06] and gender [Wilks' <unk> =.86, F(8,141) = 2.52, p <unk>.05], but no significant interaction [Wilks' <unk> =.96, F(8,141) = 0.75, ns]. Specifically, trans women were found to find it harder to be assertive, were too dependent and too caring as compared to trans men. Furthermore, trans individuals with NSSI reported that they find it harder to be assertive and sociable and that they are more aggressive than participants who do not engage in NSSI (see Table 3). --- Perceived Social Support With respect to perceived social support, there was only a significant main effect for gender [Wilks' <unk> =.95, F(3,148) = 2.55, p <unk>.06] (see Table 3). Trans men received significantly more social support from family and significant others than trans women. With respect to NSSI, participants with NSSI reported less support from their family compared to patients without NSSI (p <unk>.07). --- Prediction of NSSI in trans participants A logistic regression analysis was performed to find out which variables predicted the presence of NSSI, while taking into account all variables under study (which differentiated participants with and without NSSI). Overall, a younger age, being trans male and reporting more clinical symptoms (psychoneuroticism) were the significant predictors of NSSI in trans participants (see Table 4). --- Discussion The sample of the present study consisted of individuals referred to a national gender clinic who fulfilled diagnostic criteria for transsexualism according to the ICD-10 [1] and who were at the initial stage of their transitional process. The study investigates whether psychological symptoms, victimization, interpersonal problems, and (lack of) perceived social support in the studied population were significantly related to gender and the presence or absence of NSSI. This is the largest study of its kind to date, with 155 trans participants, of whom 66.5% were trans women and 33.5% were trans men. The study found that with regards to psychological symptoms, trans women reported significantly more intrapersonal (e.g., anxiety, depression) and interpersonal symptoms (e.g., interpersonal sensitivity) as well as lower levels of self-esteem and high levels of body dissatisfaction compared to trans men, in line with findings of previous studies [28][29][30]32,47,48]. The high levels of psychological symptoms found in trans women may be a reflection of the lack of perceived social support from family and significant others compared to trans men, as found in this and other studies [14,49]. Although this can be explained by the fact that cis males generally receive less social support from significant others than cis females [49,50], it is likely to be the result of the high levels of stigmatization of gender nonconformity among men compared to women [48]. Trans women (even more than trans men) face systematic devaluation as a result of social stigma attached to their gender nonconformity. This stigma will increase rates of psychological distress in trans individuals (minority stress model) as well as these individuals' body dissatisfaction, which can partially be moderated by social support [48], also called the buffering hypothesis of social support [25]. The higher level of stigmatization in trans women, combined with a lower level of social support, can thus partially explain the higher level of symptomatology in trans women. In relation to body dissatisfaction the study found that trans women were particularly dissatisfied with their (armpit/pubic) hair, forehead, nose, chin and skin. This could be related to the fact that these facial features tend to be more prominent in cis males and are therefore visible features of their birth gender and may be a source of concern with regard to successfully passing as a member of their desired gender. Moreover, treatments to transition, including hormone treatment and sex reassignment surgery do not exert a major change in these facial features. Furthermore, the internalisation of the "thin ideal" in our society (the message communicated through media that a woman equals thinness) may affect the way trans women see themselves as individuals. This can make trans women particularly susceptible to develop body dissatisfaction, which may affect the way they internalize themselves as female. As a consequence due to their desire to not only achieve a congruent body, but also one that conforms to the socially promoted thin ideal and potentially improve their ability to pass in their expressed gender, they may be more critical about their bodies than trans men. Interestingly, in spite of the high levels of psychological symptoms among trans women, the study found that cutting was more prevalent among trans men, a finding also recently reported by Skagerberg et al. [19] when investigating a population of children with gender dysphoria. Cutting has also been found to be significantly more prevalent among cis women than cis men [51] which may indicate that expressions of distress by both trans men and trans women follow the pattern according to birth gender. Of the 155 trans participants, 36.8% reported a lifetime history of self-cutting behavior and NSSI was more prevalent in younger participants. A closer inspection of the self-cutting behavior showed that most participants cut their arms [19] and injured themselves 1 to 5 days a month. Anger and sadness were found to most often preceded self-cutting, followed by relief after self-cutting. The most important functions of selfinjury were to suppress/avoid negative thoughts/images, negative feelings and suicidal thoughts, followed by punishing oneself. This is consistent with other research that has found that emotionregulation and self-punishment are often described as the most important functions of NSSI in general populations [8,9] as well as in LGBT populations. For example, Nickels et al. [20], in the only study investigating NSSI in a sample that included trans participants, found that, overall, in LGBT people emotional release was the most important function of NSSI, followed by stimulation and self-hate. However the study only included 13 self-diagnosed trans individuals. Additionally, NSSI as a way to express self-hate or self-punishment, can also be related to internalized feelings of shame, which is considered as a potential psychological outcome of stigma [52]. Given that transgender individuals often experience stigma (due to gender nonconformity), NSSI can be a way to express their feelings of shame. According to Longhofer [52] behavior which is clearly associated with shame, "is the inability to respond, or to respond in exclusively bodily ways (i.e., sweating, blushing) (p. 300), and maybe also include NSSI. Moreover, the fact that NSSI is also used as a way to suppress/avoid suicidal thoughts is not surprising, given that many trans people report suicidal thoughts and behaviors [15,[53][54][55]. This study also investigated whether NSSI status and its interaction with gender significantly influenced psychological and interpersonal problems. With respect to psychological symptoms, trans people with a history of NSSI scored significantly higher on all intra-and interpersonal symptoms compared to people without NSSI. Trans participants with NSSI also disliked their body more than those without NSSI. It may indicate that those trans individuals who experience higher body dysphoria cope with these feelings through self-injury. However, the cross-sectional nature of this study does not preclude a conclusion to be drawn regarding the direction of effects; that is, whether the negative attitude towards their body precedes or follows the NSSI (or both). This is a worthwhile area of further study as it may potentially lead to the development of specific therapeutic treatment aiming at decreasing self-harm and increasing coping skills. With regard to interpersonal functioning, both the level of perceived social support and interpersonal functioning was investigated in this study. Trans women reported finding it hard to be assertive and being too dependent and too caring. Those characteristics may make individuals more vulnerable to develop psychological problems, particularly when going through the transitional process as it requires good interpersonal skills to deal with change. This may indicate that interventions aiming at improving interpersonal skills may help trans individuals at risk of developing psychological problems when going through the transitional process [56]. When examining trans individuals with NSSI, the study found that they reported less social support from family members compared to individuals without NSSI. Moreover, trans people with NSSI reported finding it harder to be assertive and social, and were found to be more aggressive compared to trans people without NSSI. It is possible that an aggressive style of interaction and aggressive behavior develops as a means of self-protection and as a result of being at the receiving end of others' unsupportive or hostile reactions. Indeed, if experiences of harassment are linked to these findings, trans women with NSSI are found to have a higher probability of having experienced physical abuse due to trans status than those without NSSI. This study is the first to systematically investigate, in detail, the associations between trans people, NSSI, and intra-and interpersonal problems. The findings are important in highlighting the different experiences and coping mechanisms of trans women and trans men. The sample consisted of trans individuals who were at the start of the process of gender transition and therefore these findings may not reflect the experiences of trans people at different stages of their transition or individuals who have transitioned many years ago. Clinical experience suggests that NSSI behavior reduces as individuals move through the transitional process towards expressing one's gender identity. Future studies may want to investigate this further by comparing NSSI behavior at different stages of the transitional process, which may include treatment with cross-sex hormones and sex reassignment surgery [2,3,57]. Only one type of NSSI (cutting) was assessed and although evidence shows that this is the most prevalent form of NSSI, a more in depth study including other forms of NSSI in this population may be warranted. Finally, only the lifetime prevalence of NSSI was studied, due to a small number of recent NSSI behaviors. Future studies, therefore, need to include trans individuals with recent as well as lifetime NSSI. Overall, this study highlights that the treatment of trans individuals needs to take into account their complex intra-and interpersonal functioning. In particular, the fact that trans women have higher levels of general psychopathology and interpersonal problems. Young trans men show high levels of engaging in NSSI behaviour. Expressions of distress by both trans men and trans women tend to follow the patterns of birth gender and future research could look what effect hormone treatment would have on NSSI behaviour as hormones can affect emotional expression. Psychological treatments should be tailored to each individual bearing in mind the aforementioned findings. Since NSSI was found to be associated with an increased level of psychopathology, body dissatisfaction, and perceived lack of social support, clinical interventions, including family therapy aiming at increasing social support and at improving body satisfaction may reduce NSSI behavior and improve outcomes in this extremely vulnerable group of individuals. Many trans people would like family therapy to be part of their treatment pathway, although very few adult gender identity clinic services offer this treatment modality [58,59]. --- Conclusions Trans individuals with NSSI reported more psychological and interpersonal problems and perceived less social support compared to trans individuals without NSSI. The probability of having experienced physical harassment related to being trans was highest in trans women with NSSI (compared to those without NSSI); whereas for trans men the presence/absence of NSSI was not associated with a physical harassment history. Regarding psychological symptoms, trans women reported significantly more intrapersonal (e.g., anxiety, depression) and interpersonal symptoms (e.g., interpersonal sensitivity) compared to trans men. The probability of engaging in NSSI by trans individuals was significantly positively related to a younger age, being trans male and reporting more psychological symptoms. These findings need to be taken into account when providing psychological treatment and support for this vulnerable group of individuals.

There is a paucity of systematic research in the area of non-suicidal self-injury (NSSI) in trans people.To investigated the prevalence of non-suicidal self-injury in trans people and the associations with intra-and interpersonal problems.(according to ICD-10 criteria) attending a national gender identity clinic.

its peak in 2011, the suicide rate slowly decreased until 2017 due to various efforts by the government and society, for example, the activation of a government project named "Prevention of Suicide and Proliferation of a Culture of Respect for Life, " which is a basic plan of countermeasures to prevent suicide (2,3). However, the rate rebounded in 2018 and 2019 due to cases of celebrity suicide and worsening economic conditions for lowincome people (2,3). The number of deaths by suicide decreased in 2020 in the context of the COVID-19 pandemic and the reduced visibility of celebrity and copycat suicides compared to the previous year (2,4). The suicide rate is, therefore, affected by complex factors, with the socio-cultural background playing a significant role (5). One study of the U.S. population indicated that the culture has shifted toward tolerance of suicide (6). Changes in population composition over time accounted for <unk>50% of the increase in suicide acceptability between the 1980s and 2010s in the U.S. (6). Over time, Americans became more educated, were more likely to be single and to live in urban areas, and became less religious. These shifts were related to higher levels of tolerance toward suicide. Similar changes can be observed in the Korean population. The lifetime single-marital status rate increased from 8.0% in 2015 to 16.6% in 2020 (7). Further, the rate of the adult population who graduated college more than doubled from 23.8% in 2000 to 50.7% in 2020 (8). Higher-educated, single individuals have been linked to permissive attitudes toward suicide (6,9). Meanwhile, the number of religious people, who are known to have an attitude that suicide is incomprehensible, was 54% in 2004 but decreased to 50% in 2014 and 40% in 2021 (10,11). Additionally, news articles reporting celebrity suicides are accessible online, increasing the risk of copycat suicides among high-risk groups (12)(13)(14), and unregulated reports could make individuals perceive suicide as their right (12)(13)(14)(15). The risk of suicide-acceptive attitudes has been increasing, while the protective factors are gradually decreasing. Individual attitudes toward suicide are related to suicide intensity and behavior (16)(17)(18). To summarize, understanding the risks and protective factors related to the perception of suicide is important in establishing suicide prevention policies to prepare for future demographic changes. Previous studies on attitudes toward suicide have mainly compared specific population groups and factors, such as country, age, sex, and bereavement (16,18,19). We previously compared attitudes toward suicide between people who had experienced the suicide of an acquaintance and those who had not (17). However, few studies have classified attitudes toward suicide among different groups and investigated the correlation of sociodemographic factors. Therefore, this study aimed to identify attitudes toward suicide in the Korean population using the Attitudes toward Suicide Scale (ATTS). In addition, this study attempted to determine the sociodemographic factors in groups with different attitudes in both survey years (2013 and 2018). Furthermore, the moderating effect of survey year was analyzed to explore whether the influence of each sociodemographic factor changed over time.. Materials and methods ---.. Participants The participants' details and data collection methods have been described in our previous articles (17,20). South Korean adults aged 19-75 voluntarily participated after being informed about the study (20). Data were conducted through face-to face interviews, and a total of 2,973 people were included in the final analysis (1,473 in 2013 and 1,500 in 2018) (20). There were no missing values in the final analysis. Written consent was waived as the survey involved minimal risk. The survey was monitored and approved by the Institutional Review Board of Seoul National University Hospital (IRB No. 1405-019-577) and Kangwon National University Hospital (IRB No. KNUH-2013-06-007-001). The study was conducted in accordance with the Declaration of Helsinki. ---.. Assessments... The attitudes toward suicide scale The ATTS was developed by Renberg and Jacobsson (21) and comprises 37 items rated on a five-point Likert scale. It was used to evaluate the participants' attitudes toward suicide in 2013 and 2018. The ATTS has been frequently used in large-scale studies due to its relatively small number of questions and simplicity compared to other scales (16,18,21,22). Although it has a tenfactor structure (21), six factors with a reliability of 0.4 or more were selected (suicide as right: 0.67, incomprehensibility: 0.59, resignation: 0.64, relation-caused: 0.47, preventability: 0.43, and normal-common: 0.41) (23). The remaining four factors (noncommunication, tabooing, suicidal process, and preparedness to prevent) showed very low reliability (Cronbach's alpha <unk> 0.4) with the present data (23). Therefore, these factors were excluded, and the following six factors were used for a latent profile analysis (LPA): suicide as a right, incomprehensibility, resignation, relation caused, normal common, and preventability. ---... Sociodemographic characteristics Sociodemographic characteristics were recorded, including age, sex, education, marital status, monthly income, region, and religion. Additional questions related to suicide examined suicidal behavior ("Have you ever had any suicide-related behavior?"-no idea/idea only/plan or attempt), experience of suicidal loss among acquaintances ("Is there anyone around you who has committed suicide?"-yes/no), and interest in media reporting suicide ("Are you interested in articles reporting suicide?"-yes/no). ---.. Statistical analyses To classify groups according to Korean's attitudes toward suicide, a LPA was conducted on data from 2,793 participants using Mplus 8.0 (24). The optimal number of latent classes was determined using the Bayesian Information Criterion (BIC), sample-size adjusted BIC (saBIC), entropy, Lo-Mendell-Rubin Likelihood Ration (LMR-LR), and Bootstrapped Likelihood Ratio Test (BLRT). Analysis started with one class, and as additional classes were added, the fit indices of each model were evaluated until the optimal number of latent classes was confirmed. Categorical variables were analyzed using Pearson's chi-squared test utilizing the post-hoc analysis with "chisq.posthoc.test" R package (25,26). The following analysis was performed with IBM SPSS Statistics version 25.0 (SPSS Inc., Chicago, IL, USA). ANOVA with Turkey's multiple comparison was conducted for continuous variables to compare the classified groups. In order to identify the effect of sociodemographic factors on the groups, multinomial logistic regression analysis was performed for each survey year. All sociodemographic factors were entered into the logistic regression analysis. To determine the moderating effect of survey year on each demographic factor, Hayes' PROCESS version 3.5 macro for SPSS (Model 1) was performed adjusting age and sex as covariates (27), and additionally adjusting all other sociodemographic factors. The statistically significant level was a two-tailed p-value <unk> 0.05.. Results ---.. Comparison of sociodemographic characteristics between survey years When comparing the sociodemographics between survey years, the frequency of education, monthly income, religion, interest in media reporting suicide, and suicide behavior showed significant differences (Table 1). First, when comparing the frequency of education between 2013 and 2018, <unk>high school was lower in 2018 (n = 190, 12.7%) than in 2013 (n = 318, 21.6%), and the frequency of college or graduate school was higher in 2018 (n = 735, 49.0%) than in 2013 (n = 549, 37.3%). The frequency of lowincome was lower in 2018 (n = 217, 14.5%) than in 2013 (n = 408, 27.7%), and the frequency of high-income was higher in 2018 (n = 649, 43.3%) than in 2013 (n = 372, 25.3%). The frequency of believing in a religion was lower in 2018 (n = 633, 42.2%) than in 2013 (n = 743, 50.4%), and the frequency of having interest in the media reporting suicide was lower in 2018 (n = 622, 41.5%) than in 2013 (n = 681, 46.2%). There was a statistically significant difference between survey years when comparing the frequency of suicide behavior (p = 0.040), although this was not significant after multiple comparisons. ---.. Subtypes of attitudes toward suicide Table 2 shows the fit indices for the LPA models. The number of appropriate classes was determined based on fit indices, proportion rate, and information the classified groups presented (28). As the number of classified groups increased, the values of BIC and saBIC decreased, with the Class 5 model showing the lowest value. However, this model was excluded because it included a group with less than a 5% proportion (28). A Class 3 model showed a significant LMR-LR p-value (p <unk> 0.001), which was not significant in the Class 4 model (p = 0.083). In addition, the entropy of the Class 3 model was 0.713, the highest among the five models. As a result, the Class 3 model was selected as the most appropriate model. The average latent class posterior probabilities (Table 3), which reflect the accuracy of predicting individual's class membership, were between 0.80 and 0.90 (within the acceptable range) (28,29). The three groups classified by LPA were named according to the characteristics of the sub-scales for six attitudes toward suicide (Figure 1). Compared to the other two groups, Class 1 (10.3%, n = 305) had the highest score for incomprehensibility and preventability factors, while factors receptive to suicide (suicide as a right, resignation, relation-caused, and normal common) had the lowest score. Therefore, it was labeled the "incomprehensible" group. Conversely, Class 3 (36.9%, n = 1,098) was characterized by the highest score for suicide as a right, resignation, relationcaused, and normal common, while the lowest score was for incomprehensibility and preventability. Therefore, Class 3 was labeled the "permissive" group. Class 2 (52.8%, n = 1,570) showed the middle scores of Classes 1 and 3 in all six factors and was named the "mixed" group (Figure 1). ---.. Comparisons of sociodemographic characteristics between classes As a result of analyzing the frequency of groups classified by survey year, the proportions of the groups classified as incomprehensible [12.1% (2013) (2013), 49.9% (2018)] were higher in 2013 than 2018, and the proportion of groups classified as permissible was higher in 2018 (41.7%) than in 2013 (32.0%) (Pearson's <unk> 2 = 34.06, p <unk> 0.001) (Figure 2). The results of comparing sociodemographic characteristics between the three groups in each survey year are presented in Table 4. Monthly income, religion, and suicide factors showed significant differences between groups in 2013 and 2018. In 2013, the proportion of low monthly income (<unk>200, 10 4 won) was higher in the permissive group (32.6%), and medium monthly income (200-400, 10 4 won) was higher in the incomprehensible group (57.5%) than in other groups. The results in 2018 were not significant after multiple comparison with Bonferroni correction. Participants with religious beliefs were less often classified as permissive (44.5%) in 2013 and more often classified into the incomprehensible group (55.6%) in 2018. Among participants with no suicidal ideas, the proportions of those in the incomprehensible group [89.4% (2013), 91.3% (2018)] were higher and the proportions of those in the permissive group [70.6% (2013), 76.8% (2018)] were lower than in the other groups. In contrast, among participants with suicidal plans or attempts, the proportions of those in the incomprehensible group [8.4% (2013), 7.1% (2018)] were lower in the permissive [22.2% (2013)] group. There were no differences between the three groups in terms of the age, sex, marital status, exposure to suicidal loss, or interest in media reporting suicide factors (Table 4). ---.. Results of multinomial logistic regression assessing the three classes in each survey year Table 5 shows the results of multinomial logistic regression to identify differences in the sociodemographic characteristics of groups in each survey year. To explore the sociodemographic factors affecting acceptive attitudes toward suicide, the incomprehensible group was designated as the reference group. Age, sex, and marital status did not significantly differ among the three classes in 2013 and 2018. Medium monthly income was less likely to be associated with the mixed group in 2013 [OR = 0.60, 95% CI (0.40, 0.91)] and 2018 [OR = 0.60, 95% CI (0.39, 0.93)]. Low monthly income was less likely to be associated with the mixed [OR = 0.42, 95% CI (0.21, 0.83)] and permissive groups [OR = 0.34, 95% CI (0.17. ---. The moderation e ect of survey year To identify the moderating effect of survey year between sociodemographic characteristics and attitudes toward suicide, moderation model analyses were performed for each factor as described in Figure 3. The results are presented in Figure 4. The moderation effect of survey year for the models of education (R 2 change =0.003, p = 0.027) and monthly income (R 2 change =0.005, p = 0.027) were statistically significant. When entering all other sociodemographic factors as covariates, the moderation effect was consistently significant with education (R 2 change = 0.004, p = 0.003) and monthly income (R 2 change = 0.005, p <unk> 0.001). Other factors were not moderated by survey year. ---. Discussion This study classified participants into three groups through a LPA with sub-factors of attitudes toward suicide: the incomprehensible group with high resistance to suicide, the permissive group with acceptance of suicide, and the intermediate mixed group. Previous studies have reported that higher acceptance of suicide is associated with higher risk of suicide, while higher resistance to suicide is associated with lower risk of suicide (16,18). In the present study, the permissive group showed a higher association with suicide behavior than the incomprehensible group. Therefore, the permissive group could be regarded as a potentially high-risk group for suicide, while the incomprehensible group could be regarded as a low-risk group for suicide (5). The composition of the groups became more acceptive of suicide from 2013 to 2018. The proportion of the incomprehensible group decreased in 2018 compared to 2013, while the proportion of mixed and permissive groups increased. This increased vulnerability to suicide may reflect the increase in Korea's suicide rate in 2018, when compared to the suicide rate in 2013 (2). Previous studies have also reported increased acceptance of suicide ( 6) and increasing suicide rates in the U.S. (30). In addition, the mixed group showed higher risk of suicidal plans or attempts than the incomprehensible group. These results reflect that the mixed group, which occupied the highest proportion among the three groups (52.8%), could have a latent risk of suicide. Therefore, a strategy of targeting the mixed group to change attitudes toward suicide could help solve the problem of high suicide rates in Korea in future. According to the 2013 survey data, participants with less than high school education were more likely to be in the mixed (OR = 2.20, p = 0.019) or permissive groups (OR = 2.40, p = 0.013) than the incomprehensible group. This result might be reflected by the previous findings demonstrating that the lower the level of education, the higher the risk of suicide (31,32). However, there are differences in the influence of education level on attitudes toward suicide according to the survey country (18). For example, highly educated Russians condemn suicide, while highly educated Koreans, Americans, and Swedes perceive suicide as acceptable (6,9,18,33). Some studies have reported that education could promote individualistic values and expressions of thought. As a result, people tend to have more receptive attitudes toward suicide (6,9,34). The effect of education level was not statistically significant in the 2018 survey, and the moderating effect of survey year between education and latent class was significant. Therefore, this decreased influence of education might have been reflected by the increased level of mean education in the Korean population (8). Therefore, it is necessary to carefully observe the change in the influence of educational factors on attitudes toward suicide in the future. One previous study reported a positive association between income levels and permissive attitudes toward suicide (16). In the present study, the medium income group was less likely to be in the mixed group in both years. Especially in 2018, the low-income group was less likely to be in the mixed (OR = 0.42, p = 0.012) and permissive (OR = 0.34, p = 0.002) groups, and these results were confirmed by the moderation analysis results (Figure 4B). These results reflect that people with higher incomes perceived suicide as more acceptable in 2018. Since Korea's gross national income per capita has increased since 2000 (35), the influence of income level on permissive attitudes toward suicide should be considered important in the future. People who have experienced suicidal loss in social relationships perceive suicide as acceptable and are more likely to display suicide behavior (17,19). In the present study, experiencing suicide among acquaintances and permissive attitudes toward suicide were significantly related in the 2013 survey; however, permissive attitudes were not linked to these factors in the 2018 survey: In 2018, regardless of the presence of suicidal loss experience, there was a tendency to perceive suicide as acceptable compared to 2013. However, the moderating effect of survey year between suicidal loss and latent class was not significant (R 2 change = 0.001, p = 0.068). This tendency may be due to the increased exposure to indirect experiences related to suicide through new media, especially in 2018 (15,36). Celebrity suicide reports positively affect receptiveness to suicide more than suicide reports regarding non-famous people (15). Interest in media reports on suicide issues can provoke suicidal motivations and cause the Werther effect (13,15,37,38). Consequently, Korea's government enacted guidelines for suicide reporting in 2013, and there was no statistically significant difference in suicide deaths from the Werther effect between 2013 and 2017 (39). However, suicide deaths due to the Werther effect have increased again in Korea since 2018 (36), which is due to reduced compliance with suicide report recommendation standards for the media (26.6% in 2013 to 6.9% in 2015), increased reports of suicide-related news by new social media platforms, and sensational reports of self-harm (40)(41)(42). In the present study, while the frequency of those interested in media reports of suicide was lower in 2018 (41.5%) than in 2013 (46.2%), those with interest in such media reports were more likely to appear in the mixed (OR = 1.58, p = 0.029) and permissive groups (OR = 1.52, p = 0.048) in the 2018 survey. This may be reflected in the increased exposure to the new-media and news that indiscriminately report celebrity suicide (12,14,17,43). Therefore, there should be more effort to establish more strict reporting regulations and be cautious when reporting through new media in the future. People who had higher religiosity perceived suicide as a psychiatric problem, not an individual's own right, and an incomprehensible behavior (11). The results of this study demonstrated the association between religion and individual receptive attitudes toward suicide in both survey years. Finally, previous studies have reported a correlation between negative attitudes toward suicide and age (6,44), sex (45), and marital status (6); however, no significant relevance was identified in this study. ---. Conclusion This study classified latent classes according to attitudes toward suicide using a LPA and identified the meaningful sociodemographic characteristics of each group. Nevertheless, it has some limitations. First, data were analyzed separately for 2013 and 2018, and the effects of sociodemographic factors on attitudes toward suicide were compared by survey year. However, the participants surveyed in each year were independent groups, and changes in the influence of each factor are unlikely to reflect individual changes over time. In addition, the weight of the population was not adjusted in this study; thus, the results of our study are limited in representing the general Korean population. Second, as this was a cross-sectional study, causal inferences between sociodemographic factors and attitudes toward suicide could not be determined; therefore, further longitudinal studies are required in the future. Third, psychological factors that could affect attitudes or suicidal ideation, such as depressive symptoms, anxiety, or insomnia, were not included in this survey. Fourth, as the present study was conducted using a face-to-face interview method, participants may have underreported their symptoms and experiences. Fifth, in addition to suicide deaths, sudden deaths of close relations or member of family could affect an individual's depression level, attitude toward suicide, and suicide related behavior. However, we could not include this factor in the survey. Sixth, for determining the socioeconomic status, education and occupational groups might be more important than income level. Even though we included education level in this analysis, the occupational group analysis could not be conducted because of limited data. Despite these limitations, these results have several strengths and implications. First, to our knowledge, the present study is the first to cluster groups according to attitudes toward suicide. The association between attitudes toward suicide and the intensity of suicidal ideas has been reported in previous studies (16)(17)(18). In particular, those who have permissive attitudes toward suicide were more likely to have suicidal ideas. Therefore, considering the latent subtypes of suicide attitudes classified in this study, we expected to be able to classify potential suicide risk level, which could be a useful indicator for the high-risk suicide-related behavior group. In addition, by identifying the significance of sociodemographic characteristics for each survey year in the classified group, this study analyzed changes in the influence of the changing demographic structure. Therefore, the results of this study have clinical implications and could aid in establishing suicide prevention policies to prepare for the changing demographic characteristics in the future. --- Data availability statement The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author. --- Ethics statement The studies involving human participants were reviewed and approved by Institutional Review Board of Seoul National University Hospital (IRB No. 1810-062-979) and Kangwon National University Hospital (IRB No. KNUH-2013-06-007-001). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. --- Author contributions HL: conceptualization, formal analysis, and writing-original draft. SJR, CHKP, JHY, and KS: conceptualization and writingreview and editing. MJK: data curation and writing-review and editing. J-IP: conceptualization and methodology. YMA: conceptualization, methodology, writing-review and editing, and supervision. All authors read and approved the final manuscript. --- Conflict of interest YMA receives research support from and serves as a speaker for Janssen Korea Ltd., Lundbeck Korea Co., Ltd, and Korea Otsuka Pharmaceutical. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Introduction: South Korea has a high suicide rate, and changes in sociodemographic factors can further increase the rate. This study aims to ( ) classify participants using the Attitudes toward Suicide Scale (ATTS) through latent profile analysis (LPA), ( ) identify and compare the associations between sociodemographic factors with the ATTS in two survey years ( , ), and ( ) determine the moderating e ect of survey year. Methods: Six sub-factors of the ATTS were used for LPA with a total of , participants. Sociodemographic characteristics were compared between groups, and multinomial logistic regression was conducted for each survey year. A moderation analysis was conducted with the survey year as moderator. Results: LPA identified three groups of attitudes toward suicide: incomprehensible ( . %), mixed ( . %), and permissive ( . %). The proportion of permissive attitudes increased from ( . %) to ( . %). Participants reporting suicidal behavior were more likely to be in the mixed and permissive groups than the incomprehensible group in both years. People reporting no religious beliefs were associated with the permissive group in the two survey years. The influence of education and income levels on groups di ered by survey year. Discussion: There were significant changes between and in attitudes toward suicide in the Korean population. KEYWORDS suicide in Korea, attitudes to suicide, Korea National Suicide Survey, latent profile analysis, moderation model, Werther e ectAccording to the age-standardized suicide rate among Organization for Economic Cooperation and Development (OCED) countries, South Korea's rate is 23.5 per 100,000 people, which is more than twice the average rate of 10.9 per 100,000 across the 38 OECD countries (1). In 2020, suicide was the leading cause of death for people aged 10-39, and the suicide rate among people in their 60s and older was the highest among OECD countries (1, 2). Suicide, thus, remains a major health problem in South Korea. After reaching

Introduction Patients with cancer who uptake psycho-social support services improve their quality of life, emotional and social function, and reduce distress, depression and anxiety (Buffart et al. 2020;Goodwin et al. 2001;Kalter et al. 2018;Marchioro et al. 1996;Warth et al. 2020). For these reasons, psycho-social support services are an integral part of cancer care in many countries, and are available in both in-and outpatient facilities (Herschbach and Mandel 2011). However, despite their efficacy and growing availability, the use of psycho-social support services especially in the outpatient setting is lower than the patients' need for it (Faller et al. 2017;Frey Nascimento et al. 2019;Singer et al. 2013a, b). Inequalities in the availability of the services might create barriers to the participation. For instance, the services are more available in large certified structures (Certified Cancer Centres [CCC]), which are rare in rural areas (Kowalski et al. 2016;Singer et al. 2013a): Singer et al. (2012) reported that half of the facilities providing outpatient psycho-social support services were in cities between 20,000 and 100,000 inhabitants, and only around one-third and one-sixth were in smaller or larger cities, respectively. In addition, these structures were often combined with barriers to the accessibility, e.g. lack of parking facilities or of transport connections. Another barrier to the participation in psycho-social support services is the lack of the doctors' recommendation (Eakin and Strycker 2001;Frey Nascimento et al. 2019). Patients with cancer who receive a recommendation from their oncologist are six-times more likely to take up psychosocial support services compared to patients with no recommendation. Conversely, no association is present between the uptake of the services and the level of details of the information provided (Frey Nascimento et al. 2019). Therefore, to enhance the participation in psycho-social support services, an increase in referrals by doctors might be beneficial. To achieve this, it is important that the respective services are available in close proximity to the practices and patients, and that doctors and their patients have positive experiences with the service providers. No conclusive information is available on whether participation in psycho-social support services varies based on the migration background of the patients in Germany (Singer et al. 2022;Zeissig et al. 2015). However, patients with cancer and migration background are considered to be more likely to have higher psycho-social difficulties (Riccetti et al. 2022a;Tibubos et al. 2018), as well as higher barriers to access supportive care services than non-migrant patients (Riccetti et al. 2022a, b;Riccetti et al. 2020;Sze et al. 2015). Linguistically and culturally competent psycho-social support services are not available in every region (Schulz et al. 2018), often leading to friends and/or relatives acting as translators, with further difficulties and barriers for patients in discussing their need for psycho-social support (Hermes-Moll et al. 2022a). The aim of this study is, therefore, to investigate the availability of-and the experience with psycho-social support services for cancer patients with and without migration background from the perspective of office-based haematologists and oncologists in Germany, comparing different groups of doctors based on their socio-demographic characteristics and on the characteristics of the practices they work in. The study questions are: • Is the regional availability of overall and specific psychosocial support services for non-German speaking patients considered to be sufficient by office-based haematologists and oncologists in Germany? • Does the experience of office-based haematologists and oncologists with psycho-social support services for cancer patients with and without migration background in Germany differ based on their socio-demographic characteristics and on the characteristics of the practices they work in? --- Methods --- Study design Data collection took place between December 2020 and March 2021 via an anonymous, nation-wide online survey. The survey was part of the mixed-methods study "Psycho-oncological support in cancer patients with migration background" (Psychoonkologische Versorgung von Krebspatienten mit Migrationshintergund [POM]), which was described in detail for its purposes and results elsewhere (Hempler et al. 2021a, b, c;Hermes-Moll et al. 2022b). An email invitation to participate in the survey was sent to 581 doctors in 380 haematology and oncology practices in the networks of the Scientific Institute of Officebased Haematologists and Oncologists (Wissenschaftliches Institut der Niedergelassenen Hämatologen und Onkologen [WINHO]) and of the Professional Association of Office-based Haematologists and Oncologists in Germany (Berufsverband der Niedergelassenen Hämatologen und Onkologen in Deutschland [BNHO]). On January 2021, a second email was sent to all the practices, as reminder of the participation in the survey. This study obtained ethical approval from the Rhineland-Palatinate State Medical Association 424). --- Survey The survey was developed following a series of qualitative interviews with office-based haematologists and oncologists conducted in a previous stage of the project, and described in detail elsewhere (Hempler et al. 2021a, b). It comprised sections for: (a) socio-demographic characteristics of the doctors, (b) communication with patients with migration background, (c) cultural differences in patients with migration background, (d) experiences with screening instruments for psychological distress, (e) role of relatives and caregivers, and (f) experiences with psycho-social support services. --- Operationalization of the variables This analysis focuses on the results of the section of the survey on the experience with psycho-social support services. The section comprised: 1. The presence and characteristics of psycho-social support services in the practices. It included the presence, amount and type of psycho-social services in the practice (presence: yes/no/missing; hours: 40 h/week, 20-39 h/week, 10-19 h/week, 5-9 h/week, less than 5 h/week/ missing; type of worker: social worker/psychologist/ other/missing); 2. The opinion of the doctors on whether psycho-social support services in the practice was an advantage, and whether psycho-social support services for migrant patients should be improved (both: yes/no/missing); 3. The psycho-social support service(s) to which the doctors had referred their patients to (office-based psychotherapist/psycho-social cancer counselling centre/selfhelp group/patients association/outpatient clinics/no possibility/other/missing); 4. The satisfaction with the availability of psycho-social support services in the region, both overall and specifically for patients with migration background, considered as services in languages other than German (both: yes/ no/missing); 5. Missing languages among the psycho-social support services provided in the region; and 6. The satisfaction with the network and organization of and the connection with the outpatient psycho-social support services providers. Variables in this section were recoded as three continuous variables, ranging from "not at all satisfied" (1) to "completely satisfied" (4). The following socio-demographic characteristics of the doctors were ascertained: gender (male/female/other), age (under 49/50-59/60 or more), years of work experience (5-10/11-20/more than 20), country of birth (free-text), further education in psycho-oncology (yes/no), languages spoken other than German (one/ two/ three or more/ no foreign languages or missing), type of practice (single practice [Einzelpraxis]/joint practice [Gemeinschaftspraxis]/medical care centre [Medizinisches Versorgungszentrum-MVZ]), location of the practice (large city [<unk> 100,000 inhabitants]/ middle-size city [between <unk> 20,000 and <unk> 100,000 inhabitants]/small city [between <unk>5,000 and <unk> 20,000 inhabitants]/ village [<unk> 5,000 inhabitants]). --- Statistical analysis We reported the experiences with outpatient psycho-social support services both overall, as well as stratified by sociodemographic characteristics of the doctors and of the practices. We used Pearson's chi-squared tests for categorical data and Kruskal-Wallis tests for continuous data to explore the univariate association of socio-demographic characteristics of the doctors and characteristics of the practices they work in with their experience with outpatient psycho-social support services. To avoid over dispersion, in the group comparison the listed variables were recoded as follows: years of age (<unk> 60 years/ <unk> 60 years), years of work experience (<unk> 20 years/ <unk> 20 years), country of birth (Germany/other), type of practice (joint practice vs individual practices and medical care centres), location of the practice (large city vs small city, middle-large city, and village). --- Results --- Sample description Of the 581 doctors contacted, 55 (9%) completed at least 5% of the questionnaire and were included in this analysis. Doctors were mostly male (65%), older than 50 years old (73%), with more than 20 years of work experience (67%), born in Germany (85%) and working in cities with 100,000 inhabitants or more (65%). When asked about their proficiency with foreign languages, 40% of the doctors reported speaking one, 9% reported speaking two, and 5% reported speaking three or more (Table 1). --- Experience with psycho-social support services A total of 29 doctors (53%) reported referring patients to outpatient psychotherapists, 33 (60%) to psycho-social cancer counselling centres, 13 (24%) to self-help groups, and 4 (7%) each to patients' associations, outpatient clinics and other unspecified services. Among the doctors, 38 (69%) did not have a psychosocial support service directly in their practices, whereas 39 (71%) of them reported considering it an advantage. More than half of the respondents (56%) reported a non-sufficient presence of overall psycho-social support services in their region. On a scale from 1 (lower satisfaction) to 4 (higher satisfaction), the average reported satisfaction with network, organization and cooperation was 2.6 (standard deviation [SD] = 1.1), 2.3 (SD = 1.1), and 2.5 (SD 1.1), respectively (Table 2). --- Psycho-social support services for patients with migration background More than three doctors in four (80%) reported non-sufficient availability of psycho-social support services in foreign language in their region, and almost one in two (44%) reported that psycho-social support services for patients with migration background can be improved (Tables 2,3). Nearly half of the doctors (26 [47%], and 23 [42%]) reported missing psycho-social services in Turkish and Arabic, respectively, in their region. Other commonly reported missing languages were: Russian (12 doctors), Romanian (6 doctors), Italian and Polish (5 doctors each) (Fig. 1). --- Group comparisons Doctors with less than 20 years of work experience referred patients more often to psycho-social support services hosted in patients' associations (75% vs 25%; p = 0.02), than doctors with more years of work experience. Doctors working in practices in large cities referred patients less often to psycho-social support services in cancer counselling centres than doctors working in medium-tosmaller cities or villages (12% vs 88%; p = 0.01) (Table 4). Male doctors were more satisfied with the network with the providers of psycho-social support services than female doctors (mean score = 2.8 vs 2.2; p = 0.05) (Table 5). --- Discussion We aimed at investigating the experience of office-based haematologists and oncologists with outpatient psychosocial support services in Germany. Doctors working in larger cities reported referring their patients less often to psycho-social services hosted in cancer counselling centres, than doctors working in small and middle-large cities. This result might be attributed to the lower presence of these services in larger cities. Singer et al. (2012) reported that half of the counselling centres in the German federal state Sachsen were in middle-large cities, while only around one-third and one-sixth of these centres were in smaller or larger cities, respectively. Furthermore, these services often have barriers in accessibility (Giesler et al. 2015;Singer et al. 2012). Hence, oncologists in larger cities might be reticent in referring their patients to these services. This was also observed by Hempler et al. (2021b): doctors reported that they refer their patients more easily to centres providing psycho-social support services, which were in close proximity to the practice. Doctors with less than 20 years of work experience referred more often to psycho-social support services hosted by patients' associations than doctors with more than 20 years of work experience. Speculation might be drawn on whether younger doctors have more information on alternative providers of psycho-social support services, compared to older doctors. Another potential explanation might be that because of their fewer years of work experience, younger doctors did not yet build a network of personal contacts with psychologists or social workers providing psycho-social support services. According to Hempler et al. (2021b), doctors considered the referral to psycho-social support services to be a complicated process and, therefore, they heavily rely on personal connections with the providers of the services. The location might also play a role: younger doctors might be more present in larger cities, and, therefore, be confronted with the aforementioned lacks of services. Female doctors were less satisfied with their network with providers of psycho-social support services than male doctors. To the knowledge of the authors, no previous study reports similar results. However, it has been documented that women-in this case patients-have a more positive attitude towards psycho-social support services, due to a lower perception of stigmatization around psycho-social help seeking (Eichler et al. 2019;Faller et al. 2017;Steginga et al. 2008). Speculations can be drawn that female doctors might perceive lower stigmatization in recommending these services. In this scenario, female oncologists might consider the services more valuable and, thus, be more prone to offer it to patients in need. These larger volume of requests, could in Table 2 The presence and form of psycho-social support service in the practice, the availability of psycho-social support services in the region overall and specifically for patients with migration background, and the satisfaction with network, organization and cooperation with psycho-social support services and services providers (N = 55) a Scale between 1 (not at all satisfied) and 4 (completely satisfied) turn lead to higher exposure to structural issues, e.g. logistic and organizational difficulties. More than one doctor in three (38%) reported that psycho-social services are not sufficient in their region. Regarding psycho-social support services in foreign languages, the large majority (80%) of the doctors reported that these services are unavailable. This lack in psychosocial support services in foreign languages was previously reported by Schultz et al. (2018). --- Variables --- Implications for future research Further research could aim to look at the experiences with outpatient psycho-social support services among oncologists in Germany on a regional level. Moreover, the specific local concentration of patients with cancer and different migration backgrounds could be also considered. In addition, interventions leading to better network between oncologists and psycho-social support services' providers could be investigated. Finally, further research could look at overcoming the limitations of this study. More in detail, beside the small sample size, we included only office-based haematologists and oncologists. As psycho-social needs of cancer patients are known to vary based on the type of cancer (Krebber et al. 2014;Singer et al. 2009), a wider research including different oncological specializations should be conducted. --- Implications for policy makers and service providers Oncologists (and their patients) should be provided with available and easy-to-reach psycho-social support services in their region. Special efforts should be made to provide services in foreign languages or-at least-access to linguistically and culturally competent translators for psycho-social support services. --- Limitations This study was considered of interest as doctors' recommendation is among the most important predictors of participation in psycho-social support services (Eakin and Strycker 2001;Frey Nascimento et al. 2019). Hence, a positive experience for the oncologists in terms of availability of the services and cooperation with the providers could enhance the recommendations and the participation of the patients. The generalization of the findings of this study is limited due to the small study sample, which does not only limit the results per se but, by forcing to build aggregated groups, also disallowed for multivariate comparisons. Moreover, the definition of psycho-social support services for patients with migration background was based on the language in which the service could be provided. Therefore, no evaluation is present on whether these services could still present access barriers, e.g. lack of cultural competency. Furthermore, the participating haematologists and oncologists shared the same specialization in outpatient care and the same network. Hence, the results should be only referred to office-based haematologists and oncologists. --- Conclusions Office-based haematologists and oncologists in Germany reported a non-sufficient availability of psycho-social support services. Hence, a higher regional availability of psycho-social support services should be considered, both in terms of overall services and especially of services for non-German speaking patients. Moreover, a greater focus should be placed on services in Turkish and Arabic, followed by Russian, Romanian, Italian, and Polish. The doctors' work experience and gender, as well as the city they work in were associated with different experiences with psycho-social support services. This aspect should be considered when developing strategies in support of referral to psycho-social support services. These results should be evaluated in the explorative nature of this study and-thus-when generalizing them, awareness must be present regarding the limitations in terms of study design and study population. --- Data availability The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. --- Author contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Nicola Riccetti, Isabelle Hempler, Vitali Heidt, Kerstin Hermes-Moll, and Susanne Singer. The first draft of the manuscript was written by Nicola Riccetti and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding Open Access funding enabled and organized by Projekt DEAL. The project was financed by the German Cancer Aid Foundation (Deutsche Krebshilfe-DKH). Grant number: 70113529. The funding body had no influence on the design of the study, the collection, analysis, and interpretation of data, or in writing the manuscript. --- Declarations Conflict of interest Susanne Singer received honoraria from Lilly and Pfizer, all outside of the submitted work. The other authors declare that they have no conflict of interest. --- Ethics approval The study was carried out in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines. Ethical approval was obtained from the Rhineland-Palatinate State Medical Association. Consent to participate Data collection was anonymous and participation was on a voluntary base. Consent to publish Participants were informed that data were collected to be published in scientific journal. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Purpose We compared the perception of office-based haematologists and oncologists regarding the availability of outpatient psycho-social support services (PSSS) for patients with cancer and a migration background, as well as their different experiences with these services. Methods Data were collected via an online survey addressing the doctors' socio-demographic characteristics and their perception of-and experience with PSSS. The association between socio-demographic characteristics of the doctors and their experiences with PSSS was tested using Pearson's chi-squared test and Kruskal-Wallis test. Results A total of 55 doctors were included in this study. More than three doctors in four reported non-sufficient presence of PSSS in foreign languages in their region; one in two reported that the services for patients with migration background should be improved. Most doctors reported missing PSSS in Turkish and Arabic in their region. Doctors with less experience referred patients more often to PSSS hosted in patients' associations (75% vs 25%; p = 0.02), than doctors with more experience. Doctors working in larger cities referred patients less often to PSSS in cancer counselling centres (12% vs 88%; p = 0.01), than doctors working in small or middle-large cities. Male doctors were more satisfied with the network of PSSS' providers, than female doctors (mean score = 2.8 vs 2.2; p = 0.05). Conclusions Our results suggest that efforts should be made for a higher regional availability of overall and specific PSSS for non-German speaking patients (especially for Turkish-and Arabic-speaking patients). The experience with PSSS was associated with the doctors' work experience and gender, as well as the location of the practice.

INTRODUCTION Satisfaction with parenting refers to the extent to which parents are satisfied with their children in relation to parent-child interactions (e.g., communication, activities, praise, punishment) and child behavior (e.g., compliance, responsibilities, actions) (Bradshaw and Donohue 2014). Satisfaction with parenting is crucial in the implementation of parenting behaviors, as this satisfaction already has an impact not only on the parents themselves, but also on the overall health of the child's development, fertility and society. The number of research results on "satisfaction with parenting" as the subject of "sociological research direction" retrieval on literature collection platforms in different countries are different, among which Web of Science has 65, E-library (Russia) has 1,636; CNKI (China) has 35 (both search dates are: February 23, 2023, and all search languages are the official languages of the country). Thus, we found that Russian scholars pay more attention to the topic "satisfaction with parenting". In these studies, related to satisfaction with parenting, scholars discussed more key words such as: family structure (Rogers and White 1998), parenting pressure (Carey et al. 2009;Sevastyanova 202), parental responsibility (Henderson et al. 2016), marital satisfaction (Perlowski et al. 2019;Malenova and Borovikova 2008), life satisfaction (Mahmoud et al. 2021). It is worth mentioning that some Russian sociologists have proposed that parenting can be regarded as a labor activity, and the result of labor is children's human capital, and this kind of labor is a way of regulating the population that may help motivate the quantity and quality of the population in the future (Voroshilova 2015;Bagirova and Abilova 2017;Shubat and Bagirova 2020). It is very important to understand the parents' satisfaction with parenting to stimulate population reproduction effectively. Our research will take advantage of this concept by looking at parenting as a labor/work. Scientists have offered different perspectives on studies examining satisfaction. In a study of job satisfaction, Hoppock noted that job satisfaction reflects the psychological and physical satisfaction of workers with job-related environmental factors and can also be interpreted as workers' subjective responses to the work environment. Hoppock also suggested exploring job satisfaction by examining the experiences of respondents in their current jobs. In addition to theoretical studies, many researchers have created satisfaction models such as SCBC (Sweden Customer Satisfaction Barometer), ESCI (European Customer Satisfaction Index), ACSI (American Customer Satisfaction Index). To study the internal mechanism of the relationship between parents and children (Chinese college students and parents), the Chinese scientist Mei adjusted the ASCI satisfaction model and applied it to the field of parent-child relationships. To do this, he proposed four dimensions: user expectation, perceived quality, parent-child relationship satisfaction, and parent-child confrontation. Moreover, due to parenting satisfaction may be affected by the participation of the parenting process and the quality of the parent-child relationship (Lackovic-Grgin 2011). With the help of these research concepts, especially the measurement of parent-child relationship satisfaction and perceived quality, we will look at satisfaction with parenting in these two aspects. We assume that satisfaction with parent-child relationships can be assessed through respondents' satisfaction with parent-child interaction and parenting style in the family, and perceived quality can be assessed through their satisfaction with integrity and intimacy in the family. Study of satisfaction with parenting can yield different results for different population groups. After the reform and discovery of China, which occurred in the 70-80s. XX century, historical mass migration of the population occurred -this was mainly due to the mobility of excess labor from rural areas to the city. A group of people with special life experience arose in the country -namely left behind children. Today, some of the left behind children as adults already have their own children, and there are still a large number of left behind children in China (they may become future parents). Therefore, the study of this group of satisfaction with parenting may not only enrich the research methods for field of parenting study, but also provide some empirical research data for local population departments, especially in some areas with high number of left behind children in China. Based on the literature, we tried to construct a model of satisfaction with parenting for parents, who were left behind children in China. The way structures may be identified varies greatly due to the characteristics of the respondents, such as gender, age. The invariance of the measurement model indicated that the same underlying structure was measured across relevant comparison groups, which ensured the applicability of the model to respondents with different sociodemographic characteristics (Clench et al. 2011;Moksnes et al. 2013;Pevnaya et al. 2022). --- RESEARCH QUESTIONS AND HYPOTHESES The analysis of multigroup invariance helps to test the validity of the research model. This study aims to examine invariance of the 8-item Satisfaction with parenting between gender among parents who were left behind children in China. We hypothesized that: - --- DATA AND METHODS The study of grown-up left behind children as parents is based on data from a survey of parents living in Sichuan, China (n=431, 204 respondents were male, 227 respondents were female), conducted in 2021 and combining online + offline survey mode. The Satisfaction with parenting is a five-item scale, which employs 8-item Satisfaction with parenting on a fivepoint scale (ranging from "strongly disagree" to "strongly agree"). Satisfaction with parenting of the respondents answered in two aspects -as satisfaction with the family, in which the actors of parenting were their parents (primary family), and as satisfaction with the current family, where they themselves are the actors of parenting. Therefore, we constructed the model of satisfaction with parenting using these 8 satisfaction items (Fig. 1). Figures 1: Conceptual model of the influence of satisfaction with parenting in the current and primary family Note: S1-S8 -aspects of satisfaction with parenting in the current (S1-S4) and primary (S5-S8) families associated with: S1, S5 -family integrity; S2, S6 -family intimacy; S3, S7 -parent-child interaction; S4, S8parenting style. Double-sided arrows represent correlations; one-way arrows represent influence relationships. Firstly, using SPSS 26.0 to descriptive analysis of statistics all items of satisfaction with parenting by gender. Secondly, using Structural Equation Modeling (SEM) and Amos 26.0 for data analysis and model validation. --- ANALYSIS AND RESULTS Descriptive statistics for each item by gender are presented in Table 1, including the means, medians, mode and standard deviations pertaining to the eight variables of satisfaction with parenting are presented in Table 1. A prerequisite for conducting SEM analysis is the consistency of the model. The higher the consistency, the smaller the discrepancy between the original matrix and the matrix reproduced in accordance with the model. Consistency indicators and criteria (Hoyle and Panter 1995;Boomsma 2000;McDonald and Ho, 2002;Schreiber et al. 2006;Schreiber 2008;Jackson, et al, 2009), as well as the data of the model are given in Table 2. To determine whether the model of satisfaction with parenting to parents with left behind experience is gender-equivalent, a multigroup invariance analysis was performed on the CFA. Table 3 shows recommendation values for some model parameters proposed by different scientists, as well as the results of the research model itself. Table 3: Confirmatory factor analysis (CFA) (Stanislav and Kenneth 2008;Hair 2009;Fornell, Larcker 1981 Table 3 shows that male and female model parameters correspond to the criterion values, therefore, both factors have convergent validity (namely factor 1satisfaction with parenting in the current family; factor 2-satisfaction with parenting in the primary family). In addition, in order to test whether the correlations of the two factors are statistically significant, we tested their differential validity. For this, the AVEs of male and female model indicator were calculated, also shown in Table 3. The values indicate that there is differential validity between the factors. According to our hypothesis, we determined the model of satisfaction with parenting measurement equivalence by gender. Including tests of measurement weights (MW), measurement intercepts (MI), structural covariance (SC) and measurement residuals (MR) (Table 4). However, some researchers also recommended the use of other fit indices, such as the CFI difference (<unk>CFI), to evaluate measurement invariance -if <unk>CFI differences lower than.01 suggests some evidence of equivalence (Cheung and Rensvold 2002). Therefore, in Table 4, although the measurement intercepts model is significant, the <unk>CFI=0. Thus, all hypothesizes are not rejected. In other words, the eight satisfactions with parenting items in our study and the model of satisfaction with parenting composed of these items are not only suitable for men (fathers) with left behind experience, but also suitable for woman (mothers) with such experience. Some limitations should be considered regarding the obtained results. Due to cultural differences, population structure, parenting concept, etc., although the selected sample -namely, parents with left behind experience is representative of region China -especially the region with high number of left behind children, it is not representative of all regions with parents with such experience in other countries. --- CONCLUSIONS AND OUTLOOK This study found that the identity of the model of satisfaction with parenting of parents with left behind experience in Sichuan, China, was tested by gender grouping. There was no significant difference in the fit of the model whether male or female, which means that the model has cross-gender the invariant property of, indicating that the relationship between each item and the underlying structure is the same for both males and females. The focus of this study is to explore the comparison of the models of satisfaction with parenting for parents with left behind children in China by gender. Therefore, future research on the topic can concentrate on additional studies intended to other independent variables in the model of satisfaction with parenting of parents with left behind experience in China that have not been discussed or cross-cultural comparison between the models of satisfaction with parenting of parents with left behind experience. --- AUTHOR BIOGRAPHIES DOUDOU YAN is a postgraduate student of sociological sciences, Department of Sociology and Technologies of State and Municipal Management, Institute of Economics and Management, Ural Federal University named after the first President of Russia B. N. Yeltsin, Yekaterinburg. Her email address is yandoudoulinweiei@yandex.ru ANNA BAGIROVA is a professor of economics and sociology at Ural Federal University (Russia). Her research interests include demographical processes and their determinants. She also explores issues of labour economics and the sociology of labour. She is a doctoral supervisor and a member of the International Sociological Association. Her email address is a.p.bagirova@urfu.ru and her webpage can be found at http://urfu.ru/ru/about/personal-pages/a.p.bagirova/

The analysis of multigroup invariance helps to test the validity of the research model. Satisfaction with parenting is one of the important factors regulating population reproduction and parenting willingness. Research on satisfaction with parenting of specific social groups can provide a more targeted understanding of the specific factors that affect their parenting behavior and parenting motivation. The satisfaction with parenting of parents with left behind experience may be affected by the primary family and the current family. The purpose of the present study was to examine the invariance between gender of the satisfaction with parenting for parents with left behind experience. The sample consisted of 431 parents (204 male and 227 female) with left behind experience, which conducted in 2021 in Sichuan, China. Multigroup analysis supported measurement invariance between gender, which include measurement weights, measurement intercepts, structural covariance, and measurement residuals. These results provide evidence that the model of satisfaction with parenting in the current and primary family for parents who were left behind children is a valid parenting satisfaction measure to be used among male and female actors with such experience in some regions -in particular, some regions with high number of left behind children in China.

Background Diabetes mellitus, which has become a threat to human health, is the most important noncommunicable disease [1]. In 2013, 11.6% (approximately 1140,000) of Chinese people aged above 18 reported having diabetes mellitus [2]. More than 60% of rural diabetic patients experience delayed diagnosis and treatment [3]. Without supervision, blood sugar is controlled badly, and errors in using insulin can influence disease control [4]. Only 18.8, 16.2 and 8.0% of the people aged 35 to 74 with diabetes in China's rural areas engaged in proper awareness, treatment and control of diabetes, respectively [5]. The reasons for such low percentages include poor medical condition, low degree of education and low level of cognitive behavioral intention of residents in rural areas [6]. Delayed diagnosis and treatment of patients with diabetes have serious impacts on the quality of life and disease prognosis. Gulam showed that mild diabetes that is not treated in time may incur complications, such as infection, sepsis and gangrene, and can increase the risk of amputation and death [7]. Zubair believed that the delayed time of diagnosis and the amputation rate were significantly positively related, as delayed treatment time increases the likelihood of amputation for patients with diabetes [8]. Stone P A believed that persistent inflammation and blood glucose fluctuations may induce cardiovascular events and increase cardiovascular disease mortality, such as acute myocardial infarction and cardiomyopathy [9]. Delayed diagnosis and treatment are characterized by a delay in diagnosis and a delay in treatment. A delay in diagnosis, which is called a patient delay, occurs if the time between when a patient first observes suspicious symptoms and when the patient first visits a doctor in a medical institution spans more than 3 months [10,11]. A delay in treatment refers to a situation in which a patient with diabetes begin to receive regular treatment more than 2 weeks after receiving the diagnosis or a situation in which a patient who is in the process of receiving treatment does not receive treatment in a timely matter [12]. Most studies of diabetes have focused on complications and influencing factors, while few have considered delayed diagnosis and treatment. China has started to attach great importance to the prevention of diabetes, but there is a lack of prehospital targeted screening and early intervention measures, with impacts on prevention, diagnosis and treatment. This study was conducted to explore the inner experience of clinical treatment and the behavioral intention concerning delayed diagnosis and treatment of patients with diabetes and provides a reference for comprehensive intervention in rural China. --- Methods --- Design In this phenomenological qualitative research, interviews and observations were used to collect data for conventional content analysis through a qualitative descriptive method conducted by the first author. Content analysis can explain and classify textual data by considering individual cultural and contextual effects on phenomena. The final products of data analysis for content analysis are classifications and themes [13]. --- Participants This research adopts the convenience sampling method, and thirteen diabetes patients with delayed diagnosis and treatment were sampled with maximum variation in rural Daqing City and Tangshan City in China. All diabetes patients with delayed diagnosis and treatment had to meet the inclusion criteria provided by the first author to participate in the study. The inclusion criteria were having diabetes, living in a rural area and having experienced delayed diagnosis and treatment. Further information gathered from the patients interviewed included demographic background, age, gender, course of the disease, delay time of diagnosis and treatment, level of education and family income per month. All of the included patients with delayed diagnosis and treatment were asked and agreed to participate in this research, and all were farmers. --- Data collection The data collection was carried out from April to October 2018. In the process of the in-depth interviews, all diabetes patients with delayed diagnosis and treatment were informed of the aim of the study and agreed to participate. The respondents were encouraged to express their inner thoughts concerning the behavioral intention behind their delayed diagnosis and treatment of diabetes. All interviews were conducted in a quiet and well-lit room, comprised a face-to-face conversation between the interviewee and the interviewer, and lasted from 45 to 60 min. The researcher listened carefully and recorded the audio contents of the interview with a recording pen at the same time, and the interviewees' behaviors and expressions were also recorded in a timely manner. After the interview, we returned the transcripts to the respondents, and they were asked to check whether the transcripts were consistent with their intention in order to improve the credibility of the results. When information was repeated in the process of the interviews, the sample size was considered to have reached saturation. All diabetes patients with delayed diagnosis and treatment and living in rural areas spoke Chinese, and the researcher translated the interviews into English to obtain the results. The outline of the interview was as follows (Table 1). --- Data analysis There were 2 researchers,who had nursing master's degree, involved in coding the data. The seven-step Colaizzi phenomenology was used to analyze the interview data [14]. The qualitative research and analysis software QSR Nvivo 12.0 was used to collate and analyze the data. The contents of interviews were transcribed within 24 h after the interview, and the transcriptions were returned to the respondents to verify the text information and to improve the credibility of the results. After the interviewee verified the contents, two researchers, who continuously and repeatedly read, analyzed and coded the same textual information, compared the results with the original information and formed the final theme. --- Ethical considerations Consent for the study was obtained from the ethics committee of Daqing campus of Harbin Medical University. Before each interview, the researcher explained the purpose and content of the study to the respondents and promised confidentiality. The participants then agreed to participate in the research and signed the informed consent. Participants had the right to refuse to participate in the research at any time without reprisal. --- Results There were thirteen participants in this study: five were male, and eight were female. All patients had type 2 diabetes. The age of the participants was between 38 and 82 years old, and the mean age was 57.53 years. The disease course was between 1 and 20 years, the mean course of the disease was 9.1 years, the delay of diagnosis and treatment was between 0.5 and 13 years, and the mean delay of diagnosis and treatment was 6.1 years. Twelve patients had diabetes complications, 4 had diabetic foot, 7 had diabetic macular edema, 9 had diabetes complications with cardiovascular symptoms, and 2 had asymptomatic symptoms. Three of the participants were illiterate, 6 had primary school education, 2 had junior high school education and 2 had senior high school education. All of the participants were married, and four had experienced the death of a spouse. Six people had a monthly family income less than 500 RMB, 5 people had a monthly family income of 500 999 RMB, 1 person had a monthly family income of 1000 <unk>1999 RMB, and 1 person had a monthly family income of more than 2000 RMB (Table 2). Data analysis led to the development of four themes and ten subthemes. The themes were "Lacked knowledge of diabetes", "Negative coping style", "Dissatisfaction with the existing medical service" and "Influence of social support" (Table 3). --- Lacked knowledge of diabetes This theme includes three subthemes: an insufficient understanding of the disease, a lack of awareness of early treatment and lack of self-control and monitoring. Only 1 participant could accurately answer questions concerning relevant knowledge, and the remaining participants were unable to answer questions accurately. --- Insufficient understanding of disease Most of the participants said that they did not have basic knowledge about diabetes, symptoms associated with the The third participant (N01) said, "I didn't know about diabetes before I had this disease, I paid attention to this information after I was sick. If I had known I was a diabetes patient, I would have gone to the hospital early." --- Lacked awareness of early treatment Judgment of the severity of disease can have an effect on patients' health behavior decisions. Two respondents thought that going to the hospital sooner or later would have no effect on the treatment of the disease. Four participants thought that if the disease did not affect normal life, it did not need treatment. One participant (N08) said, "Diabetes is a chronic disease, it doesn't matter that you go to hospital sooner or later." Another participant (N12) stated, "Everyone says that diabetes is a cancer that you can't die from. If you are comfortable, then you don't need to see a doctor." A third interviewee (N10) said, "I didn't need to be hospitalized after the diagnosis of diabetes, I just went home and controlled it by myself. And I always thought that diabetes was ok. I was diagnosed with diabetes five years ago, and I was hospitalized for the first time because I fainted." --- Lack of self-control and monitoring The patients' effective self-health management involved a long and difficult process. Some people did not accept regular treatment, and they used folk prescriptions and delayed treatment. One participant (N02) said, "I took the medicine, I still felt that I was bad, so I didn't continue to take the medicine." Another participant (N07) stated, "Usually, I have time to try to walk for exercise, because it is difficult to control my diet. I feel no one can control their diet so strictly." A third participant (N09) said, "At first, the doctor told me that I should take medicine and have a strict diet, but I could not do it. I took folk prescription drugs for nearly three months, but it was no use, and I felt it was more and more serious, then I went to the hospital." --- Negative coping style This theme includes three subthemes: poor medical behavior, fear/anxiety and uncertainty about the future. All participants were not positive about going to the hospital for various reasons. --- Poor medical behavior The interviewees had a weak awareness of medical treatment in general. Two respondents said that they did not go to the hospital if diabetes mellitus had no effect on their normal life. One participant (N08) said, "We live in the countryside and are busy every day, I don't pay attention to myself, and I never go to the hospital." Another participant (N10) stated, "I had symptoms of diabetes early, but I didn't want to come to the hospital; when I had an examination, my blood sugar was 23.5 mmol/L." Two respondents said that they were in good health and did not engage in treatment by themselves. Most people did not have good habits, and they thought it was trouble and a waste of time to go to the hospital. One participant (N06) said, "I still feel very young, and the chronic disease is not going to heal, so I never want to go to the hospital; I can control the disease at home." --- Fear/anxiety After some patients were diagnosed with diabetes, they were fearful and refused to discuss their diabetes with others. One participant (N11) said, "Everybody said diabetes was difficult to cure, but people couldn't die from diabetes. So I felt fear in the face of the disease, and I didn't want to let others know that I had diabetes and I didn't go to the hospital." Another participant (N04) stated, "I don't want to be in the hospital because it's too much trouble, I'm afraid of family trouble, and I also fear missing work." The patients with diabetes felt too much great worry and anxiety about the care of their children and families. A married woman (N04) said, "I was worried about my child, I was afraid nobody would take care of him while I was in the hospital, and it would have an effect on his studies." Two respondents said that the hospital environment would produce anxiety and irritability. One participant (N13) said, "When I came to the hospital, I felt depressed, bad tempered and agitated." --- Uncertainty about the future Most of the patients had no confidence in the treatment of diabetes. Their uncertainty about the future led to delayed treatment. A patient (N04) with badly controlled blood sugar said, "I don't know whether I can control my blood sugar or not, sometimes I had no confidence." Another participant (N09) stated, "My current total blood sugar is above 10, now my legs are itchy. I don't know whether going to the hospital and spending money can cure me or not." --- Dissatisfaction with existing medical services All participants believed that the hospital environment was of poor quality. This is the main reason most people with diabetes were reluctant to seek help. --- Discomfort in the medical environment and the cumbersome medical process The medical environment is not comfortable, and the poor medical services provided can directly affect the patient's medical behavior. Three respondents said they were not satisfied with the medical environment of the hospital. One participant (N02) said, "The environment of the hospital was poor and not comfortable. There were too many patients, and they did not have good rest in the hospital." Another participant (N03) stated, "I think the hospital conditions do not meet the standard in my mind, there were not enough beds in the hospital, and extra beds were in the corridors. In particular, I didn't want to the hospital." A third participant (N08) said, "I think that the process of seeing a doctor was too much. The fee was paid on another floor, seeing a doctor meant running around all morning, a person can't see the doctor." Professional guidance provided about the disease Three respondents hoped to obtain professional knowledge to guide the treatment of their disease. One participant (N03) said, "We don't know what diabetes is, I didn't even know when I was sick. When I don't feel strong enough, then I'll go to the community hospital, but that's no medical condition." Another participant (N05) stated, "Information in rural areas is developed, but I don't know where to go to understand diabetes. Even if I'm sick, I don't know when to see a doctor." --- Influence of social support Social support is very important for people with diabetes, and patients with diabetes who have good social support have higher levels of cognitive behavioral intention. --- Support from family and others The majority of patients with diabetes learn about the disease from family, friends and people around them. Family members' lack of medical knowledge will lead to delayed treatment. Some respondents went to the hospital at the urging of those around them. A patient (N03) whose blood sugar was 19.8 mmol/L said, "I don't want to be in the hospital, but my daughter always advised me to go to the hospital." Another participant (N05) stated, "I didn't take medicine after illness, but I heard she had to take medicine to control her blood sugar, so I just came to the hospital." A patient (N06) whose diagnosis was delayed by one year said, "People say that diabetes is a chronic disease. I didn't need to go to the hospital for treatment for a long time. My diet and exercise could control the blood sugar, I received these effects, and so I didn't go to the hospital." --- Medical economic burden Three of the respondents expressed that they did not want to go to the hospital because of the heavy economic burden. One participant (N05) said, "I had to spend money when I went to the hospital. I spent thousands on an optical inspection. I can't afford it." Another participant (N09) stated, "Because of the high cost of the treatment of diabetes, although the national subsidy exempts some treatment. But I still can't solve the problem of diabetes drug costs." --- Discussion This is the first qualitative study specifically aimed at exploring the inner experience of the clinical treatment of diabetic patients with delayed diagnosis and treatment. --- Improving the understanding of diabetes and enhance awareness and education Relative to urban areas, diabetes health services for the rural patients in China are poor, which could be one of the reasons for delayed diagnosis and treatment in rural areas. Because rural patients tend to have a low education level, diabetes health education resources are scarce, resulting in a lack of knowledge of diabetes. The respondents said that before they were ill, they did not take the initiative to understand diabetes, and they did not know the first symptoms of diabetes. Eventually, this led to delays in diagnosis and treatment. The respondents did not understand the importance of early treatment for diabetes and ignored self-health management. Poor blood sugar control consciousness leads to disease development and complications involving other organs. Many people's understanding of diabetes is flawed, which results in treatment resistance. This may be due to atypical early symptoms of diabetes, which most people cannot recognize. Second, people may lack knowledge of diabetes treatment. Knowledge about diabetes has the strongest effect on health behavior, and the higher the degree of knowledge is, the better prevention and control behaviors an individual engages in [15]. Therefore, we should provide education about health, awareness and effective control of blood sugar and improved treatment for patients who have been diagnosed with diabetes [16]. People with a family history of diabetes, centripetal obesity, dyslipidemia, and lack of physical activity represent a key intervention object [17]. Medical institutions can provide free screening for diabetes, encourage people to actively seek medical attention and improve diabetes prevention awareness. --- Establish positive coping styles and motivate treatment This study found that most of the respondents did not have good habits, and they responded negatively in the face of early symptoms. They thought their condition would improve through taking their own medication or resting more. The result was that when their symptoms affected normal life, the patients were forced to seek medical treatment. Some of the respondents did not understand the state of their body and considered themselves healthy, even delaying treatment for 16 years, and when they went to the hospital for medical treatment, they had serious complications. To address these habitual negative responses, we should first change patients' negative medical attitudes. Zhang Cang found that a negative attitude towards medical treatment is the main reason for delayed treatment [18]. Therefore, nurses should inspire patients to actively engage in treatment, strengthen the importance of early treatment and improve patients' self-efficacy. --- Effective emotional and social support People with diabetes experience or whose friends work in medical-related fields engage in better medical treatment behavior and have less fear and anxiety concerning the disease. Therefore, good social support can eliminate adverse psychological reactions and promote positive treatment. Studies have shown that a good role model or a good medical treatment atmosphere can promote patient health decisions; the absence of these factors leads to delayed diagnosis and treatment [19]. Therefore, nurses should strengthen the related health guidance given to patients, provide disease knowledge education, reduce people's fear of seeing a doctor and facing the disease, and encourage patients to see a doctor [20]. At the same time, the medical staff should establish a good trust relationship with the patient, give patients great respect, and encourage patients to adopt a positive attitude towards treatment and have more confidence in continuing treatment. After patients leave medical institutions, medical staff should keep in touch and follow up by telephone on a regular basis. Such professional guidance and supervision will more effectively promote the patients' physical and mental health. --- Optimization of community health service and reduced economic burden In the interviews, when mentioning reasons for their reluctance to go to the hospital, nearly half of the respondents thought that it was troublesome to go to the hospital. We conclude that treatment involves a complicated process, there are too many patients per hospital, and the patients live far away from the hospital. The best solution to this problem is to improve community health care. At present, patients' medical understanding and patterns are changing in China, and people attach more importance to the convenience and effectiveness of treatment. The results of the study also showed that the convenience of treatment had a significant influence on residents' medical behavior [21]. Therefore, we should enrich community medical resources, improve the technical level of medical staff, and eliminate people's distrust for community health agencies and other problems. In addition, this study found that people with diabetes face a heavy economic burden, which is one of the reasons for the delay of diagnosis and treatment. Research statistics reveal that hospitalized patients with diabetes (1396 cases) spend approximately 13 days in the hospital, incurring a total hospitalization expense of 6578.88 yuan, while expenses for medicine account for the greatest cost, followed by inspection, examination and treatment. The economic burden to patients with diabetes cannot be ignored. We should continue to focus on the problem, seek to perfect medical insurance for patients with diabetes and let more people enjoy treatment. --- Limitations Because the research object of this study covers only two provinces in rural China, it is subject to bias due to region selection and research conditions. Additionally, because cultural differences exist between native English and Chinese speakers, the translation of the interviews from Chinese to English is another limitation of the study. Further strengthening of the theme extraction is necessary. Furthermore, this study was conducted on a limited number of patients with diabetes in rural China; therefore, the results cannot necessarily be generalized to all patients with diabetes in China. Because regional differences in China are relatively large, the psychological experiences of patients with delayed diagnosis and treatment of diabetes are different. --- Suggestions for future studies Future studies may add quantitative research and engage in more in-depth investigations of the causes of the delayed diagnosis and treatment of diabetes. --- Conclusions The findings of the current study show that 13 patients with diabetes suffered from problems of diagnosis and treatment delayed by a minimum of 2 to 13 years, and more than 90% of the respondents had complications. People with diabetes who live in rural areas experience a serious phenomenon of delayed diagnosis, which is particularly due to poor knowledge and habits. Therefore, nurses should start with the patient's needs and wishes, raise awareness of diabetes treatment, reduce the occurrence of delayed treatment, and improve the prevention of diabetes. --- Availability of data and material The datasets generated and analysed during the current study are not publicly available due to paticipant confidentiality but are available from the corresponding author on reasonable request. --- Authors' contributions JHH and YLY contributed the study design, conducted the interviews, analyzed the data, and led the drafting of the manuscript. HGX, ZYQ, WRQ and LL contributed to study design, data analysis and drafting of the manuscript. All authors read and approved the final manuscript. --- Ethics approval and consent to participate The study obtained the consent of the ethics committee of Daqing campus of Harbin Medical University (Number: HMUDQ201911801), and written informed consent was obtained from each participant. --- Consent for publication Consent for publication has been obtained from patients to report individual patient data. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Great changes have taken place in terms of people's lifestyles and behavioral habits. Diabetes has become a threat to human health and is the most important noncommunicable disease. More than 60% of rural diabetic patients experience delayed diagnosis and treatment. In this study, we explore the inner experience of the delayed diagnosis and treatment of patients with diabetes in rural areas and provide a reference for targeted intervention. Methods: A qualitative research design was used to examine the cognitive behavioral intention of patients in rural areas with delayed diagnosis and treatment of diabetes. Thirteen diabetes patients with delayed diagnosis and treatment were sampled with maximum variation in rural Daqing City and Tangshan City in China. The data analysis involved several levels of analysis consistent with qualitative research. Results: The following themes were relevant to diabetes patients in rural areas with delayed diagnosis and treatment delay: "Lacked knowledge of diabetes", "Negative coping style", "Dissatisfaction with the existing medical service" and "Influence of social support". Conclusions: The respondents' delayed diagnosis and treatment represent a common phenomenon. Medical personnel should provide interventions for patients and encourage them to go to the hospital on time.

Introduction Globally, accidental falls are a major public health problem and the second leading cause of unintentional injury deaths worldwide. Each year 37.3 million falls require medical attention, and adults older than 65 years of age suffer the greatest number of fatal falls [1]. At least one-third of communitydwelling people aged 65 years and above fall every year, and the incidence increases with advancing age and fragility level [2]. In Sweden, falls are the most common cause of injury in old age that leads to deaths, hospitalizations, and visits to emergency services. In the age group over 80 years, nine out of ten injuries are caused by a fall [3]. As the proportion of older people is rising globally, the individual and societal costs associated with falls will increase. The prevention of falls is therefore an urgent public health challenge. Intervention programmes that focus on balance combined with muscle strength in the lower limbs are effective in addressing risk and rate of falling [4,5]. Despite this, many older people are reluctant to adopt fall prevention exercise due to factors such as underestimation of falls risk, poor self-efficacy, fear of falling, or stigma associated with exercise for older people [6]. According to a systematic review, only 64% accepted the invitation to join the exercise programme and 19% then dropped out when they learned what the intervention entailed [7]. Moreover, around 77% of the participants in fall prevention exercise studies are women [8]. Instead of participation in exercise, many older people choose individual coping strategies, tailored to their particular circumstances, to avoid future falls, or simply discontinue activities that would involve a risk of falling [9]. Poor adherence is associated with low education (<unk> high school), living in a disadvantaged neighbourhood, being obese, having fair/poor self-rated health and having problems with walking or using a walking aid, and a history of falls [10,11]. Several factors have been identified that positively influence the uptake and adherence for fall prevention exercise programmes, including facilitators such as support from professionals or family, social interaction, perceived benefits, a supportive exercise context, feelings of commitment, and having fun. However, there is still a lack of studies investigating preferences for specific programme characteristics that may attract older people by making exercise interesting, challenging, and enjoyable [12]. Additionally, in previous studies investigating preferences, more than 75% of the participants have been women and very few studies have investigated if men and women differ in their views and preferences for fall prevention exercises [12]. Unconscious preunderstandings and expectations of how older men and women would prefer to exercise may lead to unnecessary gender bias [13]. In order to identify factors that attract both older women and men to fall prevention exercise programmes, as well as increasing the awareness and knowledge among professionals who deliver these exercise programmes, there is a need to identify if older women and men differ in their views and preferences about exercising. The aim of this study was, therefore, to explore exercise preferences and motivators of older communitydwelling women and men in the context of falls prevention, from a gender perspective. --- Methods This study is based on workshops including multistage focus group discussions [14] held with older people as part of a larger participatory action research project, aiming to develop an evidence-based, cocreated, fall prevention digital exercise programme [15]. The checklist from the 32item COREQ (consolidated criteria for reporting qualitative research) [16] was used when designing and reporting the study. The study was approved by the Regional Ethical Review Board (Dnr. 2012-170-31M). 2.1. Participants and Recruitment. Participants were recruited from seven senior citizen organisations. Thirty-eight older adults volunteered to be contacted by telephone and interviewed to inform a purposive sample consisting of community-dwelling women and men of at least 70 years of age. Sampling aimed to achieve a heterogeneous group with respect to exercise habits, living conditions, and experiences of falls, with 30% having a fall in the previous 12 months (reflecting population epidemiology). Eighteen participants (10 women and 8 men) were finally included in the study; one woman chose to end her participation after four meetings. The mean age of the participants was 74.6 <unk> 3.5 years. The majority of the participants were former white-collar workers and, in general, have self-reported being moderately physically active. Two couples were deliberately involved in the project. Four participants lived alone and 14 lived with partners (Table 1). The participating researchers formed a cross-disciplinary group from the disciplines physiotherapy, informatics, exercise physiology, and computing sciences. The group included experts in e-health, falls prevention, and gender research. All researchers were women. None of the researchers had any preexisting relationship with the participants and they all introduced themselves and their role in the project at the beginning of the process. Prior to the first meeting, the participants were divided, by the researchers, into two separate groups with similar disposition regarding age, background, and gender. One of the married couples was placed in each group. Before the first focus group discussion, the participants signed a written informed consent. --- Data Collection. Data was collected through six workshops including multistage focus group discussions [14] with each group. Each workshop was held at a community centre and lasted for 2.5 hours, including a short refreshment break. All focus group discussions were conducted in a positive atmosphere inspired by Participatory and Appreciative Action and Reflection (PAAR) methodology [17]. The PAAR methodology is a form of Appreciative Inquiry, meaning that it focuses on positive experiences rather than problems. According to this methodology the focus group discussions aimed to understand the root causes of the older participant's positive experiences of and preferences for exercises. The focus was on accomplishments, strengths, and understanding the causes of success, so that these causes could be better understood and augmented. In this way the discussions were not preoccupied by problem solving or by changing negative behavior. This appreciative form of inquiry aims to create an open, positive, and encouraging climate during the discussions. In order to stimulate the discussions within the group, a range of facilitating techniques and methods were used. The participants were asked, for example, in between the workshops to document their everyday life in different ways in order to stimulate and trigger thoughts about what inspired them to exercise and to be active. One method used for this was Photovoice, a process by which people can identify, represent, and enhance their thoughts through photographs [18]. Another example was to bring pictures from newspapers, reflecting their attitudes towards exercising and moving their body. During other sessions practical activities (e.g., strength and balance exercises) were conducted and the participants were asked to actively reflect over the exercises and to describe their views and feelings. Sometimes more detailed questions were posed and visualisations of different views were placed around the room. The participants were asked to move to the place that represented their view and were encouraged to explain and discuss their choices. Another facilitating technique used was short descriptions of five fictional characters (personas) representing different motivational profiles [19]. These personas were constructed based on data from the first sessions and the participants rated how well the descriptions defined themselves and the way they preferred to do exercises. The participants were later asked to create their own persona description, by combining sentences from the predefined examples. With the purpose of clarifying and separating differences in the participants reasoning, smaller gender divided discussion groups were organized during the fourth and fifth workshop. The main activities of each workshops have been presented elsewhere [15,20]. Between five and six of the researchers attended each workshop, except workshop three when only two researchers participated (PP and LLO). Two of the authors (BBK and AMW) had extensive previous experience from participatory action research and they were primarily responsible for facilitating the discussions, assisted by the other authors present. After each session the researchers discussed and reflected on the results and the process. Based on these reflections the purpose of the next meeting was planned. In the beginning of each new meeting the researchers summarized and presented result from the last workshop to get feedback, and then the aim of the current meeting was outlined. --- Analysis. All focus group discussions were digitally recorded and transcribed verbatim. As recommended when trying to identify similarities and differences in material, qualitative content analysis was used to analyse data and explore women's and men's reasoning about their experiences and preferences in relation to exercises [21]. An inductive content analysis approach was used since very few previous studies have investigated if men and women differ in their views and preferences for fall prevention exercises. Thus the categories were derived from the interviews and not on the basis of previous knowledge or theories [22]. Two of the authors (PP and MS) read the transcripts independently to get an overall understanding of the participants exercise preferences. The text was then transferred to the data program Open Code 4.01 [23], where meaning units related to the aim were identified and labelled with codes. The codes were organized into preliminary subcategories and categories by two of the authors (PP and MS). The categories were continuously negotiated and the texts reread within the author group until consensus was reached to ensure trustworthiness. The categories and subcategories were in addition presented and discussed at two seminars for PhD students and senior researchers at the University. In the end of this process the results were summarized and presented to the participants. Preliminary results were included in the thesis of one of the authors (PP) [20]. --- Results The participants gave many examples of physical activities they liked to perform, such as walking, cycling, working outdoors, gymnastics, water aerobics, skiing, dancing, Nordic walking, table tennis, and Qigong. They expressed many motives, as well as barriers, for starting to do exercise in the context of falls prevention. Based on their experiences, they voiced many opinions regarding how different exercise properties influence and enhance motivation and they gave examples of "tricks" to facilitate maintenance. In the majority of these opinions no clear gender separation pattern was seen. The individual differences within genders were more diverse than the differences between women and men. The results are presented in six categories with subcategories. Two categories were related to uptake and four categories concerned adherence and maintenance. An overview of the categories and subcategories is illustrated in Figure 1. Quotations within the text are annotated with participant number and the sex of the participant (W or M). 3.1. Uptake. Several factors important for uptake were mentioned in the focus group discussions. Some of them were predominantly expressed as motivators and others as barriers to starting exercise. --- Motivators and Barriers to Starting Exercise. The motives for starting to exercise, discussed within the groups, were preserving health; treating injury or disease; encouragement by clinicians or relatives; and information. Common barriers were health problems; poor self-discipline; environmental barriers; feelings of vulnerability; and societal expectations. Keeping healthy was a major motivator for both men and women. However, there were differences in how men and women expressed their reasons for this. Women talked about preserving health as a necessity in order to manage their responsibilities in their everyday life "I just have to keep on going myself, so It's just that I need to exercise for my survival" (W2), whereas men talked more in terms of keeping fit "You notice that you have to do certain things to... keep fit" (M1). An injury or disease that could improve by continuous exercise (such as being diagnosed with osteoporosis, sustaining fractures, having a rupture of the Achilles tendon, or suffering from heart disease) was mentioned by both sexes as a motivator for starting to exercise. On the other hand, commonly occurring health problems, such as arthritis and pain, were also major barriers for exercising and sometimes the participants needed to adapt their activities to their health situation. Professional advice from a doctor or other health personnel was seen as a driving force for starting to do exercises. For some this was also linked to expectations of what it means to be a good patient, as expressed by one of the men in the study "When you know that you have done what you promised to do, you feel satisfaction in not misbehaving" (M5). Relatives' influence, for example, persuasion of a spouse or a wish to do certain activities with grandchildren, was also a motivator for being physically active in the context of falls prevention that was frequently mentioned by both women and men. While information in the media inspired some of the participants to exercise: "I have started to do an exercise programme that works on strength. I saw it on a TV-show" (W9), other participants said that such information only stayed in their memory for a short time period and seldom resulted in long-term changes. A common barrier, articulated by both women and men, was poor self-discipline, even though they knew they felt better once they got started "You have to overcome certain thresholds in order to get out and exercise" (M4). Environmental barriers, such as the weather, icy or snowy roads, were also common and even though many of the participants liked exercising outdoors, they felt they were less active during the winter. Feelings of being vulnerable or fragile were also expressed as barriers to physical activity by both men and women. However, experience of fractures and fear of falling were particularly experienced as a barrier among the women and caused them to be more cautious in their activities. "Of course if you have suffered fractures a few times you are very scared after that" (W9). The men did not generally speak about fear or vulnerability, but many of them noticed that they had started to avoid doing heavy tasks, if possible, as this often led to pain or discomfort in the body for days afterwards. Other examples of situations where feelings of vulnerability or embarrassment could pose a barrier for taking part in exercises were when an exercise group was too small or predominantly one gender, such as being the only man in a group "I thought I would do it... but I, I dared not (laughs), I' d be alone. But then a few men showed up and eventually I dared to pull myself down into the water" (M1). The participants expressed different views regarding society's perception of older people. Opinions emerged that older people were often illustrated in media as passive individuals, which may affect older people's views of what is expected of them. "Well, it is not easy to find any pictures in newspapers illustrating active older people doing gymnastics and such. There is almost nothing!" (W9). In contrast, other participants expressed the view that older people nowadays are seen as active and that they have more opportunities and resources than older people in former generations. --- Adherence and Maintenance. The categories summarizing factors of importance for adherence and maintenance that emerged in the analysis of the focus group discussions were as follows: programme preferences are personal; confirmational feelings help satisfaction; spirit lifters increase enjoyment; and personal tricks maintain exercise routines. --- Programme Preferences Are Personal. Above all the older participants expressed that the exercises need to be adapted to the individual in order to stimulate adherence. When it comes to how the older participants preferred to do their training, in terms of mode, intensity, challenge, and social context, no general or gender specific patterns were discerned. Instead the discussions pointed to great variation between individuals, and sometimes even within an individual, depending on the situation. Regarding exercise mode, diverse preferences for exercise to be either conducted as a programme on specific occasions, or integrated in everyday activities were expressed. Those who were more accustomed to exercising seemed to prefer training programmes, while the more inexperienced felt that integrated exercise was preferable. The preferred degree of intensity also varied between individuals with no gender pattern; some did not like to sweat, while others thought exercising was not worthwhile if they did not feel tired and sweaty afterwards. What was emphasized was the importance of being able to hold an exercise tempo of one's own. Not being able to keep up or being prevented from exerting effort in their own style and tempo could be frustrating. "But, we (with husband) can almost not walk together (laughs)... you must have your own pace... if one of you is faster, and one is... well...." (W3). Many of the participants took part in group training and did not seem to worry about the level of challenge, e.g., not being able to keep up or perform all the elements correctly. However, they felt that the instructors should emphasize that if someone is not capable of performing a specific exercise, it is acceptable to choose an alternative exercise, as long as one continues to be active. One man even expressed that it may be motivating to see that there are more challenging levels to strive for, even if you currently cannot perform them. "But I think it may actually be stimulating when you can see that there are different levels of challenges" (M4). Some of the participants liked to work out alone while others preferred to exercise together with others. However, such preferences regarding the social context could vary depending on the activity in focus. When exercising in a group setting, a moderate group size composed of peers of the same age was preferred and the social aspect were emphasized as an important motivator. "I think that the social aspect is very important in group training... because you usually attend at specific sessions and you meet the same people. Then you start to talk to each other and you learn to know each other... that's fun" (W4). No obvious gender pattern in preferences for the social context when exercising emerged during the focus group discussions. However, in the personas descriptions constructed by the participants to describe their own motivation profile, women were more prone than men to choose sentences representing a social component to describe themselves, such as "I am social", "I like to exercise together", or "I prefer to work out in a group". --- Confirmational Feelings Help Satisfaction. Confirmation of the effectiveness and efficiency of an exercise were important motivational drivers for both sexes. "You want your efforts to pay off" (W3). Confirmation could be perceived in different ways: from a professional instructor; through visible results; by quantified results, or merely by the feeling of achievement. To have a professional and encouraging instructor was considered very important. The participants appreciated having an expert to trust and ask for advice. Their encouragement and praise could be very stimulating. Participants expressed that they felt more inspired by an instructor of the same age, someone they could identify with, while the sex of the instructor did not matter at all. "Yeah, but... the closer to one's own age, I think that gives more" (M3). Both women and men liked to achieve a secondary goal through their training which meant that they often integrated training with everyday activities to get a visible results, such as working in the garden, cleaning their home, or going out for a walk with a clear purpose in mind. One man expressed that he did not want to "...fool around doing peculiar exercises". "With my age... I do not want to just stand and... I prefer to pay attention to my everyday life and then try to integrate my exercises" (M2). However, a pattern emerged in the participants' personas descriptions suggesting gender differences. It was more common for men to choose expressions representing visible and practical secondary goals, such as "I want to see a concrete result with benefits", "I weave my exercises into everyday activities", or "I like working in the garden or my home". Many of the participants expressed that quantifying the results of their training was motivating. They wanted to keep track of their performance and see objective outcomes. Some of them used instruments for counting steps and calorie expenditure, or tracking walking or cycling routes etc., and this was perceived as stimulating. Men tended to express more fascination with opportunities to quantify their performance, as this conversation between two men illustrates "... And then I start to walk again and it's activated, it measures my speed, and how far I've walked and how many calories I have been spent" (M5). "Interesting!" (M7 comments). "Yes, fun!" (M5). However, there were also women who expressed interest in these kinds of devices. Contrastingly, there were also people who seemed to be totally disinterested in measuring their performance and were perfectly happy just knowing that they accomplished the task. One woman described how she had an exercise bike that could record several things but she was not interested in using these features "I am not doing anything (to record)... I sit and cycle for half an hour, and am fine with that" (W4). The participants expressed that an inner feeling of achievement was important for adherence to exercise. They associated the pleasant sense of relaxation in the body afterwards, and the pure satisfaction in managing to work out, with an intrinsic motivation for continuing to exercise. "Well... when you know you have done something, and you can take a shower and sit down in the armchair with a good conscience " (M5) or "You experience the same thing when doing exercises, -relaxation... when you are active, and feel that you actually manage to be active" (W5). --- Spirit Lifters Increase Enjoyment. The focus group discussions revealed the importance of spirit lifters, in an exercise context, i.e., dimensions that could contribute to enjoyment and motivate older participants to continue to be active. These included outdoor exercises; rhythm and music; exercise equipment; and humor. Outdoor activities were frequently highlighted as very motivating. Both women and men liked the idea of exercising outdoors. "I like to be out there... perhaps because I have been sitting inside all... well quite a few years now... it's great to be able to be outdoors" (W3). Many of the participants engaged in outdoor activities all year around, and some clearly expressed a dislike for exercises indoors. "I had a gym card when I was employed... I went a few times but I do not get the point... biking indoors when you can bike outdoors and get fresh air?" (M5). Many of the examples from the participants' outdoor activities referred to taking a walk, cycling, or cross-country skiing. However, at the beginning of the project strength and balance exercises were still mainly performed indoors and only a few participants could imagine doing these kinds of exercises outdoors. During the course of the study though, some of the participants became inspired by the discussions and started to add balance exercises during their regular walks. These participants could later on give examples of suitable ways to do this, which opened the eyes of the other participants to these possibilities. Another spirit lifter discussed was the use of rhythm and music, which were perceived as engaging and thought of as providing a pleasant atmosphere. "Yes, music, I like music! Absolutely! For me music is very important when exercising. I know that everyone does not like it, but I think the music means a lot" (W6). Although the participants generally liked music during their exercises, they highlighted that the music had to be tailored to the participants' age, and the rhythm had to be in line with the exercises performed. Otherwise, the music became frustrating. "Well, when I do water aerobics... it is often disco music... very rhythmic and the girls (instructors) are too darn good and move in time. But they are not in the water...and when I'm not able to do the movements...I become frustrated" (M8). The use of exercise equipment was considered effective and a motivating element in an exercise context. However, women often noted that they did not want to buy equipment to use in their own homes; it was better if they got tips on how to use things already available in a household. "Not to buy a lot of things... I believe it is better to take what you have at hand" (W1). Men on the other hand preferred specific equipment "Yes, but those things can sometimes fit... have a better grip and such... it's fun with the real stuff" (M1). The participants themselves also commented that the use of equipment might be gender related. Furthermore, the participants explained how the use of exercise equipment could contribute in the creation of a social exercise context. For example, both men and women liked to walk with poles and to meet peers who did the same, as this created a feeling of belonging to a group. --- "It's nice when you walk with poles, when you go out for an ordinary walk, and you meet unknown people you just walk on, but with poles, then you say hello, it's like you belong to a group in a sense. I think that feels nice" (M1). Humor was emphasized as an important source of inspiration. The participants expressed that it should be joyful to exercise. Humor could involve having an instructor joking and teasing, or just giving yourself a chance to enjoy. An exercise session full of laughter and happiness could create a lingering contentment afterwards. "Something I think is often overlooked is the importance of humor. I think it's very important that you get to laugh" (M2). --- Personal Tricks Maintain Exercise Routines. Several tricks or strategies were mentioned by participants to maintain their exercise routines and to prevent lapses and relapses. These tricks involved companionship; routines and challenges; dogs and nature; and safety strategies. Friends or spouses could push each other "When you are alone, then you have to put in an active effort to get going... it will be easier to find arguments against...-it looks like bad weather out there! or whatever" (M1). Having company was also inspiring, "So... well, she is the one who is pushing. And I think it is much, much nicer if we're together compared to if I am going out alone" (M4). The participants gave examples of having routines, such as specific tracks they walked, which helped them to exercise regularly. To set goals, as well as to start with one that felt achievable, was also proposed as a powerful way to get over the threshold when it felt difficult to start, because, once started, it was often easy to continue. "It may feel tough in the beginning, but if I think, -Nah, but today I will just take a trip up to the road. Then suddenly, it feels pretty good after all, and I walk on and continue for about a full hour. And that feels great, although there were setbacks when I began" (W9). A competitive element could also be an incentive for both men and women, but the vast majority created challenges solely for themselves. Other participants emphasized that they did not like competitive elements, "That you are challenging yourself... I think that is just a pain... Then it becomes a mustdo that should be avoided... I have done enough of that in my life" (M4). Dogs were mentioned as great helpers to maintenance exercise. "Yes, we've had a dog, and I miss it immensely, for it was the best exercise help I ever had" (M7). Several of the participants had a dog and felt this kept them active and provided some company. Some of the participants had even taken it upon themselves to go out with someone else's dog, just to get exercise. "Therefore, I have an extra dog that I go out with... then I have to go out... and then I feel good" (W3). Even without a dog the pure experience of nature itself could also be an incentive to go for walks. Feeling the freedom, enjoy listening to the birds, and getting fresh air were described as addictive. Some participants, mainly women, expressed that a fear of falling could inhibit their activity. But they also gave examples of safety strategies they developed in order to feel more secure and thus gave them the confidence to continue their physical activities. One example discussed was walking with poles, to have support on slippery surfaces. Another example was to bring the mobile phone to be able to call someone if and accident should occur. "Just that using a mobile phone... I always take it with me when I go out for a walk, even though I will not be long, especially if I you are somewhere where no one knows where you are" (W6). --- Discussion These multistage focus group discussions aimed to improve our understanding of the preferences community-dwelling older women and men have regarding exercise in the context of preventing falls. Older participants had many diverse preferences and confirmed that individually tailored exercise programmes, in terms of mode, intensity, challenge, and social context, are important. Moreover, important factors for exercise adherence and maintenance included the experience of individual confirmation; different spirit lifters to increase enjoyment; and personal tricks to prevent lapses and relapses. It is interesting to note that no clear gender separation pattern was seen in the majority of these factors (Figure 1). It may seem obvious that older people, just like younger people, have very different opinions on how they prefer to perform their exercises. But all too often, older adults seem to be considered as a homogenous group and recommended the same kind of training regardless of needs or preferences. Studies from Sweden have, as an example, shown that walking is by far the most suggested physical activity for older people getting exercise prescriptions in health care settings [24], although it is known that brisk walking will actually increase the fall risk in those with a history of falls [25]. Previous studies have also shown that when the prescribed exercises are not tailored to the individual, and seniors with varying degrees of physical capacity are expected to exercise together, they are frustrated and may drop out [26]. Our results emphasize the importance of taking the experiences and desires of the older individual into account when fall preventive exercises are planned and prescribed. Previous research indicates that women are more likely to take part in exercise activities to prevent falls [8]; in light of our results that is not due to substantial differences in exercise preferences among women and men. However, a recent literature review suggests that both women and men see women as more receptive to and in more need of fall prevention messages [12]. This may be due to societal norms. Our understanding of gender derives from a social constructionist perspective that gender is created in social relations and dependent on time and context [27]. This means that humans, as social beings, shape themselves in relation to prevailing norms in society for how to be a man or a woman and this is referred to as "doing gender" [28]. According to the norms, men should be independent, self-reliant, strong, tough, and willing to take risks [29]. Physical strength, competition, and risk-taking are all characteristics in "hegemonic masculinity" [30]. These masculinity characteristics often preferred among younger men could have been deemphasized with ageing and a changed life situation among the men in this study. The men were retired, some were widowers, and others were carers for relatives. These circumstances may have contributed to the many similarities in preferences expressed by men and women in this study. Some differences between men's and women's preferences for exercises were expressed in this study; for example, women talked about preserving health, whereas men wanted to keep fit; men preferred exercise equipment in training; men were more prone to emphasize the importance of quantified results as well as having visible secondary goals; women stressed the social component; and the women, but not men, expressed fear of falling as a barrier for exercise. The gendered patterns revealed in this study are not surprising, since they reflect the social gender norm. Nevertheless, these gender differences may be important for professionals to address when "marketing" exercises to prevent falls, in order to attract both men and women. An understanding of how to support motivation in a specific situation requires an understanding of how activities and contexts are experienced by the older person. The Self-Determination Theory (SDT) is a social cognition model that addresses the process through which a person acquires motivation for initiating a new health-related behavior and maintains it over time [31]. According to SDT, maintenance of behaviors over time requires that the person internalizes the behavior in order to develop an intrinsic motivation. In this process of internalization the reinforcement of three innate psychological needs: autonomy, competence, and relatedness, is critical [32,33]. For behaviors to be successfully changed and maintained individuals must come to personally endorse their importance and make an autonomous, intentional, and volitional decision to change. The person must also feel competent and confident enough to change. Finally, relatedness feelings of being included and cared for by others in the relevant context are important [32,33]. These three innate needs are clearly reflected in the categories summarizing how the focus group participants described factors important for adherence and maintenance. By relating the results of this study to the support of psychological needs, we can gain an increased understanding of how to apply key motivating elements in fall prevention exercises for older people. 4.1. Uptake. The motives for starting to do exercises (uptake) expressed by the participants in this study were mainly framed in terms of inherently satisfying goals such as to preserve health and to treat injury or disease, which means the participants valued exercise highly and identified with the behavior. A focus on such intrinsic goals (e.g., being healthy) as opposed to extrinsic goals (e.g., being physically attractive) is believed to increase maintenance over time [32]. Receiving support and receiving adequate information were also cited as important factors for taking up exercise to prevent falls. All these motives are in accordance with previously reported preferences for falls prevention exercises [12] as well as for falls prevention programmes in general [6,34]. The barriers cited as reasons for not participating in exercises were also consistent with previous studies, e.g., health problems or environmental barriers [12] and poor selfdiscipline and lack of motivation [26,35]. However, in our previous review, an anxiety about not being able to keep up with demanding exercises was noticed in many articles [12]. In the present study the participants did not seem to worry about performing correctly at all times, as long as they felt assured by the instructor that doing as good as they could was good enough. On the other hand, feelings of vulnerability, such as fear of falling, or being the only man in a group, were expressed as barriers by the participants. According to the SDT one way to help individuals integrate and internalize a behavior is to support exploration of resistances and barriers and help in the identification of possible solutions [32]. --- Adherence and Maintenance. Based on the results of this study an individually tailored exercise program seems crucial to stimulate adherence and maintenance, which is in accordance with several previous studies [26,[35][36][37][38]. Support for the importance of individualization can also be found in the SDT, which suggests that autonomy and selfdetermination are supported when opportunities for reflection and choices are allowed. A strong sense of autonomy will, in turn, contribute to internalization of values and skills for change, leading to increased intrinsic motivation and consequently also improved exercise adherence [32,33]. Those participants who were more accustomed to exercise seemed to prefer training programmes, while the more inexperienced felt that an integrated exercise, into their daily life activities, suited them better. These results seem to conform well with the results of a previous study, who described their moderately active participants as "functional exercisers" who favored activities perceived as purposeful and practical, often unplanned and low in intensity [39]. Such integrated exercises, where balance and strength training are integrated in activities of everyday routines, have been proven effective in protecting older high risk people from falling and in improving and maintaining their functional capacity [40]. It may be a promising alternative to traditional exercise programmes for many older people who are not used to exercising or who do not want to exercise in a group. Earlier studies have suggested that low intensity exercises may facilitate participation [6]. However, the participants in this group expressed preferences for both "calm" and "intense" exercise, again reminding us of the need to consider individual preferences. The importance of social interaction, and group adherence, is often found to be an important factor to enhance participation in exercise among older people [6,12]. According to the SDT social interaction will naturally contribute to feelings of relatedness, which is important for intrinsic motivation to develop [33]. Social interaction was emphasized as important by some in this study, in particular by women in their own persona descriptions, while some of the participants preferred to exercise alone. Similar findings have been found in a previous study from the UK, in which more women than men were likely to attend group sessions [41]. In another study comparing exercise habits in older people in Sweden and Ireland, the Swedish participants tended to have a higher regard for solitary exercise in comparison to older persons from Ireland, where physical activity was often seen as a means to socialize [39]. To receive confirmation of being able to exercise and that one's efforts actually give results was expressed as a very important element in an inspiring exercise context.

Background. Several factors have previously been identified to positively influence the uptake and adherence for fall prevention exercise programmes. There is, however, a lack of studies investigating if men and women differ in their views and preferences for fall prevention exercises. Aim. To explore exercise preferences and motivators of older community-dwelling women and men in the context of falls prevention from a gender perspective. Methods. Workshops including multistage focus group discussions were conducted with 18 older community-dwelling people with and without history of falls. Participants were purposively selected and divided into two groups. Each group met on six occasions over a period of five months. Participatory and Appreciative Action and Reflection methodology was used to guide the discussions. A qualitative content analysis approach was used in the analysis. Results. Older participants had many diverse preferences and confirmed that individually tailored exercise, in terms of mode, intensity, challenge, and social context, is important. Moreover, important factors for exercise adherence and maintenance included the experience of individual confirmation; different spirit lifters to increase enjoyment; and personal tricks to maintain exercise routines. The individual differences within genders were more diverse than the differences between women and men. Conclusion. Exercise interventions to prevent falls should be individually tailored, based on the specific needs and preferences of the older participant, and do not appear to require gender specific approaches. To increase adherence, intrinsic motivation for exercise may be encouraged by competence enhancing confirmations, energizing spirit lifters, and practical tips for exercise maintenance. The study provides an awareness about women's and men's preferences for fall prevention exercises, and this information could be used as guidance in designing inclusive exercise interventions.

studies have suggested that low intensity exercises may facilitate participation [6]. However, the participants in this group expressed preferences for both "calm" and "intense" exercise, again reminding us of the need to consider individual preferences. The importance of social interaction, and group adherence, is often found to be an important factor to enhance participation in exercise among older people [6,12]. According to the SDT social interaction will naturally contribute to feelings of relatedness, which is important for intrinsic motivation to develop [33]. Social interaction was emphasized as important by some in this study, in particular by women in their own persona descriptions, while some of the participants preferred to exercise alone. Similar findings have been found in a previous study from the UK, in which more women than men were likely to attend group sessions [41]. In another study comparing exercise habits in older people in Sweden and Ireland, the Swedish participants tended to have a higher regard for solitary exercise in comparison to older persons from Ireland, where physical activity was often seen as a means to socialize [39]. To receive confirmation of being able to exercise and that one's efforts actually give results was expressed as a very important element in an inspiring exercise context. Confirmation could be seen through visible results of practical activities, by assessment and evaluation of performance, or simply by feelings of achievement. Confirmation could also be obtained through feedback from a professional instructor who offers help to find the optimal exercise and conveys knowledge and confidence that the exercises performed are safe and effective. The characteristics of the instructor may play a role in influencing participants' attendance to exercise classes [42] and advisory support has previously been raised as important to strengthen participants' self-efficacy [43]. According to Bandura, the experience of mastery is the most important factor determining a person's self-efficacy, and in general people with high self-efficacy are more determined and persist longer than those with low self-efficacy [44]. In a particular activity confirmation may contribute to feelings of mastery or competence, which, according to the SDT, is important for successful behavior change [32,33]. The participants in this study gave several examples of elements in the exercise they thought of as spirit lifters. These elements had the ability to provide energy and joy to the training and bring feelings of both vitality and relatedness. Outdoor activities were such spirit lifters. Recurrent preferences for outdoor exercises among Swedish seniors of all lifestyle categories have previously been reported [39,45]. The importance of being outdoors and in contact with nature has been reported to enhance energy and induce greater vitality even when controlling for factors such as social interaction [46]. Maybe being outdoors could also contribute to important feelings of relatedness to the nature. Other spirit lifters mentioned by the participants were rhythm and music, exercise equipment, and humor. We believe all these elements may contribute to feelings of relatedness by creating a social context and building a bond between participants. However, feelings of relatedness were also seen in those who exercised alone. One example of this was how the use of walking poles created a sense of belongingness to a group of pole-walkers. --- Methodological Considerations. A strength of this study was the design with multistage focus group discussions [14], which gave rich data through the process of meeting with community-dwelling participants on several occasions. The participants had the opportunity to reflect upon their exercise preferences, both during the focus group discussions and in between the sessions. It was sometimes difficult to know if the participants talked about physical activity in general, or about falls prevention in particular. However, all participants were aware that the study focused on exercise in the context of preventing falls and this was emphasized by the researchers during the discussions. Another strength was the use of the PAAR methodology [17]. By encouraging informants to express themselves in a positive way, it has the potential to generate rich data about perceptions, feelings, experiences, motives, and attitudes. On the other hand, this appreciative approach may potentially result in an underestimation of negative thoughts and experiences, which might also be of importance. We believe that the appreciative approach of the PAAR methodology contributed to the friendly and positive atmosphere during the workshops. The attendance rate at all workshops was very high and the participants expressed that they liked coming to the meetings. Yet another strength in this study was the sampling strategy, which gave variation in history of falls and experience of exercise in both men and women in each group. However, the transferability of the findings may be limited because most of the participants were retired white-collar workers, were in general well educated, and were mainly resource-rich with previous occupational experiences. Class differences in our study group were minor, which could have contributed to the similarity in exercise preferences. It should also be noted that participants in our study had enrolled in a fall preventive study with focus on exercising and thus shared a motivation for exercise. In this respect a similar interest may had a stronger influence than gender on the results. The researchers involved were from different fields of expertise, physiotherapy, informatics, exercise physiology, and computing sciences, which enhanced both data collection and analysis [47]. The researchers brought experience from different methods and facilitating techniques that were applied during the workshops. The mix of methods helped the participants share their thoughts and contribute in various ways, which greatly enriched the data collection. Moreover, the author's diverse preunderstandings regarding falls prevention, balance and strength exercises, motivation for exercise, and gender research complemented each other and helped deepen the discussions within the author group, and thus the interpretation and analysis of the data. The cultural context of the study could affect the generalizability of the results. As norms for how to be a man and a woman differ by age and cultural context, analysis of gender in older men and women needs to consider the wider context of norms for "being old" in the respective time and culture. In Sweden, older people are encouraged to stay active in order to avoid ill health. Investigations using accelerometers have shown that 49% between 65 and 75 years are physically active with a moderate or high intensity for at least 30 minutes during an average day, with very small gender differences [45]. The project was carried out in Ume<unk>, Northern Sweden (latitude 63 <unk> N), where snow and icy roads are common in the winter period of November to April. Nature/forest is easily accessible, both in terms of walkability and of "legal rights" to walk everywhere/of public access. The results may thus not be generalizable to countries where the cultural norm of exercise, the gender norm, and attitudes to outdoor activities in old age differ. --- Conclusions Older people's preferences for exercise programmes are personal, and the individual differences are greater than the differences between men and women. Individually tailored exercise, in terms of mode, intensity, challenge, and social context, is important. In order to enhance uptake, adherence, and maintenance in fall prevention exercise professionals need to see the individual older person and apply autonomy-supportive approaches to encourage growth of intrinsic motivation. Besides individually tailored exercises such approaches should, according to older men and women, provide confirmational feelings, different spirit lifters to increase enjoyment, and personal tricks to maintain exercise routines. The study provides an awareness about women's and men's preferences for fall prevention exercises and this information could be used as guidance in designing individual tailored interventions that are inclusive, hence providing uptake and adherence for both sexes. --- Data Availability The interview data used to support the findings of this study are restricted by the Regional Ethical Review Board Ume<unk>, Sweden, in order to protect patient privacy. Data are available from the corresponding author upon request. --- Conflicts of Interest The authors declare no conflicts of interest. --- Stem Cells International Hindawi

Background. Several factors have previously been identified to positively influence the uptake and adherence for fall prevention exercise programmes. There is, however, a lack of studies investigating if men and women differ in their views and preferences for fall prevention exercises. Aim. To explore exercise preferences and motivators of older community-dwelling women and men in the context of falls prevention from a gender perspective. Methods. Workshops including multistage focus group discussions were conducted with 18 older community-dwelling people with and without history of falls. Participants were purposively selected and divided into two groups. Each group met on six occasions over a period of five months. Participatory and Appreciative Action and Reflection methodology was used to guide the discussions. A qualitative content analysis approach was used in the analysis. Results. Older participants had many diverse preferences and confirmed that individually tailored exercise, in terms of mode, intensity, challenge, and social context, is important. Moreover, important factors for exercise adherence and maintenance included the experience of individual confirmation; different spirit lifters to increase enjoyment; and personal tricks to maintain exercise routines. The individual differences within genders were more diverse than the differences between women and men. Conclusion. Exercise interventions to prevent falls should be individually tailored, based on the specific needs and preferences of the older participant, and do not appear to require gender specific approaches. To increase adherence, intrinsic motivation for exercise may be encouraged by competence enhancing confirmations, energizing spirit lifters, and practical tips for exercise maintenance. The study provides an awareness about women's and men's preferences for fall prevention exercises, and this information could be used as guidance in designing inclusive exercise interventions.

Introduction In politics and social systems, norms of social regulation, including areas of political structure, human rights, social context of self-realization of the individual, religious life, ethnic and national tolerance, freedom of information, and creativity, are being transformed, which contributes to obtaining new values, ideals. As a result, every individual constracts social and personal identity in a complex way, resulting in the forming of goals, a vision of the future, and implementation scenarios (Alexeenko, 2009;Marinov, 2008;Naumova, 1995). Student youth is the most dynamic, flexible, and active part of society, because this social group is the most subject to various social processes, including globalization and westernization. Students on the stage of building life strategies and personal values participate in communication and actively use its new forms and technologies (Zubok, 2020). Modern young people are increasingly focused on individualized scenarios and the realization of personal interests, not collective ones, which contribute to building life and career paths (Isupova & Utkina, 2016;Miheeva, 2014;Osipova & Enveri, 2016). Global tendencies of democratization and liberalization in the last decades have caused significant transformations in the lifestyle of Russian people and their values, and it has found a reflection in the construction of gender roles -changing the expression of marriage and family, former main priorities. However, besides new opportunities, there are a lot of new barriers that young people face, including the fact that young women have to resist certain difficulties and obstacles because of gender stereotypes. Contemporary Russian society is characterized by a high degree of continuity of Soviet models, where a woman got the right to work and regulate reproductive scenarios on her own. Despite the extension of patriarchal values, the state is interested in gender equality, which supports certain gender contracts and imposes a double burden on a woman -caring for the family and self-realization at work (Kotomanova, 2013;Maksimova, 2007). Undoubtedly, modern young people are already firmly integrated into market relations, and consumer practices, which links them with their Western peers. However, a special historical experience, the Soviet past, conflicting with the aggravated inequality of the post-Soviet era, creates a special ground for the formation of youth identity. Thus, this study of life strategies of female students reveals the direction of life choices of named social group, as well as the features of constructing the future by female students of Russian and Western European cultures. --- Life strategies: lines of theoretical research In the scientific definition, the concept of "life strategies" requires conceptualization and clarification. In its most general form, it is interpreted as a system of perspective ideas and orientations, as well as plans and value orientations that determine the individual's behavior and construct the future. There are several basic approaches to understanding and interpreting the concept of "life strategy" by Russian scientists. For example, K.A. Abulkhanova-Slavskaya believes that a person is an active self-regulating, and self-organizing subject. The life strategies of the individuals differ in how the conditions of life are changed and transformed by the individual values. According to K.A. Abulkhanova-Slavskaya, each person's life path is constructed individually and therefore unique. I.N. Tartakovskaya also notes that a life strategy is a conscious behaviour of an individual, the purpose of which is to achieve life goals and realize interests and values. The life strategy is determined, according to the sociologist, by the individuality and the social background within which the life of the individual takes place (Maximova et al., 2018;Osipova & Enveri, 2016). Researchers T.E. Reznik and Y.M. Reznik argue that life strategy is a dynamic category, which is characterized by a fairly built set of values, goals, and behavioral attitudes. Individuals realize personal guidelines through their behavior. The researchers emphasize that life strategy has two sides: subjective and objective. The subjective one is expressed in uniqueness, originality, and suprasituation of individual meanings and goals, and the objective one is in the presence of culturally determined patterns, norms, and values acquired by the individual in the process of socialization. Researchers distinguish three types of life strategies: the strategy of well-being, the strategy of success, and the strategy of selfimprovement. Strategies are identified based on several institutional features (Reznik & Reznik, 1995). In Western sociology, the approach to the analysis of life strategies is different: there is a theoretical approach, called Life Course Theory, also known as Life Course Approach or Life Course Perspective, which was developed in 1960 to analyze people's lives within structural, social, and cultural contexts. The origins of this approach are considered to be pioneering studies of the 1920s, such as W. Thomas and F. Znaniecki's "The Polish Peasant in Europe and America" and Karl Mannheim's essay "Problem of Generations". This approach researches the history of a person's life and various events, which influence future decisions (for example, marriage, and involvement in crime). A life path is defined as "the sequence of socially defined events and roles that an individual plays over time". In particular, the approach focuses on the connection between people and the historical and socio-economic context in which these people lived. G. Elder's work "Children of the Great Depression" ( 1974) is a classic study of this area, in which the author addressed the problem of the delayed influence of the Great Depression on the generation that grew up in its conditions, establishing a connection between the changing world and people's lives, abandoning the concept of socialization (Shanahan et al., 2004). Within the framework of this article and empirical research, the author relies on the approach to understanding the "life strategies" of T.E. Reznik and Y.M. Reznik, in which strategies are defined as dynamic systems in which systems of value orientations and comprehended behavioral attitudes are built. --- Methods and Materials Empirical research of life strategies of female students is based on semistructured interviews, which helped to determine individual opinions of interviewees as representatives of the given social group, and content analysis of Instagram accounts of female students. The research object is women aged 18 to 26 from Russian and Western European universities. So, within the framework of interviews, conversations were held with 6 girls studying at universities in different countries of Western Europe, and 6 girls studying at Russian universities. Thus, informants expressed various versions of the future they expect to have. According to analyzed interviews, it should be noted that the family still plays a very important role in lives of young woman, regardless of their country of residence: "family is an obligatory stage in life.... relatives should be with you" (In.9); "it's [family] on the top of my priority list" (In. 1). Moreover, based on the results of interviews, we can conclude that, unlike European female students, Russian young women take marriage more seriously, believing that this is an important, sometimes mandatory, stage and a natural continuation of a successful relationship with a partner. Young European women do not consider marriage as a mandatory stage, they believe that it is possible to live without civil marriage if the partner is really "your person". Girls are guided by the idea that romantic relationships and family are not always coherent. Even long-term romantic relationships are considered in the context of specific type of interactions, and not as a preparatory stage for marriage. However, there are also other opinions. In terms of the topic of parenthood and the need for motherhood, experience is very diverse -unwillingness to have children at all, the possibility of having a child (children) out of marriage, the birth of a child (children) only within marriage. No unique scenarios regarding reproductive behavior relative to the country of residence have been identified. Furthermore, when it comes to the partner, all the interviewed women are primarily looking for understanding, support, unity of goals and guidelines. Thus, it can be noted that in Russia the transformation of Soviet gender contracts, in which everyday practices of postponing motherhood for 10 years and the creation of a family were entrenched. And it is important to point out that the image of a partner in a romantic relationship is identified primarily with the role of a friend, a likeminded person, rather than a breadwinner and father. "In a long-term relationship mutual assistance, understanding, empathy are important..., compromises are needed to keep the relationship in balance,... and also a respect is needed" (In.12). "I think the same interests are important, although maybe even this is not so important, rather the same values and life guidelines" (In.1). There are similar gender frames of young women in Europe and Russia, and, at the same time, a visible transformation of the scenarios of Russian women towards the pluralization of gender relations. If during the Soviet period the dominant contracts for women were "traditional" and "working woman", then the results of the interviews point out the dominance of new female identities based on freedom of choice. Opinions about the relationship between work and family in lives of female students differ. Some women believe that "balance is a difficult process, but 79 possible" (In.4), others note that it is impossible to achieve it in full, you need to choose a priority. For Russian women, life scenarios are less clear than life paths of Western female students: girls from Europe describe their future more specifically, they indicate the importance of the following aspects -self-development, travel, new acquaintances and gaining experience: "now I am looking for an internship in Belgium or Austria, so that later I can return to France with some experience"(In.3). There are reflections about the future of Russian girls with complete uncertainty: "I "The world is so big, and life is so short" (In.6) In the following paragraphs, the data obtained by using the visual method will be analysed. As part of the study, attention was paid to the study of elements of life scenarios of female students, which they represent in the space of Instagram. The analysis of life models is revealed through the concept of visual sociology of Polish researcher P. Sztompka, according to whom the analysis of visual data, In addition, life well-being strategies are more characteristic of traditional and crisis societies, determining a lower level of development of the production sphere, an economic crisis, and authoritarian forms of government. It is typical to a greater extent for manual workers or marginalized strata. So, in Russian society, which is in a state of socio-economic crisis, well-being strategies are indeed more common among female students. Self-development strategies are more common for educated strata of the post-industrial society: artists and cultural figures, the intelligentsia. Thus, it was revealed that among the students of Russian and Western European universities, various and heterogeneous life strategies prevail. For Russian women, the strategy of life well-being is more typical, and for European women, the strategy of life success is closer. --- Conclusion As a result of the empirical study, which consisted of two stages: interviewing female students and visual analysis of their profiles on Instagram, both general discursive strategies, features and differences in the choice of life scenarios were identified. It was revealed that the prevailing life strategy of Russian women included the desire for a comfortable, stable life that satisfies all emerging needs, while for European students the strategy of active life and success has become more common, implying a life full of events, the desire for new discoveries and activities. Important components of the life strategies of Russian women are comfort, stability, financial well-being, satisfaction of basic needs and security needs. At the same time, female students from Europe focused on cognitive needs, travel, and creativity. The conditionality of such differences is more associated with the socio-economic and cultural environment of girls, which they are surrounded by.

Contemporary society provides people with various ways of personality development. Young people are most susceptible to changes that occur in economics, politics, and technology. Different types of life strategies are formed and realized by young women and men, and they eventually transform public consciousness in general. The article is devoted to the analysis of life strategies and scenarios of female students living in Russia and European countries. For this purpose, a series of in-depth interviews with girls was conducted, as well as a visual analysis of the accounts of young women on the social network Instagram. Today, the process of westernization is observable -the adoption of Western European structures, practices, and lifestyles by non-European societies and countries has become a significant cultural phenomenon. Thus, based on the data obtained, it is possible to define a life strategy as a system of ideas and guidelines of the individual, aimed at the conscious design of the future. It determines and directs the behavior of individuals, reflecting the idea of the future and life path. A life strategy includes not only orientations, but also ways to achieve and realize certain goals and expectations.

INTRODUCTION During the COVID-19 pandemic, fear and uncertainty have dominated the social landscape and jeopardized the capacity of young people to envision a positive future and maintain purpose in their lives (1,2). Mounting evidence points to the negative consequences of the pandemic on adolescent and young adults' mental health (3)(4)(5)(6)(7)(8). Simultaneously there has been an upsurge in all forms of violence associated with despair, further polarizing society (9,10). A recent Canadian study indicates that in 2017 students aged 16-18 were at higher risk of supporting violent radicalization (VR), compared to students aged 19-21 who were at higher risk in 2015 (11). This age trend suggests that societal challenges over time may render younger people more vulnerable to VR. It is therefore essential that we invest in ways to support youth to reduce violence in society. This is particularly urgent in the present context of uncertainty and life disruption caused by the COVID-19 pandemic. In a positive youth development (PYD) perspective, the present study investigates how a youth's positive identity, in terms of positive future orientation and presence of and search for a meaning in life, is associated with support for VR in a sample of Canadian college students during the COVID-19 pandemic. Violent radicalization (VR) is a complex and multidimensional phenomenon (12) defined as a process whereby an individual or a group increases support for violence as a legitimate means to reach a specific (e.g., political, social, and religious) goal (13). Although the association between support for VR and violent action is not linear, population-wide attitudes toward legitimizing some forms of violence may increase social polarization and fuel the emergence of extremist groups, thus providing a narrative to channel despair and rage in vulnerable individuals (14,15). Schools and colleges are important radicalization vectors and recruitment sites (16). Data show that 75% of lone actors in North America and Europe received post-secondary education (17). The present study focuses on late adolescents and early adults (aged [16][17][18][19][20][21][22][23][24][25] attending colleges in Québec (Canada). In Québec, colleges (known as Cégeps) are a public educational institution placed between high school and university. Their purpose is to assist in continuing education. These types of schools provide 2-3-year preuniversity programs and vocational career programs to younger students (starting from age 16) as well as older professionals; roughly 70% of students attend college directly after high school in Quebec (18). Schools and colleges are a promising location for violence primary prevention programs, as they can reach a large portion of youth (19). Unfortunately, empirical evidence on best practices and guidelines on how to support students and prevent VR are limited (20). A better understanding of risk and protective factors associated with support for VR among adolescents and young adults is crucial to inform the development of prevention programs in educational settings. Research on support for VR among young people may benefit from the adoption of a Positive Youth Development (PYD) framework aimed at fostering youth strengths across multiple domains of functioning (21). This framework postulates that the alignment of strengths and resources across contexts and developmental stages is the key for successful adaptation and transition to adulthood (22)(23)(24)(25). PYD proposes that young people are not passive in the contexts that are likely to affect their development but active participants and is based on the belief that all adolescents have the potential for thriving if their developmental assets are nurtured. PYD promotes a public health approach to the study of VR, focusing on a resiliencebased perspective to healthy development as a core component for preventive programs and policies (26). Focusing on the promotion of different paths of positive development may shed light on new non-violent ways for youth to contribute and develop a sense of belonging to their society. In addition, the promotion of PYD paths would engage youth in personal and social transformation, reducing feelings of despair and consequently disengaging them from VR processes. In line with a socio-ecological perspective, this framework acknowledges that youth-specific assets and person-context relations promoting a positive development vary across socio-cultural settings and historical times. Indeed, our global world is changing rapidly, with many societies characterized by increased violence, growing xenophobic sentiments, and socio-cultural inequalities and uncertainty, as highlighted by the COVID-19 pandemic (27)(28)(29). Prior evidence shows that self-uncertainty can motivate enhanced conviction on socially important issues and increase the appeal of extreme groups who offer a way to restore one's selfassurance and self-respect (30). In the present challenging global context, it is important to identify how to align youth strengths and their ecological contexts to promote their thriving through institutional democratic systems. In the PYD perspective, one developmental asset that may be timely and important to investigate in the present context of uncertainty and social polarization is youth's positive identity, defined as youth's ability to have a positive and optimistic vision of the future and to report a sense of meaning and purpose in life (25,31,32). Positive identity in youth may represent a form of empowerment, leading to reduction in feelings of despair, engagement in positive action for personal and social transformation, and ultimately prevent involvement in VR processes. A positive future orientation and the presence of a sense of purpose in life are crucial to human wellbeing and become especially important during late adolescence and early adulthood, when youth are exploring life and career options (33)(34)(35). A positive future orientation refers to the extent of one's positive attitudes toward the future (36) across different life domains (e.g., personal, community, and world). A growing body of literature suggests that youth who report a negative future orientation are at greater risk of engaging in violence and supporting VR than youth who are more positively oriented to the future (36)(37)(38)(39). A positive future orientation fosters hope and represents an important asset during an age of growing insecurities and anxieties, especially for late adolescent and early adults, thus contributing to a reduction of support for VR (38). Presence of meaning in life, defined as "the sense made of, and significance felt regarding, the nature of one's being and existence" (p. 81) (40) has been associated with wellbeing (41,42) and fewer problematic health behaviors (43). However, it has yet to be investigated in association with support for VR. The presence of a meaning in life is accompanied by the search for meaning in life, defined as the "strength, intensity, and activity of people's desire and efforts to establish and/or augment their understanding of the meaning, significance, and purpose of their lives" (p. 200) (44). In the North American context, higher search for meaning has been associated with lower wellbeing, especially at an older age, when the search and exploration of identities and purposes in life are less normative (44). However, findings on the association between search for meaning in life and wellbeing are mixed, suggesting socio-cultural and contextual variations (45,46). For instance, in their cross-national study Lin and Chan (46) found that the search for meaning in life in adverse circumstances appears to be more useful than in benign conditions, in support of the idea that a deficiency search is functional in that it can contribute to an improvement of one's life conditions. The authors reported that search for meaning in life and wellbeing were not significantly associated in less collectivistic and more peaceful societies (e.g., Sweden, Australia). Under very uncertain circumstances, radical discourses may constitute a tempting response for youth in search of meaning (30,47). Although this may lead in some cases to the legitimation of violence, in other situations it may lead to civic engagement and social advocacy (26,48), two important means by which adolescents can contribute non-violently to their development and society (23). The COVID-19 pandemic represents a significant adverse event and hardship for young people which is likely to elicit a quest for meaning (49), but the role of a search for a meaning in life on young people's support for VR during the pandemic has yet to be investigated. The presence of a meaning in life in the current social context can protect from feelings of despair and hopelessness (50). Simultaneously, a positive vision of the future reveals an optimistic perception of life and may be associated with the presence of meaning in life (50,51). The two constructs may interact to contribute to a better understanding of what factors are protective against VR processes in young people during these challenging times (50). In addition, the role of search for a meaning in life as risk or protective factor for VR may also depend on their vision of the future, in that a positive vision of the future may support a constructive search for meaning in life, thus reducing the risks of involvement in VR processes. In light of mounting evidence on the positive association between depression and support for VR (52)(53)(54)(55)(56) as well as the increase in depressive symptoms in youth during the pandemic (3,4), depressive symptoms are an important and timely variable to consider. Depression involves a lack of hope and pessimism that might make extremist ideologies that promote agency and empowerment especially appealing and is associated with a negative vision of the future and lack of purpose in life (57). Although the three constructs are associated in depressive states, future orientation and meaning in life describe existential states stemming from a social and cultural environment. These are distinct from depression, which refers to a more generalized pessimism and negativity based on erroneous negative emotional interpretations. The present study aims to: (1) examine the association of a positive future orientation, presence of and search for a meaning in life with support for VR in a sample of Canadian college students during the COVID-19 pandemic; and (2) investigate the moderating role of future orientation in the association between presence of and search for a meaning in life and support for VRcontrolling for severity of symptoms of depression. We expect a positive future orientation and a higher presence of a life meaning to be negatively associated with support for VR. Given the scarce and mixed findings on the role played by search for a meaning in life on support for VR, two alternative hypotheses will be tested. The first hypothesis theorizes search for a meaning in life as a risk factor for support for VR and is based on findings on the negative impact that search for a meaning in life has on wellbeing in the North-American context (44). Alternatively, search for a meaning in life could be a protective factor for VR and this would support findings that documented the protective role that search for a meaning in life can have among young people under challenging and uncertain circumstances (e.g., COVID-19 pandemic) (46). We also expect the associations between presence and search of a meaning in life with support for VR to vary depending on levels of future orientation, with students reporting lower support for VR at higher levels of reported presence of a meaning in life, especially in the presence of a more positive future orientation (50). We expect a more positive future orientation to mitigate or strengthen the association between search for a meaning in life and support for VR, depending on whether search for a meaning in life is a risk or a protective factor for support for VR. --- MATERIALS AND METHODS --- Participants A total of 3,100 participants aged between 16 and 25 years (M age = 18.6; SD age = 1.76) were recruited across 18 colleges located in different areas of Quebec, Canada. Participants' sociodemographic characteristics are presented in Table 1. Of the 3,100 students that completed the study, 68.0% (n = 2,107) identified as women, 27.5% (n = 852) as men, and 2.5% (n = 79) as transgender or gender-diverse (TGD). An additional 2.0% of participants (n = 62) chose not to report their gender. Most students were non-immigrants (i.e., born in Canada from Canadian-born parents). First-generation and second-generation immigrants had varied ethnic backgrounds: 35.0% (n = 163) came from Europe, 16.75% from Asia (n = 78), 12% (n = 56) from South America, 9.66 % (n = 45) from North America, 8.80% (n = 41) from Northern Africa/Maghreb, 8.15% (n = 38) from Sub-Saharan Africa, 8.37% (n = 39), 5.15% (n = 24) from the Middle East, and 4.51% from the Caribbean (n = 21). Twenty-three students did not report their birthplace. Given the high heterogeneity of the students and of the regional groupings in the analyses, we controlled for generational status rather than for birthplace, which were significantly correlated (r = 0.26, p <unk> 0.001). Most students reported having no religion (59.7%, n = 1,851). A substantial minority practiced Christianity (30.0%, n = 929), and a small minority practiced Islam (4.0%, n = 124) or another religion (5.7%, n = 175). Twenty-one people did not report their religious beliefs. About half of students reported never having financial difficulties in their household growing up (52.7%, n = 1,633), 33.7% (n = 1,045) reported sometimes having financial difficulties, and 13.4% (n = 414) reported often having financial difficulties. Eight people did not report this information. A total of 57.1% of students reported French as their primary language (n = 1,770), 14.7% reported English (n = 457), and 27.0% (n = 836) reported both English and French. Thirty-seven people did not report their primary language. --- Covariates Depression Depression was measured by using the 15-item scale of the Hopkins Symptom Checklist-25 (HSCL-25) (64). Items are rated on a Likert scale from 1 (not at all) to 4 (extremely), and a total score is obtained by computing the mean of all items. The clinical cut-off is set at 1.75 (score range from 1 to 4). The HSCL-25's psychometric qualities have been well-established (65). In this study, Cronbach's alpha and McDonald's Omega for the depression score were both 0.92. Depression is included in the analyses as a potential confounder in the relationship between the independent variables and support for VR. --- Socio-Demographic Variables Participants answered specific questions on their sociodemographic background, providing information on their age, gender (i.e., man, woman, transgender, and genderdiverse), religion beliefs (i.e., non-religious, Christian, Muslim, and Other), generational status (i.e., first-generation immigrant, second-generation immigrant, and third generation immigrant/non-immigrant), language (i.e., French, English, Both), financial difficulties (yes/no), and college. All sociodemographic variables are included and controlled for in the analyses because they are potential confounders in the relationship between the independent variables and outcome. --- Data Analysis Analyses were performed using R software (66). Descriptive information for the sample was summarized using means and standard deviations for continuous variables and counts and proportions for categorical variables. We used ANOVA and Chi-squared tests to examine differences in future orientation, presence of and search for a meaning in life and support for VR according to students' socio-demographic characteristics. Missing data for each study variable are reported in Table 1. Missing values for both continuous and categorical variables were imputed using multiple imputations by chained equations (n = 5) (67). Sensitivity analysis indicated that missing data and multiple imputations did not alter the observed patterns of associations. We used linear mixed-effects models to test the separate contributions of future orientation, presence of and search for a meaning in life to support for VR (three models, one per each predictor), controlling for socio-demographic variables (i.e., gender, age, immigrant status, religion belief, financial difficulties, language) and depression, while accounting for the clustered nature of the data (i.e., students nested within colleges). Students from the same institution are expected to respond more similarly than students from different institutions as there are other institutional factors that can impact the response. Therefore, our statistical analysis accounts for this intra-institution correlation by using multi-level regression analyses. Specifically, a series of linear mixed-effects models were performed using lme4 package in R (68). Prior to all analyses, scores of support for VR, future orientation, presence of and search for a meaning in life were standardized to a mean of 0 and a standard deviation (SD) of 1. To investigate the independent associations of future orientation, presence of and search for a meaning in life with support for VR, we implemented a linear mixed-effects model with support for VR as the dependent outcome and future orientation, presence of and search for a meaning in life as independent variables, controlling for depression, gender, age, immigrant status, religion, financial difficulties and language as fixed effects and colleges as a random effect. Last, we conducted moderation analyses. Future orientation was categorized into quintiles to simplify the interpretation of results from effect modification analyses and allow for a stratified analysis by levels of future orientation. We ran two separate models. In the first model, we included a two-way interaction between presence of meaning in life and future orientation, controlling for the same covariates and search for a meaning in life to explore potential effect modifications. Subsequently, we tested the two-way interaction between future orientation and search for a meaning in life, controlling for covariates and presence of a meaning in life. --- RESULTS At the bivariate level, Pearson correlations among our variables of interest indicated that search for a meaning in life and depression were both positively associated with support for VR, whereas future orientation and presence of a meaning in life were negatively associated with support for VR as well as with depression. Search for a meaning in life was positively associated with depression, and negatively associated with presence of a meaning in life and future orientation. Presence of a meaning in life was positively associated with future orientation. In terms of age, older students reported higher presence of meaning in life but a lower future orientation (Table 2). The association of sociodemographic variables with our variables of interest are reported in Table 3. --- Association of a Positive Future Orientation, Presence of and Search for a Meaning in Life With Support for VR Transgender and gender diverse students reported higher support for VR compared to students who identified as women. Second-generation immigrant students reported greater support for VR compared to third-generation immigrant and nonimmigrant students. Older students, religious students and francophone students had lower levels of support for VR compared to those who were younger, did not identify as having a religious affiliation, and were anglophone. Financial difficulties were not associated with support for VR at the multivariable level. After adjusting for potential confounders (i.e., gender, age, religion, generation, financial difficulties, primary language and depression, and after accounting for variability across colleges), future orientation was significantly and negatively related to support for VR. A 1-SD increase in future orientation was associated with a 0.10 SD decrease in support for VR (<unk>2 (df) = 24.63(1), p <unk> 0.001; <unk> = -0.10; SE = 0.02, p <unk> 0.001). This effect persisted once controlling for presence and search for meaning in life (<unk> = -0.07, SE = 0.02, p = 0.001; see Table 4). Presence of meaning in life was significantly and negatively associated with support for VR after adjusting for potential confounders. A 1-SD increase in presence of meaning in life was associated with a 0.10 SD decrease in support for VR (<unk>2 (df) = 25.36(1), p <unk> 0.001; <unk> = -0.10; SE = 0.02, p <unk> 0.001). This effect persisted when controlling for future orientation and search for meaning in life (<unk> = -0.06, SE = 0.02, p = 0.004; see Table 4). With regards to search for a meaning in life, for 1-SD increase in search for meaning in life, there was a 0.04 SD increase in support for VR (<unk>2 (df) = 5.08 (1), p = 0.024; <unk> = 0.04; SE = 0.02, p = 0.029). However, after controlling for future orientation and presence of meaning in life, the association between search for a meaning in life and support for VR was not statistically significant (see Table 4). --- Moderating Role of Future Orientation in the Association Between Presence of and Search for a Meaning in Life and Support for VR The association between presence of a meaning in life and support for VR was stronger at higher levels of future orientation. Specifically, presence of a meaning in life was significantly associated with support for VR in students who reported a future orientation score in the fourth quantile (see Table 5). Specifically, 1-SD increase in presence of a meaning in life was associated with decreases in RIS scores of 0.18 (<unk> = -0.18, SE =0.06, p = 0.003). A similar trend can be observed for future orientation scores in the third and fifth quintiles (see Figure 1), although these associations were borderline significant (p = 0.05) (Table 5). The association between search for meaning in life and support for VR did not vary across levels of future orientation (Table 5). Although the effect estimates are low and borderline statistically significant, a trend toward search for a meaning in life being a risk factor, rather than a protective factor, for support for VR especially at higher levels of future orientation can be observed (p = 0.05). --- DISCUSSION The present study aimed to examine the association of one's positive identity (i.e., future orientation, presence, and search of meaning in life) with support for VR in a sample of college students in Quebec (Canada) during the COVID-19 pandemic in a PYD perspective. The pandemic is responsible of high levels of uncertainty and life restrictions (e.g., lockdown, social distancing) which have been associated with increasing feelings of helplessness and hoplessness as well as with increased psychological distress worldwide, especially among young people (1,2). In this context, restoring a meaning in life and positive vision of the future may represent two promising ways to support young people and help them face the challenging uncertainty and losses brought about by the pandemic (69). Aligned with prior research (38), we found that a positive future orientation was associated with lower support for VR beyond the contribution of depression. Despite the social adversity and uncertainty associated with the pandemic in the past year, future orientation is confirmed in our study as an important protective factor that deters students from endorsing positive attitudes toward VR. In line with our hypothesis, higher presence of a meaning in life was also associated with lower support for VR beyond reported levels of psychological distress. Of importance, future orientation and presence of a meaning in life remained significantly associated with support for VR after including both in a model along with search for a meaning in life, suggesting that they have significant independent effects on VR. In addition, results from our interaction analysis suggest a combined effect of future orientation and presence of a meaning in life on support for VR, as we found a stronger protective association between higher presence of a meaning in life and support for VR among students who reported an average-to-high level of future orientation. Our results confirm that a positive vision of the future and sense of purpose in life represent a form of empowerment for young people that can reduce feelings of despair and prevent support for VR even in times of social crisis, during a pandemic that may have significantly eroded or changed one's life plans and purpose (26,38,50,69). Correlation analysis in our sample indicated that both a positive future orientation and higher presence of a meaning in life were associated with lower levels of depression, in line with prior evidence (50,51). Given that 62% of students in our sample scored above the clinical cut-off to our measure of depression (i.e., HSCL-25), the promotion of students' positive future orientation and sense of purpose in life represents a promising way to support young people who are presently experiencing high levels of psychological distress while at the same time contributing to the prevention of violence in our schools and societies. Of importance, scores above the clinical cut-off in the depression scale do not indicate clinical depression (70), but instead suggest important levels of distress among students, in line with growing literature before (11,70) and during the pandemic (4,5,71). This literature indicates that distress is common among young people and has increased in recent years (72). This trend was observed prior to the pandemic (70,72) and preliminary findings suggest that the pandemic has further accelerated this phenomenon of increasing mental distress among youth (73). These high levels of distress may indicate important levels of collective suffering among young people that stem from the challenges of our present times rather than from clinical conditions; as such, they should not be pathologized but instead situated and interpreted within the present social reality. Assisting students in restoring or building anew a sense of purpose and a positive vision of the future remains a promising prevention strategy under the present difficult circumstances characterized by high levels of social adversity. It is also important to comment on the associations of socio-demographic variables and support for VR. Some of these associations confirmed prior results on college students in the Québec context, notably that non-religious and older students express less support for VR (11,38,52,74). The fact that experiencing more financial difficulties is not associated with support for VR among young people confirms results from a recent meta-analysis in which socio-economic status explains only a small part of the variance of support for VR among young people (56). However, we observed some changes in the expected associations between gender, main language, immigrant generation and our outcome. Specifically, in contrast with previous research, men and women have comparable levels of support for VR (38). Transgender and gender-diverse youth emerge as the group at the highest risk of support for VR. This is in line with results of a recent survey conducted during the pandemic that highlighted high levels of support for VR as well as psychological distress among gender minorities (75). Such results point to the importance for future research to focus on this oftenoverlooked population who may be suffering more in the present context of social adversity. This also confirms prior studies that observed rapid shifts in gender roles in Western societies in recent years and underline the important role of gender in VR studies (76). The significantly higher levels of support for VR among anglophone and second-generation immigrant students in the present study may be due to the younger age of this sample compared to prior studies (11) as well as to the increasing social polarization in Québec related to language and immigration, as demonstrated by the ongoing polarizing debates around these issues in the province (77). Future studies are needed to further explore these associations. Our findings on the association between search for a meaning in life and support for VR are somewhat inconclusive. Although there is some support for our hypothesis of search for a meaning in life as risk factor for support for VR, the size of the association is small and dependent on other individual characteristics. A higher search for a meaning in life was associated with higher levels of support for VR; however, this association became non-significant once presence of a meaning in life and future orientation were included in the same model. This suggests that, in accordance with other research in North America, the search for a meaning in life can represent a risk factor for VR (44). However, this association is weak and dependent on other variables such as one's future orientation and presence of a meaning in life. This result is aligned with recent findings of no association between the search for meaning and wellbeing in more individualistic and peaceful societies (46), a description that, overall, describes Canadian society. If anything, our results suggest that search for a meaning in life is weakly associatedif at all -with support for VR and that it varies considerably across contexts and individual characteristics. Such results are in line with mounting evidence on socio-cultural variations in the role played by search for a meaning in life on psycho-social adjustment and wellbeing (45,46). Results from interaction analysis suggest that the association between search for a meaning in life and support for VR does not vary based on level of future orientation. The observed trend toward search for a meaning in life becoming more of a risk factor for VR at high levels of future orientation needs to be further investigated before any conclusions can be drawn. It may be that a stronger search for a meaning leads to positive attitudes toward extreme and violent measures especially when the person believes in the possibility of a better future. The increase in self-uncertainty brought about by the pandemic (1, 2) may have increased the appeal of extreme groups and violent means that can offer a potential immediate way to restore one's self-assurance and life expectations (30), suddenly compromised by the economic crisis, the widening of socio-cultural inequalities, and social distancing requirements. Nonetheless, the role played by the search for a meaning in life in association with support for VR was consistently small across our analyses. Future studies should also investigate the potential moderating effect of perceived quality of life and psychological distress in the association between search for a meaning in life and support for VR, which could help to shed more light on possible differential associations. Together, findings suggest that prevention and intervention efforts in the field of VR should focus more on promoting the presence of a meaning in life and one's future orientation. The role of search for a meaning in life in prevention and intervention efforts for VR is in need of further investigation; in light of the present findings a case-by-case evaluation of its role in prevention and intervention efforts may be needed as to support students' search and help them to find alternative non-violent ways to pursue their goals and make sense of their life. --- Limitations There are several limitations in the present study. First, the crosssectional design prevents us from drawing any conclusions about causality. Longitudinal studies are needed to shed light on the trajectories of associations between future orientation, presence of and search for meaning in life and support for VR. Second, we relied on an online method of recruitment that is associated with a wide variation in response rates. A selection bias of participants cannot be excluded if participation is differential. However, given the sensitivity of the topic and the pandemic context, this method of recruitment facilitated the participation of the students who would not have been comfortable in an individual interview (78). In addition, our mixed-effect models included colleges as random effects (i.e., students nested within colleges), thus allowing us to control for variability across colleges. Third, the associations between future orientation, meaning in life, depression and support for VR is likely very complex. While we have modeled depression as a potential confounder in the relationship between the independent and dependent variables, depression can also be conceptualized as mediator in the pathway between the independent variables and support for VR or effect modifier. Our cross-sectional study design limits our ability to test for depression as a mediator; our already complex model which accounts for clustering by colleges and includes future orientation as an effect moderator makes the introduction of a three-way interaction that includes depression challenging from the perspective of having the statistical power to detect relationships between variables. Longitudinal studies with larger samples are needed to explore this possibility and shed more light on the complex mechanisms involved in the associations between depression and these variables. Last, the data were collected during the second wave of the COVID-19 pandemic in Québec. Hence, results cannot be generalized to other cultural or social contexts. Given the rapidly evolving and long-lasting concerning situation of the pandemic in Québec and globally, it is important to monitor the situation over time via cohort and longitudinal studies. --- CONCLUSION Despite these limitations, our study expands knowledge on risk and protective factors for support of VR among young people during the very difficult and stressful context of the COVID-19 pandemic. Our results provide preliminary support for the relevance of a PYD approach to the study of support of VR among young people by highlighting the important role of a positive identity for the prevention of support of VR. Adolescents and early adults are among the age groups that have suffered the most during the pandemic in terms of mental health and social adjustment. Preliminary findings indicate that the pandemic has jeopardized young people's vision of the future and sense of purpose (1,2). In line with a PYD perspective, we must help students realign their personal assets with their rapidly evolving social contexts, which are presently characterized by fear and uncertainty. The developmental tasks of imagining a future for oneself as a young adult and finding a purpose in life is made more complex by the COVID-19 pandemic. Educational institutions represent a suitable target for preventive interventions aimed to support a capacity to envision future and meaning in life in spite of the pervasive social uncertainties. Our findings suggest that education efforts need to consider uncertainty and rapid social changes stemming for our rapidly evolving societies and their consequences on young generations (79). Prevention programs that address youth's sense of purpose and their capacity for complex thinking in general and about the future (31,80,81) are promising to minimize the attraction of extremist ideologies among young people and support their wellbeing during these challenging times (82,83). Life crafting interventions have also been proposed as potential beneficial programs to help rebuild meaning in life in response to the social suffering and loss of normalcy brought about by the pandemic, in line with positive psychology principles (69,84). Such interventions help people set goals and make concrete plans to move toward a more positive future. PYD interventions also offer a variety of school-based program models to engage youth in positive activities that can contribute to their sense of purpose and future orientation (80). Preventive efforts should also aim to train college teachers and career guidance officers on how to support and promote a positive future orientation and a purpose in life among students who may be experiencing difficulties during this health emergency. Written informed consent from the participants' legal guardian/next of kin was not required to participate in this study in accordance with the national legislation and the institutional requirements. --- DATA AVAILABILITY STATEMENT The raw data supporting the conclusions of this article will be made available by the authors upon request, without undue reservation. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by the Ethics Committee of the Center Intégré Universitaire de Santé et de Services Sociaux du Center-Ouest-de-l'Île-de-Montréal (CIUSSS-CODIM). --- AUTHOR CONTRIBUTIONS DM contributed to conception and design of the study, data analysis, interpretation of study findings, and writing the manuscript. GG contributed to data preparation and cleaning, preparation of tables, and drafting of the results. RF and AL contributed to methodological decisions and provided feedback on multiple versions of the manuscript. CR contributed to conception and design of the study, interpretation of study findings, and provided feedback on several drafts of the manuscript. The authors listed in the byline have agreed to the byline order and to submission of the manuscript in this form. All authors agreed to act as guarantor of the work. --- Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Copyright <unk> 2022 Miconi, Geenen, Frounfelker, Levinsson and Rousseau. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is

The COVID-19 pandemic has increased levels of uncertainty and social polarization in our societies, compromising young people's capacity to envision a positive future and maintain a meaningful sense of purpose in life. Within a positive youth development framework, the present study investigates the associations of a positive future orientation, presence of and search for meaning in life, and support for violent radicalization (VR) in a diverse sample of Canadian college students. In addition, we investigate the moderating role of future orientation in the association between presence of and search for a meaning in life and support for VR. A total of 3,100 college students in Québec (Canada) (69% female; M age = 18.57, SD age = 1.76) completed an online survey during the second wave of the COVID-19 pandemic. Results from linear mixed-effects models indicate that a positive future orientation and a higher presence of a meaning in life were negatively and independently associated with support for VR. Search for meaning in life was not associated with support for VR. The magnitude of the negative association between presence of a meaning in life and support for VR was greater among students with a more positive future orientation. Schools and colleges are in a privileged position to implement preventive interventions to support a positive future orientation and the presence of a meaning in life among young people during these challenging and uncertain times and reduce the risk of violence related to extreme ideologies in our rapidly changing society.

, and provided feedback on several drafts of the manuscript. The authors listed in the byline have agreed to the byline order and to submission of the manuscript in this form. All authors agreed to act as guarantor of the work. --- Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Copyright <unk> 2022 Miconi, Geenen, Frounfelker, Levinsson and Rousseau. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

The COVID-19 pandemic has increased levels of uncertainty and social polarization in our societies, compromising young people's capacity to envision a positive future and maintain a meaningful sense of purpose in life. Within a positive youth development framework, the present study investigates the associations of a positive future orientation, presence of and search for meaning in life, and support for violent radicalization (VR) in a diverse sample of Canadian college students. In addition, we investigate the moderating role of future orientation in the association between presence of and search for a meaning in life and support for VR. A total of 3,100 college students in Québec (Canada) (69% female; M age = 18.57, SD age = 1.76) completed an online survey during the second wave of the COVID-19 pandemic. Results from linear mixed-effects models indicate that a positive future orientation and a higher presence of a meaning in life were negatively and independently associated with support for VR. Search for meaning in life was not associated with support for VR. The magnitude of the negative association between presence of a meaning in life and support for VR was greater among students with a more positive future orientation. Schools and colleges are in a privileged position to implement preventive interventions to support a positive future orientation and the presence of a meaning in life among young people during these challenging and uncertain times and reduce the risk of violence related to extreme ideologies in our rapidly changing society.

Background Until relatively recently, the topic of menstruation has been overlooked both in international conventions on human rights [5] and in body politics in development. The first attempts to address the issue in an international context originated primarily within the Water, Sanitation, and Hygiene sector in Africa and Asia, and have largely focused on Menstrual Hygiene Management (MHM) 1 [28]. MHM programs have mainly been implemented in schools and are based on the general assumption that poor girls in low and middle income countries (LMIC) share the same situation: lack of information about the menstrual cycle, shame and discomfort during menstruation due to cultural myths and taboos, limited choices about affordable products, insufficient access to private and safe facilities to manage bleeding and menstrual products, and high school dropout rates due to difficulties relating to menstruation [8,13,21,24,25,27,31]. However, evidence to support the efficacy of MHM programs is far from conclusive [1,2,4,10]. A common limitation of such programs is that they often fail to include parents in their activities [4]. However, it is important to acknowledge that parents are hugely influenced by social norms and are responsible for making the decisions on this matter at the household level. Hence, both parents' knowledge and their social norms play a critical role in the information that adolescents receive and how they behave. As a consequence, in this paper we focus on parents' knowledge and predictions about the age of menarche. We explore whether parents have accurate knowledge about the age of menarche and whether they are able to predict or guess the distribution of the modal value of other parents' responses, in other words, the social norm [15,22]. To overcome possible social desirability bias2 in the study [14,16,26], we paid participants based on the accuracy of their answers (50 Honduran Lempiras if they hit the right answer). There are two important aspects to consider in this study. First, access to accurate, timely, and age-appropriate information about menstruation is an essential part of menstrual health [11]. Evidence shows that girls primarily obtain information from their mothers and/or other female family members [7,18,29,32]. Hence, determining whether parents are adequately informed is of paramount importance, given that if they are misinformed they might not only provide girls with incorrect information, but they might also provide it too late, which leaves girls unprepared to face their first cycle [29]. Second, most societies have established social norms about how menstruators and others are expected to behave in a given social situation [9,17]. General societal misinformation about the age of menarche could indicate that what parents consider "normal" is incorrect, which could lead them to making bad decisions regarding their daughters' health. To determine to what extent the general assumptions regarding parents' knowledge about menstruation are accurate in West Honduras, we conducted a pre-registered field experiment in Santa Rosa de Copán (a region where no MHM or similar programs have ever been implemented). Specifically, our main objective is to explore whether parents have accurate knowledge about the age of menarche; and if they can accurately predict or guess whether others parents also have accurate knowledge about the age of menarche. While the first question captures an individual's knowledge about the age of menarche, the second measures society's knowledge about the age of menarche. The rest of the paper is organized as follows. The next section presents the methods and procedures. Section III describes the sample. Section IV focuses on the results and Section V presents the conclusions. --- Methods and procedures We ran a lab-in-the-field experiment in Santa Rosa de Copán (Honduras) from May 1-14, 2019. The inclusion criteria for the study was having at least one child between the age of 6 and 9 registered at one of 11 different public schools from different districts (Osorio, El Carmen, Prado Alto, and Santa Teresa). According to the last census (2013),3 the town's population between 6 and 9 years old was 3806. Assuming one child for each household, the needed sample was 350 with a 95% of confidence level and 5% of error. Therefore, we recruited 360 parents to participate in the experiment. The recruiting process was as follows. Using the socioeconomic status (SES) of the school district, we invited 120 parents per stratum (high, middle, and low) to ensure the sample selection included households from different socioeconomic levels. However, to finish the experiment on schedule, we had to increase the observations and invite 2 and 8 additional parents from the middle and high SES strata, respectively, to complete the 360 observations. Despite this, proportions test suggests that the sample was equally distributed across the three SES categories. Participants were asked two separated questions: i) their knowledge about the age of menarche (self-report SR), and ii) to predict the modal response of the other participants regarding the same question (modal guess MG). 4 Appendix A shows the original instructions (in Spanish) and B the translation in English. It is important to highlight that SR and MG are not necessarily correlated. While SR captures an individual's knowledge about the age of menarche, MG measures society's knowledge about the same subject. Our design considers both incentives and possible order effects. We used a monetary incentive in the MG task (a monetary award was given if the mode was hit and 0 otherwise) to reduce social desirability bias. Given that the order of the questions may also contribute to bias (see [6], we randomized the question order using p = 0.5 to SR <unk> MG and 1-p to MG <unk> SR. As a result, half of the participants (n = 186) answered SR <unk> MG, and the other half (n = 174) MG <unk> SR (see Appendix A and B). We also collected participant sociodemographic characteristics to assess possible biases, primarily: sex, education, ethnic group, and socioeconomic status, as well as the composition of the household in terms of girls and boys. The field experiment was conducted by a Honduran organization, PILARH. Enumerators were trained on the objectives of the study, how to conduct the survey, confidentiality, and informed consent. Before the implementation, the questionnaire's language was reviewed by PILARH and ETEA Foundation Honduras teams. Then, it was pre-tested in the field with 24 adult participants to ensure that was culturally appropriated and comprehensible. At the same time, the pre-test served to complete the enumerators' training to avoid any bias induced by their way of introducing the questions. After the survey, PILARH entered the data in an excel spreadsheet. Using Stata, we transformed the categorical or string variables into numerical variables. Only 4 participants chose not to answer their education level, and 1 refused to provide his/her age. That is, we had 355 observations with no missing values for all the variables. Enumerators used paper-based questionnaires and received a list of households they had to visit, including the type of questionnaire (treatment) they had to implement. Face-to-face interviews were conducted in households and only one experimental subject was interviewed per household (father, mother, or guardian). The random allocation of participants into (order) treatments was made prior to the visit, therefore the enumerators had no influence on the selection. --- Sample and outcome variables The final sample was 360 participants, 50 were men and 310 women. The respondents from the socioeconomic groups were divided as follows: 31% were from lowincome, 34% from middle income, and 35% from high income households. The age of respondents varied from 22 to 78 with the following frequencies: 22-25 (15%), 26-30 (28%), 31-35 (21%), 36-40 (16%), 41-45 (9%), 46-50 (4%), and over 50 (7%). Most respondents over 50 were grandparents. The respondents cover the entire spectrum of level of education, although most are concentrated in the lower levels: 49% primary education (6 years of schooling) or less, while only 3% held a university degree or higher. As regards ethnicity: 11% were Chorti, 7% Lenca, almost 8% Maya Chorti, 70% Mestizo, and 4% were from other groups. In order to assess poverty levels, respondents were asked about access to food in the week previous to the survey: 23% responded that they did not have enough money to feed their children. Household composition was also determined to assess whether parents with at least one daughter were more informed than those with only sons: 35% of responders had only male children, whilst 65% had at least one female child, but only 21% had at least one daughter who was at least 12 or older (experience). Appendix C shows the sample's sociodemographic data. The main objective of this study was to determine whether parents had knowledge of the age of menarche and what they believe regarding other parents' knowledge. The age of menarche varies across countries and time, yet is considered healthy when it happens starts between the ages of 9 and 16. In Honduras, a recent study found that 93.3% of respondents had their first menstruation at 12 [30]. 5 As can be seen below, our sample average is 12.13 with a mode exactly equal to 12. Using the data on the mean age of menarche in Honduras and the modal value of 12 from the sample, we defined the following outcome variables: • Self-report: SRHit (takes the value of 1 if respondents answer 12 and 0 otherwise), SRUnder (= 1 if reported age is lower than 12 and 0 otherwise) and SROver (= 1 if reported age is higher than 12 and 0 otherwise). • Modal guess: MGHit (takes the value of 1 if respondents guess 12 and 0 otherwise), MGUnder (= 1 if respondents guess lower than 12 and 0 otherwise) and MGOver (= 1 if respondents guess higher than 12 and 0 otherwise). Therefore, the first set of items -SRHit, SRUnder, and SROver -determine whether parents have accurate knowledge while the second set -MGHit, MGUnder, and 5 Although the study was conducted in the municipality of Choluteca, we consider it to be a good proxy for the situation of Santa Rosa de Copán as a whole. MGOver -explores whether they think other parents are also well informed. --- Results Figure 1 shows the distribution of self-report answers, which highlights that the majority of the sample (56.11%) reported the exact value of the age of menarche, in other words, SRHit = 1. Those who over/under reported are fairly distributed across the range, in fact, SROver = 25.28% and SRUnder = 18.61%. Table 1 provides the regression results for the outcome variables. The order dummy controls for the order of the questions. The variable minority group is equal to 1 if the respondent belongs to an ethnic minority group (Lenca, Chorti, Maya Chorti) and 0 if they are mestizo. Column 1 shows that age and belonging to a minority group reduce the probability of providing the right answer, however this effect is only marginally significant (p = 0.09 and p = 0.06). 6 Interestingly, the interaction between female and minority shows positive but marginal effects (p = 0.07). Column 2 shows that respondents with higher education are less likely to overestimate(p = 0.001), but more likely to underestimate the age of menarche (Column 3, p = 0.036). 7 In turn, female shows a significant and positive coefficient in SRUnder, suggesting that women tend to underestimate the age of menarche. 8 However, the result in Column 3 must be considered with caution since the sample is not balanced by sex since only 14% of participants were men. The other variables (sufficient income, experience, minority, and task order) have no effect on any of the three outcomes. --- Result 1 Most of the sample had accurate knowledge about the age of menarche. We now focus on the results of the respondents' predictions or guesses about other respondents' knowledge as regards the age of menarche. As well as self-report responses, Fig. 1 also shows the distribution of guesses. A significant percentage of the sample, 62.78%, hit the modal value. As in the case of self-reported data, those who over or underestimate are fairly distributed along the range. In Column 4, Table 1, we estimated the probability of hitting the modal age answered by others. Respondents with experience have a lower probability of hitting the modal age (p = 0.01) and are more likely to underestimate the mode (p = 0.04). The rest of the variables are not significant. Column 5 shows that education and task order reduce the probability of overestimating the modal age (p = 0.03, p = 0.08, 9 respectively). The other control variables have no significant effects. Fig. 1 Distribution of self-report responses and modal guesses about the age of menarche 6 The mean reported by minority groups is 12.10 while the mean reported by mestizos (majority) is 12.14. However, the difference between both groups is not significant (p = 0.74). 7 For each additional year of education, the age of menarche reported decreased by 0.3%. 8 Females report an average age of menarche of 12.10 while males report an average of 12.28. This difference is not significant (p = 0.25). 9 When MG is the second question, the mean is 12.06 and when it is the first, 12.20 (p = 0.08). --- Result 2 Most of the sample accurately guessed the modal value of the age of menarche. Finally, we combined both self-report and guesses for each participant to ascertain an overall measure of their level of information. 10 Participants were labelled informed when SRHit = MGHit = 1 and misinformed when SRHit = MGHit = 0. We found that 45.56% of the sample belong to the former category while 26.67% fall into the latter. Columns 7 and 8 show that parents with experience are less likely to be informed while parents with higher education have a negative and marginal effect on the probability of being misinformed. Interestingly, the other variables have no impact on the level of information. --- Result 3 A large percentage of the sample had accurate knowledge about the age of menarche and accurately guessed the level of knowledge of the other respondents. Overall, Results 1-3 show that the parents in our sample were well informed: not only do they have accurate knowledge about the age of menarche, but they also accurately guessed the level of knowledge of the other participants. Table 3 of Appendix D shows the regression results controlling for SES categories and none of the dummy variables (middle and high SES) are significant, suggesting that people are well informed regardless their socioeconomic level. Therefore, results are robust to controlling for different socioeconomics levels. It should be noted that our results are lower bound. If we considered a more generous definition of Hit, for instance, letting subjects make an error of " <unk> 1 year" we would get even better results. In particular, SRHit would increase from 56.11% to 85.56%, MGHit from 62.78% to 95.83% and SRHit = MGHit from 45.56% to 83.05%. Therefore, as a rule, it should not be assumed that parents are uninformed. --- Discussion A common hypothesis behind Menstrual Hygiene Management (MHM) programs is that poor adolescent girls in LMIC do not receive accurate, timely, and ageappropriate information about menstruation. Studies across the world show that adult members of the family, especially mothers and other female members, are one of the primary sources of information for adolescents on reproductive health. However, MHM programs tend to consider that mothers and other family members are not informed [12]. Nevertheless, as far as we know, this is the first research that asks directly to parents if they know the age of menarche. Using data from 360 households, we determined that knowledge about the age of menarche is high among parents in Santa Rosa de Copán given that 56.11% reported the precise age of menarche that coincides with recent studies (provided by [30]. Moreover, respondents were also able to predict (in a significant 62.78%) the modal response of other participants. Interestingly, differences in knowledge are poorly explained by sociodemographics. Variables such as education and experience (at least 1 daughter age > 12), both with the expected sign, only have marginal effects. These results are in line with Mbugua [20], who found that educated mothers in urban Kenya experience sociocultural and religious inhibitions that hold them back from providing meaningful sex-education, including information about menarche, to their pre-adolescent and adolescent daughters. More educated people may know that the age of menarche has decreased over time [19,23] and therefore underestimate the current age of menarche. This could also be the reason they believe that other parents will respond higher values than those responded by themselves. However, this might also relate to prejudices within the local population. It is also important to consider that people might rely more on their own personal experience than on external information, as our results show. Further research is thus needed to understand the correlation between education level, access to information and social norms. In accordance with Baumann et al. [3], who found that caste/ethnicity was a significant predictor of menstrual knowledge and practices in Nepal, we also found that misinformation is related to minority ethnic groups, although only marginally. An unexpected result was that females (compared to males) are more likely to underestimate the age of menarche (p = 0.001). Although this result has to be considered with caution because the sample is not gender balanced, a possible explanation is that women rely more on personal experience. Another interesting result is that no differences were found between the self-reported data and the guesses about the collective modal response. The latter implies that there was no social desirability bias, which suggests that the age of menarche is not such a sensitive topic in Honduras as it is in other cultures. A possible explanation is that the age of menarche might be less problematic than other topics surrounding the menstrual cycle. In fact, another research on the topic in Santa Rosa de Copán found that the menstruation is still surrounded by misinformation, myths and gendered social norms [2]. Therefore, with the information at hand, we can't reach any conclusion about why adult people have correct information about the age of menarche in a context where misinformation and social norms about the menstrual cycle are prevalent. In order to address these issues, further research is needed. --- Conclusion This study found that the knowledge about the age of menarche is high among parents in Santa Rosa de Copán. Sociodemographic variables only marginally explain the results and there was no social desirability bias. These results show that one of the common hypotheses behind Menstrual Hygiene Management (MHM) program is not true in the study area. Nevertheless, further research is needed to understand the quality of society's knowledge about the menstrual cycle as well as to what extent parents transmit this knowledge to their daughters. --- Appendix C --- Availability of data and materials All of the main data has been included in the results. Additional materials with details may be obtained from the corresponding author. Data in Stata file is also available at the link: https:// www. dropb ox. com/s/ 2yzuu qfz0t b1ti7/ data_ menar che_ paper. dta? dl=0. --- Appendix A (Spanish) -Cuestionario sobre menstruaciones y derechos sexuales y reproductivos Pregunta auto-informada I01. <unk>A qué edad cree usted que una ni<unk>a tiene su primera menstruación? ___________ (n<unk>mero > 0). --- Conjetura modal I02. Hemos preguntado a otras personas de Santa Rosa a qué edad tienen las ni<unk>as su primera menstruación. <unk>Cuál cree usted que ha sido la edad que han dicho más veces? Piénselo bien porque si acierta, le vamos a pagar 50 lempiras (margen de error <unk> 1) ______________ (n<unk>mero > 0). Al final de la encuesta, realizamos el pago si el sujeto hab<unk>a respondido la edad modal de 12 a<unk>os. También pagamos si la edad respondida era mayor o menor en 1 a<unk>o. --- Appendix B-Menstruation and Sexual and Reproductive Rights questionnaire --- Self-report question --- I01. At what age do you believe girls have their first menstruation? _________ (number > 0). --- Guessing game I02. We have asked other people in Santa Rosa at what age they believe girls have their first menstruation. With what age do you think people responded most frequently? Think carefully about your answer because if you get it right, we will pay you 50 lempiras (error margin: <unk> 1 year) __________ (number > 0). At the end of the survey, we paid respondents if they gave the mode age of 12. We also paid them if the age they gave was higher or lower by 1 year. --- Appendix D Table 3. --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s13690-023-01030-5. --- Additional file 1. --- Authors' contributions All authors were responsible for the structure of this paper. MA conceived, designed and performed the experiment, interpreted the data, and drafted the manuscript. DJ designed the study and performed the experiment, run the statistical analysis and drafted the manuscript. PB designed the study, performed the experiment, and did the critical revisions of the paper. The authors all approved the final versions for submission. --- Declarations Ethics approval and consent to participate The study was approved by Loyola Andaluc<unk>a Ethics Committee. All participants signed an informed consent document. The field study was pre-registered in AsPredicted before execution. The documentation can be consulted here: https:// aspre dicted. org/ ps766. pdf. --- Competing interests Not applicable. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background Access to accurate, timely and age-appropriate information about menarche is an essential part of menstrual health. Reliable evidence shows that girls primarily obtain information from their mothers and/or other female family members, therefore, it is important to determine parents' knowledge and their predictions about other parents' knowledge of the age of menarche.To this end, we performed a pre-registered study with data collected from 360 households in Santa Rosa de Copán, Honduras. We implemented a novel procedure to avoid social desirability bias whereby participants answered two separated questions: i) their knowledge about the age of menarche (self-report) and ii) to predict or guess the modal response of the other participants regarding the same question (modal guess). Participants were paid according to accuracy. Both questions appeared randomly in the survey.Recent studies indicate the age of menarche at 12 years old and 56.11% of the sample gave the same response while 62.78% hit the modal value. We estimated the impact of different sociodemographic variables and found only marginal differences. Interestingly, people with formal education and women tend to respond with lower predictions.Parents' knowledge about the age of menarche is high in the study area. The study also found that there was no social desirability bias.

Introduction Following the first infections in China toward the end of 2019, corona virus continued to propagate across the world. No continent or country remained unaffected by the virus. 1 About 1.52 million deaths, more than 65.54 million people infected, and 45.38 million recoveries were reported across the world. 1,2 On 11 March 2020, the World Health Organization (WHO) declared the virus a global emergency and the pandemic impacted the entire world hugely. The pandemic continued to have economic, social, and political impact. 4 The impact of COVID-19 continues as an unsolved public health problem with potentially serious economic and social consequences. The outbreak of COVID-19 is particularly difficult for poor African countries due to their scarce economic resources, weak health system, and social crises. 2,22 The survey participants reported that about half of the global income and wealth losses occurred due to corona virus, the average losses being 5293 and 33,482 dollars, respectively as well as drops in employment, equity markets, consumer spending, and personal income. 9 Many countries discontinued long-distance travel, closed or reduced civil service working hours, and government and non-government institutions, particularly, schools were partially/fully closed, or resorted to home based studies. 1 Measures taken by developed countries to curb the pandemic involved closing borders and closing offices, which led to increased recession in the world economy in general and in the economy of developing countries in particular. 2 In Africa, the pandemic progressed slowly and its harmful economic impact kept increasing. Key sectors of the African economy already experienced a slowdown because of the pandemic. 2 Movement restrictions and quarantine measures resulted in less trade off, decrease in access to food, and decreased economic activities making the prices of basic items higher. As a result, people are forced to adopt negative handling mechanisms such as reducing food consumption, borrowing money for food. 11 In this respect, female-headed households rather than male-headed ones suffered more adverse effects due to lack of shelter. 11 One of the main preventive measures of COVID-19 that most countries used was lockdown, ie, staying at home. However, lockdown-related stress caused psychiatric problems, worsened emotional tension, and increased food and drink intake causing a dangerous health problem. Low and middle income countries, in contrast faced dietary problems leading to even worse poverty and malnutrition as a result of the corona virus. 23 While school closures had acutely negative effects, mental health of children was exacerbated by lack of peer support and alternative means for mitigating risks. 11 The corona virus crisis stretched poor health systems on the continent. The corona virus patients overcrowded the health facilities and patients with high burden diseases like AIDS, TB, and malaria were unable to obtain sustained care, which in turn led to more morbidity and mortality. 2 Moreover, COVID-19 pandemic created shortage of medicines and health equipment. Weakness and slowness of the health care system led to irrelevant inability to ensure health care for all pandemic patients. 24,25 Because of the drastic measures to reduced their medical supplies. Consequently, developing countries like Ethiopia faced difficulty in treating their patients after the pandemic gradually reached a critical stage. 2 Due to the COVID-19 pandemic, social stress and gender-based violence are growing exponentially. Many women who are being forced to "lockdown" at home face their abusers when services that support them became increasingly inaccessible. 3 In France, Singapore, and Argentina, cases of house fight increased by 30, 33, and 25%, respectively. 8 Since the corona virus outbreak began, such violence has also increased in Africa. 3 Stress, the interruption of social and protective networks, and decreased access to services aggravate the risk of violence. 4 As the world battles with the corona virus pandemic, emerging evidence indicates a sharp rise in genderbased violence in Ethiopia. For women and girls, this would have life-threatening consequences and profound impact on their opportunities and life trajectory. 10 Numerous factors such as misinformation, lack of treatment, shortage of equipment, and lockdownassociated depression, anxiety, etc, aggravate vulnerability to the pandemic. 6 Demographics such as occupation, age, gender, marital status, and occupation are factors that moderate covid-induced stress. COVID-19 led to job losses in Ethiopia. Such measures had especially bad impact on laborers and patients whose life standard relies on the hand to mouth. Owing to COVID-19 pandemic related crises, shortage of food and medical treatments became severe, leading to increased reports of hunger, stress, sexual violence and the like. This study, therefore, aimed at assessing the prevalence of corona related sexual violence, hunger, stress, indoor fighting as well as determining associated factors. Findings from such a study would enable policymakers and health institutions to update and adjust their covid-19 coping strategies. --- Methods --- Study Design and Data A cross-sectional study design involving 1288 randomly recruited participants from six Amhara Metropolitan cities was implemented. The sample size was determined using binary logistic regression model with power of 80% and a significance level of 5%. 12 After the sample size was determined, the sample was allocated for each metropolitan city using proportional stratified sampling. A high number of participants recruited from sub-cities of the metropolitan area participated in the study. A questionnaire that required participants to evaluate their experience of complications of corona virus such as sexual violence, hunger, stress, indoor fighting and others was used to collect data for the study. Based on a reference questionnaire employed in prior research, 12,13 a questionnaire that investigated important socioeconomic, cultural, biological, and demographic determinants significantly associated with sexual violence, hunger, and indoor fighting because of corona virus was designed. The first draft of the questionnaire was prepared in English and translated into Amharic using standard translating procedures by experts so as to ensure semantic and content equivalence. A sample of forty residents selected through convenience sampling was recruited to ensure that the contents of the questionnaire was relevant to the community of city dwellers. The investigator made sure that the respondents wore masks while the questionnaire was disseminated, and sanitized their hands. The eligibility criteria for participation in this survey was age of 18 years or older. --- Variables in the Study In this study four complications of corona virus, namely sexual violence, hunger, indoor fighting, and stress were considered as the outcome variables (Table 1). Each participant was asked if they experienced sexual violence, hunger, stress, and/or indoor fighting during the pandemic with binary response of either yes (1) or no (0). On the other hand, twelve independent variables that determine the characteristics of the participants were considered (Table 1). --- Statistical Methods --- Binary Logistic Regression Logistic regression is a statistical model used to determine the effects of a set of independent variables on categorical outcome variables. 14 When the categorical values of the outcome variable are dichotomous, a particular logistic regression model called binary logistic regression is used. Since the outcome variables considered in this study contained dichotomous categorical values: yes=1 and no=0, binary logistic regression model was employed to analyze the data. For any binary outcome variables Y (yes=1, and no=0) and X = <unk>x 1,x 2,...,x p <unk> factors, the probability that --- 3565 the sexual violence, hunger, stress or indoor fighting of the "i th " individual belongs to "yes=1" of outcome variable given all possible factors X is given by P(Y i = 1|X). Then, the binary logistic regression model is given by: 14 logit P Y i 1<unk>4 1jX 1<unk>2 <unk> <unk> 1<unk>4 log P Y i 1<unk>4 1jX 1<unk>2 1 <unk> P Y i 1<unk>4 1jX 1<unk>2 1<unk>4 <unk> <unk> X <unk>; i 1<unk>4 1; 2; :::; 1288 (1 ) Where <unk> is the intercept, X is matrix of all possible factors and <unk> its corresponding effect. Data entry and organization were done using SPSS software. The data were then exported to SAS software, and statistical analyses such as fitting binary logistic regression and estimating adjusted odds ratio (aOR) with a 95% confidence interval were made using SAS software version 9.4, and interpreted. --- Independent t-Test A statistical test was applied to check whether the mean difference between two groups was significant. 15 The monthly income difference of participants between, before, and during corona virus was evaluated using independent sample t-test. --- Ethical Considerations Ethical clearance was obtained from the Science College Research and Community Service Committee (SCRCSC) of Bahir Dar University, with a reference number of SCRCSC/102/02/12. Written informed consent was obtained from all participants. An identification number rather than names was assigned to keep personal information of participants in the study confidential, before completing the questionnaire all participants provided informed consent and confirmed that they participated on a voluntary basis. --- Results --- Exploratory Data Analysis The independent and outcome variables and the frequency of participant within corresponding categories of the variable were presented in Table 1. From 1288 participants of this study, 61.2% were males, 77.5% were orthodox christians, 18.6% came from Bahir Dar metropolitan city, 48.6% had single marital status, and 42.3% had an educational level of college and above. Moreover, while 26.2% of the participants were displaced, 39.9% were unemployed because of corona virus. The majority of the participants (72.0%) obtained updates about corona virus via media (TV, radio, etc.). The average age of the participants was 29.9 (years), and the average monthly income before and during corona virus was 3945 and 2484 birr, respectively. 25.2% of the study participants experienced hunger, 33.8% faced sexual violence, 8.4% faced indoor fighting, while 48.8% had stress related to COVID-19 (Table 1). Table 2 depicted average monthly income of participants before and during corona virus per each metropolitan city in Amhara region, Ethiopia. As compared to before the onset of COVID-19, in all of the cities, monthly income of participants decreased during corona virus. In Table 3, whether there was significant difference between average monthly income before and during COVID-19 was checked using independent t-test. Thus, significant difference was noticed in the average monthly income between before and during COVID-19 (p-value=0.001<unk>0.05). While Tables 2 and3 showed the economic recession due to corona virus, Table 4 showed the impact of corona virus on the health sector. As the demand of corona virus patients increased, 28.5% of the participants realized that treatment of non-corona virus patients decreased during The bivariate analysis depicted the presence of association of independent variables with corona virus complications such as sexual violence, hunger, stress, and indoor fighting, Table 5. Sex, marital status, residence, education level, displacement status, work status, and permanent residence were significantly associated (p-value<unk>0.05) with sexual violence. The majority of participants who faced sexual violence were female (95.2%), single (53.1%), illiterate (42.3%), unemployed due to COVID-19 (45.7%), and urban dwellers (85.7%). City of residence, educational level, and work status were significantly associated with hunger. 43.5% of the participants who experienced hunger were unemployed due to COVID-19. Participants' employment status was significantly associated with stress and indoor fighting. Many of the participants that faced stress (38.5%) and that were involved in indoor fighting (42.7%) were those who were unemployed due to COVID-19. Table 6 depicts the association between outcome variables, COVID-19-induced complications, such as sexual violence, hunger, stress, and indoor fighting. The p-value which was less than 0.05 indicates that there was a significant association between the two outcome variables. It showed that sexual violence was significantly associated with stress and indoor fighting. Moreover, a significant association was seen between hunger and stress, as well as between stress and indoor fighting. This study measured sexual violence using four "yes" or "no" question items. A participant who responded "yes" to at least one question was considered a victim of sexual violence. Among a total of 435 sexual violence victims during COVID-19, 32.8% of the violence was committed without the willingness of the spouse. After the closure of schools due to corona virus, 113 (26.0%) were obligated to get married (Table 7). --- Model Fitting The estimated effects of independent variables on sexual violence, hunger, stress, and indoor fighting due to corona virus using binary logistic regression model was fitted in Table 8. The estimated odds ratio was computed for each covariate. The odds ratio (OR) is a natural measure to describe whether independent variables had significant effects on categorical outcome of interest. 16 An OR different from one indicates that the independent variable had a significant effect on the outcome of interest. The fitted binary logistic regression model depicted that the sex of participants, city of residence, source of information, and monthly income during COVID-19 were important determinants of sexual violence against participants. The estimated odds of male participants that experienced sexual violence were 0.523 times the estimated odds of female participants. This indicates that female participants were more susceptible to sexual violence than males. For a unit increase of monthly income during COVID-19, the estimated odds of participants experience of sexual participants were more vulnerable to hunger than participants with an educational level of college graduate and above. Participants' sex, residence, work status, permanent residence, and monthly income before COVID-19 had significant effect on the stress level of the participants. Metropolitan residence, work status, and monthly income were important determinants of indoor fighting during corona virus. The estimated odds of stressed male participants were 1.294 times the estimated odds of female participants indicating that, as compared to female participants, male participants were more likely to become distressed. As compared to participants unemployed before the corona virus, participants unemployed due to COVID-19 were more likely to be stressed (aOR=1.688). The estimated odds of rural participants to become stressed were 0.801 times the estimated odds of urban participants. This indicates that rural participants were less likely to be stressed than urban participants. While the estimated odds of participants to become stressed decreased by 6.8% (aOR=0.932), per a unit (in birr) increase of monthly income before COVID-19. City of residence, work status and monthly salary during COVID-19 were significant determinants that affected indoor fighting among participants. The estimated odds of participants who were unemployed during corona virus being a victim of indoor fighting were 2.033 times the estimated odds of those who were unemployed before COVID-19, which indicates that participants unemployed during corona virus were more likely to have more indoor fighting than their counterparts. The estimated odds of participants to be victims of indoor fighting decreased by 38.0% (aOR=0.620), for a unit (in birr) increase of monthly income during COVID-19. This indicates that compared to participants with lower income, participants with higher income were less likely to be involved in indoor fighting. --- Discussion Since the outbreak of corona virus in China, several studies have been conducted on the virus and how it impacted the developed countries economically, politically, socially, and culturally. However, in developing countries such as Ethiopia, limited research has been done on the impact of the virus. This study, was thus aimed at assessing the prevalence of sexual violence, hunger, stress, indoor fighting, and other COVID-19 complications in Amhara metropolitan cities of Ethiopia. In addition, factors associated with the complications of corona virus were also determined. Our results give early insight into the urgent need to provide continued care and remedial action on sexual violence, hunger, stress and indoor fighting due to COVID-19. This work will help international bodies intervene to help stop the crisis. As outcome variables had binary outcomes (yes or no), in order to determine the factors significantly associated with them, binary logistic regression model was interpreted using odds ratio (OR). 16 To evaluate the association between these complications of corona virus, Chi-square d test was applied. The prevalence of sexual violence, hunger, stress, indoor fighting, and other complications did not decrease. Nearly one third of the participants (33%) included in the study were victims of sexual violence during corona virus. The figure in this finding was higher than that of figures from studies in Addis Ababa (12.7%), Medawalabu (10.9%), and Hawassa (10.2%) metropolitan cities of the country before corona virus began. 17 A study in 5 reported higher incidence of physical intimate violence during COVID-19 pandemic compared with the prior three years. This might be due to the differences in age, sociocultural factors, and other characteristics between study participants beyond the impact of corona virus. The majority of the sexual violence occurred due to a partner while the rest was committed by someone other than a partner. This finding was consistent with findings from a study that involved only women 18 in which it was reported that more than half of the women experienced violence from their partner including sexual violence. Among participants who were victims of sexual violence, 113 (26.0%) were students forced to get married by their parents, and without their consent because the schools were closed due to the pandemic. Compared to sexual violence against men (4.8%), the incidence of sexual violence against women was extremely high (95.2%), a finding which is in line with studies by. 17,18 This study also demonstrated that sex of the participants, city of residence, source of information about COVID-19, and monthly income during COVID-19 were important determinants of sexual violence. As compared to participants from Debre Tabor city, participants from Woldya city were more prone to experiencing sexual violence. This could be due to the fact that Woldya city is a border city interconnecting people from different places and is a site of varied commercial exchange. Participants with low monthly salary during COVID-19 had higher likelihood of experiencing sexual violence. This finding is in line with another finding from another study on women 4,18 which found that, compared to higher income women, poorer women experienced increased sexual violence. Even though at the level of Sub-Saharan Africa, Ethiopia has made remarkable economic strides in becoming one of the efficient economies, at a global level, it is still one of the poorest nations with a hungerstricken population. 19 Distancing measures were taken and people were encouraged to stay home, stopping their work. 4,21 Nonetheless, corona virus pandemic aggravated the burden and prevalence of hunger. This study revealed that more than a quarter of participants (25.2%) experienced hunger. The authors think that the number of hungry participants could be even higher because many participants feel ashamed to tell the truth about their experience of hunger. The authors noticed that even most of the street workers employed in small and micro works were prohibited from working by the government to reduce the spread of the pandemic, and suffered inside their homes. As most of them were head of the family rather than just individuals, they faced collateral damage. A woman interviewed arbitrarily about the current complications of COVID-19 while she was preparing coffee on the streets replied "for me I do not mind if I spend the whole day and night without any food but I always worry about whether I would afford to feed my child and to pay the rent for my living room". Educational level, work status, monthly income before and during COVID-19 were significant factors that were associated with hunger. Participants with an educational level of college and above were less likely to be hungry compared with participants who did not have any education. This finding is consistent with findings from several studies in developing countries, 19,20 which indicated that educational level is an important factor associated with under-nutrition. Compared with low monthly income participants, participants with higher monthly income before and during COVID-19 were less likely to face hunger. Nearly half of participants (48.8%) experienced stress due to corona virus pandemic. In studies where impact of corona virus was reported, 8 it was noticed that continued COVID-19 pandemic induced fear and worry in daily life. Unlike findings from a study by, 6 sex of participant was a significant factor associated with stress, while marital status had no relation with stress. Moreover, work status, participants' permanent residence, and monthly income before corona virus were important determinants of stress. Participants unemployed due to COVID-19 were more likely to be stressed than employed and participants and those unemployed before COVID-19. As in many countries, Ethiopia adopted quarantine, social isolation, stay at home orders, etc following COVID-19 incidence. 7 However, despite these measures, social impact of COVID-19 such as indoor fighting increased. This study found that 8.4% of the participants experienced indoor fighting during the pandemic. Monthly income during COVID-19 and work status were important determinants of indoor fighting. Beyond COVID-19's social impact, this study also found that in government hospitals, lack of treatment for non-COVID-19 patients and economic recession were noticed. This aligns with findings from studies by. 2,9 It also indicated that 26 corona virus impacted mental health. This was noticed toward the onset of the pandemic. The traumatic experiences and affective disorders significantly affected quality of life. 27 In general, in addition to the impact COVID-19 caused, higher prevalence of sexual violence, hunger, stress, and indoor fighting ensued due to corona pandemic, thus, the government, health institutions as well as stakeholders should engage in actions that salvage people from the diverse damage and impact of Covid-19. --- Conclusion The prevalence of sexual violence, hunger, stress, and indoor fighting due to COVID-19 pandemic was considerably high. Work status of participants was an important factor that was associated with hunger, stress, and indoor fighting. Monthly income during COVID-19 and sex of participants were significantly associated with sexual violence. Moreover, in government hospitals especially, there was lack of treatment for non-COVID-19 patients. Economically, covid-19 brought a recession. These all call for timely remedial action from the government and other stakeholders. In order to reduce the damage of COVID-19 pandemic would put into question the survival of many among the society whose life hangs on a hand to mouth subsistence. --- Data Sharing Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. --- Abbreviations --- Ethics Approval and Consent to Participate The study was conducted in full accordance with the World Medical Association Declaration of Helsinki and approved by the Ethics Committee of Science College Research and Community Service of Bahir Dar University, Ethiopia (SCRCSC/102/02/12). --- Author Contributions Tesfaw had the idea of the study. All authors contributed to data analysis, drafting or revising the article, have agreed on the journal to which the article will be submitted, gave final approval of the version to be published, and agree to be accountable for all aspects of the work. --- Disclosure The authors declare that they have no competing interests. --- Risk Management and Healthcare Policy --- Dovepress --- Publish your work in this journal Risk Management and Healthcare Policy is an international, peerreviewed, open access journal focusing on all aspects of public health, policy, and preventative measures to promote good health and improve morbidity and mortality in the population. The journal welcomes submitted papers covering original research, basic science, clinical & epidemiological studies, reviews and evaluations, guidelines, expert opinion and commentary, case reports and extended reports. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www.dovepress.com/testimonials.php to read real quotes from published authors. Submit your manuscript here: https://www.dovepress.com/risk-management-and-healthcare-policy-journal Risk Management and Healthcare Policy 2021:14 DovePress

The onset of corona virus (COVID-19) pandemic brought about multiple crises with financial, social, and political dimensions, and problems involving absence or scarcity of medication, food, and other amenities. COVID led to reports of severe hunger, stress, sexual violence, and associated complications. This study was aimed at assessing the prevalence of sexual violence, hunger, stress, indoor fighting, and related complications caused by the corona virus, and determining factors associated with it. Methods: A cross-sectional design study involving a total of 1288 participants recruited from Amhara Metropolitan cities was implemented. Each participant completed a questionnaire regarding experience of sexual violence, hunger, stress, indoor fighting, and other complications faced due to COVID-19. A statistical analysis was done by fitting binary logistic regression model, independent t-test and Chi-squared test. Results: From the total participants, 1288 (26.2%) were displaced, while 39.9% were unemployed due COVID-19. About half of the participants (48.8%) experienced general distress, 33.8% faced sexual violence, 18.2% experienced hunger, 8.4% faced indoor fighting. The majority (81.7%) of non-corona virus patients in government health-care institutions faced lack of treatment. For every unit increase of monthly income during COVID-19, the estimated odds of participants experiencing sexual violence decreased by 31.3% (aOR=0.687). Educational level, work status, and monthly salary before and during corona virus were important factors of hunger. Conclusion: COVID-19 pandemic led to considerably high prevalence of sexual violence, hunger, stress, and indoor fighting. Work status of participants was an important associated factor of hunger, stress, and indoor fighting. Sex of participants and their monthly income were significantly associated with increase in sexual violence during COVID-19. In addition to a general economic recession, a decrease in treatment of non-COVID-19 patients was noticed, particularly in government hospitals.

Introduction In nations with Internet access, social network sites (SNSs) are well established. Facebook is the most popular social media platform, with 2.85 billion monthly active users worldwide followed by YouTube, WhatsApp and Instagram (Statista, 2021). The number of Internet users and SNSs in China has exploded in the last decade, owning primarily to the mobile Internet. There are 1.01 billion Chinese Internet users, with 99.6% using the mobile Internet, and 271,000 social messaging apps (CNNIC, 2021). WeChat Moments, Qzone and Weibo are the top three SNSs in China (CNNIC, 2017) and are among the top 10 most popular social Social network site use and users' wellbeing media platforms in the world, with over 2.37 billion monthly active users (Statista, 2021). Users can post (including text, photographs and videos) and update their status on WeChat Moments and Qzone, but they can only browse and engage with content provided by online friends or public accounts (including likes, comments and private messages). Weibo allow users to create different types of posts, as well as view and engage with others' profiles and public posts. Moreover, according to a recent poll conducted in 46 regions, global netizens spend an average of 144 min each day on social media, with Chinese netizens spending 117 min (Statista, 2020). This shows that an increasing number of people in China and the world have integrated SNSs into their daily lives. A great deal of studies from various disciplines have investigated the influence of SNSs on people's adjustment, with a focus on the relationship between SNS use and well-being, which is frequently quantified using life satisfaction as a major indicator (e.g. Kim and Lee, 2011;Oh et al., 2014;Verduyn et al., 2017). It is a common assumption that using SNSs leads to decreased well-being or other undesirable consequences. Excessive or problematic use patterns, fear of "missing out," and upward social comparisons have been linked to SNS addiction, social isolation or overload, low self-esteem and some negative feelings such as depression, anxiety and envy (e.g. Satici and Uysal, 2015;Wang et al., 2018a;Schmuck et al., 2019). However, some encouraging findings suggest that using SNSs may help people find fulfillment in various ways. This is because enhanced perceived social support, relational certainty and social capital from online friends can lower psychological stress, promote problem-solving and improve happiness among SNS users (e.g. Nabi et al., 2013;Verduyn et al., 2017;Lian et al., 2020;Brailovskaia and Margraf, 2019). As the co-existence of these two distinct connections, researchers are looking into potential mediators in the relationship between SNS use and well-being. These studies were based on diverse SNS usage behaviors, which are broadly divided into two types: active and passive. This explains the inconsistency. Generally, passive SNS use refers to the observation of other people's lives by browsing their profiles, status and posts, whereas active use comprises interactions with other SNS friends in a private or public setting, such as posting on one's wall, updating status and commenting (Burke et al., 2010;Frison and Eggermont, 2015a). Directed communication may encourage content creation, information exchange and social interactions among SNS users, whereas content consumption will more likely minimize social engagement and boost comparison. Active SNS use is positively connected with well-being, whereas inactive SNS use has the opposite effect (Verduyn et al., 2017;Wang et al., 2018a). Another cause may be the various motivations underlying SNS use behaviors (Lin et al., 2017;Wang et al., 2019;Brailovskaia and Margraf, 2019). Although extensive research has explored why people use SNSs (Yang and Brown, 2013;Syn and Oh, 2015;Wen et al., 2016), few studies have examined the consequences (Boyle and Johnson, 2010;Krishnan and Hunt, 2015;Pittman and Reich, 2016). Most discussions on the relationship between SNS use and well-being have addressed motivations or needs driving SNS use behaviors. SNS users who consume content, for example, are often motivated by a desire to relax or escape from the real world (Wang et al., 2018a); those seeking information and entertainment are not likely to make social comparisons (Schmuck et al., 2019). Initiating interactions or active posting on SNSs is associated with a desire for belongingness and self-presentation (Lian et al., 2020). Therefore, the study aimed to investigate the associations between motivations for active SNS use and well-being, as well as the mediating role of self-presentation and perceived social support, which were consistently linked to active SNS use and positive outcomes (Meng et al., 2017;Pang, 2020). Self-presentation is central in active SNS use, whether for informationsharing, communications with others or just posting a photo or status update (O'Donnell et al., 2021). Self-presentation on SNSs could benefit users' well-being directly or indirectly through the perception of social support on SNSs (Pang, 2020). As a vital source of human well-being, AJIM 75,1 perceived social support represents the outcome of social interactions or the psychological resources provided by social interactions (Wang et al., 2019;Pang, 2020). Although diverse types of SNS use can boost perceived social support (Wang et al., 2019), when SNS users actively, rather than passively, engage in social interactions, perceived social support can lead to enhanced well-being (Meng et al., 2017). As SNSs have become an indispensable part of people's lives, this study will aid in improving their understanding of the mechanisms underlying the relationship between active SNS use and well-being, encouraging their healthy use of SNSs, and promoting SNS interface design and user management. --- Literature review 2.1 Motivation for SNS use and well-being According to the uses and gratifications theory, people who utilize mass media to fulfill psychological needs or desires are engaged, goal-oriented and motivated (Katz et al., 1973;Rubin, 1994). The theory is more suited to the social media context, which allows users to actively publish and exchange information rather than passively receiving it (Papacharissi, 2002;Basilisco and Jin, 2015). The internal force that drives people to utilize SNSs to meet their needs or desires is the motivation. Some researchers suggested that considering the motivations for SNS use would shed more light on its relationship with well-being (Nabi et al., 2013;Wen et al., 2016). Satisfaction of SNS users' needs or desires by the functions they employ, rather than the frequency or length of use, enhances well-being (Guo et al., 2014;Manuo glu and Uysal, 2020). Furthermore, SNS usage has been strongly associated with interpersonal utility, information-seeking, entertainment and other motivations in prior studies (e.g. Giota and Kleftaras, 2014;Basilisco and Jin, 2015;Wen et al., 2016). A few studies have linked SNS users' motivation with their well-being. Wang et al. (2014) found that university students who used Qzone for social interaction (i.e. status updating, comment giving and blog writing) had a high level of well-being, while those who used it for entertainment (i.e. passively browsing friends' pages and playing games) did not yield the same result. Similarly, Guo et al. (2014) found that among Weibo and Facebook users, using the SNS functions for social activities and information (i.e. interacting with friends, meeting new people, looking for news, seeking information) increased the level of users' perceived well-being, whereas using them for entertainment (i.e. browsing the posts, having fun, passing time online) did not. Conversely, Baek et al. (2014) conducted a survey among Facebook users and discovered that interpersonal motives significantly negatively predicted well-being only among users with secure attachment, whereas information and entertainment motives had no effect on well-being. According to a new study on adolescent Facebook users (Jarman et al., 2021), social interaction motivation was not correlated with well-being, while information-sharing and other motivations (including passing time and escapism) were negatively correlated with it. Researchers primarily focused on three motivations of using SNSs in the aforementioned studies: social interaction, information and entertainment. However, there is no consensus on the relationship between these three motivations and well-being. One possibility is that the effect of motivations is explored across multiple SNS usage behaviors. For example, when social and information motivations were positively connected with active SNS use, they improved well-being (Wang et al., 2014;Guo et al., 2014;Baek et al., 2014). However, when they were not connected with active SNS use or were strongly correlated with passive SNS use, they did not promote well-being (Wang et al., 2014;Guo et al., 2014;Jarman et al., 2021). Another possibility is that there are some discrepancies in the definitions of motivation. Information motivation includes two meanings: seeking and sharing. Information-seeking was positively associated with well-being (Guo et al., 2014), while information-sharing had the opposite effect (Jarman et al., 2021). When entertainment motivation is equivalent to the Social network site use and users' wellbeing incentive of passing time and escapism, it is connected with passive surfing (Wang et al., 2014;Guo et al., 2014) and lower well-being (Jarman et al., 2021). To regulate these influences, researchers must describe the motivations for SNS use and clarify the corresponding use behaviors in their investigations. --- Mediation of perceived social support Many people use SNSs to get social support, which can help them cope with psychological stress caused by unfavorable life events and improve their overall well-being. Perceived social support has gotten more attention than actually received social support in the literature (Nurullah, 2012). Perceived social support relates to the belief that social networks can provide help, and generally includes emotional support, instrumental assistance and information-sharing (Lakey and Cohen, 2000). In the last several decades, the manner in which social networks provide support has changed dramatically, shifting away from face-toface encounters, letters and phone conversations to emails, SNSs, and other new media. With the rise of SNS users, online and offline social networks are no longer distinct, but significantly overlap. Families, relatives, friends and classmates are on the contact lists of 70% of Chinese SNS users (CNNIC, 2014). Previous studies have demonstrated that SNS users obtained considerably more social support from their online rather than their offline friends (Indian and Grieve, 2014) and that perceived social support from SNSs predicted users' wellbeing (Olson et al., 2012). Perceived social support should be a key mediator in the relationship between SNS use and well-being. According to some studies, it mediated the relationship between the number of SNS friends and well-being (Valkenburg et al., 2006;Kim and Lee, 2011;Nabi et al., 2013). Moreover, the amount of time spent on SNSs and other commonly utilized SNS features were positively related to users' perceived and received social support, which decreased stress and improved well-being (Meng et al., 2017). As a result, perceived social support may also play a role in the link between motives for using SNSs and well-being. Some common types of motives for SNS use (e.g. interpersonal utility, information-seeking and entertainment) were positively connected to SNS use time and online social support (Giota and Kleftaras, 2014). Other types (such as intrinsic and external, exhibitionism and voyeurism) were also associated with the number of SNS friends (Wen et al., 2016), content generation and content consumption (M€ antym€ aki and Islam, 2016). --- Mediation of self-presentation on SNSs Self-presentation is important for relationship building and development in offline and online social settings (Ellison et al., 2006;Boyle and Johnson, 2010). Similar to impression management and self-disclosure, it refers to the action of people revealing themselves to others and attempting to influence others' and their own impressions (Leary and Kowalski, 1990;Boyle and Johnson, 2010;Yang and Brown, 2016). Compared with face-to-face encounters, there are more possibilities and instruments for users to present themselves online with diverse objectives, such as information, entertainment and contact with friends and families (Papacharissi, 2002). Prior studies mainly compared two types of selfpresentation on SNSs (Kim and Lee, 2011;Niu et al., 2015): first, positive self-presentation, which refers to the selective broadcasting of personal information, thoughts and feelings (e.g. only posting "happy" status updates or stunning images, and sharing good or favorable information); and second, honest or authentic self-presentation, which means the nonselective posting of personal information, ideas and sentiments in a true-to-self manner (e.g. uploading happy and negative feelings or unaltered photos). --- AJIM 75,1 Self-presentation on SNSs may mediate the relationship between motivations for SNS use and well-being. First, online self-disclosure acts as a link between online communication and well-being (Valkenburg and Peter, 2009). This benefit was evident in teenagers who used the internet to maintain existing connections, rather than those who used it to indulge in solitary activities. Second, SNS users, according to the functional model, employ multiple functionalities for self-disclosure with varying levels of intimacy, depending on their objectives (Bazarova and Choi, 2014). On Facebook, where wall posts are a type of public communication, the most popular incentives for self-presentation are relationship growth and social validation, with information and entertainment playing a minor role. When interacting with diverse people, from strangers to acquaintances, selective disclosure can help achieve the intended results while reducing the danger of information control loss. Third, self-presentation on SNSs is linked to well-being. According to Lee et al. (2011), the level of self-disclosure on Cyworld had a positive effect on the users' well-being through increased social capital. Self-disclosure on SNSs was not just a confession, but also an anticipation that SNS friends would reciprocate with their selfdisclosure and social support. Authentic self-presentation does not generally imply disclosing all unpleasant experiences or sentiments, but only some portion thereof. This process aids SNS users to create genuine social ties, and promotes well-being (Reinecke and Trepte, 2014). Although the type of self-presentation was not considered in these studies, it is plausible to assume that more highly motivated SNS users will present themselves more deliberately and consciously, whether positively or authentically, resulting in enhanced well-being. Furthermore, a mediating factor between self-presentation on SNSs and well-being may be perceived social support. Kim and Lee (2011) found that positive self-presentation on Facebook had a direct effect on users' well-being, whereas honest self-presentation affected Facebook users' well-being through perceived social support. The following studies have confirmed that honest self-presentation on SNSs affects life satisfaction through positive online feedback (Liu et al., 2016) or enacted social support (Zhang, 2017). Some other studies have found an association of positive or authentic self-presentation on SNSs with audience feedback (Yang and Brown, 2016;Bareket-Bojmel et al., 2016). Given that audience feedback was measured based on self-reports of SNS users, it could reflect the level of social support that they perceived online. According to these findings, positive and honest self-presentation may also have a chain mediation with perceived social support in the relationship between motivations for SNS use and users' well-being. --- Research questions, hypotheses and models 3.1 Research questions and hypotheses 3.1.1 Three main motivations of active use and well-being. In order to better understand the relationship of motivations for SNS use and users' well-being, this study limited SNS use behavior to message posting, which is a typical active usage of SNSs. According to the research of Syn and Oh (2015), message posting on SNSs can be motivated by 10 different factors, the first two being social engagement and learning. The desire to connect with others through SNSs is referred to as social interaction motivation in this study. Learning motivation-information motivation in this study-refers to seeking information, assistance and opinions from people on SNSs. When people use SNSs to maintain existing interpersonal ties, the higher their social motivation, the happier they are while posting (Wang et al., 2014). By bridging social connections and getting information support, active SNS use for socialinformation motivation correlates positively with life satisfaction (Guo et al., 2014). Information-seeking was a predictor of sharing information on SNSs, which may satisfy the current informational needs of oneself and others (Lee and Ma, 2012). Furthermore, many studies on SNS motivations indicated that entertainment is also one of the most important motivations for SNS use (e.g. Lee and Ma, 2012;Wang et al., 2014). Regarding the need for Social network site use and users' wellbeing escapism, emotion release and anxiety relief, entertainment motivation was not a significant predictor of news-sharing in social media (Lee and Ma, 2012), and did not have a favorable impact on users' well-being (Guo et al., 2014). Regarding the intrinsic motive of enjoyment or fun, entertainment motivation could positively predict online social participation (Nov et al., 2010) and well-being (Wen et al., 2016). Given that SNS use is often driven by multiple motives, the overall intensity of the three positive motivations based on an active use behavior may positively predict users' wellbeing. Therefore, the first goal was to examine if the three main motivations of posting on SNSs (social interaction, information and entertainment) were positively associated with the posters' well-being. H1a. The overall intensity of motivations for posting on SNSs is positively associated with the posters' well-being. H1b. Social interaction motivation for posting on SNSs is positively associated with the posters' well-being. H1c. Information motivation for posting on SNSs is positively associated with the posters' well-being. H1d. Entertainment motivation for posting on SNSs is positively associated with the posters' well-being. 3.1.2 Mediation of perceived social support between three main motivations of active use and well-being. As mentioned above, there is evidence that perceived social support might mediate the relationship between motivations for SNS use and users' well-being. Specifically, the three main motivations for SNS use may all have relations with perceived social support. Internet use (including Facebook) for relationship maintenance and information-seeking was shown to be positively associated with perceived social support from online and offline sources, but the use of the internet for entertainment/escape and social recognition was not (Leung, 2010). Furthermore, recent studies on Facebook and WeChat usage revealed that motives for social engagement and information acquisition have favorable effects on social capital and perceived social support (Frison and Eggermont, 2015b;Ahmad et al., 2016;Wang et al., 2019). In this study, posting on SNSs for social interaction and information is typically directed toward communication, and posters will perceive more social support through feedback, such as likes, comments and forwards. Posting on SNSs for enjoyment, not for passing time or escapism, may satisfy posters through online supportive feedback (Giota and Kleftaras, 2014). Therefore, the second goal was to investigate whether perceived social support might be a mediator in the relationship between the three motivations (social interaction, information and entertainment) for SNS use and well-being. H2a. The association between the overall intensity of the three main motivations for SNS use and users' well-being is mediated by perceived social support. H2b. The association between social interaction motivation for SNS use and users' wellbeing is mediated by perceived social support. H2c. The association between information motivation for SNS use and users' well-being is mediated by perceived social support. H2d. The association between entertainment motivation for SNS use and users' wellbeing is mediated by perceived social support. 3.1.3 Mediation of self-presentation between three main motivations of active use and wellbeing. As mentioned above, self-presentation may play two mediating roles in the relationship between motivations for SNS use and well-being, however, the three motivations for SNS use AJIM 75,1 and two self-presentation strategies may have different connections. An incentive for social interaction can lead to positive and honest self-presentation. Users with higher motivations for connecting with friends and networking had more frequent self-presentation, such as expressing personal interests and insights (Papacharissi, 2002;Boyle and Johnson, 2010). Furthermore, Chinese college students with higher social motivation had more selfpresentation on WeChat, such as updating self-introductions and sharing information; online self-presentation and social support had a chain mediation in the relationship between social motivation and the sense of belonging (Pang, 2020). However, self-presentation is uncommon in using SNSs for entertainment (Wang et al., 2014;Pang, 2020). Users who are motivated by entertainment are more likely to engage in passive use behaviors such as perusing friend sites or playing games, causing overindulgence and limiting their desire to give personal information (Wang et al., 2014;Pang, 2020). However, in this study, entertainment motivation was assessed based on an active usage behavior. It is unclear whether users who post for fun on SNSs may disclose themselves and obtain positive feedback. As self-presentation is also less essential when seeking information, there may be no link between information motivation and SNS users' well-being mediated through self-presentation and perceived social support. Overall, the last two goals were to investigate (1) the simple mediating effect of positive or honest self-presentation in the relationship between the three motivations (social interaction, information and entertainment) for SNS use and well-being; and (2) the chain mediating effect of positive/honest self-presentation and perceived social support. H3a. The association between the overall intensity of motivations for SNS use and users' well-being is mediated by positive or honest self-presentation. H3b. The association between social interaction motivation for SNS use and users' wellbeing is mediated by positive or honest self-presentation. H3c. The association between information motivation for SNS use and users' well-being is not mediated by positive or honest self-presentation. H3d. The association between entertainment motivation for SNS use and users' wellbeing is not mediated by positive or honest self-presentation. H4a. Positive or honest self-presentation and perceived social support has a chain mediation in the relationship between the overall intensity of motivations for SNS use and users' well-being. H4b. Positive or honest self-presentation and perceived social support have a chain mediation in the relationship between social interaction motivation for SNS use and users' well-being. H4c. Positive or honest self-presentation and perceived social support do not have a chain mediation in the relationship between information motivation for SNS use and users' well-being. H4d. Positive or honest self-presentation and perceived social support do not have a chain mediation in the relationship between entertainment motivation for SNS use and users' well-being. --- Research models Two conceptual research models were proposed based on the aforementioned demonstration (see Figures 1 and2). The first considered the impact of overall SNS usage motivation on users' well-being and the mediating roles of perceived social support and two types of self- Syn and Oh (2015). After translation into Chinese, the instructions were slightly revised to address the conditions in China. The sentences that began with "I share wall posts on Facebook because..." or "I tweet because..." were revised to "I post on SNSs (such as WeChat Moments, Qzone and Weibo) because...." "Facebook" and "Twitter" were replaced by "social network site." The questionnaire originally included 30 items covering 10 motivations. Each item was measured with a Likert-type scale ranging from 1 (completely disagree) to 5 (completely agree). Confirmatory factor analysis (CFA) was conducted to examine the construct validity. The 10-factor model fit well (<unk> 2 5 502.064, df 5 360, root mean square error of approximation (RMSEA) 5 0.023, standardized root mean square of residual (SRMR) 5 0.032, comparative fit index (CFI) 5 0.969, Tucker-Lewis index (TLI) 5 0.962). Factor loading of the items ranged from 0.551 to 0.764. This study adapted three factors (social engagement, learning and enjoyment) as social interaction, information and entertainment motivations for posting on SNSs. For each factor, composite reliability (CR) values were 0.761, 0.710 and 0.743; average variance extracted (AVE) values were 0.515, 0.451 and 0.492; Cronbach's <unk> values were 0.812, 0.805 and 0.811, respectively. For the three factors, the CR value was 0.885, the AVE value was 0.462 and the Cronbach's <unk> was 0.853. --- Perceived social support. The perceived social support scale (Zimet et al., 1988) included 12 items reflecting the support from families, friends and others. Participants were asked to answer the items according to their experience on SNSs (such as WeChat Moments, Qzone and Weibo). Each item was responded to on a Likert-type scale ranging from 1 (completely disagree) to 7 (completely agree). The three-factor model fit well (<unk> 2 5 112.784, df 5 51, RMSEA 5 0.040, SRMR 5 0.027, CFI 5 0.980, TLI 5 0.974). Factor loading of the items ranged from 0.620 to 0.780. For each factor, CR values were 0.810, 0.766 and 0.802; AVE values were 0.517, 0.450 and 0.503; Cronbach's <unk> were 0.809, 0.765 and 0.801. For the factors, the CR value was 0.920, the AVE value was 0.490 and the Cronbach's <unk> was 0.865. 4.2.3 Self-presentation on SNSs. Self-presentation on SNSs was measured with the positive self-presentation index and the honest self-presentation index (Kim and Lee, 2011;Niu et al., 2015). One scale had six items and the other had four items. A 5-point Likert-type scale was adapted for each item (1 5 completely disagree, 5 5 completely agree). The two-factor model fit well (<unk> 2 5 106.695, df 5 34, RMSEA 5 0.053, SRMR 5 0.032, CFI 5 0.972, TLI 5 0.963). Factor loading of the items ranged from 0.665 to 0.761. For each factor, CR values were 0.842 and 0.821, AVE values were 0.470 and 0.535 and Cronbach's <unk> values were 0.841 and 0.821. --- 4.2.4 Well-being. Well-being was measured with five items from the Life Satisfaction Scale (Diener et al., 1985). A 7-point Likert-type scale was adapted for each item (1 5 completely Social network site use and users' wellbeing disagree, 7 5 completely agree). The one-factor model fit well (<unk> 2 5 5.509, df 5 5, RMSEA 5 0.012, SRMR 5 0.010, CFI 5 1.000, TLI 5 0.999). Factor loading of the items ranged from 0.626 to 0.743. The CR value was 0.816, the AVE value was 0.471 and the Cronbach's <unk> was 0.815. --- Analytical strategy The software Mplus 7.4 for the structural equation model of latent variables was used to test our hypotheses. The following indices and criteria are often adopted to determine if the model fit is adequate: The <unk> 2 /df ratio is less than 3.0, the CFI or TLI is more than 0.90, the RMSEA is less than 0.08 and the SRMR is less than 0.05 (Schreiber et al., 2006). Bootstrapping was used to test the values and significance of indirect effects in the model, because it is superior for reducing errors. The indirect effect size is statistically significant if the 95% bias-corrected confidence interval for the parameter estimate does not contain 0, indicating a mediating effect (Preacher and Hayes, 2008). --- Results --- Common method variance assessment Harman's 1-factor analysis was used to examine the possibility of common method bias. For all variables, we used an exploratory factor analysis (EFA) without model rotation. The EFA results showed 15 different components with a single factor accounting for 14.28% of the total covariance explained. CFA was also used to investigate a one-factor model. The model did not fit well (<unk> 2 5 5641.934, df 5 594, RMSEA 5 0.106, SRMR 5 0.113, CFI 5 0.388, TLI 5 0.350). These results indicated that the data obtained did not have any serious common method bias. --- Descriptive statistics and correlations The descriptive statistics, gender t-test and correlation analysis of the variables, which are all the average scores of their respective items, are shown in Table 2. Excluding positive selfpresentation, which had a strong negative connection with honest self-presentation, and information motivation, which was not statistically correlated with well-being, all the variables were significantly correlated in a positive way. Furthermore, some had significant age correlations and gender differences. Hence, age and gender were added as covariates in the structural equation models that followed. --- Examining structural model The variables in the models were latent constructs that were estimated using the values of the items. Motivations for SNS use and perceived social support were second-order latent variables, while positive or honest self-presentation and well-being were first-order. First, we built a simple regression model with the motivation for SNS use as the independent variable, well-being as the dependent variable and age and gender as controlled variables. The results showed that the model fit well (<unk> 2 5 180.776, df 5 99, RMSEA 5 0.033, SRMR 5 0.038, CFI 5 0.963, TLI 5 0.956). Motivations for SNS use significantly, directly predicted wellbeing (<unk> 5 0.195, p <unk> 0.01, 95% CI 5 [0.082, 0.297]). Second, the aforementioned model incorporated perceived social support, positive selfpresentation and honest self-presentation as mediators between motivation for SNS use and well-being, while controlling the effect of age and gender on well-being. The multiple chain mediation model (Model 1) fit well (<unk> 2 5 1069.718, df 5 648, RMSEA 5 0.290, SRMR 5 0.040, CFI 5 0.950, TLI 5 0.946). The direct prediction of motivation for SNS use on well-being was not significant after the mediating variables were added (<unk> 5 <unk>0.066, 3 shows the regression coefficients for all the pathways in Model 1. The results of indirect effects from motivation for SNS use to well-being in Table 3 were consistent with the findings in Figure 3, indicating that the mediation of perceived social support and that of positive self-presentation, as well as the chain mediation of honest selfpresentation and perceived social support, were established. However, neither the mediation of honest self-presentation nor the chain mediation of positive self-presentation and perceived social support were established. Another multiple chain mediation model (Model 2) was used to examine the indirect effects of the three motivations for SNS use on well-being. They were interrelated independent variables, as well as first-order latent variables. The model fit well (<unk> 2 5 1050.885, df 5 640, RMSEA 5 0.029, SRMR 5 0.039, CFI 5 0.951, TLI 5 0.946). As shown in Figure 4, social interaction motivation predicted honest self-presentation and positive self-presentation significantly, entertainment motivation predicted honest self-presentation and well-being significantly, and information motivation predicted perceived social support significantly. The results of Model 2 were consistent with those of Model 1 regarding the links between positive and honest self-presentation, perceived social support and well-being. Table 4 shows the indirect effects of the established mediation paths. --- Discussion --- Positive effect of motivations for SNS use on well-being The results showed that the overall intensity of motivations was a significant predictor of well-being. This finding confirmed Hypothesis 1a and suggested that SNS users with higher AJIM 75,1 motives felt happier when publishing messages on SNSs. An explanation is that message posting is an active-SNS-use behavior. According to the uses and gratifications theory, highly motivated SNS users are more likely to engage in active SNS usage activities to meet their needs or desires. Participants in this study were virtually active SNS users, with 93.5% having used at least one SNS for two years and 92.4% spending more than 30 min each day on SNSs. SNS users' positive usage behaviors were associated with their well-being (Verduyn et al., 2017). Another explanation is that all the motivations examined in this study were positive. Prior research revealed that most seemingly positive motives resulted in high life satisfaction or other good consequences (Guo et al., 2014;Wen et al., 2016). Undesirable motives, such as escape, fear of missing out and passing the time, were linked to several negative behaviors or outcomes (Throuvala et al., 2019;Jarman et al., 2021). Although social interaction and entertainment motivations were significantly positively associated with well-being in this study, information motivation was not. These results supported Hypothesis 1b and 1c, but did not support Hypothesis 1d. As prior studies showed, social interactions on SNSs can help the users build and maintain social networks, making them feel accepted and trusted, and have higher self-worth (Yang and Brown, 2016) and life satisfaction (Lian et al., 2020). Also, the intrinsic motivation (enjoyment or fun inherent to the behaviors) for using SNS positively predicted well-being (Wen et al., 2016;Manuo glu and Uysal, 2020). In this study, posting hobbies and funny experiences satisfied enjoyment or amusement needs, which can bring positive emotions and directly improve subjective wellbeing. However, information motivation had no direct correlation

Purpose -This study aimed to investigate positive associations between three main motivations (social interaction, information and entertainment) for the use of social network sites (SNSs) and users' well-being, as well as the multiple mediating effects of perceived social support, positive and honest self-presentation. Design/methodology/approach -A sample of 759 active users of SNSs (WeChat Moments, Qzone and Weibo) aged 14-43 years was measured with online questionnaires. Correlation analysis and structural equation modeling were implemented to examine the corresponding hypotheses. Findings -The results showed the overall intensity of motivations was positively associated with users' wellbeing; perceived social support and positive self-presentation played intermediary roles and honest selfpresentation and perceived social support had a chain mediation effect. However, the motivations of social interaction, information and entertainment indirectly affected users' well-being through three different mediation paths. Originality/value -Although some studies have investigated the effects of motivations (including social interaction, information and entertainment) for SNS use on users' well-being, there has not been a consistent conclusion. The findings may shed light on the motivations for SNS use and how they may affect people's wellbeing in the digital era, thereby promoting their healthy use of SNSs as well as improved interface design and user management of SNSs.

al., 2014;Wen et al., 2016). Undesirable motives, such as escape, fear of missing out and passing the time, were linked to several negative behaviors or outcomes (Throuvala et al., 2019;Jarman et al., 2021). Although social interaction and entertainment motivations were significantly positively associated with well-being in this study, information motivation was not. These results supported Hypothesis 1b and 1c, but did not support Hypothesis 1d. As prior studies showed, social interactions on SNSs can help the users build and maintain social networks, making them feel accepted and trusted, and have higher self-worth (Yang and Brown, 2016) and life satisfaction (Lian et al., 2020). Also, the intrinsic motivation (enjoyment or fun inherent to the behaviors) for using SNS positively predicted well-being (Wen et al., 2016;Manuo glu and Uysal, 2020). In this study, posting hobbies and funny experiences satisfied enjoyment or amusement needs, which can bring positive emotions and directly improve subjective wellbeing. However, information motivation had no direct correlation with SNS users' well-being, although there was some evidence in prior studies (Guo et al., 2014 Social network site use and users' wellbeing 2020). Information motivation mainly refers to information acquisition from others, which will not be satisfied without feedback. Overall, these results further indicated that the impact of SNSs depends on why and how people use them. The reasonable and active usage of SNSs can promote users' psychological functioning or social adaptation. 6.2 Mediation of perceived social support in the relationship between motivations for SNS use and well-being The mediation effect of perceived social support existed between the overall intensity of motivations for SNS use and well-being. This finding was expected by Hypothesis 2a and was consistent with previous research (Nabi et al., 2013;Giota and Kleftaras, 2014;Oh et al., 2014). When highly motivated SNS users were exposed to more helpful social interactions (such as giving advice, showing empathy or teaching something new), they felt greater support from online friends. Furthermore, posting on SNSs is a quick and cost-effective approach to request assistance, ensuring that users receive fast and meaningful assistance. It can assist them in resolving problems and improve their perception of their capacity to cope with difficult situations (Frison and Eggermont, 2015b). There is a difference in the relationships between the three main motivations for SNS use and perceived social support. Although information motivation had no direct effect on SNS users' well-being, it indirectly affected well-being through perceived social support. This finding confirmed Hypothesis 2c. As mentioned earlier, information motivation mainly refers to information acquisition in this study. It is positively associated with social, emotional and information support online as SNSs are a tool for maintaining and increasing social capital (Wang et al., 2019). The results rejected Hypothesis 2b and 2d, because social interaction motivation and entertainment motivation had no direct effect on perceived social support. However, these two motivations affected perceived social support through honest selfpresentation, as discussed below. Interesting or self-entertaining posts can increase interpersonal attraction and social interactions. Meanwhile, interpersonal interactions on SNSs are helpful for bonding and bridging social connections, providing SNS users more support online when they need it (Wang et al., 2019). In summary, although the paths are different, the three motivations for SNS use enhance users' perceived social support online and promote users' well-being. --- Mediation of positive self-presentation in the relationship between motivations for SNS use and well-being The results showed that the mediation effect of positive self-presentation existed between motivations for SNS use and well-being while the mediation effect of honest self-presentation did not. The results partially supported Hypothesis 3a. Consistent with prior studies (Kim and Lee, 2011;Gonzales and Hancock, 2011), highly motivated SNS users would more frequently present their positive aspects in the posts and thus feel more satisfied with their lives. A reason is that SNS users can experience well-being by presenting their ideal selves on SNSs. Users often review their profiles and posts, even their comments under the posts. According to Gonzales and Hancock (2011), updating and viewing profiles could improve their self-esteem, specifically, when they selectively presented their data, because it could make them focus on their positive information and evoke positive emotions within themselves. Another reason is that positive self-presentation on SNSs can also reflect a positive self-image (Kim and Lee, 2011). According to the positive illusion theory (Taylor and Armor, 1996), exaggerated self-perceptions help cope with self-threatening events or stressful situations and promote happiness. Specifically, only social interaction motivation could affect well-being through positive self-presentation; information and entertainment motivations could not. These findings partially supported Hypothesis 3b, 3c, and 3d. People who were developing intimate AJIM 75,1 relationships would want to obtain a commitment from others by presenting their ideal selves (Boyle and Johnson, 2010). Positive self-presentation on SNSs is helpful to establish or maintain interpersonal relations, and the presenter experiences greater well-being (O'Donnell et al., 2021). As the hypothesis, information motivation and entertainment motivation did not show a direct effect on positive self-presentation on SNSs. Users who post on SNSs for information acquisition and enjoyment rarely try to manage others' impressions of themselves (Wang et al., 2014;Pang, 2020). Their focus is on information or fun and not the purpose for sharing information about their life and identity (O'Donnell et al., 2021). Conversely, information-sharing may correlate with self-presentation because it has often been considered as a kind of communication activity or social exchange (Jarman et al., 2021). 6.4 Chain mediation of honest self-presentation and perceived social support between motivations for SNS use and well-being Unlike positive self-presentation, honest self-presentation showed a chain mediation effect on the link between the overall intensity of motivations for SNS use and well-being. This finding partially validated Hypothesis 4a. Higher incentives encourage SNS users to provide a more realistic image in their posts, which will be helpful to establish real interpersonal connections (Boyle and Johnson, 2010); nonetheless, they cannot directly lead users to feel better until they receive positive feedback. Self-disclosure that is unfiltered and profound is known as honest self-presentation. The degree of self-disclosure was due to the objects, context assessments and personal worth, such as building vital relationships and seeking emotional support (Greene et al., 2006). People tend to make more frequent, intimate and honest disclosures when they are stressed. Disclosing some difficulties or negative aspects for reciprocal selfdisclosure or social support from others can shield people from life stress and increase their well-being (Lee et al., 2011;Liu et al., 2016;Zhang, 2017). Hypothesis 4b partially received the support from one of the findings that there was a link between social interaction motivation and well-being through the chain mediation of honest self-presentation and perceived social support. This result is consistent with prior studies (Pang, 2020). First, honest self-presentation on social media is generally for social connection (Wang et al., 2018b). Users with social motives would like to include personal information in their posts (Boyle and Johnson, 2010). Second, when SNS friends realize someone needs help, they are more inclined to offer support through honest self-presentation rather than positive self-presentation (Kim and Lee, 2011). Candid admissions about one's struggles and bad feelings can help one gain interpersonal trust and get emotional support or advice from SNS friends, which are beneficial to well-being. However, failure to receive encouraging responses from SNS friends may aggravate the user's depressed mood (Moreno et al., 2013;Park et al., 2015). Between entertainment motivation and well-being, the results demonstrated a chain mediation of honest self-presentation and perceived social support, but not between information motivation and well-being. Hypothesis 4d was confirmed, but Hypothesis 4c was rejected. The findings differ from those of Pang (2020), who found no link between entertainment motivation and online social support, but did not specify whether online selfpresentation was honest or positive. This study found that posting on SNSs for pleasure can also be used to present oneself in an implicit way (Yang and Brown, 2013). Although SNS users may be engaged in the entertainment functions and have no intention of disclosing their personal information, this study looked into the impact of posting for enjoyment, which also displays one's interests or preferences. Funny photos or unpleasant experiences shared for pleasure on social media can inadvertently disclose the uploaders' actual selves. They will feel happier with themselves and others as a result of subsequent encouraging feedback from others. Information motivation, on the other hand, was unrelated to any type of self-Social network site use and users' wellbeing presentation. This is most likely because information-seeking motivates users to focus on information-exchanging on social media sites for current and future information requirements rather than self-related information presentation (Lee and Ma, 2012). --- Theoretical and practical implications This study presented a model for understanding the underlying process between active use and enhancement of well-being in SNS users with various motivations. The findings confirmed that self-presentation and perceived social support are two significant mediators in the construction of three pathways from motivations for using SNSs to well-being. The mediating roles of positive and honest self-presentation are different. Each of the three main motivations for using SNSs (social interaction, information and entertainment) has unique pathways to well-being. This study backed up the uses and gratification theory by demonstrating how SNSs can fulfill a variety of needs. The study's findings can help people learn to utilize social media healthily. Why and how people use SNSs determines their value as a communication channel. Not only should educational institutions or the media tell the public about the problems or negative consequences that improper use of SNSs might create, they should also instruct the public on how to utilize SNSs effectively and productively to benefit. Everyone's active use facilitates the creation of a favorable online social or public opinion environment. As a result, SNS users should clarify their goals and needs (mainly social engagement, information-seeking and entertainment), engage in more active use behaviors (particularly message posting) and connect favorably with online friends via likes, comments, forwards and other features. According to the findings, SNSs should optimize their functional modules to support selfpresentation and social interaction, such as rewarding with points, improving user group division, increasing content management and guiding social interactions in a supportive way. SNSs can also tailor their functional modules according to their target users' multiple motivations. Furthermore, SNS administrators can encourage and favorably steer users toward self-presentation and social interaction on SNSs by supporting or hosting activities. SNS use can also improve their sense of belonging to the site while satisfying their needs. Their sense of belonging makes them conversely engage in more postings, reading and issue conversations. A great example is the "Ice Bucket Challenge" for Amyotrophic Lateral Sclerosis. This activity not only met its public welfare goals but also improved users' self-presentation, perceived social support and well-being, as well as the data and social impact of SNSs. --- Limitations and future research There are a few limitations in our study that should be discussed and addressed in future research. First, the hypotheses were examined with cross-sectional data. However, the hypothesized model was tested using a structural equation model of latent variables, controlling the measurement error effect. The models will be a good foundation for further investigation. Future studies should consider using a longitudinal or experimental study design to investigate the causal links between variables. Second, the average daily usage time of SNSs was significantly connected with the motivation for SNS use, but not with the other variables. Although this finding is consistent with that of some studies which found that time spent on a SNS was not related to users' well-being (Lee et al., 2011;Nabi et al., 2013;Guo et al., 2014;Baek et al., 2014), future studies should measure it as a continuous variable to examine its direct or moderated effect on other variables. Third, age and gender had some effects on research variables and were controlled in this study, suggesting that they could be used as moderators in future research. Finally, this study simply defined SNS use as message posting AJIM 75,1 and looked into three positive motivations for message posting. Researchers can look at the negative motivations (such as passing time and escape) for passive SNS use and how they relate to SNS use outcomes in the future. --- Conclusion This study found that the higher the overall intensity of the motivation for using SNSs, the greater the users' well-being. There were three intermediary paths between them: the simple mediation of positive self-presentation (indirect path A), the simple mediation of perceived social support (indirect path B) and the chain mediation of honest self-presentation and perceived social support (indirect path C). Specifically, social motivation encourages users to make more positive self-presentations in posts, leading to their improved happiness (indirect path A), or to make more honest self-presentations, enhancing perceived social support and further improving well-being (indirect path C). Users' happiness cannot be directly increased by posting messages for information acquisition; rather, their life satisfaction can be improved only when their posts receive more feedback and increase in perceived social support (indirect path B). Users' happiness can be immediately boosted for the enjoyment while posting, and their unintended authentic self-presentation and subsequent feedback from others can also further enhance their life satisfaction (indirect path C). The prior findings suggest that a combination of positive motivation and active usage behavior on SNSs can have a favorable impact on users. There are three common mediation paths between SNS users' posting motivation and their well-being and different motivations have varied influence mechanisms on the latter.

Purpose -This study aimed to investigate positive associations between three main motivations (social interaction, information and entertainment) for the use of social network sites (SNSs) and users' well-being, as well as the multiple mediating effects of perceived social support, positive and honest self-presentation. Design/methodology/approach -A sample of 759 active users of SNSs (WeChat Moments, Qzone and Weibo) aged 14-43 years was measured with online questionnaires. Correlation analysis and structural equation modeling were implemented to examine the corresponding hypotheses. Findings -The results showed the overall intensity of motivations was positively associated with users' wellbeing; perceived social support and positive self-presentation played intermediary roles and honest selfpresentation and perceived social support had a chain mediation effect. However, the motivations of social interaction, information and entertainment indirectly affected users' well-being through three different mediation paths. Originality/value -Although some studies have investigated the effects of motivations (including social interaction, information and entertainment) for SNS use on users' well-being, there has not been a consistent conclusion. The findings may shed light on the motivations for SNS use and how they may affect people's wellbeing in the digital era, thereby promoting their healthy use of SNSs as well as improved interface design and user management of SNSs.

Background Maternal and child health (MCH) has been globally recognized as one of the key indicators in measuring health system performance. The Millennium Development Goals (MDG), set the target to reduce under-five mortality rate and maternal mortality rate by two thirds and three quarters respectively by 2015 [1,2]. To rectify the unfinished MCH goals in the MDG era, the Sustainable Development Goals (SDG)-3, established the ambitious targets of global maternal mortality reduction to less than 70 deaths per 100,000 live births, and to ending preventable deaths of children under 5 years of age by 2030 [3]. The MCH goal was also linked with the SDG-2 where the United Nations (UN) Member States committed to promote food security and improve the nutritional status of vulnerable populations, including children under 5 years of age, adolescent girls, pregnant and lactating women, and the elderly [4]. The World Health Organisation (WHO) proposed 11 indicators for MCH monitoring-maternal mortality, prevalence of stunted children, exclusive breastfeeding for 6 months after birth, and skilled birth attendants, to name a few [5]. Amongst these indicators, De Onis and Branca suggest that childhood stunting is 'the best overall indicator of children's well-being and an accurate reflection of social inequalities' [6]. The WHO also highlighted that stunting is a result of long-term malnutrition, poor diets and nutritional deprivation-leading to growth retardation, delayed mental development, poor school performance and reduced intellectual capacity [7,8]. The consequences of stunting are catastrophic, not only in terms of health but also in terms of economic outcomes. The World Bank suggested that a 1% increase in loss of height is associated with a 1.4% loss in economic productivity [9]. The global prevalence of stunting is extremely high. Around 155 million children under 5 years of age are experiencing stunting [10]. The situation is most severe in Africa and Southeast Asia, which account for about 34% of the global stunting prevalence [11]. A number of factors contribute to stunting: political economy; health and health care systems; education, society and culture; agriculture and food systems; and water, sanitation and environment [12]. Poor hygiene and low breastfeeding rates were associated with stunting [13], and low birth weight and prolonged breastfeeding (more than 12 months) without adequate and appropriate supplementary feeding, and diarrhoea were among major risk factors [14]. Factors that contribute to stunting can be categorized as (1) child conditions, such as non-exclusive breastfeeding and discontinuity of breastfeeding before 2 years, slow fetal growth during pre-natal period, and having been born prematurely [9,15]; (2) maternal conditions, including poor nutrition and health status of a mother during pre-natal, peri-natal, and post-natal phases, and inadequate preparation of a mother during pregnancy [9]; and (3) socio-economic factors, both macro-level determinants (lack of sanitation and an unhealthy environment for child rearing), and micro-level determinants (household poverty, improper childrearing practices, and poor maternal education) [14]. Thailand has been successful in improving health outcomes and providing financial protection for its population in spite of relatively low healthcare spending [16]. The infant mortality rate was markedly reduced from 26 in 1990 to 11 per 1000 live births in 2010 [17]. Despite this, attempts to combat childhood stunting are not progressing well. Although the early 1990s marked a success in reducing stunting prevalence from over 22% in 1987 to around 12% in 1995, the stunting trend has stagnated and never fallen below 10% since then [13]. Child stunting was found to be the second most concentrated MCH problem amongst the poor, following only the problem of underweight children [18]. Moderate and severe stunting stood at 10.5% in 2016. The prevalence amongst households headed by a non-Thai speaker was more pronounced at 16.3% [19]. The path to tackling the problem of stunting is not paved with roses, as a case study from Thailand clearly reflects. In response to the repeated violations of the Code of Practice on Marketing of Breast Milk Substitutes by milk-substitute industries [20], in 2017 the Thai Ministry of Health (MOPH) introduced a draft Bill to regulate such interferences by the industries. However, there was strong resistance from certain paediatricians and the Medical Council who amplified the downsides of prolonged breastfeeding. Though several studies on childhood stunting describe the magnitude of stunting, comprehensive assessment of its social determinants is lacking [21,22]. A better understanding of contributing determinants will help policy makers to devise proper comprehensive multi-sectoral policies to alleviate stunting prevalence. Therefore, this study aimed to assess potential factors contributing to stunting amongst Thai children below 5 years of age, in particular those who were breastfed beyond 12 months. --- Methods --- Operational definition Prevalence of stunting is defined as the percentage of children under age five whose height for age is greater than two-fold of standard deviations at the direction below the international reference median for children aged 0-59 months. Note that for children up to two years old, height is measured by recumbent length unlike older children where height is measured by stature while standing [23]. --- Study design and data source This study applied a cross sectional quantitative design. Data were obtained from the Multiple Indicators Cluster Survey (MICS) in 2012, which was jointly conducted by the International Health Policy Program (IHPP), the MOPH, the National Statistical Office (NSO), the Thai Health Promotion Foundation and the United Nations Children's Fund (UNICEF) in 2012. Multi-stage stratified cluster sampling was applied. Enumeration areas (EAs) and households served as primary and secondary sampling units respectively (approximately 20 households in each EA). The overall sampling frame is presented in Table 1. A total of 24,119 households participated in the survey from a target of 26,850 (response rate = 98.5%). Amongst these households, 9716 children were recruited. As this study focused on factors that significantly contribute to stunting, particularly the effects of prolonged breastfeeding, we limited our analysis only to children aged over 12 months (as prolonged breastfeeding applied the cut-off at 12 months), numbering 7018 in total. --- Data collection, questionnaire design, and variable management Face-to-face interviews with mothers and/or legal guardians were performed in each household by the NSO field staff. Each interview took an average of 60 min to complete. Field staff entered data directly into the mobile tablets. Revisiting was done if mothers at the visited household were absent during the first round of survey. The standard MICS questionnaire was used [24]. All variables were those collected earlier in the prior rounds of MICS [9,14,15]-for instance, history of breastfeeding; maternal education; and location of a household, and also other variables such as geographic region; housing characteristics; economic status; maternal antenatal care history; and health status of a child before the survey period. The variables of our focus were (1) current age (months), height (cm), and weight (kg) of a child for measurement of moderate and severe stunting; (2) pregnancy age at delivery (less than 20 years, 20-29 years, 30-39 years, and equal to or more than 40 weeks); (3) maternal perception on the size of her baby at birth (small, average, large, and 'cannot clearly determine the size'); (4) history of diarrhoea during the last two weeks before the survey (yes v no); (5) educational background of mother (no education, primary school completion, secondary school completion, diploma, bachelor degree, and above bachelor degree); (6) household domicile (municipality [representing urban] v outside municipality [representing rural]); (7) geographic region of the household (Bangkok, Central, Northern, Northeastern, and Southern); (8) antenatal care (ANC) history (4 visits or more, fewer than 4 visits, and 'cannot remember visit number'); (9) breastfeeding history (less than 12 months v 12 months or more), and; (10) household asset index (poorest, poor, moderate, rich, and very rich asset quintiles). Note that the World Bank recommends using asset index as the indicator for measuring household's economy over monetary income [25]. The index was calculated by principal component analysis, taking into account all important durables, land possession, household infrastructure, and domestic commodities. --- Data analysis We used STATA software version 14 (serial license number: 401406358220) to analyse the data. The analysis was divided into three steps. Firstly, descriptive statistics were done to provide the overview of the data. Secondly, a univariable analysis by Chi-square test was done to determine the association between each covariate and stunting. Lastly, multivariable logistic regression was applied to assess the likelihood of stunting after accounting for the effects of all covariates at the same time. All covariates from the univariable analysis were included in the multivariable analysis regardless of the significance level, since we hypothesized that these variables were likely related to stunting, as supported by much literature in the introduction section. As stunting can be a result of complex interactions between various social determinants, we then divided the multivariable analysis into two strands: first, the analysis without interaction terms and second, the analysis with interaction terms. In this regard, a new variable representing the interactions between breastfeeding duration and a household's socio-economic profile was included in the final model. Robust standard error was used to adjust for potential cluster effects within a household. Crude and adjusted odds ratios (OR) with 95% confidence interval (95% CI) were presented. Statistical significance was measured at the 95% confidence level (P-value <unk> 0.05). --- Results --- Descriptive statistics and sample characteristics A total of 7018 children under 5 years old were enrolled in this study. Around three quarters of them were born from mothers who had completed at least high-school education and gave birth at age between 20 and 39 years. The majority of those mothers for whom ANC information was available had adequate ANC visits. About 88.74% of the children were not currently breastfed. Only 30.8% of the children had been breastfed for more than 12 months. Most mothers (71.0%) did not have adequate information about the size of their child at birth. However, for those mothers who could determine the size of their child at birth, the most common report was 'average' (21.4%), followed by 'large' (4.9%). Over 95.7% of the children did not have recent experiences of diarrhoea and approximately 14.1% of the children experienced stunting. Regarding household attributes, around three quarters were ranked in the 3rd to 5th quintiles of the wealth index. Those located in a municipal area slightly outnumbered those living in rural areas. The samples were quite equally shared between the different geographic zones except Bangkok (Central 21.0%, Northern 22.0%, Northeastern 23.5%, Southern 24.2% and Bangkok 9.3%) as depicted in Table 2. --- Univariable analysis The results of univariable analysis are described in Table 3. Clearly, a high maternal education background and increased age at delivery had an inverse relationship with the probability of childhood stunting. Prolonged breastfeeding was positively associated with stunting (OR = 1.57, 95% CI 1.35-1.81). Stunted children tended to be the younger age groups (less than 36 months), and were those who had a small size at birth, as reported by their mothers. Furthermore, the finding shows that children brought up in the better-off households, and those living within municipalities, had lower stunting probability compared to those from poor socio-economic backgrounds. Note that geographic residence was not significantly associated with stunting. --- Multivariable analysis Table 4 presents the outcomes of multivariable analysis (both with and without interaction effects). In the analysis with no interaction effects, higher maternal education and the more advanced the age of mothers at the time of delivery lowered the probability of stunting. The perceived size of children at birth as 'average' or 'large' yielded smaller odds of stunting compared to those with a low birth weight who were reported by mothers as'small' in size. Children in the better-off households were less likely to face stunting. Prolonged breastfeeding was positively associated with stunting. Other covariates, such as number of ANC visits and history of diarrhoea within the last two months, appeared to have insignificant effects on stunting. The interactions between the duration of breastfeeding and household economic levels contributed to stunting with varying degrees of statistical significance. By using the interaction between 'poorest wealth index' and 'duration of breastfeeding of less than 12 months' as a reference, it appears that wealth provided a protective effect against stunting regardless of the duration of breastfeeding; but amongst mothers in the poorest quintiles, beyond-12-months breastfeeding tended to contribute to stunting relative to those with shorter duration of breastfeeding. The direction of odds ratio and significance level of other covariates in the analysis with interaction terms did not show a remarkable difference from the analysis without interaction terms. --- Discussion This study sheds light on risk factors of stunting amongst Thai children under-five years of age through the analysis of the 2012 MICS data. The household's economic status was the strongest influence on stunting. Children in the poorest quintile were about two-fold more likely to experience stunting than those in the richest quintile. The discovery from this study supports findings from some prior research. For instance, the UNICEF data and the Global Nutrition Report suggested that stunting prevalence amongst children in the poorest households was more than double those in the richest households [10,26]. Furthermore, studies in Burundi and Nepal found that economically poorer households had a greater association with stunting and severe stunting than the richer ones [27,28]. Likewise, in Nigeria, low household wealth index was a risk factor for severe stunting [29]. However, when it comes to breastfeeding, the findings should be interpreted with caution. While Tiwari et al. from Nepal [28] and Akombi et al. from Nigeria [29] reported that a long duration of breastfeeding contribute to stunting, our study found that prolonged duration of breastfeeding by itself is not a strong determinant for stunting, but rather, the interaction between breastfeeding duration and wealth status. Also, this point is largely overlooked in most existing literature on this subject. Children with beyond-12-months breastfeeding in the poorest households were most prone to stunting while children in the well-off households were less likely to suffer from stunting regardless of their breastfeeding duration. Theoretically, long-term breastfeeding creates substantial benefit to the health of mothers and children. Previous studies [27,[30][31][32] found that continued breastfeeding positively affected children's development reducing the risk of autism spectrum disorder (ASD) and increasing cognitive development. In addition, mothers who breastfeed have a lower risk of chronic diseases [32]. A potential explanation for this phenomenon is that the poorest families might not have means to sufficiently provide an adequate amount of appropriate complementary food for their children after the first six-months of exclusive breast feeding. As a result, breastfeeding was continued as the only viable choice, but its nutrient content alone is not sufficient to match the increased demand of children as they grow up. This coincides with the WHO's recommendation, suggesting that exclusive breastfeeding should be performed for the first 6 months. After that, complementary food should be provided in addition to breastfeeding. Unfortunately, the current version of the MICS questionnaire does not allow for detailed analysis of complementary food acquisition. To unpack this complexity, UNICEF should, in the future, amend the MICS survey questionnaire to reflect the magnitude and profile of supplementary food feeding and feeding management. Additionally, some literature mentions the effects of the practice of six-month exclusive breastfeeding as a factor that contributes to optimal child growth [33,34]. This may also explain the stunting phenomenon in this study as the exclusive breastfeeding rate of Thai infants during the first six months of life was consistently low (MICS 3 in 2009 and MICS4 in 2012 show that the exclusive breastfeeding rate was only 5.4% [35] and 12.3% respectively) [24]. For other covariates, those with low birth weight or who were born premature were about 30-40% more likely to develop stunting than those with a normal or large body size at birth. Surprisingly, children residing in more affluent areas, like Bangkok, were more likely to face stunting than those in other regions. This finding is in contrast to some of the international literature on this subject which has found childhood stunting to be more prevalent in rural than urban households [6,26,36]. The child-rearing behaviour of mothers in Bangkok, who were mostly employed in the service and manufacturing sectors and as a result had limited time, which subsequently affected the quality of child rearing, and the nuclear-family characteristic of most households in Bangkok where little or no support from relatives is provided to mothers, are possible explanations. Despite a rigorous sampling technique and a large number of participants, which are the strengths of the study, certain limitations remain. Firstly, the questionnaire did Without this information, the association between duration of breastfeeding beyond 12 months and stunting can be misinterpreted, and send the 'wrong signal' to the society about the downside of breastfeeding. It can be used wrongly by the 'pro-breast milk substitutes' advocates to refute the benefits of breastfeeding. In fact, there was a question about food security in the MICS questionnaire, that is, 'Did your child have the following food in the last 24 hours?'. The respondents were asked to answer 'yes' or 'no' for each food item on the list, for instance, water, condensed milk, rice, and fruit. However, it seems that the question was not well crafted, making it difficult to serve as an indicator for complementary food consumption. Therefore, it was dropped from our analysis. Secondly, the sampling process relied on the household registry from the Department of Provincial Administration. Although this is a normal process, used by other NSO surveys, there is a downside to this approach. That is, people whose households do not have a registry number, especially disadvantaged populations such as undocumented migrants, homeless people, and slum dwellers, were likely to be excluded from the sampling frame. As a result, the results might underestimate stunting prevalence as the analysis may be prone to missing the poor households. Thirdly, the question about perception on the size of babies at birth provided the only available measurement of a child's health status at the time of delivery, as information about the exact weight of a child at birth was not available in most households. The question was prone to recall bias and measurement error. Further studies that collect the true actual size of the babies at birth are recommended. Finally, there were other factors that were not captured by the questionnaire, for instance: the child rearing practice by mothers, household environment, and familial genetic problems. All these limitations were difficult to rectify through a single quantitative survey, warranting future studies to unpack these complexities. --- Conclusion and policy implications In conclusion, this research found that childhood stunting was significantly associated with the following risk factors: poor economic status of a household, small size of a child at the time of delivery, and residential location in Bangkok. The study also suggested that prolonged breastfeeding beyond 12 months when in combination with poor economic status of a household potentiated the risk of stunting. Household wealth was protective factor against stunting in children who were breastfed for both short and prolonged periods of breastfeeding. Children living in the least well-off households were more prone to experience stunting than others. This study points to target interventions to identify and support the poorest households, using multi-sectoral actions such as poverty reduction, ensuring food security, health education and the empowerment of women. As long as household poverty remains an impediment, health education and supplementary food provision alone will not be sufficient to radically tackle the malnutrition of children. --- Availability of data and materials The data that support the findings of this study are available from the NSO but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the NSO. --- Abbreviations Author's contributions All authors designed the study. NC, CT, RS, KT and SL were responsible for data analysis. NC, RS and CT crafted the first draft of the manuscript. KT SL and VT revise and finalize the manuscript. All authors contributed toward data analysis, drafting and critically revising the paper and agree to be accountable for all aspects of the work. --- Ethics approval and consent to participate The data set used by this study is one of the national routine household surveys conducted by the government agency, namely the National Statistical Office (NSO). As mandated by the Statistics Act 2007, (see http:// web.nso.go.th/en/abt/stat_act2007.pdf) the NSO is not required to obtain signed consent forms from respondents. However, Article 15 of the Statistics Act ensures confidentiality of data collected by the NSO fieldwork and also other users of the dataset. In this study, the researchers strictly followed the confidentiality requirement in Article 15 of the Act; all individual information was strictly kept confidential and not reported in the paper. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Childhood stunting, defined as the height-for-age standardized score lower than minus two, is one of the key indicators for assessing well-being and health of a child; and can be used for monitoring child health inequalities. Thailand has been successful in improving health and providing financial protection for its population. A better understanding of the determinants of stunting will help fill both knowledge and policy gaps which promote children's health and well-being. This study assesses the factors contributing to stunting among Thai children aged less than five years. Methods: This study obtained data from the Multiple Indicator Cluster Survey Round 4 (MICS4), conducted in Thailand in 2012. Data analysis consisted of three steps. First, descriptive statistics provided an overview of data. Second, a Chi-square test determined the association between each covariate and stunting. Finally, multivariable logistic regression assessed the likelihood of stunting from all independent variables. Interaction effects between breastfeeding and household economy were added in the multivariable logistic regression. Results: In the analysis without interaction effects, while the perceived size of children at birth as 'small' were positively associated with stunting, children in the well-off households were less likely to experience stunting. The analysis of the interactions between 'duration of breastfeeding' and 'household's economic level' found that the odds of stunting in children who were breastfed longer than 12 months in the poorest household quintile were 1.8 fold (95% Confidence interval: 1.3-2.6) higher than the odds found in mothers from the same poorest quintiles, but without prolonged breastfeeding. However prolonged breastfeeding in most well-off households (those between the second quintile and the fifth wealth quintile) did not show a tendency towards stunting. Conclusions: Childhood stunting was significantly associated with several factors. Prolonged breastfeeding beyond 12 months when interacting with poor economic status of a household potentiated stunting. Children living in the least well-off households were more prone to stunting than others. We recommend that the MICS survey questionnaire be amended to capture details on quantity, quality and practices of supplementary feeding. Multisectoral nutrition policies targeting poor households are required to address stunting challenges.

Introduction I was not born in Hatch End, Middlesex. I was actually born in the Royal Free Hospital, Islington but my parents moved there shortly after my birth and just before the outbreak of the World War II. 1 By then it was a small but developing suburban area between Harrow and Watford, a station on the Bakerloo Line from Queen's Park or the main line from Euston. The house that we moved into was part of a small, leafy private estate, the roads lined with almond and crab apple trees. It was located in a small cul-de-sac, Felden Close, and seemed to be about ten minutes' walk from most things that we might need: the station, the shops, the doctors and the school. Only the Methodist Church required a longer walk or a short bus ride to Pinner. In this paper I want to use some notions derived from Lefebvre, perhaps at some distance, to explore my experiences of Hatch End and its place in my biography (Holdsworth and Morgan 2005: 73-90;Lefebvre 2002). I consider Hatch End in terms of practical space, symbolic space and imaginary space and the interactions between them. --- Practical Space Some of the practicalities I have hinted at already. Most amenities (shops, the post-office, the station, a pub, the Anglican Church etc) were within easy walking distance, an important fact in the days when car ownership was only beginning to take off. We never possessed a car and my parents never learned to drive. The station (an attractive building) boasted some six (or was it more?) working platforms. 2 This was commuter country, and it was a common sight in the mornings to see commuters dodging between platforms to attempt to get the fastest train into central London. In addition to the train there were also reasonably frequent buses to Pinner and North Harrow. 'Conveniently placed' might be a term used to describe both Hatch End itself and our house within it. --- Symbolic Space The development of Hatch End, previously little more than a small village, was clearly part of the development of the commuter belt around London. Workers of various kinds could be expected to be in the centre in under an hour, aided by what were called "Workman's" tickets. But these practicalities revolved around symbolic considerations. These can be briefly listed: --- Countryside In common with developments from the mid nineteenth century onwards, Hatch End reflected a belief in the symbolic value of the English countryside, with an abundance of greenery and open spaces. There was some reality in this. Working farms were within walking distance and, in my early childhood, it was possible to walk along the lanes and across the fields to the edge of Watford. Residents seemed to go along with these ideas; the shopping area strung out along the main Uxbridge Road was frequently referred to as 'the village'. --- Family The newly developed semis reflected an organisation of domestic space which, in the 1950s, came to be identified with the ideal of companionate marriage. Parents were able to sleep apart from their children and the houses and gardens were constructed as shared spaces. --- Gender At the same time the development of Hatch End and places like it, reflected the wider and gendered division between home and work. There was a predominance of men in the daily commutes to London while during the weekday, Hatch End was predominantly, apart from shopkeepers, a place of women and children. --- Class As with other suburban areas there was some distinction between the 'established' (or relatively established) and 'the outsiders'. So, our new estate (which my mother told me was sometimes referred to as a slum) contrasted with the larger and older houses on "The Avenue". Our immediate neighbourhood was probably quite mixed in terms of occupation and social origins but despite these differences there was a key theme of'respectability'. This sense of respectability probably became reinforced immediately after the war with the development of 'the prefabs' (on a supposed temporary basis) which housed families from the centre of London. They were considered rough and we respectable. --- Modernity From a present-day vantage point, the identification of the new estates in Hatch End and similar places with modernity might seem a little quaint. Our house, for example had a coal fire and a stove in the kitchen to heat the water; there was just a two-bar electric fire in the dining room. Yet at the time, the clean uncluttered lines of the new houses, the space, the electric light and the gas street lights must have seemed at the front edge of the modern age. Some houses had garages and those that did not had a drive that was wide enough to take a car at some unspecified stage in the future. To summarise, the symbolic space of Hatch End at the time I was growing up was anchored around notions of the countryside, modernity and family. Gender divisions remained apparent but there was some blurring of class lines in the face of a concern with respectability and finer divisions of status. --- Imaginary Space Trains stopping at Hatch End were greeted with a sign which read: 'Hatch End: For Pinner'. As a child I remember feeling resentful about this sign. Why did not Pinner Station (which had fewer platforms than Hatch End) have a sign which read: 'Pinner: for Hatch End'? But did I really feel resentful? I probably didn't. Even as the thought passed through my mind, I almost certainly realised that Pinner had justice on its side. Pinner looked like a real place, the sort of place someone might go to rather than pass through. The High Street rose up past some old shops and dwellings to a striking parish church and a shady green. Pinner had several pubs and cafes and a cinema as well as small bus station and an Annual Fair which went back for centuries. It was older, more established, more affluent and appeared to have a strong sense of local identity, expressed in a regular publication called, appropriately enough, 'The Villager'. Pinner was very much located in Metroland as immortalised in John Betjeman's BBC film of 1973. It was the Metropolitan Line that took commuters from Pinner to Baker Street in about 30 minutes. In contrast, Hatch End seemed to be on the edge in several aspects. This is not the 'Edgelands' as graphically described by Farley and Roberts (2011). It was not quite Metroland; it was probably not old enough, middle-class enough or quirky enough to attract the attention of a Betjeman. Although, especially in my early years, there were clear elements of countryside, it did not feel like the countryside. It lacked real drama and houses all too quickly came into view. Perhaps the grandest piece of open space was virtually out of bounds as it belonged to Pinner Hill Golf Club, although you could pick blackberries on the bordering bushes. If Hatch End was not properly country, it was not properly urban either. True, London was not far away but it did not belong to us. I felt slightly envious of the Cockney accents which I heard in the Ealing Comedy Hue and Cry or, for real, from some of the inhabitants of 'the prefabs'. 3 Hatch End speech was a mixture of a softened form of Cockney with echoes of rural Hertfordshire, genuine or assumed posh and God knows what else. The nearest towns were Harrow or Watford, which had more or less what was needed (choice of cinemas, swimming pools and department stores) but little real urban excitement. Later, in the RAF, I listened to some of my fellow National Servicemen, talking in accents which were clearly anchored in solid places about Edinburgh, Glasgow, Manchester or Sheffield. Hatch End provided slight competition. When asked where I came from, I would usually say 'near London' or'somewhere between Harrow and Watford'. When I spoke of Hatch End it was with large ironic quotation marks. These wider questions of geographical identity did not concern me much in my childhood. Hatch End seemed to be good enough, able to provide me, and my friends, with a degree of freedom and opportunities for excitement and the development of the imagination. In the immediate post-war period, there were a few half-bombed houses to explore or the little network of underground shelters that were still accessible from the recreation ground. Friends or friends of friends could identify the ponds that were good for frogspawn or newts or the best sites for blackberries or bluebells. One area which was clearly identified by the boys of my generation was known as 'The Bumps'. These were the remains of some ancient earthworks (or so we were told) known as Grimsdyke. My school took its name from these earthworks. These were a set of small steep hillocks, then largely covered with brambles and other weeds or small trees. They were where a pre-war housing estate came to an end and were close to the main railway line. It was an area adaptable enough for improvised games, picnics or, on summer evenings, sex (or so I was told). One day, during the summer holidays I learned from a friend that some boys from Pinner Green had declared war on Hatch End and that the site of the battle was to be 'The Bumps'. I can remember walking towards the battlefield with John Webb. 'It is going to be very dangerous' he said as we picked up dustbin lids to act as improvised shields. There was a goodish crowd of boys already there when we arrived, most of whom I did not recognise. I am not sure how long it took for us to realise that the advertised battle was not going to take place and that the enemy hordes were not to come advancing across the fields. Some of the older boys started wrestling among themselves and the rest of us drifted away to other pursuits. The battle that did not take place can be seen in terms of construction of masculinities, the environment of post-war Britain or in various other ways. But, on reflection now, it was also a fleeting attempt to transfer a 'No Place' into 'Someplace'. Growing up involved the recognition that Hatch End was not much of a place. When I went to Harrow Weald Grammar school, I found that some pupils there saw Hatch End as 'rather posh'. However, people in Pinner tended, I was told, to see it as rather down-market. I realised that the other parts of Harrow, Wealdstone or Harrow Weald seemed to have more of an edge to them, with common points of reference (a football ground, swimming pools, cinemas, shops). And with each venture into worlds beyond the place where I grew up, this sense of being on the margins increased. --- Discussion We all come from some-place but some are more placed than others. To come from an Edinburgh, a Liverpool or the East End of London is to come from a place with a strong public identity and a tightly knit nexus of overlapping stories with which most people can identify and in whose telling most people can participate. We may be talking about places with some, real or imagined, historical depth: My own golden age was spent in a picturesque village on the banks of the River Dart. Most villagers then were from old Devon families of farmers and fishermen who ancestors' gravestones in the churchyard stood witness to their antiquity... 'It [another village recorded by Richard Hillyer] was a village unchanged since Saxon times, where work was governed by the 'immemorial rhythm' of the seasons and social life revolved mostly around church and chapel and cricket pitch. It was an essentially feudal world in which the lives of the poor were dictated by the various local gentry' (Bowker in Hillyer 2013: 5). 4 'Some-place' therefore has some historical depth and a set of overlapping stories. Or it may be a place with a strong public identity, even if, at times, the number of visitors is greater than the number of residents. Or, in some unhappy cases, that identity may be thrust upon it: think of Dunblane, Lockerbie or Hungerford. 'No-place' or no-particular place has none of this. We may think about coming from a particular place, possibly of belonging to a particular place in terms of the distributions of social and cultural capital. For those who come from the kinds of communities described in these quotations, we are talking about the social capital that bonds, to the past as well as to present, close-knit social networks. To those who come from, largely urban, areas with a high public profile and wide recognition we are talking more about a kind of cultural capital that has a value on the open market. Cockneys, Scousers and Dubliners are possibly assumed to be more knowing, more streetwise and wittier than the rest of us. Or consider Brighton with its historical associations with the Prince Regent and dirty weekends and Brighton Rock and its more recent links with Gay Pride and Green politics. To come from Brighton is to be able to draw upon a stock of cultural capital that is denied to residents of other, possibly equally attractive, seaside resorts. However, many people born in this country have neither the deep historical roots of rural or longestablished communities nor the cultural resources of some of our urban areas. (I am excluding here, those who are nomads either within or across our national borders). They come from 'no particular place'. These may be suburban areas on the edge of large cities or small towns that many people pass through on the way to somewhere else. While such places may have important historical roots, many of these may have been obscured as a result of the Second World War and post-war development or various strands of modernisation. From the garden on my second home, now on the borders of Hatch End and Pinner, you could see what looked like a French chateau built by a local eccentric with a liking for such things. It was called Tooke's folly or, simply, the Towers. It was pulled down before I completed my teens with just the clock tower being preserved. Almost certainly Hatch End had historical roots although there was, apart from the farms on the outskirts, little evidence of them by the time my parents moved there. No doubt there was plenty of material for local historians even here. I have shown how, as an outer suburb, it was composed of a mixture of practical and symbolic space. The trouble was that the symbolic space was not always that positive in the light of numerous critiques of suburbia with its associations of dullness and conformity. Much work, therefore, was left to the elaboration of imaginary space; the rather dull looking 'Railway Tavern' became a local, Anglicans went to the local church and thespians put on regular performances at the un-prepossessing St Anselm's Hall. For a young boy, unaware of sneers about subtopia, there was plenty of material for the construction of imaginary space. The rows of shops on either side of the main Uxbridge Road did not look like a'real' village but this did not matter. We had The 'Rec' and The 'Bumps' and, in the immediate post-war period there were half bombed-out houses and underground air-raid shelters to provide adventure and opportunities for dares and competitive masculinity. If the adventures of Swallows and Amazons or even the Famous Five seemed rather remote, there was at least one role model. William Brown, the hero of numerous books by Richmal Crompton, lived in a not dissimilar setting even if his family were much more securely located in the middle-class than was the case with me or my friends 5. If William and the Outlaws could do transformative work on his suburban setting, then why couldn't we do the same? William's construction of imaginary space was incorporated into our construction of imaginary space. Growing up entails the realisation that these carefully constructed imagined spaces are, after all, 'no particular place'. Today I, like many other people, pass through Hatch End at some speed on the stretch of railway line between Watford and Euston. You catch a glimpse of the depleted station, the rather shabby looking Commercial Traveller's School, a supermarket which was not there when I was growing up and the usual back-gardens and allotments. As Goffman (1959) described in his discussion of role-distance, imaginary involvement becomes detached irony. But still 'no particular place' can be, autobiographically speaking, a rather special place. --- Michael Erben was for many years Director of the Centre for Biography and Education at the University of Southampton (UK). He was a founder member of the BSA Auto/Biography Study Group and has published widely in the area of biographical studies and narrative. His most recent book, coauthored with Hilary Dickinson, is Nostalgia and Auto/Biography. He is now largely retired and is, contentedly, an independent scholar while holding an Honorary Fellowship at the University of Southampton and preparing (with Jenny Byrne) an extensive study of British postwar (1945-51) lives.

This paper is concerned with what is now the north-east London suburb of Hatch End and the feelings thoughts and memories it had for the author during his formative years. It deals with the quotidian and the reflections it can arouse. There is no attempt to produce a formal local history of Hatch End but rather there is employment of the author's autobiographical memory to record how temporal change influenced (and to a degree formed) him as an individual. The mix of perspectives to perform this task involves, in the main, considerations of physical and symbolic space from (a) solid and identifiable features of Hatch End (housing, public buildings, transport links, recreational spaces, etc) to (b) the meanings of countryside, family life, social class and modernity. For the author Hatch End does not carry with it the freight of being a place with deep historical roots or of having a high public profile. In general critiques of suburbia Hatch End could be classified as dull and uninteresting. In fact, the author ventures that Hatch End could be regarded as "no particular place". This is not however how the author personally regards Hatch End. For him there was in Hatch End plenty of exciting material for the construction of a self in a real place. As he says, "no particular place" can be, autobiographically speaking, a rather special place.David Morgan (1937Morgan ( -2020) ) was a co-founder of the Auto/Biography Study Group and its great, loyal, and cherished friend for some thirty years. The piece below has its origin in a paper given at the Auto/Biography Summer Conference of 2014 at Wolfson College, Oxford. The present version was among David's papers and his widow, Professor Dame Janet Finch, asked Michael Erben to make it ready for publication. The piece has required only light editing (slight changes coming from notes made by David and notes by Michael Erben of conversations with David over the two years 2014-16). Those many who are familiar with David's style and scholarship will be able to hear his voice as they read.

Introduction As a cooperation between Graz University of Technology and University of Graz, the Gruendungsgarage is inter-university organized, interdisciplinary oriented and provides a convergence of scientific theory and entrepreneurial practice in the format of a course. Every semester students and employees of all universities apply for one of the ten coveted places to further develop their innovative start-up ideas within one semester at the Gruendungsgarage and, at best, to subsequently found a start-up. The participating start-up teams are professionally accompanied and supervised by the scientific staff of both universities as well as by qualified mentors who are professional experts in their field. Within the structured course of one semester at the Gruendungsgarage, basic knowledge about starting a business is acquired, the participants work on their own business plans and get coaching on the further improvement of their business ideas. Workshops on business model development, design thinking, online marketing, intellectual property rights, legal and tax consulting complete the preparation for the everyday business life of company founders. After the first pitch of the start-up ideas, each team defines development goals together with their mentor. The improved start-up projects as well as the goals <unk> achievement are presented at the interim and final presentation in front of sponsors, investors, faculty members and personalities from the regional start-up scene. Figure 1 shows the course of the Gruendungsgarage within one volume, corresponding to one semester. Digitalization is progressing and is undoubtedly a megatrend with far-reaching implications. Investigating business models of the industry 4.0 Kaufmann [2] states, that products no longer consist only of hardware, especially the relevance of software is rising sharply and many products are being expanded by controls and communication modules in order to communicate. The technical possibilities and developments offer opportunities for business model innovations and new market participants arise [2]. Setting the focus on manufacturing, digitalization enables faster processes in product development, a more flexible production and increased efficiency regarding the use of resources. Therefore, costs can be significantly reduced while reliability, quality and process sustainability increase [3]. Digitalization also offers plenty of chances for start-ups and future entrepreneurs developing new business ideas. It is therefore of interest in this paper to examine the start-up projects of the Gruendungsgarage and to answer the questions as to whether the start-up projects are affected by digitalization and how the start-up projects can be classified accordingly. With regard to the aspects of digitalization, the start-up projects participating in the volumes I to X (ten semesters) of the Gruendungsgarage are examined in more detail in the present paper. Based on an initial, practice-oriented assessment, an attempt to classify the start-up projects is made. After analyzing the successful start-up projects of the Gruendungsgarage, implications for the entrepreneurship education, especially in the field of engineering education, are derived. --- Classification Concept and Allocation of the Sample Start-ups In the context of the digital transformation in society, economy and science, the start-ups (already founded or in the founding process) of the Gruendungsgarage are analyzed for the first time. A practice-oriented classification of the digitalization of start-up projects with a focus on the product, the process and the business model is carried out, adapting the concept of Matzler et al. [4]. After an investigation of several studies Rachinger et al [5] state, that digitalization affects and triggers changes in businesses and business models threefoldly -when refining established business models, transforming established business models and developing novel business models [5]. Matzler et al. [4] also describe the impact of digitalization on three levels: At the level of products and services with customer value, the more efficiency-driven processes leading to more profound changes and at the level of the business model, combining data and data-driven services, creating new customer value and a new business logic. Classifying a start-up <unk>s business model predominantly into analogue, hybrid (a combination of digital and analogue) or digital is a new approach and therefore, only a first attempt is made in the present paper referring to the following definition of a digital business model by Veit et al. [9]: "A business model is digital if changes in digital technologies trigger fundamental changes in the way business is carried out and revenues are generated" [9, p.48]. In the present paper, the data generation and assessment of the digitalization on product, process and business model level was conducted on the basis of the projects documentation of the Gruendungsgarage and compared with the information available at the start-ups <unk> websites as well as personal response from the founded start-ups and start-up projects. Based on the current state of information on the founding projects, these are classified as predominantly digital, hybrid or analogue at each of the three levels and evaluated in a tabular form. --- Classification of digitalization by examples To classify the product into digital, hybrid and analogue at product and process level, the product was defined as a physical result or service for the customer and the process of service provision in terms of production and sales. An assessment of the business model was made with reference to the dimensions "Who, What, How and Why" of the business model by Gassmann [6] and the definition of Teece [7], according to which a business model describes the design and mechanisms for the creation, delivery and capture of value. Stated by Vendrell-Herrero et al. [8] digital technologies are drivers for change in the competitive market as companies develop new ways of generating, delivering and capturing value. Using nine examples of founded projects, the classification into the levels of digitalization is demonstrated in Table 1. The clustering into predominantly digital, hybrid or analogue answers the questions in which form value is offered (product), how value is provided and distributed (process) and how revenue is generated (business model). In Table 2 three examples of successfully founded and established start-ups of the Gruendungsgarage Graz are investigated in more detail. Studo, the app from the startup project Moshbit is classified as digital on all levels of focus: digital product, digital processes and digital business model. Drone Rescue Systems is evaluated hybrid, as the product, the processes for delivering value, and the business model consist of both, analogue and digital components. Mehlspeisenfräulein with individually handmade confectionery and pastries is considered an analogue example, whereby the use of digital media for communication, social media and operating a website are left aside in this classification. Taking a closer look at the start-up projects, in Table 2 the business model patterns according to Gassmann, Frankenberger and Csik [6] are used to sketch the business models. Not every single business model pattern that is identified has compulsory digital aspects. The patterns are adjusted and combined by the start-up projects and attributed to the classification in this first analysis based on the information available. --- Analysis of the digitalization over ten semesters After ten semesters of the Gruendungsgarage a positive summary can be given: Over the past five years, 91 teams have participated in the academic start-up accelerator program Gruendungsgarage, of which 25 start-up project ideas have resulted in founded start-ups with more than 130 jobs created. A further 16 projects are currently in the start-up founding phase and are about to enter the market. The analysis and classification of the total of 41 projects of the Gruendungsgarage (25 founded start-ups plus 16 promising start-up projects) in terms of digitalization was carried out as these have already gained a foothold in the market or have the best prospects of establishing themselves. The overview of the number of analogue, hybrid and digital products, processes and business models shown in Figure 2 provides information about the degree of digitalization of the start-ups and start-up-ready projects of the Gruendungsgarage. --- Fig. 2. Levels of digitalization of start-ups and start-up projects of the Gruendungsgarage The 41 analyzed projects of the Gruendungsgarage clearly show a focus towards digitalization. While at product level 19 of the start-up projects are classified as analogue, 7 as hybrid and 15 as digital, this ratio changes dramatically at process level. About the half of the teams (49 %, 19 teams) already founded or currently in the founding process rely on analogue products or services, but integrate the aspect of digitalization at the process level. Only 12 % implement a largely analogue business model. All of the start-ups use digital technologies for communication and web presence, but this fact is not considered as a digitalizing element of the business model. A total of 88 % of the projects of the Gruendungsgarage (37 % digital and 51 % hybrid) are characterized by digitalization in their business model. In the case of the start-up projects with digital products (37 %), the dominant focus on digitalization is asserting itself to the same extent at the process level (37 %) and is also reflected in the business model (37 %). This phenomenon corresponds to the result considering the start-ups that have already been founded only, where 36 % have digital products, digital processes and digital business models. Figure 3 shows the classification of the start-up projects <unk> business models over the course of all ten volumes (semesters) of the Gruendungsgarage. Six out of the ten volumes are characterized by start-up projects with digital and hybrid business models only. Although no clear trend can be derived from the graph, it is shown that aspects of digitization have played a role across all volumes. --- Conclusion and Implications The focus towards digitalization is revealed by analyzing the start-ups and current projects of the Gruendungsgarage. The analysis shows that analogue products are increasingly extended by services in order to achieve unique selling points. These services are usually provided in digital form, for example by storing and evaluating data or networking devices. Therefore, nowadays an early consideration of the topic of digitalization seems indispensable in the founding process of a business. Digital enterprises are empowered by new information and communication technology systems that integrate the following three technological keys according to Loonam et al. [10]: virtualization (for example cloud technology), mobility (such as social media, internet of things or mobile devices), and embedded analytics (like big data). These technologies coupled with embedded enterprise solutions enable the digital enterprise [10] and can also been seen as a point of origin to be promoted for developing new start-up ideas in the academic field. Müller et al. [11] show that the lead of American companies regarding the degree of digitalization compared to German companies is also reflected in the implications of their digitalization activities. For example, performance goals can be achieved, sales can be increased and costs reduced. For start-ups, it is interesting to note that digitalization enables to open up new markets and enhance the company <unk>s image [11], a fact that could also be taken into account when promoting the Gruendungsgarage. Another tendency is that many business ideas require programming knowledge or digital skills, which the idea providers are often lacking themselves. Therefore, in addition to basic user knowledge, other topics of digital education (raising awareness on the relevance of digitalization as well as basic knowledge in programming) should be integrated into curricula and cross-study training formats. The masterplan of the Austrian Federal Ministry of Education, Science and Research for the digitalization of the Austrian educational system aims to provide orientation for three fields of action: software, hardware and teaching [12]. Starting with the school year of 2018/2019, curricula of all secondary level I schools in Austria cover basic education in digital competencies based on successful projects on "Digital Basic Education" [13]. Nevertheless, a continuation of the digital education on an academic level seems reasonable. Lindner [14] provides starting points for the entrepreneurship education of seven different target groups. The start-up teams participating in the Gruendungsgarage can be assigned to the target group four of entrepreneurship education programs as the attempt to implement, the behavioral intention and the business idea are present. Hence, the Gruendungsgarage should continue to focus on the further specification of the business idea and competence development for founders. Sedelmaier and Landes [15] investigated the skills needed by requirements engineers at present and propose a set of competencies as a basis for the development of learning situations, which could also be of interest in the entrepreneurship education at the Gruendungsgarage. With entrepreneurship education as a cornerstone in engineering education, Holzmann et al. [16] provide insights from the Entrepreneurial Campus Villach, which could be considered as a case to elaborate on similarities and differences comparing it to the programs already offered in Graz. In engineering education on entrepreneurship, especially in the Gruendungsgarage, it is worth considering how to better support teams with digital business ideas or startup ideas that require digital components. For example, the range of workshops can be extended and the pool of mentors broadened to include more people with expertise in digitalization. A further initiative could be launched to match people willing to set up a business with programmers. In addition to making entrepreneurship education even more attractive, the inter-university and interdisciplinary cooperation contributes significantly to the success of the Gruendungsgarage and will therefore be continued.

Founded in 2013 as a playground for implementation-oriented start-up ideas, the Gruendungsgarage is now an established academic start-up accelerator and an integral part of the regional start-up scene. Starting with a brief introduction to the program of the Gruendungsgarage as a best practice example with five years of experience in academic entrepreneurial education, a model with a practical-oriented focus for classifying the degree of digitalization in start-up projects is presented. Successful start-ups and promising start-up projects from the Gruendungsgarage are classified according to their degree of digitalization and illustrated by examples. Finally, implications for the practice in engineering education on entrepreneurship are derived and suggestions for future research are identified.

Human mental life is profoundly situated. The human mind, for better or worse, has always been inextricably intertwined with the plasticity (or stability) of our changing sociomaterial environments. This process of metaplasticity applies to human ontogeny as it does to human evolution. [1][2][3] Different academic disciplines, constrained by the theoretical assumptions, the time scales and the analytical units that they use, are able to observe, and would probably emphasise, different aspects of the relationship between mind and matter. Still, the underlying message is the same: our everyday material habits and forms of material engagement (past, present and future) are inextricably linked with our cognitive and emotional livesthe major dynamic being the plasticity of the brain linked to the affordances of our body and the action possibilities offered by the things we make and use. We cannot understand human intelligence (cognition and affect) in isolation from its changing material and social surroundings. Material things matter and need to be taken seriously. 4,5 This realisation of the interrelatedness between human experience, affect and materiality has, potentially, major implications for how we approach and understand important issues of mental health. Yet, those implications are not well understood. The important question concerning the influence of materiality on mental health and human well-being has not been adequately addressed and requires work across disciplinary boundaries. Although the agency of things [6][7][8] and the vitality of matter 9 have been central themes of research in the humanities and the social sciences, research in neuropsychiatry has, by and large, side-tracked the thorny issue of material engagement. Given what we now know about the importance of materiality in human social and cognitive life, [10][11][12] this form of epistemic neglect is unacceptable and needs to be overcome. I do not wish, and there is no space in this brief editorial, to discuss the reasons for that blind spot. Instead, I want to suggest a way forward. Suffice it to say that the main reason psychiatry avoided investigating (at least in a direct and systematic sense) the full variation of the interactions between cognition and material culture, and their effects on mental health, is not because it sees those interactions as unimportant or uninteresting. Rather, it is because some traditional assumptions about the mind (for instance, the view that anything mental must refer to, and is explained by, processes internal to the individual) make it hard to study them. 13,14 Those assumptions are changing, replaced by the more productive idea that the boundaries of the mind'should not be set a priori, but should be responsive to the nature of the phenomena under study'. 15 The boundaries of the human mind have always been a problem, but perhaps never so problematic as they are today. [16][17][18][19][20] There is unrealised potential here for creating new approaches to care and intervention that can complement existing practices and broaden, or in some cases challenge, prevailing assumptions in mental health. However, in order for this potential to be realised, a more nuanced, casespecific and evidence-based understanding of the material dimensions of human mental health is needed. 21 How can we measure the effects that different material ecologies, 12 practices and technologies have on human mental life? What constitutes evidence in that context? How can we best interpret existing evidence and use it to effect changes that could help improve our emotional and mental wellbeing? Largely, we still lack effective ways of answering those questions. Combined expertise and a range of specialisations are needed to develop both the theoretical and practical means that will allow us to make progress with these challenging issues. --- The material engagement approach The aim of this essay is to propose one possible path to this end, a path that could lead to better integration of in-depth anthropological and philosophical qualitative analysis with the quantitative methods and measures of psychiatry and neurophysiology. Specifically, I propose a material engagement approach combining methods from anthropology and embodied cognitive science that allows us to observe, to record and to compare the actual processes by which different materials and forms of material culture affect the people who experience and use them. [22][23][24] The material engagement approach provides a theoretical framework tailored to facilitate a heightened responsiveness to the details of situated action and the affordance of the materials involved (the concept of affordance stems from Gibsonian ecological psychology, where it denotes the relational opportunities for perception and action between an organism and its environment). 25,26 The distinctive feature of the material engagement approach is the commitment to a view of thinking as a process that is distributed, enacted and situated, as well as assembled, from a variety of non-localisable mental resources spanning the boundaries of the individual brain and body. Material engagement theory as an explanatory path is based on three interrelated working hypotheses, which can be summarised as follows. [22][23][24] (a) The hypothesis of extended mind, which explores the constitutive intertwining of cognition with material culture; (b) the hypothesis of enactive signification, which explores the nature of the material sign not as a representational mechanism but as a semiotic conflation and cohabitation through matter that enacts and brings forth the world; and, finally, (c) the hypothesis of material agency, which explores agency not as a human property but as the emergent product of situated activity (Fig. 1). Taken together, these hypotheses enable us to restate the problem of cognitive boundaries, offering a crossdisciplinary relational basis for understanding the material ecology of mental health that integrates the'materiality turn' in humanities 27 and social sciences with new embodied and enactive trends in philosophy and cognitive science. 28 Traditional anthropological participant-observation combined with visual methods (e.g. photography/video capture and elicitation) and object interviews can be used for the study of everyday naturally occurring situated material activities and somatic modes of attention. 29,30 The creative methods of art and medical humanities are also needed to investigate how different forms of material engagement achieve their effects. [31][32][33] Some of the key issues include: (a) the role of arts and creative industries in psychotherapy; 33,34 (b) the development of evidence-based understandings of the therapeutic effects and qualities of different art materials and digital media; (c) the role of museums in mental health [35][36][37][38] as a space for creative interventions; and (d) the cross-cultural, 39 social and gender 40 dimensions of mental health. Such a cross-disciplinary anthropological understanding will allow us to harness the power of creative industries more effectively and to develop new tools that will not just meet current needs but also help to generate new ideas on how to design a landscape of creative affordances for future experimentation. There is currently a lot of emphasis on the ways technological innovation (e.g. digital media, internet of things, big data, self-tracking) can be fostered to benefit older people. [41][42][43] Still, there has been very little work trying to understand how these new innovations fit or compare with more traditional material ecologies and creative habits. 44 Doing so could prove beneficial for all stakeholders involved. To illustrate those claims, I will focus on the example of dementia. --- The case of dementia Dementia is one of the biggest mental health challenges modern society faces. 45 The profound changes in people's lives brought about by dementia and by long-term caregiving are well known. This process of self-transformation is often described as a diminishment of or loss of self. 46,47 All people who live with dementia (patients and family/carers) are affected. It is common to see dementia solely as the product of brain pathology and to leave people's bodily interactions with the material world unattended. But perhaps it is also important to pay closer attention and try to understand better how dementia extends beyond the brain of the individual. 20 The gradual loss of self is not solely the product of neurological factors. An impoverished material environment and the gradual loss of connection with significant others (humans and things) that provide a familiar, stable and comforting self-referential space also contribute to a diminishing capacity for memory, interaction and communication. One important question in this regard concerns how the evocative nature of material culture and its various cognitive, emotional and affective dimensions can be harnessed to guide the design of new care strategies and material settings that promote patients' agency, autonomy and ability in social interaction and communication. Are there ways to alter the person's material environment that can result in positive changes in embodied self-conceptualisation? 48,49 One way to answer that question could be by examining the changes in materiality experienced during periods of major transition, for instance, during the transition from home to 'care home'. This is a challenging period, often marked by uncertainty, anxiety, confusion and distress, which can be explained in terms of changing cognitive ecology and the stripping away of personal possessions that help to constitute one's personal memories and self-identity. 50 The philosopher Daniel Dennett, in his book Kinds of Minds (1996), describes this cognitive impairment when elderly people are in hospital, bereft of the usual contextual cues and affordances of their familiar home environments. 51 More recently, Shaun Gallagher, 26 adopting an enactive perspective, proposes an affordance-based approach to therapy that places special focus on the changing physical, social and cultural environments. He argues that a variety of mental health problems and neurological and psychiatric disorders can be understood in terms of changes in a subject's 'affordance space'which is the lived space defined relative to the capacities (sensorimotor, social, cultural, cognitive) of a specific agent. Significant objects and personal possessions that have been acquired and accumulated over the life course constitute material memories and material anchors of selfidentity. The powerful affective linkage of material possessions to one's identity explains the positive association between memorabilia and mood, as well as their observed influence on how elderly individuals achieve a sense of continuity. To lose these objects, often abruptly and unwillingly, owing to changes in residential circumstances, such as the transition from independent living to a nursing home, can be a'major cause of loss and grieving for elders'. 52 Understanding the cognitive life of things and the importance of object memories can allow us to develop strategies of care for the things that can actively enhance or even create a sense of continuity, stability, and comfort, especially during those transitional stages when one moves from home to a long-term care facility. 53 Interestingly, in many care homes and hospices nowadays, residents are allowed to bring their own furniture and personal objects. One could track what they choose to take with them and how they actually use it. 54 Are there any preferred types of objects? Do the objects carried along retain their original meaning and usage or do they become something different? Do individuals always keep what they want to remember? Do they always remember what they keep? What is the evidence that personal objects can facilitate patients' transition to their new and often alienating surroundings (e.g. providing cognitive and emotional support)? Can we provide informed guidance on how objects can be used more efficiently and appropriately in personal memory practices? There have been a number of studies looking at how old people moving into long-term care engage with their new material surroundings, 55 but few of them have focused specifically on how dementia and other cognitive impairments may affect residents' ability to think and to feel with, through and about things. For instance, Buse and Twigg explored the potential of material objects to evoke narratives and elicit memories through the sensory and material dimensions of dress. The study focused on three main dimensions, i.e., kept clothes, discarded clothes and absent clothes, and used ethnographic methods to explore everyday experiences of clothing and their influence on maintenance of personal identity and biographical continuity. The study nicely illustrates the significance of dress, often unnoticed, in supporting personal identity in dementia care. 56,57 Combining traditional anthropological methods of participant observation with in-depth phenomenological analysis could help to answer questions of how different people experience this radical shift in their material ecology. Ethnographic attention to patients' changing bodily engagements, habits and material attachments during their transition to institutional care could provide a thorough first-hand examination of that. Cognitive 58 and visual ethnography 59 offer particularly apt methods to record and analyse everyday episodes of situated action and distributed intelligence, paying attention to the details of specific contexts. To investigate those processes, an analytical vocabulary cutting across brain-body-world divisions is required. Focusing primarily on non-verbal aspects of the dyadic or triadic interactions between patients, carers and things, one could analyse the sequential organisation of those interactions using a combination of pictures, written commentaries and graphic elements to illustrate features of material engagement such as movement, direction, eye-gaze, gesture, attention and action coordination. It is important to recognise that every person's experience of dementia is different. People with dementia have their own unique life histories and actively make meaning of their lives through their attachments to and interactions with specific places, objects and people. People should be encouraged to spend time with the objects of their attention, to observe closely and with all their senses, and to draw what they observed or find other ways to voice their opinion. 60,61 One could look specifically at self-narratives evoked and constructed through shared discussions of artefacts and objects. Also of interest is to explore, more specifically, the role of biographical objects as memory scaffolds and affective probes of autobiographical memory through their attachments to family histories and self-identity, offering perhaps a complementary pathway to non-verbal facilitation of communication, promotion of independence, and rediscovery of personal identity. Communication is one of the major problems in dementia care, and participatory ethnography offers an appropriate means of interaction. To study people with dementia, it is essential to give them a voice. Material engagement interventions facilitate and support complex interactions, offering alternative non-verbal avenues of expression and communication. People with dementia may understand, do and express with and through things far more than they can express with language. What would be important to clarify in this connection is the extent to which memory loss in dementia and other neurodegenerative diseases might be partially compensated by the use of selected biographically salient objects and material scaffolds for self-narrative and autobiographical memory. For instance, studies on bodily and object ownership have shown that owned objects can trigger a strong memory trace and are often treated as psychological extensions of self. 62 One could explore the effects that such strong associations between self and owned objects may have in the context of dementia. Another promising avenue of research could be to investigate the cognitive mechanisms activated when people with dementia think with and through things in an embodied, performative manner, and to compare them with the cognitive mechanisms activated when they think about things in the internalist representational manner. Understanding those processes is important, because whereas most forms of dementia seem to affect the latter capacities for thinking about the world (in the sense of abstract recollection, selfnarrative, communication and language use), the former enactive capacities remain largely intact and could be re-used and developed to compensate for the lost ones. For instance, memory interventions in the form of memory books, photo albums and so-called memory boxes have repeatedly demonstrated their effectiveness as external memory aids and powerful communicative means. 63 The precise cognitive function of those external memory prompts and cues remains largely unknown, but it has been suggested that they work precisely because they do not require conscious cognitive effort in order to trigger retrieval of related semantic information from memory storage, effecting positive changes in the conversational behaviours of people with a wide range of severity of cognitive impairment. 64 There is little doubt that things and material objects evoke and trigger memories through associations with important life experiences and events, although exactly how that happens and how best those forms of material memory should be described is less understood. Beyond the widely recognised memories of objects, we should be considering also the possibility of object memories. Objects embody memories in their forms, life histories and uses that can provide powerful media for recollection, selfexpression and self-identification. The mundane things we own constitute memories of our past, material traces available to be re-enacted through our bodily engagements with the material world. In that sense, things provide a durable network of material signification, transposable dispositions and bodily habits which can be harnessed to enhance cognitive abilities or compensate for memory loss when biological memory is damaged. Like a blind's person stick, mundane things and media can help people to re-orient and re-structure their cognitive landscape (neural and extraneural) in a variety of ways. 65 A material engagement approach could help us to explore whether there are ways we could structure and adapt the material ecology of any given environment in order to create scaffolds, material anchors 66 and action affordances 26,67 able to effect a more efficient utilisation of brain networks or stimulate the person's ability to recruit alternate brain networks that could enhance cognitive reserve. 68 A broader but related question in this regard could be whether different forms of material engagement can dynamically influence and mediate the relationship between brain damage and its clinical outcome. Here, one could refer to similar findings on collective remembering and cognitive reserve in old couples. 69 --- Conclusion In this essay, taking the example of dementia, I sketched the outline of a material engagement approach to mental health. I proposed that a collaboration of archaeology, anthropology, philosophy and neuropsychiatry could help us understand how people with dementia are actually engaging with their local material environments and the effects that the latter has on people's agency, cognitive abilities and selfhood across the full dementia care pathway. Such a collaboration could also form the basis of a longitudinal mixed method exploratory investigation of the possibilities for action and interaction that different objects and forms of material culture provide for people with dementia in common everyday situations and in different care settings (their homes and residential care). Ultimately, the objective is to broaden recognition of the material aspects of dementia. That is, understanding better the ways by which different forms of material engagement can help people with dementia to: (a) remember and/or forget; (b) enact, express their agency and preserve their selfidentity; and (c) communicate and interact with others, especially family members and care providers. Hopefully, this will stimulate further cross-disciplinary research and broader engagement with material culture studies, which, in the context of current dementia and mental health research, are often seen as marginal and of limited value (despite the success of well-established tradition in artbased and occupational therapies). Achieving these aims requires concerted action from a wide variety of organisations, collaboration and cooperation from the public. This exchange between practitioners, researchers and the public would generate further awareness, promoting new evidence-based cross-disciplinary research on the role of creative practices and material culture in mental health. It could also create transferable insights into the influence of creative practices in mental rehabilitation and in psychotherapy that could be used as alternative/additional solutions to traditional treatments of mental illness across the life course. The benefits of the proposed material engagement approach could be broad, offering important insights and practical recommendations for the well-being and long-term care of the aging population, with dementia or not. Understanding how different forms of material engagement affect people's cognitive abilities and selfhood across the full dementia care pathway would also lead to specific evidence-based recommendations for designing interventions, material scaffolds and assistive techniques that could help people maintain and improve their mental well-being. This could also inform and facilitate safe and responsible design of healing environments, offer suggestions on how their effects should be measured, and suggest policy changes to provide a mental healthcare environment fit for the future. Fostering deeper understanding of the ecology of mental health and developing simple interventions enabling people to improve their psychological well-being is increasingly important in today's environment, with mental health leaders seeking innovative and cost-effective ways to optimise the quality of care. Concerted action and cross-disciplinary research is required to ensure that all gaps in research are filled. Understanding how different forms of material engagement affect people's mental health presents us with such a challenge. --- About the author Lambros Malafouris is a senior research fellow at Keble College and the Institute of Archaeology, University of Oxford, UK.

Consensus is growing, in many areas of the humanities and social sciences, that aspects of the material world we live in have causal efficacy on our mindsthe major dynamic being the plasticity of the brain linked to the affordances of our bodily engagements with things. The implications of that on how we approach and understand important mental health issues have not been adequately addressed. This paper proposes a material engagement approach to the study of the processes by which different forms of materiality achieve their effects. Focusing on the example of dementia, I propose that a collaboration between archaeology, anthropology, philosophy and psychiatry could help us to fill this gap in our knowledge, allowing us to understand the exact effects of everyday objects, personal possessions and forms of material engagement on people with dementia.

Introduction As a concept, climate change relates to long-term alterations of weather patterns and environmental conditions due to increasing global surface temperatures. The UNFCCC (2022) perceives climate change as a significant fluctuation of global surface temperatures associated with anthropogenic influences at a specified time. Additionally, the IPCC (2021) links climate change directly with changes in climatic properties recorded over decades due to anthropogenic and natural variability. The latest findings from scientific studies strongly link the consequences of unsustainable human activities with increasing global warming and exacerbated negative impacts (Ramanujan, 2021;IPCC, 2021) that threaten the survival of the poor and marginalised groups in rural livelihoods. Climate change began as a scientific and environmental concern, and later became a global debate and a key defining factor in equity and socio-economic development policy agenda (Burleson, 2015;IPCC, 2021). Moreover, perceived knowledge and definitions of climate change lack the incorporation of local knowledge and experiences. Hence, most of it has been irrelevant and misinformed about the climate change agenda (MacKay et al., 2020). However, the current global climate change agenda enshrined in the Paris Agreement (2015) and goal number 13 of the Agenda 2030 (SDGs, 2015) has influenced the promotion of local climate actions and stakeholders' mapping to enhance resilience (IPCC, 2021). In Africa, Agenda 2063 of the AU (2015) further sets forth the climate change agenda. In Tanzania, the climate change agenda is designated by relevant policy initiatives ascribed to environmental and various sector policies (NCCRS, 2021). According to IPCC (2021), climate change mitigation strategies refer to interventions to limit and remove greenhouse gases from the atmosphere. Unlike mitigation, climate change adaptation strategies are preferable in developing countries as they combine with socioeconomic activities that enhance livelihood resilience. The sustainable development agenda dates back to the Brundtland Report (1987)-Our Common Future-which clearly describes sustainable development as balancing the significant needs of the present and future generations (Narksompong et al., 2015). The adoption of the sustainable development agenda received global acceptance in 1992 through the United Nations Agenda 21, and recently through the Agenda 2030, which comprise of 17 goals to mobilise global partnerships to end poverty and ensure environmental protection, peace and prosperity (UN, 2020). Africa's Agenda 2063 further extends the 17 global development goals to guarantee locally relevant socio-economic and environmental dimensions of sustainable development (Nkrumah, 2021). The UN (2020) defines youth as persons between 15 and 24 years old. However, existing literature reveals varying concepts and definitions of youth (URT, 2007;Narksompong et al., 2015;MacKay et al., 2020), implying little consensus on best practices and overall youth engagement landscapes thereof (O'Brien et al., 2018;Ho et al., 2015). Narratively, youth is portrayed as a unique stage in human development subjected to diverse challenges and cascading influences (O'Brien et al., 2018). In Africa, specifically in Tanzania, youth range between 15 and 35 years old (AU, 2019;URT, 2007). Despite mismatched perceptions of age, existing socio-economic strategies acknowledge youth as an energetic, well-informed, knowledgeable, skilled and innovative workforce (Narksompong et al., 2015). Engagement strategies are described as response measures taken to enhance resilience and socio-economic development (Narksompong et al., 2025). Owing to heavy reliance on diminishing livelihood assets, and increased vulnerability due to extreme weather conditions, the climate change agenda becomes relevant for rural livelihoods and the proper functioning of socio-economic systems and development (O'Brien et al., 2018;IPCC, 2021). Therefore, this study describes engagement strategies in relation to relevant livelihood activities that enhance coping mechanisms, adaptation and mitigation responses (Nkrumah, 2021). Furthermore, youth engagement entails any inclusion or partnerships with young people in socio-economic interventions as critical stakeholders, vulnerable victims and change agents (Narksompong et al., 2015). Despite mismatched perceptions of the age range, most socio-economic strategies acknowledge contributions from the youth (MacKay et al., 2020). An effective youth engagement strategy is expected to prepare the youth to be inspired, wellinformed, empowered, responsible citizens and active drivers of changes relevant to enhancing the climate change agenda (Makondo et al., 2018;Samaddar et al., 2021). However, most strategies exert mismatched impacts owing to the adoption of irrelevant socio-economic contexts (O'Brien et al., 2018;MacKay et al., 2020). As a people in transition, the youth presents an unlimited influence on the climate change agenda and sustainable development (AU, 2019;Narksompong et al., 2015) owing to their increasing awareness and access to digital information (Ojala et al., 2017) of the undesirable impacts of climate change (IPCC, 2019). However, O'Brien et al. (2018) and Bandura et al. (2019) argue that mismatched and misplaced strategies deliberately promote youth exclusion, prolonged exposure, and vulnerability amid ongoing climate change scenarios. As such, sustained youth exclusion is thought to perpetuate a lack of confidence, misinformation, mistrust, inconsistency, inaction, dissent, resentment and anxiety amongst them (Nkrumah, 2020). These same factors also influence vulnerability and misinformation on engagement strategies in climate change and sustainable development among the youth (O'Brien et al., 2021;Ho, 2015). Furthermore, the irrelevance of most top-down youth interventions has derided bottom-up approaches and indigenous knowledge easily accessible to most youth (O'Brien et al., 2018). Rao et al. (2016) and Mackay et al. (2020) have noted youth's growing anxiety, dissent, mistrust and passiveness on climate change and sustainable development agendas. This paper answers critical questions on the awareness and effectiveness of strategies for rural youth engagement in these agendas using the Maasai-dominated agro-pastoral community within Naitolia village, in Monduli, Tanzania. --- Conceptualization of the Research Gap Youth engagement in the climate change agenda and sustainable development in rural areas has attracted low research interest (Solar et al., 2023). As such, the role of rural youth in strategic interventions that promote low-carbon development pathways and climate resilience is insignificant. Previous studies-such as Narksompong et al. (2015), Nkrumah (2021), O'Brien et al. (2015) and Zimba et al. (2021)-portrayed youth as passive stakeholders and mere project beneficiaries; instead of being active leaders, promoters, active agents and powerful actors to influence low-carbon development pathways and a climate resilient future. According to Simons (2022), vulnerable and marginalised rural youth have lacked effective strategies to influence engagement in climate change agenda and sustainable development in the face of evident adverse impacts of climate change. The role of local youth in climate change over time has also received insignificant investigation (Barford et al., 2021;Simons, 2022). Additionally, the role of policy experts, decision makers, practitioners and the general public in promoting the inclusion of local youth in the climate change agenda and sustainable development has also lacked critical investigations. According to Ho et al. (2015 andZimba et al. (2021), existing youth engagement strategies rely on diverse stakeholders' interventions. However, Kosciulek (2022) and Samaddar et al. (2021) observed that top-down strategies have relied on governments (public); while bottom-up strategies have relied on individuals, families and the community. Nkrumah (2021) andO'Brien et al. (2020) also observed that engagement strategies by the private sector have demonstrated both top-down and bottom-up characteristics. Most youth engagement options include philanthropy (charity), volunteering, politics, community service, economic activities, fine art, education systems, skills development pathways, employment and innovation (Nkrumah, 2021). Nevertheless, O'Brien et al. (2015) and Zimba et al. (2021) highlighted that youth engagement has experienced different socio-economic challenges influenced by natural risks. As such, the youth have faced systematic marginalisation, under-representation and exclusion; with limited capacities to influence the formulation and implementation of strategies. Also, Nkrumah (2021) has portrayed local youth as passive recipients and mere beneficiaries of policies and interventions (tokenism), rather than key strategic stakeholders (Kosciulek, 2020). On their part, Narksompong et al. (2015) andO'Brien et al. (2018) have observed the inevitable quest for a paradigm shift to facilitate and embrace effective youth-based strategies. Additionally, Ho et al. (2015) noted that emergent engagement strategies still face several crosscutting and context-specific challenges that slow down intentions to enhance livelihood resilience and sustainable development. Kosciulek (2020) depicted that diverse variables such as biophysical attributes, exposure to disaster risks and climatic changes, historical backgrounds, socioeconomic divides, democracy and local politics, education, technology and skills development pathways, as well as attitudes: all these often determine the role of youth in the climate change agenda and sustainable development. Supporting this, Barreda (2018), Nkrumah (2021) andO'Brien et al. (2018) also highlighted the role of global climate strategies in denouncing climate misinformation and promoting efficient socio-economic development. Furthermore, Kosciulek (2020) andO'Brien et al. (2018) also postulated that the fate of young people rely on engagement strategies that directly link and influence development goals and climate change actions. Even though Rupia (2020), Kitasho (2020) and IPCC (2021) highlighted evidence of youth engagement in sustainable development, it remained unclear how local youth engaged with climate change; and whether the engagement in sustainable development was for all the 17 goals (Barreda, 2018). To bridge the identified gap, the main objective of this paper is to assess the effectiveness of youth engagement strategies in the climate change and sustainable development agenda. The guiding questions were: Are youth aware of strategic engagement in climate change and sustainable development agendas? Are existing engagement strategies effective in enhancing the role of youth in climate change and sustainable development agendas? Bridging the identified gap could be critical to eliciting anxiety and effectively implementing future youth engagement strategies in climate change and sustainable development agendas (Idowu et al., 2019;Mabhuye et al., 2021;Mackay et al., 2020). --- Contexts of the Study Area --- Description of the Study Area This study was conducted at Naitolia village in Mswakini ward, about 75km from the Monduli district headquarters in the Arusha region, Tanzania. The village lies within the prominent Randilen Wildlife Management Area (RWMA) at latitudes -3.60246 and longitudes 36.09109 (Pearson et al., 2017). Administratively, Naitolia village consists of Ormang'wai and Engusero subvillages, covering about 178km 2 along the Arusha-Babati-Karatu highway (Follows, 2018;TPP, 2020). --- Demographic and Socio-economic Characteristics The human population at the Naitolia village was about 1,800 (TPP, 2020) at the time of the study in June 2021. The Maasai tribe dominated the study area, with patriarchy, pastoralism and part-time crop farming as the mainstay socio-economic activities that exclusively relied on climate-sensitive natural resources (Pearson et al., 2017). Being based on the traditional pastoralism economy, meat and milk constituted the leading staple food. However, due to increased climatic changes and immigration, most native households have slowly turned to agropastoralism. Rain-fed crop farming for maize and leguminous varieties such as chickpeas (dengu), beans, black beans (ngwala) and soybeans (TPP, 2020) is undertaken on a part-time basis. Also, the residents engage in subsistence beekeeping projects, the sale of traditional handicrafts, and motorcycle transportation (bodaboda), among other mixed livelihood activities (TPP, 2020;Follows;2018). On the other hand, age-based systems and initiation ritesamong other firmly held traditional norms, culture and knowledge-can determine youth's stake in socio-economic activities (Kitasho, 2020). --- Biophysical Resources The study area exhibits a semi-arid Maasai steppe landscape with a bimodal rainfall pattern (TPP, 2020;Follows, 2018). The typical annual rainfall of about 400mm-800mm, and the annual temperature of 23oC-30oC, are suitable for drought-tolerant plant species and migrating wildlife (Pearson et al., 2017). Generally, the study area experiences unpredictable rainfall, prolonged droughts, water and food insecurities, and human-wildlife conflicts intensified by ongoing climate change crises (TPP, 2020). --- Justification of the Study Area Compared to other villages in the district, Naitolia village was prioritised for the study because it falls under the Tanzania Partnership Program (TPP); alongside tourism value-chains and diverse stakeholder interventions aimed at enhancing climate resilience and sustainable development (TPP, 2020;Pearson et al., 2017). Kitasho (2020) and NTRI (2019) opine that diverse stakeholder interventions, mixed livelihood activities, exposure to frequent human-wildlife conflicts, and perceived vulnerability to climate change: all tend to shape awareness, experiences, local knowledge, ambitions and strategies to embrace climate change and sustainable development agendas. As such, the research projected the suitability of local respondents within the prescribed age-set system for the study. --- Methodology --- Research Design This study employed a mixed research design to assess the effectiveness of youth engagement strategies in climate change and sustainable development agendas at Naitolia village in Monduli district (Bryman et al., 2021;Dawadi et al., 2021;Creswell et al., 2020). The mixed design allowed an extensive understanding of engagement trends and promoted the voices of marginalised local youth (Barreda, 2018;Ahmadi et al., 2022;Creswell et al., 2020). Quantitative respondent questionnaires were used to collect data on the influence of youth engagement strategies (independent variables) on climate change. and sustainable development agendas (dependent variables). Concurrently, trends of youth engagement strategies in climate change and sustainable development agendas were also explored using qualitative data collection tools administered to respondents at the study site. Furthermore, triangulation was used to merge and cross-check the validity of analyses and presentation of findings from quantitative and qualitative approaches (Bryman et al., 2018;Dawadi et al., 2021). This mixed approach resulted in a holistic comprehension of cross-cutting youth-related perspectives, while offsetting weaknesses and gaining strength associated with individual research approaches (Creswell et al., 2020). Pragmatically, the literature review revealed less application of a mixed research design as a standalone option in similar contexts; hence, it was the preferred design (Kitasho, 2020;Barreda, 2018). --- Sampling Methods and Sample Size The current study employed purposive non-probability techniques to sample out Naitolia village from other villages in Tanzania, Arusha region, Monduli district and Mswakwini ward. Dominated by the Maasai ethnicity, the study area exhibited patriarchy, age-set systems, agro-pastoral livelihoods, climate vulnerability, and diverse stakeholder interventions. Only youth aged between 15 and 35 (URT, 2007) from the village were targeted for this study. As researchers, we relied on quota and purposive non-probability sampling techniques (Bryman et al., 2018) to obtain a sample size of 147 youth respondents through active engagement in various stakeholder interventions, membership in youth groups, and contrasting socio-economic characteristics that existed during the research. The use of non-probability sampling was suitable to recruit respondents due to the lack of the total population of youth in the study area to determine a sampling frame (Bryman et al., 20218). All 147 respondents successfully participated in the questionnaire phase. Convenience and snowball sampling were used to further isolate 45 apt respondents for focus group discussions (FGDs), and 12 respondents for indepth interviews. --- Data Types, Sources and Collection Both primary and secondary data were collected for this study. Primary data involved information on socio-economic and demographic variables, awareness levels, perceptions, experiences, and skills; as well as prospects, worries, responses, measures and contributions relevant to the engagement of youth in the climate change agenda and sustainable development journey (Barreda, 2018;Bryman et al., 2021;Creswell et al., 2020). Semi-structured questionnaires, with open and close-ended questions, were administered to 147 youthful men and women respondents during weekly group meetings at the village offices to gather specific responses on demographic characteristics, socio-economic aspects, anxiety, perceptions and experiences on existing strategies for youth engagement in climate change and sustainable development. We also conducted three FGDs with 45 respondents and undertook informant interviews with 12 respondents to saturation levels at the village offices. Semi-structured questions were asked to reveal respondents' awareness of climate change meaning, causes, effects, prospects, anxiety and strategic measures to enhance climate resilience and sustainable development. Awareness of climate change and sustainable development was gauged based on respondents' knowledge, attitudes, behaviours, beliefs, experiences, anxieties and prospects (Barreda, 2018). Direct field observation-which involved active listening and watching additional respondents' behaviours and characteristics-was also sustained throughout the primary data collection process (Creswell et al., 2020). Online searches and desk reviews of the most recent and trustworthy published and unpublished sources (Bryman et al., 2018) yielded secondary data to meet the study objectives. Only suitable secondary data from relevant government agencies, grassroots stakeholders and experts were prioritised. Typical secondary data on local governance, biophysical variables, socio-economic and demographic characteristics, youth engagement strategies, stakeholders mapping and global interventions provided helpful guidance to the study. --- Data Analysis Qualitative and quantitative data were analysed separately. Thematic content analysis was used to analyse qualitative data from FGDs, in-depth interviews and direct observations (Creswell et al., 2020;Ahmadi et al., 2022). The thematic content analysis allowed the researchers to interact deeply and familiarise themselves with collected qualitative data (Creswell et al., 2020). The IBM SPSS (Statistics) software, version 20, was used to perform statistical analysis for categorical datasets from respondent questionnaires (Kitasho, 2020;Capsticka et al., 2022). Descriptive statistics provided the summary and simplification of primary data for interpretation and discussion. The binary logistic regression model analysed the awareness of strategic youth engagement in climate change and sustainable development agendas (Barreda, 2018;Ahmadi et al., 2022). --- Results and Discussion --- Socio-economic and Demographic Characteristics of Respondents Most respondents (97.1%) were youth aged 25-35, while only 2.9% were aged between 15-24 years. Men constituted the majority of the respondents at 54.3%, compared to passive and vulnerable women at 45.7%. Most respondents (84.8%) were married, with only 13.3% being unmarried; while as few as 1.9% were divorced. Most youth had early active marital relations and unquestioningly engaged in family-related socio-economic responsibilities (Theodory, 2014;Rao et al., 2016;Mabhuye et al., 2020). Depending on local weather and climatic conditions, most respondents engaged in mixed livelihood activities (56.7%) and farming (12.5%), compared to about 20.2% who relied on pastoralism as a core activity. This is consistent with Kabote (2018), Kitasho (2020), Yanda et al. (2021), and NTRI (2021); who highlighted voluntary transitions from core pastoralism to engagement in diversified activities as a means to enhance livelihood resilience in the area. Most respondents (85.6%) had attained various levels of formal education (primary education (42.9%), secondary education (14.3%), certificate (20%), diploma (6.7%), and degree level (1.9%)), compared to about 14.3% respondents with no formal education (illiterates). Formal education is linked to increased awareness and engagement in climate change and sustainable development among the respondents (Barreda, 2018;Corner et al., 2015). However, additional findings indicated that formal and informal education systems favoured men. As a result, most men appeared better informed than female respondents. Furthermore, most respondents (93.3%) were natives born in the study village, compared to a small minority (6.7%) who migrated into the village from elsewhere. The importance of local inclusion in climate change related studies as a means to acquire first-hand local knowledge and pertinent experiences on interconnecting variables was also uncovered by Narksompong et al. (2015) and IPCC (2019). Additionally, Zimba et al. (2021) and UNFCCC (2019) opined that including local youth in climate change and development studies offer respondents an opportunity to discuss the relevance of strategies, policies, and interventions. Since this study involved youthful male and female respondents with prior knowledge, exposure and experiences on the topic, analysing the socioeconomic and demographic variables provided insights on significant social processes, perceptions, anxieties and progress influencing awareness of engagement strategies in climate change and sustainable development. --- Awareness of Climate Change and Sustainable Development Most (90.6%) of the questionnaire respondents confirmed awareness of climate change, its impacts and response measures in the area, compared to 4.7% unaware respondents; while 4.7% did not know what to say. Additionally, the majority of focus group and interview respondents linked development challenges-such as water insecurity, food insecurity, low livestock productivity, invasion of weeds and diseases, reduced pasture, unreliable onset and cessation of rainfall, droughts and increased human-wildlife conflictswith the impacts of ongoing climate change in the area (Kimaro et al., 2018). The presence of stakeholders' interventions at various levels indicated typical response measures taken to address the climate change and sustainable development agendas (Barreda, 2018). Researchers noted that increased knowledge and exposure influenced awareness, engagement attitudes and the roles of youth in the area. Kimaro et al. (2018), Makonnen (2018), andYanda et al. (2021) 2018) also correlated comparable findings, highlighting the influence of indigenous knowledge on rural livelihoods. Furthermore, the researchers noted that increased awareness levels enhanced livelihood diversification and maximised local resilience. This indicates the potential and absolute necessity of incorporating local knowledge in stakeholders' strategic interventions to enhance awareness and youth engagement in climate change and sustainable development (Barreda, 2018). --- Emerging Anxiety about Climate Change in the Area Approximately 90.5% of the questionnaire respondents confirmed anxiety, compared to 9.6% of non-anxious respondents. This indicates that increasing climate change awareness increases anxiety and the drive to respond. One focus group discussant precisely asserted: "Everyone in this village is worried due to the prevailing drought and the lack of pasture. Last year, and even this year, we lost so many livestock. We have lost our relatives due to stress, migration and reduced livestock economy." Another discussant highlighted the risks associated with the diversification of socio-economic development: "Most of us no longer rely on pastoralism alone. We are trying to explore different livelihood activities in order to diversify our income. In the process, we expose ourselves to so many unexpected risks -the wildlife, diseases, divorces, deaths: but what else should we do?" These findings are also consistent with Narksompong et al. (2018), IPCC (2021), Greene et al. (2020), Barford et al. (2021) andO'Brien et al. (2018): all echoed despair, resentment and dissent among the youth as a result of climatic change that weaken adaptive capacities and livelihood resilience. --- Respondents' Contribution to Climate Change About 95.2% of the questionnaire respondents confirmed factors contributing to local climatic changes. Specifically, the majority of FDG and interview respondents linked uncontrolled farming, improper livestock husbandry, charcoal production, deforestation and land degradation with the ongoing climate changes. One focus group discussant narrated: "Unsustainable socio-economic activities like overstocking, shifting cultivation and deforestation cause drought and excess heat." An increased understanding of the role of human activities in climate change among the youth due to interventions from existing stakeholders was also elucidated by Mung'ong'o et al. (2016), Barreda (2018), UNFCCC (2019) and IPCC (2021). --- Awareness of Strategic Youth Engagement in Climate Change and Sustainable Development Descriptive analysis showed that most respondents (81.0%) confirmed awareness of local strategies for youth engagement in the area, compared to 19% who lacked awareness. Additionally, focus group discussants highlighted poverty eradication, food security, health and well-being, quality education, gender equity and water security as substantial goals for local resilience. This indicates that the local youth are prioritised among the 17 global sustainable development goals (Nkrumah, 2021). Furthermore, the binary logistic regression model (Table 1) showed that the level of youth awareness of engagement strategies in climate change and sustainable development was significantly predicted by mixed livelihoods activities (p = 0.048), primary education (p = 0.015), secondary education (p = 0.019), certificate (p = 0.039), diploma (p = 0.028) and non-formal education (p = 0.020) at 5% significance levels. These findings are consistent with the findings by Corner et al. (2015), Partey et al. (2020), Kitasho (2020) and Marie et al. (2020); who also reported the influence of livelihood diversification and education on the knowledge, attitudes, behaviours, beliefs, experiences, anxiety and prospects held by respondents on engagement strategies in climate change resilience and sustainable development. The findings also supplement Ojala et al. (2017) andMakondo et al. (2018), who highlighted the influence of formal and informal education on access to climate messages, increased awareness, and engagement in diverse livelihood activities. Respondents also hinted at polygamy and early marriages as strategies to enhance resilience and development due to the perceived increase in manpower and economic gains from dowries. These findings are consistent with Allegretti (2014), Mung'ong'o et al. (2016), Kabote (2018) and Kitasho (2020); who highlighted the influence of implicit local knowledge, initiation rites and gender-blind systems on youth engagement in socio-economic development. --- Explicit Engagement Strategies The FGD and interview findings revealed that explicit intervention involved selfhelp, micro-finance (soft loan) services, and business incubation schemes under the district government in an attempt to implement the national youth policy (NYP, 2007). The presence of active NGOs and CBOs in the area also complemented government efforts to enhance local resilience and sustainable development. Researchers observed that most private interventions aimed at wildlife conservation, tourism promotion, crop protection, rangeland management and alternative income-generating activities. Notably, establishing the RWMA and raising awareness on land use rights influenced the role of youth in wildlife management, tourism and resource utilisation (Follows, 2018;TPP, 2020). One interview respondent asserted the following: (Corner et al., 2015;Ojala et al., 2017). Furthermore, respondents thought that emerging non-state actors embraced implicit and explicit strategic interventions to enhance youth engagement in sustainable development and climate change agendas (Kitasho, 2020;O'Brien et al., 2018). However, FG discussants revealed that climate misinformation, competition and duplication of efforts among actors negatively influenced the implementation of priority development goals. These findings complement the findings in the literature that highlight substantial inefficiencies due to misinformation; and mismatched, delayed, and unsystematic youth engagement strategies (Nkrumah, 2021;O'Brien et al., 2018). The fact that this study was conducted in a specific geographical context with homogenous cultural characteristics presents a possible bias to the generalizability of the final findings to youth in other contexts. The study also concedes typical limitations associated with time and resource constraints, the choice of methodology, sampling and data analysis techniques. Future studies should address these limitations and seek alternative methodologies to attain valid and reliable outcomes replicable in different locations and cultural settings. --- Conclusion and Recommendations

This paper employed a triangulation of methods in assessing the effectiveness of youth engagement strategies in climate change agenda and sustainable development within Naitolia village, in Monduli district Tanzania. Nonprobability sampling techniques were used to determine a total of 147 youthful respondents aged between 15 and 35 years from Ormang'wai and Engusero sub-villages. Qualitative data were collected using FGDs, in-depth interviews and direct field observations; while quantitative data were captured through administered questionnaires. The results reveal that 81.0% of the respondents knew of the explicit and implicit engagement strategies in the area. The binary logistic regression model showed that mixed livelihoods and education systems significantly influenced the respondents' awareness and engagement trends in priority development goals. Most engagement strategies embraced a combination of explicit and implicit strategies to enhance livelihood resilience. However, implicit strategic interventions were popular, while explicit strategies were considered ineffective and tokenistic due to mismatched execution of misaligned 'one-size-fits-all' policies. In spite of increasing awareness on climate change, existing youth engagement strategies are still misplaced, tokenistic and ineffective in guaranteeing sustainable development in a rapidly changing climate. The paper recommends proper mapping and inclusion of key stakeholders to raise public awareness and resource harmonization for enhanced youth engagement in climate change and sustainable development. Specifically, it recommends concrete paradigm shifts in governance, education systems, skills development pathways and financial accessibility to create enabling environments for effective youth engagement and empowerment in climate change, and beyond priority development goals.

Introduction A wave of publications around lived experiences and theorisation of disabilities and chronic illnesses in academia (see Brown, 2021;Brown & Leigh, 2020;Dolmage, 2017;Kerschbaum et al., 2017;Price, 2011) highlights an increased interest in ableism in academia within the English-speaking countries across the world. Disability scholars whose theories of disability identify social, cultural, economic, and political dimensions (Goodley, 2014;Saltes, 2013;Shakespeare & Watson, 2001) have long questioned views of disability as representations of flawed bodies that diverge from determined norms. Yet, the pathological conceptions remain intact and detrimentally shape the higher education experiences of faculty, staff, and students with disabilities. Disability in higher education remains viewed through medicalised approaches and thus shapes how universities institute policies and practices (Merchant et al., 2019;Waterfield et al., 2018). From literature and statistics, we know that numbers of academics disclosing their specific needs are low in comparison to students or the general public disclosing their needs (Brown & Leigh, 2018). Disclosure is a complex, sensitive, and very private event that results in individuals weighing benefits and costs and identifying whether the risks may not be too great (Brown, 2020). A major challenge in disclosure of disabilities to employers relates to requests of accommodations (Lindsay et al., 2018). Employers do not readily agree to adapt organisational structures (Van Laer et al., 2020) which constructs disabled employees/faculty as inferior and characterised as deviant, unproductive, need of care (Dobusch, 2017;Mik-Meyer, 2016). As part of her research work as well as to raise awareness of the experiences of disabled staff in academia, Nicole organised and facilitated a workshop-type conference in London entitled Ableism in Academia in March 2018. This conference was planned and arranged to be as accessible and inclusive as possible (for full details, see Brown et al., 2018). The purpose of the conference was to provide a space for disabled staff in academia to share experiences, whilst simultaneously allowing for data collection in the spirit of action research as a lived practice (Carson & Sumara, 1997). Theorisations of experiences alongside recommendations to make academia more inclusive have already been published (Brown, 2021;Brown et al., 2018;Brown & Leigh, 2020). This article thus explores ableism as experienced by members of the academy in the UK, through analysis of the contents from the lightning talks, the workshop posters, and feedback from the conference. The aim of this article is to provide an insight into the lived experience of ableism in academia as a stepping stone towards practical recommendations. To this end, we commence with a brief introduction into ableism and disability in higher education before outlining the details of the research design, data collection, and analysis. The nuanced experiences of disabled academics in higher education are discussed as well as their collective understandings of these experiences as constructed through normalisation and able-bodiedness. Thematic representations are explained with regard to marginalisation in academia, being silenced in academia, and perspectives on improving disabled academics' experiences. Finally, we share implications to further research and practices to be applied at individual and institutional levels. --- Positionality, language, and terminology Before continuing, we feel it is important to include a brief section reflecting on our positionality within our work, but also the kind of language and terminology we use. Nicole, the first author and the main conference organiser, identifies herself as a white woman of dual nationality and with a chronic condition. She is profoundly deaf and has been diagnosed with fibromyalgia, Sjogren's syndrome, and a vertigo-condition similar to Meniere's disease. Although Nicole can pass as someone with no disabilities, she uses her personal experiences with accessibility issues due to her lived experience of chronic conditions and profound hearing loss within the scope of advocacy and lobbying. Nicole's interest in ableism is therefore simultaneously of a professional and personal nature. Karen, the second author recognises the dynamic and shifting nature of identity. She describes herself as an Indo-Caribbean immigrant woman, who experiences being othered as a result of intersectional marginalising characteristics related to race/ethnicity, socioeconomic status, gender, ability, religion, and citizenship. Karen works as a tenure track faculty member researching the marginalisation of disabled people in educational institutions. Through the critical disability studies lens, her work explores how institutional, socio-political, and cultural influences shape their experiences and trajectories. Words, labels, terminologies, and language, in general, play an important role in communicating connotations and identity markers or in ensuring clarity in legal or policy contexts, for example. The debate around person first language is particularly contentious as using one over the other supports or denies identification with and belonging to a specific group (see Dunn & Andrews, 2015;Sinclair, 2013), and it really is up to individuals to let us know how they identify and consequently, how they would like to be addressed. Person-first language relates to description of a person with a disability, such as "person with autism", whereas some prefer the use of language that reflects an identity-first approach, thus "an autistic person". We have deliberately chosen not to apply the person-first language, because through our research into the lived experiences of disability, we realised that for many individuals, using the disability label is connected with activist work and raising awareness for the cause of disabled people. For the purpose of this article, we use the term disability in its broadest sense to describe the experiences of disabled persons, as well as those with chronic conditions or neurodivergences, such as dyslexia, dyspraxia, dyscalculia, attention deficit disorders, and syndromes or autism. We are fully aware that disabilities, neurodivergences, and chronic conditions are not one and the same, and yet, we do feel that in terms of ableism impacting all of those individual groups, we may take the liberty in this article to use disability to encompass them all. --- Ableism and disability in higher education Disability studies challenge the socially constructed notion of normality and problematises the abled/non-abled dichotomy. Differences are broadly understood, accepted as natural experienced in multiple and dynamic ways and not only connected with but intrinsic to human experience (Ferguson & Nusbaum, 2012). Ableism "is a set of beliefs that guide cultural and institutional practices ascribing negative values to individuals with disabilities whilst deeming able-bodied and able-minded individuals as normal, therefore superior to their disabled counterparts" (Annamma et al., 2013(Annamma et al., :1279)). In this sense, ableism is an "umbrella ism for other isms" (Wolbring, 2008:523) with disability being "cast as a diminished state of being human" (Campbell, 2001:44). Thus, bodies and minds must conform to dominant standards or become devalued and flawed. In other words, "Disabled people are the 'able-bodied' gone wrong" (Garland-Thomson, 1999:49). Within higher education, the organisational spaces are equally entrenched in power relations between disabled and non-disabled. A recent study illustrated the manner in which "ableist organizational space enable, and becomes dominated by, a routine employee practice that reflects and normalizes the way able-bodied employees can, and unconsciously do, interact with a space designed for them" (Van Laer et al., 2020, p. 13). The experiences of these employees in academia are lived and experienced in ways that are etched in ableist structures. Naturally, categorisations are problematic, as a group of employees within academia is not and cannot ever be considered homogenous. Yet, perceptions of otherness resulting from oppressive attitudes, policies, and practices are experienced which questions their value, independence, comfortability, and safety in this space, as is evidenced across a variety of disciplines and contexts (e.g. Bê, 2019a;Branco et al., 2019;Broderick & Lalvani, 2017;Dunn, 2019;Lynch & Macklin, 2020;Mireles, 2020;Singer & Bacon, 2020). Research also indicates the importance of understanding the perspectives of faculty in academia in relation to disability. Mori<unk>a et al. (2020) found that disability, viewed as a valuable asset and not relegated to its established deficit position from a medicalised perspective, can be valuable in recognising that all students are important and can learn when the conditions and attitudes are properly in place. Higher education can be spaces of inclusivity and affirmative practices where disabled individuals can thrive. Disability is a judgment-based determination and the product of social, political, economic, and cultural practice (Baglieri et al., 2011). Variations from an arbitrary construction of normal become evaluations of abnormality or deviance. Across the higher education sector, the complexity of disability seems to be overlooked because of the engrained norms of normality. The concept of normal privileges some whilst marginalising others. It creates rifts and borders that function to delineate preferred characteristics (Bê, 2019b;Inckle, 2018;Pionke, 2019). --- Research approach and process As mentioned, Nicole spearheaded a conference on the topic of ableism in academia with two aims. The first aim was to offer a conference that would be accessible for academics with disabilities and chronic illnesses to demonstrate best practice in relation to conference organisation (for full details, see Brown et al., 2018). The second aim was to use the conference as an opportunity for gathering data within the scope of a constructivist inquiry in relation to academics' experiences of ableism in academia to develop strategies for best practices for higher education institutions. The conference therefore enabled presenters and attendees to exchange, share, and discuss experiences of being disabled and othered in academia. Constructivist, practice-based inquiries are commonly applied in educational contexts (Clarke & Erickson, 2003). As with all research, there is no one unified or right way to conduct practicebased inquiry in educational contexts, which may include action research (Sumara & Carson, 1997), reflection in action and reflection on action (Schön, 1987), case study explorations (Merriam, 1988), or living educational theory (Whitehead, 1989). What binds all these forms together is their focus on specificities of contexts. The aim of these inquiries is to use constructivism as a framework for gaining in-depth understanding of very specific, contextualised, localised circumstances that then leads to more generally applicable conclusions. Thus, practice-based inquiries make abstract, theoretical and general what is usually concrete, experiential and particular. The overarching research question for this study was "What are the lived experiences of disabled academics in contemporary higher education?" The data collection process was organised around the two elements of the conference: peerreviewed lightning talks and a workshop for all delegates. After a keynote presentation, the morning session entailed 13 lightning talks of approximately 5 to 10 minutes each, in which academics outlined their personal experiences of disability and ableism within academia. The topics predominantly covered experiences of invisible disabilities and neurodivergences, whereby there was a specific mention of autism, depression, microscopic colitis, grief, dyslexia, voice impairments, pain conditions, and colour blindness. Throughout the morning, the delegates were able to discuss and contribute to the lightning talks publicly and privately in plenary debates and via social media platforms, some of which were restricted to conference delegates only. The afternoon was dedicated to a workshop during which delegates considered and discussed five subsidiary questions in small groups. Groups were developed organically from the way the room had been organised in a cafe style layout, and from how individuals with particular needs would have selected their preferred seating. The small groups therefore consisted of four to six delegates, with the option of asking for a conference helper to facilitate the discussion. All groups were required to provide a summary of their discussions and responses to the questions via a poster or via email. Where required, conference helpers acted as scribes for this activity, for which flipchart paper and pens were made available. The five questions were as follows: (1) How does your disability/disadvantage affect you in the workplace and what practical effects does it have on your ability to perform your role? (2) What does your employer do to help you, and what more could they do? (3) What forum(s) is or are there at your institution for discussing matters related to ableism? (4) What forums are there in higher education to deal with these matters (from Unions to Higher Education Funding Council for England)? ( 5) What could/should be done to encourage members of academia to disclose their concerns/disabilities? The conference was inextricably linked with the data collection process through interactions within the conference, through connected social media activities via the hashtag #AIA2018 and the twitter handle @AbleismAcademia. Consequently, the lightning talks, the workshop, the discussions, the recordings from the conference, the tweets, and the follow-up feedback were all considered data sources. --- Participants This is not a typical formal study, this is an organic study, where we worked with the delegates, who all had access needs in some respect. The research participants were conference delegates. Eighty delegates attended the main conference in person, with a known 30 delegates attending a break-out event. The break-out event meant that some participants could engage in debates and discussions locally whilst watching the conference livestream and communicating with the main event via twitter and similar online platforms. All delegates seemingly had a vested personal and/or professional interest in disabilities and ableism in academia. In this sense, academics with disabilities were definitely over-represented in comparison to typical conferences or research studies. Based on the delegates' self-identification, the male to female ratio was 1:4. Delegates were of different ages and held a variety of professional roles that ranged from doctoral scholarships, temporary part-time research or teaching positions, permanent teaching and research fellowships, and professorial and tenured positions. The main concern of all those involved in and associated with the conference related to the experience of disability itself. Many contributors and delegates felt that there was a strong need for an open debate around how disability is viewed in academia. At the same time, however, openly stating, disclosing a condition or challenge was considered problematic. Participants felt they would make themselves vulnerable, open themselves up to criticism and potential discrimination or risk to be stigmatised for being different. --- Ethical considerations Bearing in mind the use of practice-based inquiry as an approach to the conference, institutional ethical approval for the use of the data, and the direct approaching of individuals for follow-up conversations or interviews was sought and formally granted within the specific context of the organiser actively working with participants to generate meaningful data that can inform policy and practice. Individuals provided written consent for their data and contributions to be used within the scope of a research project and dissemination via publication and further conference presentations. However, as some delegates had not disclosed their conditions or needs at their own institutions for fear of stigmatisation and discrimination (Goffman, 1990(Goffman, /1963)), the data needed to be protected and kept as anonymous and confidential as possible. On the other hand, there were delegates who had openly disclosed their health status, who were comfortable with their disabled and chronically ill identities, who wanted to have their voices heard and who were therefore vocal in participating. Everyone who wanted to be part of the discussions and contributed via twitter did so with the understanding that their contributions were public. If someone wanted to remain anonymous in the discussions, they had the opportunity to do so via the organised padlet, which was not included in the data analysis. Where tweets were part of the analysis, Nicole personally contacted all contributors and asked for consent to include the data. The approach to ethical consideration was therefore focused on combining ethics of care with ethics of justice to result in a single process of "reconciliation, reciprocity, diversity and responsibility, and with an awareness of power" (Edwards & Mauthner, 2012:12). Within the context of the practice-based inquiry and in the spirit of collaborative learning during the conference, ethical approval was therefore negotiated with individuals at every step during the conference, the subsequent analysis, and write-up to offer individuals control. This flexible approach to research is not uncommon in "community-based participatory research, or emancipatory methods and methodologies" (Kara, 2018:41) and is commonly associated with methodologies developed to address various forms of oppression (Kovach, 2015:46). --- Data analysis Qualitative data analysis is more than merely following specific steps to achieve meaning. Indeed, analysis is a craft that requires "reading beyond data" (James, 2013:574) and falling back upon "the repertoire of implicit knowledge that researchers themselves possess" (James, 2013:574). Therefore, data analysis can never be an objective process of having themes emerge. It is a very subjective, personal, and active process of the researcher making sense of data and pulling out specific themes and meanings (Morgan, 2018). In this project, data analysis was an iterative process of searching for and identifying themes by an active data manager who constructs themes from the data in a "transparent, reflexive and critical" way (Brown, 2019:497). The first step was to collate all the data, the twitter feed comprising of 978 tweets from three months before the conference to three days after the event, the workshop answers on flipchart paper, the transcripts from the lightning talks, the recorded livestream, and the discussions (which totalled eight hours of footage) in Excel data spreadsheets. Thematic analysis was used (Braun & Clarke, 2006) in its intended reflexive form of research practice (Braun & Clarke, 2019) through commonly used coding practices (e.g. Bazeley, 2013;Bazeley & Jackson, 2019;Salda<unk>a, 2009) to develop the initial codes: symptoms, management of symptoms, labelling and diagnosis, relationship with others, experiences of difference, negotiating differences, understanding of ableism, examples for improvements, and recommendations for action. These codes were then used to generate overarching themes around disabled academics' experiences and perceptions. After all data were re-considered in connection to these bigger themes in an iterative analytical process, they were broken down again into more detailed categories in relation to descriptions of experiences, from which we formulated the resulting themes: feeling marginalised due to their perceived deficits, being silenced, and what academic organisations can do to improve their experiences as disabled people. --- Findings and discussion In the following section, we discuss the three dominant themes in an interpretive, analytical manner with some substantiating quotes from named or unnamed delegates. With this specific approach, we honour delegates' personal wishes regarding anonymity and protection of identities or having their quotes attributed to them in public. --- Marginalised in academia Participants felt strongly about otherness and being othered and consistently commented on how specific ways of working or living were not accepted and acceptable in academia. For them, their lived experiences of being disabled or ill represented a "different physical reality" (a delegate), which it is necessary to make understood. Disability may result in consequences that will affect productivity and attendance. For example, negotiating everyday pressures, alongside managing a condition and soundscapes of bright rooms in busy buildings, leads to increased levels of tiredness and fatigue. One participant aptly shared, Invisible disability in the academy is exhausting, peers & work conditions constantly overlook my needs. They have difficulty grasping fluctuations & often it's easier to just shrug off my needs. I can so relate to Clara's talk #AIA2018 #Spoonies s_golightley @s_golightley As this statement shows, otherness and othering in this context are not necessarily intentional, but are a reality of life when managing bodies and emotions. For those who have disabilities, othering and marginalisation happen because academia is built on an "assumption of a universal model of health" (a delegate), which delegates described as the inherent assumption of everybody being healthy and well. Academics are faced with particular expectations around health, more specifically the expectation that an academic is to function able-bodied at all times. For those academics who have disabilities, the fluctuations of managing their conditions mean that they cannot meet those expectations. As a result, they are seen as less productive, underachieving, and failing in their roles as academics, which in turn others them. At the same time, when adjustments are put into place to accommodate individuals' needs, this being treated differently is also considered a form of othering. Moreover, the stigma that is commensurate with othering is detrimental to the social and professional experiences of disabled persons in academia. The problems described here are due to the "inequalities, negative attitudes, misrepresentations, and institutional practices that result from the process of stigmatization" (Garland-Thomson, 1999:32). In this regard, a person is devalued because they do not fall within the normalised conception and expectations of academia. Due to their differences, perceived problematic characteristics, they are constructed as inferior to their healthy and normal counterparts, such as associates or colleagues. Marginalisation in academia is even more complex, once the individual experiences of those with disabilities are concerned. These individuals are marginalised for their differences, but they also feel marginalised among an entire group of disabled academics. Delegates' understanding of what is normal is so internalised that some participants not only measured themselves against the societally acceptable, standardised norms but also against internalised criteria of disability. In absence of obvious visible identifiers (such as a wheelchair), one problematic assumption is of nondisability, leading both to individuals 'passing' as abled, and to unacknowledged diversity of disability present in staff and student populations. (...) It can be professionally harmful if a system either over-or underestimates the predictability of an individual's symptoms, and so demands too much or expects too little. Carla Finesilver, a delegate in her speech When participants exchanged their experiences of being different and othered in the workshops and discussions, some individuals almost apologised for their issues not being "as serious as yours" (a delegate). Despite the environment and atmosphere having been one of acceptance and mutual understanding, some individuals felt invisible and hypervisible at the same time. They felt they stood out "like sore thumbs" (a delegate) for not being disabled enough for their issues and conditions to be outwardly recognisable. They felt marginalised within a group of marginalised individuals. Regardless of each's perception of their position within the continuum of disabling conditions, the complexity and nuances of their oppressive realities in academia is the common thread which binds them together. Challenging the dominant conceptualisations of normality within institutional structures is rather difficult due to the ableist mind-set and practices that are associated with working in this context. --- Silenced in academia Where the lived experience of disability related to academia, individuals specifically referred to the prevalent culture of overwork and productivity. Participants consistently highlighted the time and effort required for self-advocacy in order to gain access to reasonable adjustment, time and effort that other academics can spend on preparing grant proposals and publications. The burden of proof to provide necessary evidence for support lies with the individual, but at no point is there a consideration of how attending to doctors' appointments and completing relevant forms disrupts the everyday academic routine. In the competitive, neoliberal environment of present-day academia time and effort spent on managing a condition and its effects means to lose out on important opportunities for being productive. Due to the constraints inherent in academic institutions, as a result of neoliberal ideology at work, the limited and quantified accountability measures that demonstrate productivity (i.e. publications, citations, student evaluations) often stymie the efforts of historically marginalised groups (Ramlackhan, 2019). Academics generally are required to demonstrate their research impact and disseminate their findings, ideally to an international audience. That conferences pre-COVID-19 did not commonly allow for remote presentations, most often for financial reasons, meant that academics with disabilities were not able to attend and therefore not able to contribute to scholarly debates in that way. One of the most impactful tweets highlighting the ableist structures within academia relates to the conference organisation: This is the first conference I have attended in 8 years. A delegate Live tweeting conferences is also invaluable for those of us who physically cannot make them in person! I'm eternally grateful to those who can do that so I can at least be there virtually if not in reality. Debbie White @medievaldebbie The ableist environment of academia, as described by participants, demonstrates that "individual bodies must conform to institutional standards, rather than restructuring the social environment to accommodate" their varying needs (Garland-Thomson, 1999:51). The capacity of individuals to conform in order to meet the required demands of working in academia is narrowly framed and constrains their ability to function appropriately and effectively. This conference was a different experience for some as it allowed them to discuss "uncomfortable truths that academia would prefer weren't talked about" (workshop bullet point). This statement encapsulates the experience of being silenced when it comes to exploring the issues and concerns of the disabled. However, disabled academics also feel silenced in relation to their scholarly work. Academics who have openly disclosed a condition are expected to become disability activists and advocates and as such are recognised for these activities rather than their research foci (see also Brown & Leigh, 2018). As a consequence, the kinds of scholarly debates they are privy to are of specific natures and kinds, and they feel silenced as contributors to scholarly knowledge in their own right. Being perceived undesirably by only that which makes them different, such as after disclosure of a condition understood as disabling, reifies the "negative stereotypes and cultural values that surround disability..." (Vehmas & Watson, 2014:640). Furthermore, "the problem is not the disabled person; the problem is the way that normalcy is constructed to create the 'problem' of the disabled person" (Vehmas & Watson, 2014:640). This emphasises the importance of making apparent the entrenched assumptions of disability and disabled persons in academia in order to bring about real change and action that can shape the experiences, institutional structures, and perceptions of people within their social and professional contexts. --- Perspectives on improving disabled academics' experiences This theme was more prominent than others, probably because of the way that the conference specifically aimed at theorising ableism in academia and developing a strategy to improve individuals' experiences. The key recommendations for policy and attitudinal changes therefore related to a clearer vision of inclusion in academia and more awareness through networking opportunities. On individual levels, delegates highlighted how some practices modelled in the conference need to become standard, such as live-captioning or the use of microphones, recording, and streaming. The conference was considered modelling best practices and enabled individuals to engage in academic work in an unprecedented way: Thanks for the opportunity to attend via live stream from the sofa with a hot water bottle in E. Yorkshire #AIA2018 Kerry Pace @DiverseLearners The overall sentiment was, however, that institutional and indeed sector-wide changes are needed to improve the experiences of academics with disabilities. Such changes, delegates commented, would only be possible with more awareness in relation to the lived experience of chronic illness and disability in academia. This is because many instances of ableism, discrimination, and marginalisation occur unintentionally but due to lack of awareness and understanding of individuals and institutions. [We need] forums to discuss (student/staff forum, anonymous questions via whiteboard or forum, nothing really for staff except informal networks, equality/diversity committees for those who have access) workshop bullet point To this end, delegates highlighted, more research and activism is required. It is therein that the delegates saw the issues, because academics with a vested interest in improving the situation for disabled colleagues may not necessarily have the energy or means to pursue activism. At the same time, it was felt that education on its own would not be enough and that policies would be required to enforce cultural and attitudinal changes in the long term. Ultimately, delegates talked about moving away from what is expected as normal and standard to an environment where different forms of working and living are embraced to such an extent that "no adjustments are required" (workshop bullet point). Delegates described the working environment of academia as male-dominated and more conducive to and appropriate for men. This negative attitude towards the working atmosphere in academia relates to the masculine, strong, competitive reinforced (Hey & Bradford, 2004) through the new managerialism in higher education (Deem, 1998(Deem,, 2001;;Deem & Brehony, 2005;Waitere et al., 2011). In gendered discourses of the neoliberal universities, men are engaging in real work, whereas women are forced in to lower paid or unpaid positions that involve caring and nurturing duties (Lovin, 2018). Some academics openly highlighted their scholarly and personal interest in feminism and their feminist worldviews underpinning their interpretations of what happens in academia. Others, however, who did not necessarily relate to a feminist identity, still pointed out the differences between disabled men and women, with those at the intersection of disability and femininity finding it more difficult to work and live within academia. And yet, many continue to romanticise working in academia (Lovin, 2018). There is an important difference between visible and invisible conditions and disabilities. Depending on the kind of challenges individuals face, they may or may not have the choice to openly disclose their disabilities. This management of the self (Goffman, 1990(Goffman, /1963) ) is not always straight-forward, as sometimes tools and adjustments "out" a person for being different. And where individuals may not want their immediate colleagues or line-managers to know about specific needs and potential requirements, they may decide to withhold relevant information to avoid being "othered". The interdisciplinary nature of the conference enables individuals to explore their experiences of ableism and safely disclose specific concerns. Individuals felt that this type of event provided the platform for discussion, away from one's own disciplinary field and expert network. That the speakers were not surrounded by their expert colleagues, but a room full of strangers with similar experiences, seem to have made the process of disclosure more intimate, more controlled, and less daunting, even though everyone knew about the livestream and the recording being available via YouTube. The general accessibility of the conference provided a safe environment, in which "otherness" became the norm and so was not "otherness" after all. Delegates used pillows, sunglasses, blankets, and socks, for example to make themselves comfortable in the physical space of the conference room. It was acceptable to get up, lie down, remove oneself from the room, and return at any time. One delegate commented on the relevance of "awareness around people's needs and sensitivities". --- Conclusion We used a Disability Studies in Education lens (Connor et al., 2008) to explore the experiences and challenges of being disabled in academia. Disability Studies in Education is an offshoot of disability studies that emphasises the importance of inclusive institutional structures, advocacy, and equitable access as well as a critique of the societal structures that have functioned to create the normal. Whilst recognising the "hegemonic ideas and practices [that] reify the ideology of normal" (Annamma et al., 2013(Annamma et al., :1280)), Disability Studies in Education "argues for an examination of the institutional and cultural practices...that shape dis/ability" (Collins, 2013:284). The concept of Disability Studies in Education purports that disability "is a socially and politically constructed response to perceived difference" (Collins, 2013:284), and the deficit beliefs from dominant ableist thinking that are inherently understood and accepted as normal, for example, as found within educational environment of academia. Shifting the focus from limitations, such as attitudes, environments, practices, and beliefs, and their negative connotations (Ferguson & Nusbaum, 2012) of disabled people in academia, is a necessary step that can lead to changes in policies and processes that challenge previously conceived notions that constructs disability as negative and problematic (Werner, 2015). This type of change cannot occur with a systematic examination of the structural and cultural policies and practices that impede disabled people in their efforts to be successfully in academia. Moreover, this is not only an organisational issue but a moral and political one as well. In sum, this article's contribution lies in (a) the elucidation of the lived experiences of ableism in academia through the application of a constructivist inquiry within the scope of the disability studies in education lens, and (b) in using this approach as a stepping stone for framing recommendations for practice and policy. With the advancement of equality, diversity, and inclusion initiatives within institutions of higher education, of which disabled academics should benefit, the following are shared to highlight a few imperatives: Leadership matters. The power and responsibility of leaders, such as Department Chairs or Deans, are critical to creating working conditions that value and enhance individuals and to build community. These leaders are needed to institute policies, allocate funds, and provide resources in support of disabled academics. By sending clear and consistent messages of the importance of equality and inclusivity in actionable ways, these advocates will strengthen the physical space, emotional and mental well-being, and individual capacities of disabled people in higher education. Nurturing excellence. Within the neoliberal agenda and confines of academia where incessant productivity is at the helm, disabled academics can be supported best with flexible scheduling, collaborative structures, and accessible processes. During the COVID-19 pandemic, institutions of higher education have made significant adaptations to working environments so that working from home proved to be a viable option to the traditional office work space. Although not everyone has benefitted from working from home, these changes have proven to increase productivity for disabled people. In March 2018, the Ableism in Academia conference was one of its kind because of its flexibility, inclusivity, and accessibility offering disabled academics a unique opportunity to engage in knowledge exchange and research dissemination. Flexible and adaptable processes and procedures need to continue if we are to maximise their strengths and excellence. For example, trainings or professional development for leadership personnel, academics, and staff focused on understanding disabled experiences and ways to support them is needed. Individuals working in higher education contexts with disabled academics first must acknowledge their views and perspectives about disability and configure ways to support colleagues and co-workers that value and enable their contributions. Above all, therefore, an attitudinal shift is required that will allow for disabled people to engage in and with higher education as a matter of course. Whilst we aim towards a Universal Design for Learning to support disabled students (Bracken & Novak, 2019), a more ubiquitous universal design that truly focuses on inclusion and equality rather than reasonable adjustments would benefit everyone working, studying, living, and breathing the academy. An attitudinal shift is also required for an improved understanding of how disabled faculty, staff, and students engage in and with academia. Within disability studies and ableism studies, there is a strong narrative relating to the disability experiences and discourses of who is and should be able and allowed to research and report on disabilities. The slogan "nothing about us without us" comes to mind (e.g. Bryden, 2015;Charlton, 2000;Yeo & Moore, 2003). However, research into academic's lived experiences of ableism and disabilities is complex. On the one hand, the involvement and participation of those directly affected is needed; on the other hand, experiences of

To understand the experiences of the disabled in academia, a fully accessible and inclusive workshop conference was held in March 2018. Grounded in critical disability studies within a constructivist inquiry analytical approach, this article provides a contextualisation of ableism in academia garnered through creative data generation. The nuanced experiences of disabled academics in higher education as well as their collective understandings of these experiences as constructed through normalisation and able-bodiedness are presented. We show that disabled academics are marginalised and othered in academic institutions; that the neoliberalisation of higher education has created productivity expectations, which contribute to the silencing of the disabled academics' perspectives and experiences due to constructions of normality and stigmatisation; and that it is important to enact policies, procedures, and practices that value disabled academics and bring about cultural and institutional changes in favour of equality and inclusion.

co-workers that value and enable their contributions. Above all, therefore, an attitudinal shift is required that will allow for disabled people to engage in and with higher education as a matter of course. Whilst we aim towards a Universal Design for Learning to support disabled students (Bracken & Novak, 2019), a more ubiquitous universal design that truly focuses on inclusion and equality rather than reasonable adjustments would benefit everyone working, studying, living, and breathing the academy. An attitudinal shift is also required for an improved understanding of how disabled faculty, staff, and students engage in and with academia. Within disability studies and ableism studies, there is a strong narrative relating to the disability experiences and discourses of who is and should be able and allowed to research and report on disabilities. The slogan "nothing about us without us" comes to mind (e.g. Bryden, 2015;Charlton, 2000;Yeo & Moore, 2003). However, research into academic's lived experiences of ableism and disabilities is complex. On the one hand, the involvement and participation of those directly affected is needed; on the other hand, experiences of marginalisation and othering need to be navigated alongside stigma and matters of disclosure. Hence, there is significant tension between the conflicting need of disclosing to be involved in relevant research and of not-disclosing to maintain one's existing public, scholarly identity. As we have seen from the findings section, within the context of higher education, academics are keen to be involved in scholarly debates and to engage in activist work, but not at the expense of their career trajectories and potential opportunities to pass as non-disabled. As with all research, this report should not be considered an end-point, but as a resource to draw on for future work. Considering our own conclusions, we have identified three main avenues for further research: First, the practical recommendations from this article alongside previously published work could be used for the development of equality, diversity, and inclusion training, which should then be evaluated for its effectiveness in raising awareness, changing attitudes, and creating long-term change. Second, we see an important connection to be made between the existing work around ableism in academia and the most recent developments around flexible, different, and creative ways of working that have arisen during the COVID-19 pandemic. And third, building on the constructivist inquiry applied in this project, there is a real opportunity here through and evaluation study to contribute to methodological advancement within the scope of participatory, democratic ways of working for data generation and analysis. The framework of the constructivist inquiry applied during the conference enabled individuals to be part of the conversation and shape the discourses, whilst simultaneously remain anonymous, if they so wished. The research here was not done to, on, or with them, but very much by them. The result is a disabled-consciousness conceptual approach (Ramlackhan, 2021) that frames ways of thinking, being, and doing within the professional spaces of academia and that outlines how to construct these spaces with a disabled-conscious mind-set. It is this kind of approach that enables and empowers and that as individuals and institutions we need to strive for. We all need to become more aware and conscious of what it means to be disabled in higher education and of how we can be allies in that environment. The main recommendation resulting from our work therefore is learning to listen. Data transparency Not applicable. Funding The conference Ableism in Academia was financed through sponsorships from University College London, UCL Institute of Education, UCL Arena, University of Kent, University of Leeds, and the University of Nottingham. --- Declarations Ethical approval Institutional ethical approval for the use of the data and the direct approaching of individuals for follow-up conversations or interviews was sought and formally granted. --- Competing interests The authors declare no competing interests. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. --- Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

To understand the experiences of the disabled in academia, a fully accessible and inclusive workshop conference was held in March 2018. Grounded in critical disability studies within a constructivist inquiry analytical approach, this article provides a contextualisation of ableism in academia garnered through creative data generation. The nuanced experiences of disabled academics in higher education as well as their collective understandings of these experiences as constructed through normalisation and able-bodiedness are presented. We show that disabled academics are marginalised and othered in academic institutions; that the neoliberalisation of higher education has created productivity expectations, which contribute to the silencing of the disabled academics' perspectives and experiences due to constructions of normality and stigmatisation; and that it is important to enact policies, procedures, and practices that value disabled academics and bring about cultural and institutional changes in favour of equality and inclusion.

A moderate midlife body weight is associated with favorable health outcomes including lower morbidity and extended longevity (Breeze et al. 2006). Consequently, the dramatic increase in the overweight and obese population since 1960s has been a major public health concern in the United States (Wang and Beydoun 2007). In fact, more than two thirds of US adults are overweight (BMI>=25) and about 38% are obese (BMI>=30) (Flegal et al. 2016), and American women today are obese for a longer duration of their lives (Lee et al. 2010). Additionally, recent reports suggest that the obesity rate is highest among middle-aged American women (early 40s to mid 60s) relative to women in all the other age groups (Mendes 2010). Sociologists argue that body weight variations are conditioned by social contexts, yet, the social determinants that are associated with body weight among midlife American women remain understudied. One important social context for midlife women that may shape body weight is their coresidential and non-coresidential relationships with their adult children. Nest-leaving of adult children is a historically key experience for midlife parents. Over the past two decades, however, demographers have noted drastic changes in home departure behavior, including high rates of what are known as "boomerang" children who return to the maternal home after departing, as well as the "never-left" who have remained in the maternal home longer than expected. Boomerang and never-left coresidential patterns appear to buffer young adults from economic shocks (e.g., economic recession, unemployment) and do not negatively affect young adults' emotional and psychological well-being (Copp et al. 2015;Fingerman et al. 2016). Yet, what remains unknown is how these changing coresidential patterns affect the well-being -including body weight-of the other family member in the coresidential home: mothers. A life course perspective argues intergenerational coresidential patterns are very likely to affect mothers' body weight due to the generations' linked lives (Umberson et al. 2011). A stress perspective further suggests that midlife mothers may experience coresidential patterns with their adult children as stressful, which in turn may affect body weight (Milkie, Bierman, and Schieman 2008). Yet, virtually no empirical research examines the potential relationship between these emerging coresidential patterns and midlife maternal body weight. In order to fill this research gap, this study asks: How are intergenerational coresidential biographies during mother's midlife associated with one key health outcome: mothers' body weight trajectories? Body weight index (BMI) is a key indicator of health at midlife and is sensitive to life course changes including the parent-child relationship, and thus provides the ideal lens to examine the proposed effects of coresidence (Umberson et al. 2011). This study sheds light on the intricate link between notable changes in family demographics and an important dimension of health among midlife women. --- BACKGROUND Adult weight trajectories follow a slight upward trend across the life course (Adams and Schoenborn 2006), wherein a relative moderate body weight (non-overweight) with a stable (or slightly increasing) trajectory suggests good and stable health whereas a dramatically fluctuating body weight suggests significant health change in response to various negative life events (Baltrus et al. 2005). Previous research has established that motherhood is strongly associated with body weight changes across the life course, with this research focusing on the effects of parenting young children on parental body weight. For example, Laroche and colleagues (2013) find that living with young children (under five years old) is associated with accelerated weight gain among mothers relative to fathers and childless women across their young adulthood (18 to 30). This is likely because parenting young children is associated with a higher-calorie diet (e.g., higher levels of fat, sugar, and meat) and less physical activity (Berge et al. 2011;Elstgeest, Mishra and Dobson 2012;Nomaguchi and Bianchi 2004). In one of the few studies that examined the effects of parenting and mothers' body weight as children age into adulthood, Umberson and colleagues (2011) found coresidence with adult children was associated with higher body weight at the baseline, further showing that the departure of adult children is associated with greater weight gain for both fathers and mothers. Yet, Umberson and colleagues focused on an older cohort of parents (born between 1890 and 1962) and did not examine how young adults' emerging coresidential patterns were related to mothers' body weight change for the current cohort of midlife women. Due to drastic changes in both coresidential patterns and body weight trajectories for the current cohort of midlife mothers, a focused look on body weight and motherhood for midlife parents today is imperative. More specifically, this prior research (Laroche et al. 2013;Umberson et al. 2011) conceptualizes the association between body weight and parenthood as a gradual and incremental process by modeling body weight as body mass index (BMI). Additionally, the effects of stress on BMI are well defined and provide an additional plausible pathway through which young adult coresidential trajectories are likely to impact mother's wellbeing (Torres & Nowson, 2007;Block et al., 2009). Therefore, our study examines mother's BMI changes across midlife in relation to mother-child coresidential patterns. In order to theorize this relationship across different coresidential patterns, below we outline life course and stress process perspectives that likely underlie coresidence and maternal body weight. --- Intergenerational Coresidence and Mother's Body Weight Intergenerational coresidence trajectories are efficient indicators of children's transitioning pathways to young adulthood (Houle and Warner 2017;Sassler, Ciambrone and Benway 2008). These patterns are uniquely established during mother's midlife, and thus are likely key to maternal body weight changes during this life stage (Grossman 2005;Newman 2012). Three primary types of intergenerational coresidential patterns have gained prevalence in the past two decades in the U.S. and are likely to have differential impact on maternal body weight: (1) children who left the parental household and never returned between ages 18-30 (i.e., the "gone-for-good"), (2) children who exited and came back at least once between the ages of 18-30 (i.e., "boomerangers"), and (3) children who did not exit parental household between the typical years of exit, 18-30 (i.e., prolonged coresidence, or the "never-left") (Burn and Szoeke 2016;Copp et al. 2015;Newman 2012). We choose age 18 and 30 as the beginning and end points to define coresidential patterns among young adults for a number of reasons. Age 18 is selected as a starting point of a coresidential trajectory because this is the modal age at which U.S. young adults graduate high school and either begin to work full-time, attend college, or both (Payne, 2012). According to a life course perspective, age 18 is also considered to be the age at which emerging adulthood begins (Arnett, 2007). Age 30 is determined to be the endpoint for our measures of coresidential histories because by age 30, most young adults who attend college have either graduated or dropped out as well as made the first transition out of or back into their parents' household, if they are going to leave at all (Parker 2012). Next, we theorize that these three intergenerational coresidential patterns serve as a key dimension of parent-child tie for midlife mothers as these patterns distinctly structure maternal-child bonds with important implications for body weight. At midlife, these three intergenerational coresidential patterns theoretically influence mother's body weight in different ways due to the different contexts that each coresidential pattern brings. For example, body weight is very sensitive to perceived stress and psychological distress (Torres and Nowson, 2007), and women's body weight is especially responsive to family-related strains (Block et al., 2009). Intergenerational coresidential patterns are associated with distinct life events and transitions of young adults, which entail different stressors that further influence mother's body weight. For example, current research shows that mothers' mental wellbeing decreases when adult children encounter problems or negative life events (e.g., divorce, unemployment) via a process of intergenerational stress proliferation (Milkie et al. 2008;Greenfield and Marks 2006;Kalmijn and De Graaf 2012). Below we theorize how the three coresidential patterns may be differentially associated with mother's body weight in different ways via the contextual changes and stress processes associated with each coresidential pattern. The gone-for-good.-Despite the increasing delay in home departure (Furstenberg, Rumbaut, and Settersten 2005;Newman 2012), almost all young adults agree that the ideal time to leave parental household is by age 25 (Settersten 1998). In a report that uses data from NLSY97-Millennial cohort of emerging adults in the U.S., 47% of men and 46% of women leave parental household after 18 and do not return by age 24 (Payne 2011). The majority of these young adults move in with a spouse or cohabiting partner, with the remainder either live alone or enter semi-autonomous living such as dormitories and military barracks, or move in with other family members (Goldscheider and Goldscheider 1999). Many young adults also relocate geographically away from their parents because the educational or employment opportunities are in another location, and this transition is viewed as a positive development from a mother's perspective (Goldscheider and Goldscheider 1999). In general, independently-living young adults are more likely to be employed and earn more than those who live with parents (Payne and Copp 2013). The gone-for-good coresidential pattern may be most favorable for midlife mothers' body weight. Gone-for-good children take the desired and expected pathway of departing from the home and not returning within a normative timeline, and thus mothers may feel a reduction in stress due to a sense that they have appropriately prepared their child for a successful launch into adulthood. Additionally, the gone-for-good children may decrease mother's chances for body weight increase relative to other groups because of the reduction of daily stressors that come with intergenerational coresidence with adult children-stressors that are important biosocial mechanisms that influence body weight (Pearlin et al. 2005). For example, the transition to an empty nest is associated with reduced daily hassles (Dennerstein et al. 2002), reduced parent-child conflicts (Ward and Spitze 2007), and decreased work-family conflicts (Erickson, Martinengo, and Hill 2010), all of which are important stressors that can link with body weight. Additionally, parents tend to have more time and resources for self-care and leisure after adult children transition to independence (Henderson et al. 1996), and parents may not be influenced by unhealthy eating and exercise habits by their non-coresidential children. A stress perspective further suggests that in the absence of their child, mothers to a gone-for-good young adult child may be better able to attend to their own wellbeing such as engaging in regular physical activity and healthy eating. In other words, a gone-for-good coresidential pattern means fewest stressors and more resources to promote healthy behaviors; as a result, mothers of the gone-for-good may have the most moderate midlife body weight trajectory among all three groups. The boomerangers.-In this study, boomerang children are defined as those who have left the maternal home after 18 but return before the age of 30. According to a nationwide survey of 2,048 adults conducted in 2011, 40 % of 18-to-24-year-old boomerang after the initial departure (Parker 2012) -this does not include boomeranging during college enrollment, but does include children who return home after dropping out of or completing college. Postponement in young adult role transitions (e.g., transition to marriage and home ownership) and deficiency in young adult's economic resources-such as income loss, a lack of employment opportunities, or the greater cost of sustaining independent livingare associated with higher likelihood of returning home (Houle and Warner 2017;Sassler et al. 2008). Compared to mothers of the gone-for-good, mothers of the boomerangers likely encounter more stressors that may impact body weight. For example, boomerang mothers may feel stress and distress because they feel they did not adequately prepare their child to live independently. Moreover, boomerangers' mothers may experience additional stressors and health habit adjustments as children's transition back to the nest, such as the negotiation of housework division and financial responsibilities (Sassler et al. 2008). These additional stressors and time constraints accompanied by children's return may in turn increase mother's unhealthy behaviors such as irregular food intake and disrupted routines in physical activity, increasing body weight. Children may also bring back to the home unhealthy eating and exercise habits, which may in turn influence maternal body weight. Consequently, midlife mothers of the boomerangers may have higher body weight than mothers of the gone-for-good. The never-left.-Recent research on intergenerational continuous coresidence into children's young adulthood highlights a third less common, but increasingly important subgroup of adult children: those that have not left the parental household prior to age 30. For instance, using a sample of 1,113 older mothers aged 58-to-97, Wiemers and colleagues (2017) find that 6.85% of women had never-left adult children in their household at the time of the survey. Although there is little known about this group of young adults and their parents, young adults who are male, unmarried, unemployed, economically needy, and those who are working on their post-secondary education are more likely to be never-or later-leavers (Burn and Szoeke 2016;Goldscheider and Goldscheider 1999;Sassler et al. 2008). Because this is a particularly non-normative status, mothers may feel the stigma and strain of housing a never-left adult child, increasing stress and in turn body weight. Additionally, the conditions of continued coresidence with adult children who never leave may shape body weight trajectories. Research has shown that mothers tend to prioriti e children's wellbeing and engage in unhealthy behaviors more (e.g., eating less healthy food and working out less) when children need more support (Reczek et al., 2014). Compared to mothers of the gone-for-good, mothers of the never-left are exposed to the daily stressors for a longer duration of time and likely engage in more unhealthy behaviors, which can take a toll on their body weight. Taken together, due to differential levels of stress proliferation and varying motherhood contexts across coresidential patterns, we hypothesize that: H1: Mothers of the gone-for-good will experience the lowest BMI score at baseline and the most moderate increase in their BMI trajectory among the three groups. Furthermore, while mothers of the boomerangers adjust to additional stressors and constraints associated with children's transitioning back home, mothers of the never-left may need to manage an extended situation wherein children lack progress in transition to adulthood. As a result, mothers of the boomerangers and mothers of the never-left likely have distinct body weight trajectories reacting to different stressors. Additionally, although both of these non-normative coresidential patterns may be interpreted as mother's setbacks in launching adult children and subsequently increase maternal stress and time constraints, young adults' return rather than continual coresidence may present itself as a more acute stressor, inducing more dramatic maternal weight gain across midlife. We thus hypothesize that: H2: Mothers of the boomerangers and mothers of the never-left will have similar baseline BMI scores but the former group may gain more weight across midlife than the latter group. --- DATA AND METHODS --- Description of the Data This study uses data from two prospective cohort studies -the National Longitudinal Survey of Youth 1979 and the corresponding 1979 Young Adult Survey-to investigate the impact of intergenerational coresidential biographies on mother's BMI across midlife. In 1979, 12,686 NLSY79 respondents, aged 14-22, were randomly selected and followed over time. Participants were interviewed annually between 1979 and 1993 and biennially since 1994. The most recent wave of data available is from 2014. In 1986, data collection efforts began on all children born to NLSY79 mothers to gather information on a wide range of social, economic, and psychological characteristics. The NLSY79 children, once turned 14, participate in NLSY79-YA survey conducted biannually until 2014. In addition to the initial intention of studying the baby boomers' labor market experiences as they transitioned from adolescence into adulthood, the NLSY79 are among the best datasets with which to examine how the health of both mothers and children change over the life course. For the current study, we combined data from female members of the original NLSY79 cohort with information from their young adult children (NLSY79-YA) to examine how patterns of coresidence shape mother's body weight during midlife. --- Measures Dependent variable-The outcome measure is body mass index (BMI), which is calculated by dividing self-reported maternal weight (in kilograms) by height (in meters) squared. Self-reported weight and height have established validity in nationally representative surveys, largely due to the ease and accuracy with which respondents from a wide variety of backgrounds can recall them (Gorber et al. 2007). We modeled BMI as a continuous variable, which allowed us to examine the trajectory and degrees of change in BMI as it varies incrementally over time (Umberson et al. 2011). Because change in BMI for the middle-age women is primarily driven by change in weight, we thus use "body weight" or "weight" to refer to the dependent variable in subsequent sections. Independent variable-To construct our primary independent variable, coresidence trajectories, we relied on information from NLSY79-YA respondents who were between the ages of 18 and 30 years. In each survey year, young adult respondents were asked who they reside with, and were given the option to signify whether they were living in the household of mother, with both parents, in their own dwelling unit, in dormitory, fraternity or sorority, or in other living arrangements. To minimize coding error in the field, NLSY79 staff constructed a residence type variable from a series of residential questions and recommended it to researchers interested in residential patterns. We deployed answers of this residence type variable from young adults since they first entered the survey of NLSY-YA starting at age 14. Young adults who reported to have lived in a household with their mother or both parents between age 14 and 18 were considered to be children that have grown up in a maternal household and are thus included in the sample. As NLSY79-YA survey is to collect information from children born to all the NLSY79 women, 99.9% of the young adults in the data grew up in a maternal household. Young adults who did not grow up in a maternal household were deleted from the sample. Young adults' first exit was captured when they reported the first episode of residence in a household without their mother since age 18 1. A return to the maternal household after the first exit was considered a boomerang event. Based upon this year-specific measure, we categorized NLSY79 young adults into three categories to describe their coresidential histories during emerging adulthood. Those who left their maternal home once after age 18 and never returned before age 30 were classified as "gone-for-good." NLSY79-YA respondents who exited from their maternal home after age 18 and transitioned back by age 30, either once or multiple times, were placed in the "boomerang" category. As for college-attending young adults who reported to be "living in dormitory, fraternity, or sorority" and those that were "living in own dwelling unit" in the original residence type variable 2, we followed prior research (Parker 2012; South and Lei 2015) and considered them living outside of maternal household. These young adults were thus placed in either the "gone-for-good" or the "boomerang" category depending on their coresidential history. YA children who consistently reported living in their mother's household between the ages of 18 and 30 were designated as part of the "never-left" group. Although most of the never-left adult children might ultimately become gone-for-good or boomerangers as they aged, we argue that it is important to keep these three categories separate for a number of reasons. First, the never-left group was designed to capture young adults' continued or prolonged residence in the maternal household when their mothers were at midlife. Following the life course perspective that takes the timing of events seriously for 1 A close scrutiny of young adult's residential trajectory shows that about 30% of the young adults in the sample exited maternal household prior to age 18. Sensitivity analysis that includes an early-exit variable shows qualitative similar results as those presented in the text. 2 1% of young adults in the sample reported to be "living in dormitory, fraternity, or sorority" and 9% reported to be "living in one's own dwelling" while enrolled in regular school. the focal person, our study examines one particular life event-mother-child coresidential pattern during mother's midlife and how this relates to mother's wellbeing for one cohort of women. One of the main coresidence patterns is a never-left pattern. Second, the experience of housing an adult child who has never exited for a midlife mother may come with unique psychosocial stressors and, thus, result in worsening health for the mother, including but not limited to unhealthy weight gain and an increased risk of obesity. In subsequent sensitivity analyses (available upon request), we divided the "boomerang" category into two additional groups-those who remained within their mother's residence upon returning home and those who, returned to the maternal home after leaving but at some point in the future left their maternal home for a second time. Findings remained remarkably consistent across models. Control variables-We accounted for a number of potential confounders in multivariate regression analyses that could be associated with both intergenerational coresidence patterns and mother's BMI. We included sociodemographic and SES-related control variables for NLSY79 mothers. They are as follows: race/ethnicity (Non-Hispanic White, Non-Hispanic Black, Hispanic), marital status (never married, married or cohabiting, other), region of residence (northeast, north-central, south, and west, to address the influences of housing prices on coresidential patterns), highest grade completed, family income (adjusted for inflation and reported in 2014 dollars, and transformed to address positive skewness), number of persons in the family which was top coded at six plus, employment status, and health insurance coverage (private, public, self-pay). In order to account for additional stressors that are not linked directly to the stressors of coresidence, we included a set of control variables that represented young adult's major life events and transitions. These variables include sex (male/female), marital status (never married, married, cohabiting, other), parental status (yes/no), educational attainment (below high school, high school, college or above), and employment status (neither in school or working, in school, working, in school and at work). Most of these measures are time-varying; the exceptions are mother's race/ethnicity, education, and young adult's gender. Therefore, any findings we show are more likely to be related to coresidence dynamics rather than external forces that sort individuals into distinct coresidence patterns. Analytical Approach-We deployed multilevel regression models (i.e., growth curve models) to examine how young adult's coresidential biographies influence midlife mother's BMI. Multilevel analysis is appropriate for this study because of its ability to capture how BMI fluctuates within as well as across respondents over time. As a result, multilevel models are well positioned to distinguish between-individual variability from within-individual variability. To this end, we estimated a series of multilevel regression models predicting how BMI among NLSY79 mothers fluctuates across midlife based upon the following equation: y ij = <unk> 1 + <unk> 2 <unk> j + <unk> 3 t ij + <unk> 4 <unk> j t ij + <unk> 5 x 5ij... + <unk> n x nij + <unk> 1j + <unk> 2j t ij + <unk> ij where y ij represents BMI of mother j at time i, <unk> 1 is the population-averaged intercept (or the initial BMI score across all mothers), <unk> j captures mother-child coresidential pattern, t ij is the age of mother j at time i, x 5 -x n stands for a series of additional time varying or time invariant covariates, <unk> 1j is the random intercept, <unk> 2j is the random slope, and <unk> ij refers to the individual-and time-specific error term. Note that the time unit for the analysis is mother's age. Robust standard errors were calculated using the Huber/White correction estimate and clustered at the highest level of aggregation (the original NLSY79 family). Regression models were estimated using the mixed command in Stata 15.0. We used multiple imputation methods with chained equations to deal with issues of missing data. These procedures use a diverse set of predictors to estimate multiple sets of responses for each missing observation. Imputed values include a random component based on draws from the posterior predictive distribution of the missing data under a posited Bayesian model which, under the missing-at-random assumption, provide unbiased estimates of variance (Allison, 2001). According to the suggestion in extant literature that three to five datasets are appropriate for models with up to 20% missing, we produced five distinct datasets (van Buuren, 2012;Royston, 2005). Guided by previous research (Von Hippel 2007), we first ran the imputation model including observations with missing values on BMI, but then deleted respondents who had missing BMI from subsequent descriptive and multivariate analyses. The proportion of missingness for all variables fell below the suggested cut-point of 5% (Schafer 1999) except for family income, for which 13% of observations lacked valid responses. To address concerns related to the high proportion as well as the nonrandom distribution of missing data on this variable, we performed sensitivity analysis excluding family income from all imputations and subsequent regression analyses. Results were consistent across model specifications. The NLSY79 contains information on 3,301 mothers who had at least one child 18 years or older by 2014, the last year for which data is available. All the interviewed young adults, some of whom were siblings with the same mother (discussed below), had valid information regarding their living arrangements. We excluded 10 NLSY79 mothers whose young adult children were either institutionalized or had never lived with mother during childhood. 36 NLSY79 mothers were dropped from the analytic sample due to missing values on BMI. Furthermore, 214 of values for weight and/or height from 82 NLSY79 mothers were extreme, defined here as greater than 3 standard deviations from the mean across all NLSY79 mothers in a given year and were likely attributable to respondent error or interviewer miscoding. To address this issue, we replaced these extreme values with person-specific mean BMIs. We conducted a series of sensitivity checks and found that this strategy to deal with the outliers did not influence our results. All told, these inclusion criteria generated an analytic sample of 3,255 mothers and 7,197 young adult children for whom we have prospective data from 1994 through 2014. Thus, the unit of analysis in the current study is person-years (n=27,903). The NLSY79 young adult survey contains information on multiple children born to the same mother. In order to estimate the association between mother-young adult coresidential patterns and fluctuation in maternal BMI across midlife, we chose to include data on all siblings. An alternative would be to randomly select one child per mother, include only oldest children, or include only youngest children. Retaining all siblings in the analytic sample does not require that we fully understand all of the underlying complex family dynamics that could give rise to differences in the young adult coresidential and maternal health relationship. In some families, the experiences of the oldest child might more strongly influence a mother's risk of high body weight, while in others it could be the experiences of the youngest child. Randomly selecting a focal child for each family would thus introduce other sources of unknown bias if the selected child did not correspond to the one that most forcefully impacted maternal health. Further, it is more likely that it is some combination of mother-child corresidence patterns with all siblings within a given family that has the biggest impact on mother's body weight changes. By including all children born to each NLSY79 mother, we are able to maximize the amount of information regarding young adult transitions for each NLSY79 respondent. This approach mirrors scholarship that has examined the intergenerational transmission of health outcomes and behaviors (Classen 2010;Weden and Miles 2012) and research that has studied the effects of parenting experiences on maternal as well as children's health (Colen and Ramey 2014; Zilko et al. 2010). To determine if our decision to include multiple YA children for each NLSY79 mother resulted in biased findings, we conducted sensitivity analyses in which we randomly selected one YA child from each NLSY79 family. These results, which are presented in Appendix 1, closely resemble those generated by regression models inclusive of all siblings and suggest that our results are robust to this type of model specification. --- RESULTS --- Descriptive Findings Descriptive results for the full sample and the sample stratified by parent-child coresidential biographies are presented in Table 1. Significance tests (p <unk> 0.05) were conducted to assess differences across intergenerational coresidential patterns. The majority of children in the full sample (61%) adhered to coresidential trajectories that could be classified as gone-for-good (GFG), while 22% and 17% fell into the boomerang (BMG) or never-left (NL) categories, respectively. Mean BMI for the full sample of NLSY79 mothers was 29.10 kg/ m 2. Comparing across the three coresidential groups, we find mothers of the BMG and the NL to have similar BMI scores of 29.65 kg/ m 2 and 29.51 kg/ m 2, respectively, which were significantly higher than BMI scores among mothers of the GFG (28.79 kg/m 2 ). Mothers of young adults with distinct coresidential biographies appear to differ on several key characteristics. Compared to mothers of the GFG young adults (YAs), boomerangers' mothers were slightly younger, less likely to be non-Hispanic White, and less likely to be married or cohabiting. Mothers of the NL young adults were also more likely to be African American and Hispanic, less likely to be previously married, and more likely to live in the Northeast than mothers of the GFG category. In comparison to mothers of the GFG, boomerangers' mothers had lower SES (e.g., fewer years of completed education, lower family income, and higher rate of public insurance), whereas mothers of the NL had similar or even higher SES. YAs also differed from one another by intergenerational coresidence. Compared to the GFG YAs, a smaller proportion of boomerangers and the never-left were married or had children and completed fewer years of schooling. --- Multilevel Regression Results Results from multilevel regression analyses that predict BMI trajectories among NLSY79 mothers are shown in Table 2. We centered mother's age to generate coefficients that represent mothers' mean BMI score at 40 years, an age that is typically considered to be the beginning of midlife (Lachman et al. 2015). Models 1A and 1B control for an extensive set of maternal sociodemographic covariates, while Models 2A and 2B adjust for additional key child characteristics that could potentially confound the association between YA coresidential trajectories and maternal BMI. Models 1A and 2A use mothers with GFG children as the reference group, while Models 1B and 2B use mothers with NL children as the comparison group. According to Model 1A, mothers of the NL and BMG were, on average, 0.620 kg/m 2 (p <unk> 0.05) and 0.467 kg/m 2 (p <unk> 0.05) heavier than mothers of GFG, respectively, at age 40. This translates to a weight gain of 4 pounds and 3 pounds, respectively, for an adult woman with a height of 5 feet 4 inches and an original weight of 173 pounds, the average height and weight of American women aged 40-49 between 2011 and 2014 (CDC, 2014). Results in Model 1B reveal that the baseline weights between the NL and BMG mothers were not significantly different (b = -0.153, p > 0.05). Model 1A also reveals how average BMI scores for all three coresidential groups fluctuated across midlife. BMIs for GFG mothers increased at a rate of 0.142 kg/m 2 per year (b = 0.142, p <unk> 0.001). Results for interaction terms (coresidential group*age) reveal significant differences in the rate of change over time between GFG and NL mothers, but not GFG and BMG mothers. NL mothers could only expect their BMIs to increase by 0.068 per year (0.142 -0.074). In other words, a GFG mother with a height of 5 feet 4 inches and an original weight of 173 pounds would gain an average of 8 pounds over 10 years, whereas her counterpart with a NL child would gain about 4 pounds in the same period of time. The small magnitude of the coefficient for the boomerang*age interaction suggests that BMG mothers gained weight at a similar rate across midlife as GFG mothers. Results from Model 1B show that BMG mothers' BMI scores also increased more than NL mothers across midlife (b = 0.073, p <unk> 0.001). We controlled for YA's as well as maternal sociodemographic characteristics in Models 2A and 2B, Table 2. Compared to results in Model 1A, Model 2A shows that the coefficient of the difference in BMI at age 40 between NL and GFG mothers increased from 0.620 to 0.884, suggesting that differences in YA children's sociodemographic characteristics suppressed NL and GFG mothers' BMI difference at age 40. However, the difference in the rate of change in BMI between these two groups decreased slightly from -0.074 to -0.066 kg/m 2 per year (p <unk> 0.001) once we accounted for additional child controls. In other words, a GFG mother with a height of 5 feet 4 inches and an original weight of 173 pounds would gain an average of 4 pounds over 10 years, whereas the weight of her counterpart with a NL child would remain stable in the same period of time. Subsequent analyses (available upon request) reveal that BMI scores of GFG and NL mothers converged during their early 50s

Midlife mothers report their children returning to the maternal home after departing (i.e., boomerang children) and remaining in the maternal home longer (i.e., never-left children) than the past half century. Over the same time period, the percent of Americans considered overweight and obese have increased. Yet, we know very little about how such delays affect the body weight of mothers. The current study uses the National Survey of Youth 1979 (NLSY79) and its corresponding young adult sample (NLSY79-YA) across 20 consecutive years (N=7,197) to determine if extended coresidence with an adult child is associated with midlife mothers' body weight changes. Results from multilevel regression models show that compared to mothers whose young adult children left home and never returned ("gone-for-good"), mothers of the "never-left" had higher body weight at 40 but similar body weight at 50. Mothers of the boomerangers had higher body weight relative to mothers of the "gone-for-good" across midlife. Mothers of the boomerangers and mothers of the "never-left" had similar weight at age 40 but the former group had more weight gain across midlife. These findings lend new insight into how different patterns of mother-young adult coresidence likely affect the health of mothers, and suggests the effects of recent demographic trends such as "failure to launch" on family formation and functioning should be viewed holistically with a more inclusive sociological lens.

NL mothers across midlife (b = 0.073, p <unk> 0.001). We controlled for YA's as well as maternal sociodemographic characteristics in Models 2A and 2B, Table 2. Compared to results in Model 1A, Model 2A shows that the coefficient of the difference in BMI at age 40 between NL and GFG mothers increased from 0.620 to 0.884, suggesting that differences in YA children's sociodemographic characteristics suppressed NL and GFG mothers' BMI difference at age 40. However, the difference in the rate of change in BMI between these two groups decreased slightly from -0.074 to -0.066 kg/m 2 per year (p <unk> 0.001) once we accounted for additional child controls. In other words, a GFG mother with a height of 5 feet 4 inches and an original weight of 173 pounds would gain an average of 4 pounds over 10 years, whereas the weight of her counterpart with a NL child would remain stable in the same period of time. Subsequent analyses (available upon request) reveal that BMI scores of GFG and NL mothers converged during their early 50s. Regarding comparisons between GFG and BMG mothers, our results remained remarkably consistent across Models 1A and 2A, suggesting that child characteristics might explain more of the BMI difference between NL and GFG mothers compared to that between BMG and GFG mothers. Model 2A suggests that accounting for child characteristics, BMG mothers had a higher BMI score than GFG mothers at age 40 but these two groups experienced similar rates of weight gain across midlife. Results from Model 2B suggest that compared to NL mothers, BMG mothers had similar BMI scores at age 40 (b = -0.423, p > 0.05) but they experienced more weight gain over time (b = 0.061, p <unk> 0.01). Predicted values from Model 2A were plotted in Figure 1. NL mothers had higher BMI scores than GFG mothers at the beginning of midlife (age 40). Due to differences in weight gain over time across coresidential groups, the discrepancy in BMI between GFG and NL mothers gradually decreased and ultimately converged in their early 50s. BMG mothers, on the other hand, had significantly higher BMIs at age 40 than the GFG mothers but experienced similar weight gain over time. NL and BMG mothers had similar body weights at age 40, but BMG mothers gained significantly more weight across midlife than NL mothers. --- Sensitivity Analyses We ran multiple sensitivity analyses to test the robustness of our findings. First, we considered whether our results were driven by maternal and/or child health status. YAs with mothers in poor health might be less willing or able to leave maternal home as other YAs, falling in the NL or BMG category, if they were needed to help care for a sick or aging mother. However, previous research has shown that mother-child coresidential patterning is more dependent on the needs of YA children than their mothers at this particular life course stage (Fingerman et al. 2012b). To this end, we ran additional sensitivity analysis that included measures of mothers' and YAs' self-rated health and found our results to be qualitatively similar (Appendix 2). Second, we assessed the extent to which maternal race/ethnicity, education, family income, and employment status predicts changes in BMI scores across midlife and whether this alters our findings about intergenerational coresidence and maternal BMI (Appendix 3). According to these analyses, the maternal race/ethnicity and family income were predictive of mother's BMI changes across midlife, yet the findings about international coresidence and mother's BMI remained similar to those presented in Table 2. Third, to determine if the association between children's coresidential patterns and mothers' BMI was driven by younger NLSY79 YAs who had not yet had time to fully establish a pattern of home leaving, we conducted sensitivity analyses that included only mothers whose YA children had at least surpassed age 22 years by the first wave of data collection (Appendix 4). These results are comparable to those presented in Table 2. --- DISCUSSION The rise of overweight and obese statuses in the general U.S. population has drawn significant research attention, yet the underlying social determinants of body weight at midlife remain under-addressed. Research is especially needed on the body weight of midlife women because women tend to be obese for a more prolonged period of time than men, and midlife is the life course stage with the highest obesity rate. Midlife motherhood is likely a key determinant of body weight given the increasingly diverse set of young adult children who take multiple pathways to independence, and the present study is the first to examine how mother-adult child coresidential patterns are associated with mother's body weight trajectories at midlife. Results from multilevel models show that intergenerational coresidential patterns are associated with both baseline BMI differences as well as differing BMI trajectories among midlife mothers. Below, we discuss our findings and highlight our main contributions and implications for future research. First, our findings suggest that intergenerational coresidential biographies matter for midlife mothers' body weight both at baseline and over time, even when all covariates including adult children s life events are included. In support of hypothesis 1, findings show that mothers with gone-for-good children experienced the lowest baseline BMI; at 40, having a child who has left the home and not returned is associated with lower BMI than midlife mothers whose children boomeranged back into the home or whose adult children had never left the home. But mothers of the boomerangers continued to have higher body weight than mothers to the gone-for-good until late 50s. This suggests that mothers of the boomerangers experience more disadvantage relative to our predictors of BMI. Upon the return to maternal household, boomerang children may increase a mother's feelings of maternal inadequacy, which may relate to maternal body weight change. Additionally, this transition may (re)introduce daily stressors, including routine hassles, parent-child conflicts, and work-family strains to their mothers' lives (Dennerstein et al. 2002;Ward & Spitze, 2007;Erickson et al., 2010), which in turn may increase maternal body weight. Alternatively, as the rate of change in body weight between early 40s and late 50s did not vary between mothers to the gone-for-good and mothers to boomerangers, it is possible that the driving force of such difference is a pre-existing selection factor. For example, it may be that other sustained stressors or SES characteristics unable to be included in the study underlie these processes, all of which warrants future research. Second, and inconsistent with Hypothesis 1, we found that the body weight of mothers of the never-left and the gone-for good converged in mothers' early fifties. Such convergence may reflect the fact that a constant coresidential status-regardless of what that status ismay reduce the negative effects of any coresidential pattern. For example, while mothers may at first be stressed by an adult child failing to leave the nest, mothers may get used to the presence of the never-left and over time experience diminished effects of stress on BMI. In fact, current work has shown that the young adults who stably reside with parents report similar mental wellbeing as those who remain stably independent (Caputo, 2018), which suggests that the never-left may not introduce additional stressors to their mothers by being mentally distressed themselves. Moreover, coresidential stigma perceived by never-left's mothers has lessened, as extended coresidence with adult children has become increasingly common for the past two decades (Fry, 2016). Similarly, current research has found that after the Great Recession that occurred in the late 2000s, coresidence with adult children is no longer associated with lower marital quality among parents as it once was (Davis et al., 2016). As a result, mothers to the never-left young adults, many of whom transitioned to adulthood during the post-recession years and were attending school to increase their employability at the time, may develop more understanding of their children, experience less strain over time, and thus gain less weight over time than mothers of the gone-for-good. Third, we found support for Hypothesis 2, which theorized the mothers of the never-left and the boomerangers would have similar body weight at early 40, but that the mothers of the boomerangers would gain more weight by their mid 50s. This once again suggests that boomerangers' mothers may experience the worse effects on their body weight. One major difference between mothers of the never-left and mothers of the boomerangers is that the former group had a continued intergenerational coresidence with the focal child whereas the latter did not. Previous literature establishes that instability in family life increases stress and decreases wellbeing (Cooper et al., 2009). A young adult's return may shift the diet and exercise habits of mothers so as to accommodate young adult's schedule and habits; this may induce weight gain that would otherwise not take place if these coresidential changes did not occur. --- Limitations Although the current study contributes to our understanding of the health effects of intergenerational coresidence patterns, findings should be interpreted in light of certain limitations. First, while all NLSY79 mothers in our analytic sample have entered midlife, reflecting the study's primary aim of examining predictors of women's health and wellbeing during this critical life course period, not all of their YA children have reached age 30 by the latest available wave (2014). This is primarily a result of heterogeneity in fertility timing that led a small proportion of NLSY79 mothers to postponed childbearing. Thus, for a small subset of NLSY79 YAs, we were only able to capture coresidential trajectories for some but not all of the time period between 18 and 30 years of age3. We recognized this issue and conducted sensitivity analysis (Appendix 4) that included only mothers with YA children aged between 22 and 30. We found the results comparable to those presented in Table 2. Yet, extending mother's age up until later life so as to include a more detailed intergenerational coresidence trajectory for most mother-YA children pairs (if data permits) can invite other conundrums, one of which is that at later life, intergeneration coresidence is shifting to meet the needs of aging mothers' instead of the children's. This requires additional theoretical frameworks and is beyond the scope of our study, but warrants attention from future research. Second, this study cannot fully control for selection into the three different coresidential groups. As is evident from our descriptive results, mothers of boomerang YAs were more likely than gone-for-good YAs to be African American, less educated, and had lower family incomes, while mothers of never-left YAs reported more years of completed schooling and higher family incomes. Although we did control for these factors, along with other, key maternal and child sociodemographic characteristics that could potentially confound the association between intergenerational coresidence trajectories and maternal BMI in multivariate models, standard regression techniques -even those as rigorous as the ones employed in the current study -are ill equipped to completely account for selection into different patterns of home leaving for YA children. Unfortunately, despite our longitudinal study design, fully addressing issues of selection is beyond the scope of the current paper since doing so would require a markedly different analytic approach (e.g. one that uses a natural experiment to approximate random allocation into the three types of YA coresidential patterns). Future research should assess how selection plays a role in the relationship between intergenerational coresidence and midlife mothers' body weight trajectory. Third, our measure of YA coresidential transitions does not perfectly capture moves in and out of the maternal household due to data collection taking place every other year as opposed to every year; thus, we might not capture short-term (within two years) fluctuations in residential status. However, our study follows the best practice to capture coresidential patterns with the contemporary survey data, that is-using individuals' self-report of residence rather than household roster (Houle & Warner 2017). We call for future research to address this measurement issue with more detailed data on transitions of intergenerational coresidence. Lastly, we were only able to assess the impact of intergenerational coresidential trajectories on mother's BMI, not father's BMI. This was largely a result of how the initial NLSY79-YA sample was constructed. All children born to female, but not male, NLSY79 respondents were included in the young adult cohort and followed over time. Thus, the extent to which gone-for-good, boomerang, and never-left children disproportionately affect their midlife father s body weight, and subsequent risk of obesity, remains an important empirical question that should be tackled by future research. --- Conclusion Despite the limitations, this study expands prior research on the links between the motherhood contexts of midlife and women's well-being. Taken together, these findings extend prior literature, most of which finds that the normative behaviors of adult children are positively associated with parents' psychological wellbeing (e.g., Fingerman et al. 2011;Pudrovska 2009). This study further reveals that the normative leaving-home behavior of young adult children also has substantial impact on mothers' physical health. Of note, we find that intergenerational coresidential patterns have sustained physical health effects for mothers across midlife. This is of particular interest because midlife is a life course period during which the risk of cardiovascular disease among women rapidly increased due to a plethora of factors including rapid weight gain (Lee et al. 2010;Choi et al. 2012), hormonal fluctuations secondary to the onset of menopause (Matthews et al. 2009), and worsening health behaviors especially those associated with physical activity (Shaw et al. 2010). This study contributes to the existing literature by emphasizing that social exposures are just as important to consider as health related exposures when trying to fully understand rising rates of obesity among midlife women. This study provides one new puzzle piece to the effects of the changing demographics of family life for midlife mothers -whose children take increasingly diverse paths to independence -on their midlife health.. --- Figure 1. Predictive BMI by Intergenertional Coresidence Patterns. Note: The bivariate differences between gone-for-good and boomerangers or never-left are included in the table. Table 2. --- Results from Multilevel Regression ---.

Midlife mothers report their children returning to the maternal home after departing (i.e., boomerang children) and remaining in the maternal home longer (i.e., never-left children) than the past half century. Over the same time period, the percent of Americans considered overweight and obese have increased. Yet, we know very little about how such delays affect the body weight of mothers. The current study uses the National Survey of Youth 1979 (NLSY79) and its corresponding young adult sample (NLSY79-YA) across 20 consecutive years (N=7,197) to determine if extended coresidence with an adult child is associated with midlife mothers' body weight changes. Results from multilevel regression models show that compared to mothers whose young adult children left home and never returned ("gone-for-good"), mothers of the "never-left" had higher body weight at 40 but similar body weight at 50. Mothers of the boomerangers had higher body weight relative to mothers of the "gone-for-good" across midlife. Mothers of the boomerangers and mothers of the "never-left" had similar weight at age 40 but the former group had more weight gain across midlife. These findings lend new insight into how different patterns of mother-young adult coresidence likely affect the health of mothers, and suggests the effects of recent demographic trends such as "failure to launch" on family formation and functioning should be viewed holistically with a more inclusive sociological lens.

RESEARCH Victims of intimate partner violence (IPV) may experience num erous physical, sexual, psychological, behavioural and even fatal consequences (Table 1). [1] These include an increased number of visits to emergency departments, and an increased risk of gastrointestinal, cardiovascular, gynaecological and psychiatric disorders, including depression, substance abuse and posttraumatic stress disorder. [1] Women who refuse to have sex without a condom have an increased likelihood of becoming IPV victims, and if forced to do so face the additional risk of sexually transmitted infections, including HIV, and unplanned pregnancies. [1] The effects of IPV may not be immediately apparent, and may only manifest with targeted questioning or after an examination. It is precisely because of the complexity of these implications that it is important for the clinician to know how to screen for and manage these individuals. The World Report on Violence and Health [1] defines IPV as 'actual or threatened physical, sexual, psychological, emotional, or stalking abuse by an intimate partner'. An intimate partner can be a current or former spouse or a nonmarital partner, such as a boyfriend, girlfriend or dating partner, and can be someone of the same or the opposite sex. The South African (SA) Domestic Violence Act 116 of 1998 [2] expands on this definition and speaks of a 'complainant' as any person who is or has been in a domestic relationship and has been or allegedly has been subjected to an act of domestic violence. The term domestic relationship is allencompassing and includes customary marriages, various types of relationships and even flatmates. The definition of domestic violence is further expanded to include economic abuse, intimidation, harassment, stalking, damage to property, entry into the complainant's residence without consent, and controlling behaviour. [2] Looking at the global prevalence rates, a review of more than 50 populationbased studies from 35 nations found that 10 52% of females reported that they had been physically abused, and 10 30% had experienced sexual violence from an intimate partner at some point in their lives. [3] The reported IPV prevalence of 20 71% in subSaharan Africa has been thought to be an underestimate due to underreporting and poor standardisation of methods. [3] A national study from SA found a 19% lifetime prevalence of victimisation among female participants and a 27.5% prevalence of men perpetrating violence in their current or most recent relationship. [4] More than 50% of female homicide victims were killed by their intimate partners. [4] Furthermore, an SA study on female homicide indicated that every 6 hours a woman was killed by her intimate partner. [5] University students are considered an elite sector of society, and it could be assumed that attending a tertiary institution would render individuals less likely to be perpetrators or victims of IPV. [6] Some of these students study towards careers in which they will need to deal with the victims of IPV. There is a paucity of data relating to SA students who are victims of IPV, particularly in the fields of medicine and social work. The objective of this study was to ascertain the prevalence of and gender differences in IPV in a group of students studying to be healthcare workers in an SA tertiary institution. --- Methods Approval to conduct the study was obtained from the University of the Witwatersrand Human Research Ethics Committee (ref. no. M120670). --- Study participants A sample of male and female students from the medical school and department of social work at a university in Gauteng Province, SA, were asked to complete a quantitative anonymous questionnaire. The sample consisted of all 1st to 6thyear medical students and all 1st to 4thyear social work students. These two faculties comprised RESEARCH 1 593 students, and the decision was made to use the entire group as a sample. Connor et al. [7] found that students who received education about IPV while at medical school reported having greater confidence and perceived preparedness to deal with IPV victims. We therefore selected this population because they would be exposed to IPV and would be likely to screen for it. --- Data collection process The consent process consisted of a brief presentation by the first author (KS) detailing the study prior to handing out the questionnaire. The information sheets, consent forms and questionnaires were handed out after lectures or exam sessions. Participation in the study was voluntary, and students were told they could leave questions unanswered if necessary. An electronic marking sheet was used, and the questionnaires were anonymous with no name or student number reflected anywhere on the response sheet. Completed questionnaires were then placed into a sealed box. After the survey, each student was given a form with the contact details of a dedicated psychological counselling service available to them if they experienced discomfort as a result of the survey process. --- Data collection tool Two recommended screening tools from the US Centers for Disease Control and Prevention (CDC) were combined to create 25 multiple choice questions, which were adapted to the SA context. These tools were the computerbased IPV Questionnaire and the Women Abuse Screening Tool (WAST). The WAST had a Cronbach's alpha of 0.75 estimated for the reliability of the tool and a correlation with the construct validity of the Abuse Risk Inventory of 0.69. [8] The data collection tool comprised four domains, of which this article covers only two -demographics and the prevalence of the different types of abuse. The first six questions requested demographic information from the participants, including the degree for which they were enrolled (medicine or social work), year of study, age group, gender, racial group and relationship status. Questions 14, 15 and 16 were about relationships and were derived from the Women Abuse Screening Tool (Table 2). [8] Questions 11 13 and 17 22 were derived from the questions on the computer based IPV questionnaire listed in Table 3. [8] Two questions involving handguns from the original IPV questionnaire were not included in this questionnaire. Questions 24 and 25 were original questions (Table 4). --- Statistical analysis The results of each sheet were tallied electronically per question. Data were recorded on an Excel 2013 spreadsheet (Microsoft, USA), and Statistica version 12 (Statsoft, USA) was used for all statistical procedures. Descriptive statistics were reported as totals and percentages for categorical data. Comparative statistics were done using a <unk> 2 test or Fisher's exact test as appropriate, with a pvalue of <unk>0.05 being considered statistically significant. For the purposes of definition, participants were deemed to have been victims of physical abuse if they answered in the affirmative to the questions referring to being pushed, kicked, or otherwise physically hurt, victims of sexual abuse if they had been made to have sex (oral, anal, vaginal) when they did not want to, and victims of emotional abuse if they answered in the affirmative to the question directly asking the respondent whether he/she had been emotionally abused. The definition of emotional abuse was further expanded to include whether a partner gets very jealous or tries to control the respondent's life, tries to keep them away from family or friends, says insulting things or threatens them, is disagreeable, or may try to hurt them or other family members. --- Results Responses were obtained from 1 354 of the 1 593 students (85.0%). The responses collected were similar for each year of study. Most respondents were aged 20 24 years (61.9%) and were female (67.8%); 45.9% were black, 32.7% white, 16.6% Indian and 4.8% coloured (Table 5). Fiftythree percent of the respondents were in a relationship, engaged or married. Of the 718 respondents who reported that they were in a relationship, 123 (17.1%) indicated that they had a jealous partner. Of the remaining 636 respondents who indicated they were not currently in a relationship, a very similar proportion (n=127; 19.9%) indicated that they had had a jealous partner in the past. The following responses pertain to respondents in a relationship (n=718): 40 (5.6%) often and 250 (34.8%) sometimes felt that their --- RESEARCH partner tried to keep them away from family and friends, while 250 (34.8%) reported that their partner was sometimes insulting or threatening towards them and 123 (17.1%) were afraid to disagree with their partner because they might hurt them or other family members. Two hundred and three respondents (28.3%) felt that there was some tension in their current relationship, 310 (43.2%) experienced some difficulty when settling arguments with their partners, 555 (77.3%) sometimes felt lowered selfesteem after settling an argument, and 124 (17.3%) were afraid not to agree with their partner. The prevalence of any type of IPV (sexual, physical or emotional abuse) among all the respondents was 42.6% (577/1 354). Emotional abuse was the most common type of abuse (35.0% of respondents). With the expanded definition of emotional abuse (as above) the prevalence increased to 54.9%. Physical abuse (20.0%) and sexual abuse (8.9%) were reported less often. Thirtyfive females (6.5% of respondents who had suffered IPV) indicated that they had been emotionally, physically and sexually abused. Most of these females (28/35, 80.0%) indicated they were not currently in a relationship. [8] Question Response --- Table 2. Components of the WAST In --- RESEARCH With sexual abuse, although no trend according to age was apparent among male students, the percentage of females reporting sexual abuse increased by 6.1% between the 17 19yearold (9.8%) and >24yearold (15.9%) age groups. Older females (>24 years) appeared more vulnerable to emotional (43.4%) and sexual abuse (15.9%) (Fig. 1). Responses of females relating to questions pertaining to whether the partner was isolating or insulting, to whether there was tension in the relationship, and to settling of arguments and disagreeing with a partner did not appear to indicate a major problem, with less than half of the respondents reporting that these were issues. However, overall 59.5% of female respondents in a relationship aged <unk>25 years indicated they had a jealous partner. Furthermore, overall 68.0% had arguments resulting in the individual 'feeling bad' often (14.4%) or sometimes (55.2%). This response was also more common in the two younger age groups. Of all the victims of sexual abuse, 45.0% were at least able to negotiate protected sex. Overall, 58.7% of the respondents knew where they could get help and counselling as victims of abuse. The perpetrators (14.0% of total respondents) of IPV tended to be males in the white and Asian ethnic groups, females among coloured respondents, and without an apparent gender difference (p=0.55) among black respondents. There were no agerelated differences (males p=0.87 and females p=0.69) among the perpetrators, although younger males were more often perpetrators in the 17 19year age group. Only three perpetrators of sexual abuse indicated that they themselves had been sexually abused. --- Discussion Future healthcare workers tend to provide superior management to victims of IPV when they seek assistance for themselves first. Christofides and Silo [9] found that practitioners who reported their own, a friend's or a family member's experience with IPV had an increased 'quality of care score'. This could be the result of their ability to identify and empathise with victims. Kim and Motsei [6] stated that healthcare workers who were victims of IPV tended to have the same cultural values as the victims they treated and counselled, underlining the 'genderbound constructs' within which they operated, which extended from their personal to their professional capacities. Sugg and Inui, [10] however, reported that doctors were 'emotionally inactivated' and therefore not able to manage patients with similar experiences, and might experience personal distress. This study determined the prevalence of IPV in an SA tertiary institution population, representative of the ethnic makeup of the national population. The study is unusual as it obtained responses from medical and social work students in all years of study and from respondents of both genders. These students, whose future professions include counselling of victims of abuse, would be expected to be aware of this problem. The total prevalence of IPV at the institution in question was 42.6%, comparable to a study from China that found a prevalence of 37.1%. [11] The university with the highest reported prevalence of IPV victims was in Nigeria, with a figure of 58.8% for female students. [12] Studies at other universities found varying prevalences (in descending order) from 28.7% (Tennessee), 22% (Wisconsin) and 10% (Vanderbilt) in the USA, to 7% (Ontario, Canada). [1316] A study at numerous institutions in the UK found an overall prevalence of 14.2%. [17] In line with previous studies, the present study found a higher prevalence of female (particularly in the older age groups) than male IPV victims. A report from the Office of Justice in the USA [18] stated that female students in fulltime education were at a higher risk of sexual violence than the general female population. Several risk factors were noted: 'living on campus, being unmarried, getting drunk frequently, and experiencing prior sexual victimization'. [19] In an SA study by Gass et al. [4] it was found that the single most common risk factor for IPV victims of both genders was witnessing parental violence. Specific risk factors pertaining to male victims included 'low income and lack of closeness to a primary female caregiver', whereas risk factors for female victims included 'low educational attainment, childhood physical abuse, and adultonset alcohol abuse/dependence and intermittent explosive disorder'. [4] IPV has been strongly linked to intergenerational cycling of violence and risk exposure across the life course. [4] An interesting observation in our study was the low use of condoms. Peltzer [19] --- RESEARCH used condoms over a 3month period. This is concerning given the high national HIV incidence and that this was a welleducated group of individuals who should be aware of the risks of unprotected sexual intercourse. An interesting and unexpected finding was the high rate of female perpetrators, particularly among black and coloured students. The National Intimate Partner and Sexual Violence Survey by the CDC expanded the definition of IPV to include rape while under the influence of illicit drugs, undesirable touching, expressive aggression, conceiving by coercion, refusing to use a condom, and stalking. [20] With the new definition they found an equal rate of perpetration between the genders. Female aggressors are also at increased risk of becoming a victim later on. [21] Gass et al. [4] reported that risk factors in male perpetrators were likely to include cohabitation, low income and early and adultonset mood disorders, whereas risk factors in female perpetrators included low educational attainment and early onset alcohol abuse/dependence. Medical schools need to have appropriate support mechanisms in place for victims and potential victims of IPV, and students should be given a list of local resources. According to Ambuel et al., [14] students need to be allowed to process their feelings and experiences within their academic spaces. There should be a safe environment that fosters open, respectful sharing of experiences and ideas, appropriate supervision, and encouragement of responsible selfcare including access to psychotherapy and support groups. To this end, our institution has dedicated psychology and counselling services available. Broader institutional and policy changes should also be implemented in academic programmes to enable early detection and management. [14] --- Study limitations Concepts such as physical, sexual or emotional abuse could have been better defined, as the questions assumed that students had prior knowledge or understood the scope of abuse as defined by current legislation. This may have affected their ability to answer the questions. Certain components of emotional abuse as defined by the SA Domestic Violence Act were not specifically included, such as economic abuse, intimidation, harassment, stalking, damage to property, and entry into the complainant's residence without consent. The participants may also have had different perceptions of what defines sexual abuse, and the fact that a person did not agree with the statement does not mean it did not happen. If the data collection tools had covered these various other forms of IPV, labelled domestic violence under the Domestic Violence Act, the prevalence rate might have differed. --- Conclusion The extent of emotional, physical and sexual abuse among university medical and social science students sampled was unacceptably high, both as victims and as perpetrators. As a consequence of their own experiences of abuse, these individuals may have difficulty in managing patients who have been subjected to abuse. To determine possible regional differences in this unacceptable practice, it is recommended that similar surveys be undertaken at other centres of higher learning.

This open-access article is distributed under Creative Commons licence CC-BY-NC 4.0. Background. Intimate partner violence (IPV) is actual or threatened physical, sexual, psychological, emotional or stalking abuse by an intimate partner. Despite the high prevalence of IPV in South Africa (SA), there is a paucity of data on university students training in fields where they are likely to have to manage the aftereffects of such events in their personal capacity in the future. Objectives. To ascertain the prevalence of IPV in an SA tertiary institution population with a diverse demographic profile. Methods. Students from the faculty of health sciences and the faculty of humanities, social work department, completed an anonymous questionnaire. Students were made aware of psychological counselling available to them. Results. Responses were obtained from 1 354 of 1 593 students (85.0%) (67.8% female, 45.9% black, 32.7% white, 16.6% Indian, 4.8% coloured). Of the respondents, 53.0% indicated that they were in a relationship. The prevalence of any type of IPV (sexual, physical or emotional abuse) among all respondents was 42.6%. Emotional abuse was reported by 54.9% of respondents, physical abuse by 20.0% and sexual abuse by 8.9%. Thirtyfive females (6.5% of respondents who had suffered IPV) indicated that they had been emotionally, physically and sexually abused. Fourteen percent identified themselves as perpetrators of abuse, but only three perpetrators of sexual abuse reported having also been victims of sexual abuse. Most respondents (58.7%) knew where to get help.The extent of IPV among the medical and social work students sampled was found to be unacceptably high, both as victims and as perpetrators. As a result of their exposure to IPV, these individuals may have difficulty in managing patients who have been subjected to abuse.

Background To be cared for, and to die at home is important to many cancer patients and their family caregivers (hereafter: caregiver) [1]. To be able to fulfil home death, support from caregivers is important [2]. They know the patient well and have the experience of how to help the individual. Many of the caregivers exert significant responsibility-and care tasks to their loved ones and perform a central part of the total care tasks given by the society [3][4][5]. However, to care for a cancer patient at home at the end of life is a considerable burden for the caregivers [6,7]. Therefore, support to caregivers from healthcare professionals is equally important [8,9]. Caregivers experience physical, psychological, social and financial challenges [2]. Many are not prepared for the demands and challenges they encounter [10]. The caregivers have to manage their own psychological worries, as well as the patients' worries throughout the disease trajectory. At the same time, they have to provide help with daily tasks. Therefore, lack of focus on caregivers' needs may lead to poor mental health and low quality of life [11]. The feeling of safety, stability, and being free from fear or anxiety through the disease trajectory gives security for caregivers [12]. Effective communication between healthcare professionals and caregivers empowers caregivers to fulfil their responsibilities and provide security to enable home care and home death [13]. This facilitates a smooth transition of care and allows the families to prepare for end-of-life phase and death [14]. Long distances, variable infrastructure and shortage of healthcare professionals-and services are barriers that make the access to specialized palliative care difficult in rural areas [15][16][17][18]. Palliative care is in many rural areas provided through standard nursing home services and general district nurses [19]. In Norway, palliative care for patients staying at home is provided by general practitioners and home care nurses. Palliative care beds are available in designated inpatient units in nursing homes in the municipalities [20][21][22]. District healthcare professionals therefore play an essential role in the provision of palliative care for cancer patients and caregivers with long distances to hospital and specialized palliative care services [19,23]. Former Sogn og Fjordane county (merged with Hordaland county to Vestland county Jan 1, 2020) is a rural region in Western Norway. This is a region of more than 18,000 km 2 with a small population of around 110,000. The region has two local hospitals and one central hospital. The region's only palliative care team, as well as the oncology and haematology departments and the physicians responsible for cancer treatment, are located at F<unk>rde Central Hospital in the middle of the region. These consultants often use telephone contact or videocalls to conduct consultations with the patients, but when specialised treatments and examinations are necessary, the patients have to travel to F<unk>rde Central Hospital [24]. The longest drive from a municipality in Sogn og Fjordane to F<unk>rde Central Hospital is three and a half hours by car [25]. The region is characterized by fjords, mountains, and islands and has many narrow and winding roads with great danger of avalanches and mudslides, which often results in closed roads. Some municipalities are small, located on the mainland and have a short distance to the local hospital or have several nursing homes. Other municipalities are large, consist of several islands, and people may have to take a ferry or a boat to their destination, which also is dependable on the weather, or generally have a long travelling route. All this affects access to healthcare services. The overall aim of this study was to explore factors that determined the feeling of security of caregivers of dying patients with advanced cancer that cared for the patient at home at the end of life in the region of Sogn og Fjordane in Norway. The following research questions were addressed: What factors made the caregivers feel secure when they cared for the patient with advanced cancer at home in the rural region of Sogn og Fjordane? In the case that the patient died at home,what facilitated home death? In the case that the patient did not die at home,what were the barriers for home death? --- Method --- Study design This is a qualitative retrospective study using semistructured in-depth interviews with bereaved. --- Participants and recruitment Participants were recruited between September and December 2018. Cancer coordinators and nurses working in ten municipalities and in the specialized palliative care team at F<unk>rde Central Hospital in Sogn og Fjordane received an email with written information about the study and a consent letter to present to potential participants. They contacted potential participants they had been involved with within the last 2-12 months. After the cancer coordinators/nurses had approached potential participants, the interviewer (ASB) contacted by telephone those who gave their permission to be contacted and gave them more information about the study. Interviews were conducted consecutively as the caregivers provided written consent to participate. More than one caregiver could be recruited per patient. In that case, they were interviewed together. Inclusion criteria were parents, children or spouse 18 years or above of a deceased cancer patient; between 2 and 12 months since the loss of their loved ones; experience from being a caregiver of a cancer patient who did not permanently live in an institution at the end of life; experience with planned or unplanned home death or institutional death; previous or current contact with municipality cancer coordinator/nurse or palliative care team at F<unk>rde Central Hospital and ability to understand information in Norwegian or English. There were no exclusion criteria. --- Data collection and management The interviews were conducted using a semi-structured interview guide developed by all of the researchers [26]. Three of the researchers (BA, KS, ATB) had experience from the participant population and from previous qualitative research [27][28][29][30][31]. The questions in the interview guide addressed informants' experiences and thoughts from caring for the family member at home at the end of life, the experiences and interaction with healthcare services and what made them feel secure or not (Table 1). The interviews were audio recorded after consent from the informants, and transcribed verbatim. All the transcriptions were read and analysed independently by at least two of the authors [26]. Consensus regarding natural units, categories and subcategories during the analysis of the interviews was reached by discussion between the researchers. The demographic data about the informant and patient (Tables 2 and3) were collected by a questionnaire filled in by the informants before start of the interview. --- Analysis The qualitative data analysis software NVivo 12 pro was used to import, manage and analyse the transcribed interviews. The analysis was performed according to the procedure described by Kvale and Brinkmann [26]. First, natural units were identified in the transcripts, and condensed into meaning units. Then codes were extracted from the meaning units. The codes were categorised inductively, i.e. the categories were derived from the data, not from the interview guide. Finally, subcategories were identified under each category, and illustrative quotes were selected. --- Ethics The Regional Committee on Medical and Health Research Ethics (REK) determined that the study was in the category health services research and therefore did not need approval (2018/1129/ REK s<unk>r-<unk>st). Health service research is not included in the Health Research Act and is therefore not obligated to get approval from REK [32]. The research protocol was approved by the Norwegian Centre for Research Data (NSD, ref.nr 61,366 AMS/LR) before start of the project. --- Results Ten bereaved caregivers from nine families where recruited. They were all women, and they were all native Norwegians. Characteristics of the informants and the patients are given in Tables 2 and3. Three main categories of factors were identified that made the caregivers feel secure when they cared for the patient with advanced cancer at home, or the opposite, lack of these factors made them feel insecure: Personal factors, healthcare professional factors and organizational factors (Table 4). --- Personal factors The informants talked about several personal factors that influenced the feeling of security when caring for the advanced cancer patient at home at the end of life. These factors varied from family to family and resulted in a different set of perceptions, attitudes and behaviour towards the feeling of security. To be physically present and available for the patients in last phase of life was important to all the informants. One of the informants said it like this: "It was important to me to be with him... to be there, all the time. And it wasn't hard, it was natural to be there. It was safe for me to be there with him" (Interview 6). The informants had different attitudes towards place of care and place of death. The most common reason Table 1 Interview guide summary (Supplementary file 1) • What was bothersome, rewarding, difficult by caring for the cancer patient at home? • Why did you choose to care for the patient at home? • What made you as a caregiver feel secure or insecure? • What was demanding? • What kind of expectation did you have? • How was the access to HCS and was this a problem? • Did you get support from others, and was this support crucial for caring for the patient at home? • What kind of support was most important? • Did you talk about death and preferences? • Was home death important for you and for the patient? Was the decision difficult? • How did you experience HCS? • What kind of offers and information did you get from HCS? • What was/ would have been the most important factors to take the patient home to die? given for home death, was that it was the patient's wish to be at home. It was important for the caregivers to fulfil the last wish of the patient, even though it affected their own physical health, mental health or social life. "I knew he wanted me there, and that he felt safe because of this and calmed down. But really, I feel I went into vacuum. I lived on autopilot, I did what I was asked for, what was expected of me. I don't know how I did it (...)I was unsure of a lot of things, and it was very stressful, certainly. But knowing that he felt safe, that was the most important thing to me" (Interview 5). At a hospital, it was said to be a restless pace, focus, and atmosphere. It was experienced as exhausting for both patient and caregiver. This made the informants feel stressful and they felt a lack of privacy and peace. At home, it was a calm atmosphere and more space to move in. This gave a greater feeling of freedom which contributed to the feeling of security. "(... ) because it is a completely different traffic in hospitals, and noise and-it's tiring. (... ) something had to be measured, something-blood tests or temperature or blood pressure or something... and if they were two-three patients in the room, then there was a lot of traffic" (Interview 6). However, if the patient's situation was complicated with many issues and depending on many measures, some informants felt more secure when the patient was admitted to an institution with healthcare professionals (HCP) around all the time. "He wanted to be at home, but it didn't work. He had so much tubing and drains and analgesic pump and epidural in his back and-(... ) Yeah, I didn't have the competence to manage that" (Interview 3). To have someone else to share the responsibility with of following the patient through the disease trajectory, discuss decisions concerning the patient and have the opportunity to share thoughts and worries, was said to be particularly important for the caregivers' feeling of security: "I declined the offer of attendance allowance; I wanted another sister to do this. I didn't want to be alone in this [i.e. follow the patient in the last phase]. I wanted more people to be involved because I had-... *sigh* I had been doing this alone for so long" (Interview 5). To be able to be open and realistic about the patients' remaining time and that death was approaching was said to be a difficult topic, but important for the family concerned. By talking about this the informants could get to --- Living situation Lived in the same residence as the deceased N (%) --- (50%) Lived in another residence than the deceased N (%) --- (50%) Relationship to the deceased Spouse/cohabitant N (%) 5 (50%) Child N (%) 4 (40%) Parent N (%) 1 (10%) --- Place of death Home N (%) 3 (33%) Hospital N (%) 3 (33%) Nursing Home N (%) --- (33%) Contact with the palliative care team N (%) 8 (89%) know the patient's last wishes. Several of the informants wished that HCP brought up this topic: "I tried to hint, but I was so scared that it would be misunderstood, that he would think I would get rid of him or something, or that he would understand something else than what I really meant" (Interview 8). Informants settled around the county said that they were used to the long distances and had adapted to this. However, what was common for most of the informants and emphasized as an important factor for security was that they had a driver's license: "It was of great help to have a driver's license, because then I could come and go whenever I wanted to. Plus, then-I travelled often to our son because-it wasn't always easy to come home to an empty house [i.e. when her husband was at the hospital]. You may have a need to be with others" (Interview 4). --- Healthcare professionals The need for qualified and helpful healthcare professionals (HCP) was most often said to contribute to the feeling of security. Qualified and helpful HCP comprise of the following topics: the need for knowledge and competence among HCP, clear distribution of roles and responsibility between caregivers and HCP, the need for information from HCP and the importance of speaking Norwegian. Specialised healthcare services like the palliative care team was appreciated, as was contact with the local cancer coordinator and the GP. The informants felt safe when HCP had the knowledge to notice if the patient was in pain or in other ways unwell, and competence to provide medication or other measures to ease the situation. It was substantial for the caregivers that HCP not only approached the patient, but also turned to the caregivers, so they felt seen in this phase. Several of the informants expressed a need for talking to the HCP when HCP visited the patient because they had spent most of the day alone with the patient. They had therefore many questions and needed confirmation that what they were doing was right. "The home nurse didn't ask, just went straight to the room. Didn't ask how the night had been, very little dialogue. They didn't care about my mother as a caregiver. She needed care. Someone to talk to. Someone to discuss medication with, that asked how the night had been and how the medication should be throughout the day, how much morphine we should administer" (Interview 1). The informants experienced that caring for the ill patient at home carried a lot of responsibilities. They felt like they had to take the responsibility for providing the right medication, get the necessary equipment, and make sure that the patient got the right care. Some expressed that they had to monitor the patient all the time because they were afraid that HCP would overlook something or not see changes when they were visiting, like changes in pain level or signs of infections. This led to less trust in HCP and a high level of stress, exhaustion, insecurity and less energy and concentration in the need of just having the role as a family member. "It felt safe and secure for us to know that they were visiting her [i.e. home nursing care]. But when changes occurred, they were slow. And in that phase of changing, it's very tough to be a caregiver, because you see things, feel things and know things, and the rest [i.e. HCP] doesn't follow" (Interview 2). Lack of information, unclear information, and information given too late was said to be a big problem that contributed to a great deal of insecurity among the informants. The informants said that since they never had experienced going through a palliative-and terminal phase before and were lacking knowledge about the patients' course at the end of life, they did not know what to ask for or what to expect. They expressed that they Also, how the information was given was decisive. When the informants received brochures or other written information to read without HCP's explaining, they often did not read it because they did not have time to read a brochure and/or felt that the information was not valid for them yet or that the information was given too late. This made the informants feel that HCP's were unrealistic to where the patient was in the phase of the disease or that the informants themselves had denied accepting the outcome of the disease. "The cancer coordinator came with these brochures... *laughter* about dying at home and everything and left them up there with my mother. I just took them away. Took them home with me, hid them. Like, "we shall not talk about this"" (Interview 5). Some informants said that the information they got after treatment was not useful for them because they did not understand what it was saying: When the caregiver and HCP had an open communication and common understanding of the situation, this led to a feeling of security: " "We had many conversations with the doctors and nurses at the hospital before we took him home [i.e. the patient] where we could ask any questions we wanted. I felt that we got very good information about the situation and of taking him home at the end of life" (Interview 9) To be able to understand what they were told by the HCP and to feel that they were being understood was important for the caregivers' feeling of security. Several informants said that it was uncomfortable, and they felt insecure to be involved with HCP who did not master the Norwegian language: Some of the informants felt more secure when the palliative care team took over the treatment and care of the patient. However, they wished that the team came into the picture earlier in the palliative phase. Some of the patients had lived with incurable cancer for several years without contact with the palliative care team until the last months or days of life. "(... ) if "It was a shame that they were presented the last few days of mom's life (... ) They should have been present earlier (... ) They don't meet us [i.e. the family] at our best, the last few days" (Interview 2). --- Organizational factors Good organization of the palliative healthcare service (HCS) in Norway was important for caregivers to feel secure when carrying out palliative care at home. Subjects that were repeated in the interviews were interaction and collaboration between HCS in the municipalities and the hospital, the need for equipment and facilitation at home, support from The Norwegian Labour and Welfare Administration (NAV) and thus facilitation for use of attendance allowance. Many of the informants said that they could not have managed the caregiving without the service of home nursing care. However, it was considered as a burden and perceived as difficult to be involved with home nursing care when the caregivers experienced poor organization of the service: "They [i.e. home nursing care service] came with different competence: everything from unskilled persons to nurses. It seemed as if they had not considered what this patient needed of competence and continuity and what information they needed before they came into the home. Such an ill patient-I was very disappointed. But you must understand: the ones who came did their best, right? It wasn't anything wrong with them, right? There was something wrong with the system" (Interview 1). It was also said that with home nursing care it was unpredictable when someone came and gave painkillers or provided other important measures: "(... ) you couldn't ring a bell and get help when you needed to, it's not the same with home nursing care, right? They are on their rounds and there are large geographical areas and they can't get there so fast, so you feel much more alone. So, the insecurity with how we would get help when we needed it was-when could they come, right?" (Interview 1). The informants were dependent on access to competent healthcare services 24/7 to care for the terminally ill patient at home. When this was not possible, and when necessary measures could not be carried out at home, it created a barrier for home care. For example, home nursing care services in some municipalities could not administer intravenous interventions because of the lack of resources and/or competence. In these cases, admissions to nursing homes were necessary and contributed to caregivers' feeling of security, even though it was not the patient' or caregiver's first choice in place of care. "When she [i.e. the patient] came home [i.e. from the hospital], she had to have intravenous pain relief, but the home nursing care couldn't come to our home and administer this, so she had to go to the nursing home every day" (Interview 9). Good collaboration between the GP and home nursing care was said to be important for the implementation of home care and home death. Knowing that the family and home nursing care got help right away from the GP contributed to a feeling of great security, instead of having to contact the hospital or emergency room with unknown doctors who did not know the patient. "We always had the home nursing care in hindsight who was prepared to help us whenever we needed. It wasn't a problem to call the GP either, if it was something, no matter what time of the day, but we didn't call him, but we had the opportunity, and that was a safety for us to know" (Interview 6) A major concern the informants had in terms of having the patient at home was access to equipment. Cooperation between HCP at the hospital and in the municipality was said to be important in facilitating home care and get the necessary equipment in place. "They were incredibly good at the hospital [i.e. HCP]. We had an occupational therapist who submitted applications, facilitated the house and received the equipment we needed. They contacted the municipal physical-and occupational therapist who came home, and... We got what we needed." (Interview 4) To be involved with NAV was repeatedly said to contribute to insecurity. NAV is the Norwegian social security system that manage unemployment benefit, pensions, attendance allowance and care allowance [33]. Informants experienced resistance and slow processing time from this authority, which led to greater insecurity in the care of the patient at home and the informant's economic situation before and after the patient's death: "They [NAV] said that they had to have a confirmation that my husband really was so sick (... ) It wasn't unfamiliar that he was sick, within the NAV-system, but this is the way it is. I haven't heard from NAV since. They wait until they don't have to pay. That's what happens" (Interview 8). Although attendance allowance was said to be of good help when the informants had received information of this right, the system was said to be difficult and not adjusted to today's cancer care or the caregiver's everyday life. "(... ) but it's a demanding system because you have to plan when to take out days and everything in that form, and you don't know that, because it's dependable on the health-both for my mother and my father [i.e. the caregiver and the patient]-so, therefore, it was barely used" (Interview 1). --- Discussion Personal factors among patient and caregivers and factors related to healthcare professionals and organization of healthcare were identified in this study to make caregivers feel more secure when caring for advanced cancer patients at home at the end of life in Sogn og Fjordane. The informants especially highlighted the importance of support from healthcare professionals and good organization of palliative healthcare services. Good communication between HCP and family/patient about patient matters and what to expect at the end of life, and between HCP so that HCS were well coordinated, was highly appreciated and contributed to security. Contact between healthcare professionals, the patient, and the caregivers must be based upon understanding and trust [34]. To achieve this, it is essential that HCPs have knowledge and competence in the field, can provide timely and necessary information, and acknowledge and support the caregivers in their tasks. In Norway, every healthcare professional who work within the primary-or specialist healthcare is expected to have basic competence in palliative care [22]. Basic palliative care is part of HCP's primary education [22]. In addition, nurses working with palliative patients should have additional competence in palliative care equivalent to level B. Level B is necessary competence to work with patients with palliative care needs and is provided through additional courses in palliative medicine and palliative care after primary education [22]. In rural districts, the recruitment of nurses with additional competence in palliative care may be difficult [35,36]. This may result in palliative care services of lower quality in rural districts and increased likelihood of admission to hospital [16,17]. It is also essential to have a clear distribution of roles and responsibility between HCP and the caregivers [8,[37][38][39][40]. Caregivers are often not used to care for severely ill patients and therefore have no previous knowledge about caring tasks, cancer or end-of-life care, which makes them dependent on HCP and HCS to fulfil the task [39,41]. This is consistent with the findings in our study. Informants described that they felt safer when HCP with more experience or more education in palliative care or cancer care were involved; they could better answer questions and confirm that the caregivers did right. Another finding in our study, is the importance of having Norwegian-speaking HCP in palliative care when the patients are native Norwegians. Our informants reported that poor language dissemination made them feel unsure about the HCP's knowledge and competence. The barrier of language can prevent skilful nurses from being understood by their patients and colleagues. This can affect the patients' and caregiver's safety [42]. Earlier research and the present study revealed that the caregivers received information about the disease course when they requested it themselves [43]. Informants commented that they did not know what to ask about because they had not been through this before. This led to uninformed decisions, which is also reported by others [38,43,44]. Our informants expressed that they felt unsafe and unprepared for the role of caregiving when they were uninformed about what to expect or had an unrealistic expectation of care at home. It is shown by others that late information and the lack of information give the caregivers a feeling of uselessness, helplessness, insufficiency and a lack of control which again contributes to a great sense of insecurity [37-39, 41, 44, 45]. Our informants expressed that they were dependent on timely information about what they were getting into to be able to make rational and safe decisions. To talk about the disease, prognosis, and death makes the caregivers more prepared for the end of life, the funeral and the time after death [46]. Still, many are afraid of hurting the patient by addressing these topics [44,47]. Dealing with death and safeguarding the caregivers in the palliative phase and in the bereavement period are central elements in palliative care [48]. However, previous studies as well as this study found that these elements often are deficient in today's organization of palliative care [44,46]. Different caregivers have different needs and cope with information in different ways [44,46]. Therefore, knowing the exact time and way to provide information to caregivers is not always clear. A valid evidence-based Carer Support Needs Assessment Tool (CSNAT) has been developed due to lack of practical tools to assess and evaluate caregivers' support needs in the palliative phase [49]. By using this tool, individual caregivers' needs for information and support can be improved and caregiving strains are reduced [49,50]. According to Norwegian law, information about the patient's health condition and the provided healthcare can only be given to the closest relatives, who often is the caregiver, if the patient gives consent [51]. Information distribution is therefore dependent on the patient. This safeguards the patient's privacy. Still, according to Norwegian law, HCPs have a general duty of giving guidance to the caregivers [52]. Caregivers need information, advice, and guidance from HCP to be able to care for the patient at home [38], and this does not require consent from the patient. However, the caregivers are often in need of more than general information and guidance when they care for the patient at home, such as information about the patient's prognosis or special considerations that must be taken. If the caregivers do not get this information, it can affect their possibility to prepare for death and thus make the experience of the palliative phase more difficult. For HCP, it may be an ethical dilemma that important information cannot be given to the caregiver due to confidentiality and privacy issues. The duty of guidance may therefore be neglected because HCP's are afraid of breaking their duty of confidentiality. One way to overcome this, is to inform the patient and include the caregiver early in the disease trajectory [3]. We found in our study that if the caregiver and patient achieved a common understanding about the situation, the caregivers felt more secure and had fewer worries and unanswered questions. This supports involving the caregiver in addition to the patient when planning for the last phase of life. Advanced Care Planning (ACP) is a communication process between the patient, the family and HCP where the patient plan for a time when he/she cannot make decisions for themselves. This includes reflection, deliberation and determination of the patient's values, and wishes and preferences for treatment at the end of life [53]. None of our informants had experience with ACP. There is increased attention to the use of ACP in Norway [54], and this will maybe strengthen the involvement of caregivers. Support from HCP and HCS helps mitigate the burden caregivers experience, which allows the patient to stay longer at home [13]. Continuity in home nursing care, close follow-up from the GP and contact with the Palliative care team at F<unk>rde Central Hospital were central elements in the organization of palliative healthcare that contributed to the feeling of security among the informants and facilitated home death. The informants reported that they felt security knowing that HCP came regularly. This is in line with results from another qualitative Norwegian study. This study, as ours, states that support from HCP is important when carrying out home care and home death for cancer patients [55]. It was important for the caregivers that HCP could perform necessary measures and that they could contact competent HCP whenever needed [55]. It also supports the need for available and predictable home nursing care to achieve continuity and thus security for the families who are caring for the patient at home. Poor communication, interaction and organization between different healthcare professionals, healthcare services, the Norwegian Labour and Welfare administration (NAV) and families are barriers that can make it difficult to meet patients' desires to stay home as long as possible and to die at home. Caregivers are not only dependent on HCP and HCS to fulfil home care and home death, but also other authorities like NAV. NAV helps with economic conditions for the patients who receive palliative care and their caregivers, as the arrangement of attendance allowance [22]. Attendance allowance allows a carer to care for the patient at home at the end of life. Knowing their economic rights give caregivers less concerns regarding their own working situation and makes it easier for the caregiver to be more present with the patient. The informants in our study experienced that NAV was difficult to reach and gave the caregivers unnecessary strain and insecurity, in line with a previous study [43]. Financial issues are often a barrier of completion of home palliative care [56,57], because the caregiver has to leave work to manage the care at home [38]. Knowing about and receiving financial support, such as attendance allowance, may help to overcome this barrier. Most of the informants in this study had access to a private car which contributed to a feeling of security caring for the patient at home, knowing they more easily had access to healthcare. Many people are dependent on owning a private car in Sogn og Fjordane because of the long distances and fewer offers of public transportation than in urban areas. Sogn og Fjordane thus lies in the upper half of the ranking of registered cars in Norwegian counties [58]. A study done in England in 2011 showed that distance was not a factor in the experience of hospital-accessibility [59]. The same study from England showed that the experience of access to HCS was significant related to socioeconomic status, and thus carownership [59]. Earlier studies have shown that the availability of a specialist palliative care team, both physically and for telephone contact, is important for the feeling of security at home [9]. The informants in this study reported that they missed that the team had been involved earlier. A study by Johnson et al. [60] showed that oncologists mainly refer patients with advanced cancer to specialized palliative care services for physical symptom-related reasons, and not for other reasons like psychological or social issues [60]. In addition, the oncologists stated that they were trained to take care of the physical symptoms of these patients, something that led to an even later referral to specialized palliative care services [60]. McDonald et al. (2017) found that caregivers of patients with advanced cancer who received early palliative care (6-24 months before death) were more satisfied with care than those receiving only standard oncological care [61]. These findings and the results in this present study support that early involvement from a specialist palliative care team and a more available team may contribute to better services for the caregivers as well as for the patients. --- Strengths and limitations The qualitative research interview seeks to understand the world from the informant's perspective and aims to produce knowledge within a topic [62]. Using a semistructured approach, the interview provided room for the informants to deepen their views and provide additional information that they experienced as relevant to their situation. Dependability and confirmability are major factors in understanding the implications of this study, and considerably effort was dedicated to examining these issues. Content analysis was used to identify similarities, differences, and patterns in the experiences of informants, and conclusions were deduced from the collected material without a predetermined hypothesis. All the informants were from different municipalities with different distances to F<unk>rde Central Hospital. All the interviews were analysed by minimum two researchers. Many of the questions in the interview guide were formulated to address things that were positive in the palliative phase. During the interviews, the informants repeatedly drew attention to factors that were perceived as negative and contributed to additional insecurity instead of help in the palliative phase. This may have affected my analysis process. The opposite of factors that give insecurity are not necessarily factors that provide security and it could therefore be challenging to answer the research questions. All the informants were women. One can only speculate on why there were no men included in the interviews. The interviewer was not involved in the actual recruitment of participants. Historically, women have had the main role of caregiving, and even today, women have easier to take on care tasks [63,64]. One can therefore imagine that there was greater access to female informants during the recruitment process. It is uncertain how much the results were affected by the informants' gender, but it cannot rule out that it had an impact. However, we could have controlled the recruitment process more by having more dialogue with cancer coordinators/nurses throughout the recruitment period about what kind of informants we wanted for this study. --- Conclusion and implications This study from former Sogn og Fjordane county confirms the importance of support from healthcare professionals and a good organization of the palliative healthcare for giving the caregivers security in caring for the cancer patient at home in the last phase of life. It also confirms the importance of good coordination of care between the different services. Distance to hospital was not found to be a barrier for home care and home death. Many of the factors contributing to a feeling of security are factors that healthcare professionals can assess, map, and provide help with, in their interaction with the caregivers. --- Availability of data and materials The audio-taped and transcribed interviews are not

Background: For cancer patients and their family, an important factor that determines the choice to die at home is the caregivers' feeling of security when caring for the patient at home. Support to caregivers from healthcare professionals is important for the feeling of security. In rural areas, long distances and variable infrastructure may influence on access to healthcare services. This study explored factors that determined the security of caregivers of patients with advanced cancer who cared for the patients at home at the end of life in the rural region of Sogn og Fjordane in Norway, and what factors that facilitated home death. Methods: A qualitative study using semi-structured in-depth interviews with bereaved with experience from caring for cancer patients at home at the end of life was performed. Meaning units were extracted from the transcribed interviews and divided into categories and subcategories using Kvale and Brinkmann's qualitative method for analysis. Results: Ten bereaved caregivers from nine families where recruited. Five had lived together with the deceased. Three main categories of factors contributing to security emerged from the analysis: "Personal factors", "Healthcare professionals" and "Organization" of healthcare. Healthcare professionals and the organization of healthcare services contributed most to the feeling of security. Conclusion: Good competence in palliative care among healthcare professionals caring for patients with advanced cancer at home and well-organized palliative care services with defined responsibilities provided security to caregivers caring for advanced cancer patients at home in Sogn og Fjordane.

role of caregiving, and even today, women have easier to take on care tasks [63,64]. One can therefore imagine that there was greater access to female informants during the recruitment process. It is uncertain how much the results were affected by the informants' gender, but it cannot rule out that it had an impact. However, we could have controlled the recruitment process more by having more dialogue with cancer coordinators/nurses throughout the recruitment period about what kind of informants we wanted for this study. --- Conclusion and implications This study from former Sogn og Fjordane county confirms the importance of support from healthcare professionals and a good organization of the palliative healthcare for giving the caregivers security in caring for the cancer patient at home in the last phase of life. It also confirms the importance of good coordination of care between the different services. Distance to hospital was not found to be a barrier for home care and home death. Many of the factors contributing to a feeling of security are factors that healthcare professionals can assess, map, and provide help with, in their interaction with the caregivers. --- Availability of data and materials The audio-taped and transcribed interviews are not publicly available due to private details about the participants, but are available from the corresponding author on reasonable request. --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12904-020-00705-y. Additional file 1:. Interview guide Abbreviations REK: Regional committee on medical and health research ethics; NSD: Norwegian centre for research data; HCP: Healthcare professionals; HCS: Healthcare services; NAV: The Norwegian labour and welfare administration; CSNAT: Carer support needs assessment tool; ACP: Advanced care planning Authors' contributions ASB designed the study, led the recruitment of participants, performed and transcribed the interviews, and wrote the manuscript. ASB, KS and ATB developed the interview guide. BA, KS and ATB analysed the interviews together with ASB, and contributed writing the manuscript. ATB and ASB finalized the manuscript. All authors read and approved the final manuscript. --- Ethics approval and consent to participate The Helsinki declaration was followed. The Regional Committee on Medical and Health Research Ethics (REK) determined that the study was in the category health services research and therefore did not need approval (2018/1129/ REK s<unk>r-<unk>st). Health service research is not included in the Health Research Act and is therefore not obligated to get approval from REK [32]. The research protocol was approved by the Norwegian Centre for Research Data (NSD, ref.nr 61366 AMS/LR) before start of the project. All participants signed an informed written consent to participate. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: For cancer patients and their family, an important factor that determines the choice to die at home is the caregivers' feeling of security when caring for the patient at home. Support to caregivers from healthcare professionals is important for the feeling of security. In rural areas, long distances and variable infrastructure may influence on access to healthcare services. This study explored factors that determined the security of caregivers of patients with advanced cancer who cared for the patients at home at the end of life in the rural region of Sogn og Fjordane in Norway, and what factors that facilitated home death. Methods: A qualitative study using semi-structured in-depth interviews with bereaved with experience from caring for cancer patients at home at the end of life was performed. Meaning units were extracted from the transcribed interviews and divided into categories and subcategories using Kvale and Brinkmann's qualitative method for analysis. Results: Ten bereaved caregivers from nine families where recruited. Five had lived together with the deceased. Three main categories of factors contributing to security emerged from the analysis: "Personal factors", "Healthcare professionals" and "Organization" of healthcare. Healthcare professionals and the organization of healthcare services contributed most to the feeling of security. Conclusion: Good competence in palliative care among healthcare professionals caring for patients with advanced cancer at home and well-organized palliative care services with defined responsibilities provided security to caregivers caring for advanced cancer patients at home in Sogn og Fjordane.

Studies suggest that the rates of gambling-related problems may be higher among certain populations (Binde 2005;Raylu and Oei 2004). The 2001The -2002 National Epidemiological Survey on Alcohol and Related Conditions found the prevalence of gambling-related disorders to be almost double in African Americans (2.2%), Native Americans (2.3%), and Asian Americans (2.3%) relative to Caucasians (1.2%) (Alegr<unk>a et al. 2009). Additionally, a cross-sectional study evaluating gambling problems among 96 Southeast Asian refugees found that rates of pathological gambling among that sample were as high as 10 to 25 times that of the general population (Petry et al. 2003). One less-studied cultural group that may also be predisposed to gambling-related disorders is that of Iranian heritage. Gambling has historical relevance for this group; games of chance have been, and continue to be, an important part of the Persian culture. Archaeologists have found predecessors of dice and board games that date back to 3,000 B.C. in excavation sites throughout Mesopotamia where the Persian Empire once reigned (Schwartz 2007). Specifically, backgammon and possibly poker were developed in this region (Martin and Mason 2006;Schwartz 2007). These continue to be popular games amongst the Persian people. In 400 B.C., reports by an ancient Greek historian even describe a Persian king who was an avid gambler, often betting the lives of his servants over games with dice (Schwartz 2007). After the introduction of Islam in the seventh century, gambling became strongly forbidden by religious law (Rosenthal 1975). According to the Koran, "wine, gambling... are an abomination wrought by Satan" (verse 5:90). The identification of gambling in the Koran suggests that communities in that era recognized the presence of problematic gambling and it's potential detriment to the individual and society. Written records from this period even recount tales of irresponsible gambling and it's associated sequela (Rosenthal 1975). Throughout Iranian history, the legality of gambling depended on the current regime in power. Before the 1979 Islamic Iranian revolution, gambling was permitted, and the Shah (or king) owned four out of the eight casinos in the country (IRAN: The Shah Mollifies the Mullahs 1978). With the enforcement of Sharia Law in post-revolution Iran, however, gambling became strictly prohibited. After the 1979 revolution, which replaced the Iranian monarchy with an Islamic theocracy, a wave of Iranians became dissatisfied with the policies of a new religious regime. A period of heavy emigration from Iran ensued, and many Iranians decided to begin anew in the United States. Currently, nearly half a million United States residents (469,569) identify themselves as having Iranian heritage, and almost half of them (232,427) reside in California (American Community Survey-American FactFinder 2009). Unlike many other immigrant populations that came to the United States seeking economic opportunity, many Iranian immigrants left their country with financial security. In fact, compared to other immigrant groups, the Iranian-American community is in a relatively high socioeconomic class. In an analysis of the American Community Survey, Bozorgmehr and Douglas (2011) found that Iranian-Americans were on average well educated (60% have graduate or professional degrees), had high mean incomes ($84,270 males, $35,928 females), and mostly immigrated to the U.S. post-revolution (64%). Due to the atypical socioeconomic status of this population, generalizing findings from other immigrant or refugee groups to Iranian-Americans is challenging and likely inaccurate. Further complicating the study of Iranian-Americans is the fact that in most health-related epidemiological studies they are classified as white or Caucasian. Resultantly, there is a paucity of knowledge about this population's physical and mental well-being. Studies of gambling among Iranian immigrants are even scarcer, with only one small qualitative study having been done in Canada. This study found that while Iranians stigmatized problem gambling, they still recognized recreational gambling as an acceptable social pastime (Zangeneh et al. 2004). Given that cultural norms, values, and attitudes can predispose populations to disordered gambling, and that the prevalence of disordered gambling varies between cultures (Binde 2005;Raylu and Oei 2004), we hypothesize that the heavy influence of gaming on the Persian culture may put Iranian-Americans at increased risk for developing gambling disorders. In this preliminary study, we sought to begin characterizing gambling behavior in the Iranian-American community and to start exploring the various cultural influences that may play a role in the development of gambling disorders in this population. The results may suggest an association between Iranian ethnicity and disordered gambling, and may help build a better understanding of how culture influences the development of gambling disorders. --- Methods --- Setting On September 25 and 26, 2010, data was collected from a convenience sample at an Iranian-American festival, Mehregan 2010-Persian Festival of Autumn, in Orange County, California. According to their website, "Mehregan is a festival originating in ancient Persia and revived annually to celebrate the deep history and culture of the Persian ethnicity" (Shokooh, Farrokh 2006). This 2-day festival celebrates Persian cuisine, music, dance and art; it is the largest gathering of Iranians in the United States. Five thousand attendees were expected at the festival; the authors sought to sample as many participants as possible. Participants were recruited using posters advertising free baked goods as incentive to complete the study questionnaire. Eligibility criteria included an age minimum of 18 years and the ability to read and write English. Individuals who chose to participate gave verbal informed consent, and all study procedures were approved by the UCLA Institutional Review Board for Human Subject Research. --- Gambling Disorder Screening Interested participants completed a 10-min self-administered questionnaire that inquired about gambling behavior and motives for gambling. Gambling disorders were assessed using 10 questions corresponding to the Diagnostic and Statistical Manual, Fourth Edition (DSM-IV) criteria for pathological gambling. The questions asked about preoccupation, tolerance, withdrawal, loss of control, escape, chasing, lying, illegal acts, and risked relationships as they pertained to gambling (APA 2000). Pathological gambling refers to individuals who meet five or more of the criteria; problem gambling refers to individuals who meet one to four of the criteria; and non-disordered gambling refers to individuals who do not meet any of the criteria (Shaffer et al. 1999). This is a valid and reliable diagnostic screener with high sensitivity and specificity (Stinchfield 2003;Zimmerman et al. 2006). From a public health perspective, both pathological and problem gambling are associated with negative individual and societal consequences, suggesting that examining disordered gambling is a significant and important endeavor (Hodgins et al. 2011;Shaffer and Korn 2002). --- Gambling Motives Questions from the Gambling Motives Questionnaire (GMQ) were also included in the study questionnaire to assess gambling motives (Stewart and Zack 2008). The GMQ is a valid and reliable questionnaire consisting of 15 questions asking respondents whether they almost never, sometimes, often, or almost always gamble for given reasons. Answers are scored from 5 to 20 for three subscales (enhancement, social, and coping) indicating the component most likely responsible for the respondent's gambling behavior. --- Data Analysis Gambling behavior was categorized by severity into three groups as follows: non-disordered gambling (meets no criteria for pathological gambling), problem gambling (meets one to four criteria for pathological gambling), and pathological gambling (meets five or more criteria for pathological gambling). This approach to categorizing gambling behavior is consistent with that used in previous studies (Lorains et al. 2011). Sample demographics were compiled and preliminary analyses (Chi-Square tests, Fisher Exact tests, and independent sample T-tests) were conducted to determine whether there were any demographic differences between the gambling groups. Gambling behavior (frequency and type of gambling) was also compiled, and the distribution was reported. Mean gambling motive scores and mean differences between gambling severity groups were compiled for respondents who reported gambling at least once per year. The motive scores were stratified by gambling classification category (pathological, problem, or non-disordered gambler). Paired sample T-tests were conducted to explore differences in gambling motive subscale scores. Survey questions with missing data were included in all analyses as "no answer", a separate response category. SPSS 19.0 was used for all analyses. --- Results 182 volunteers from the Iranian-American festival participated in this study. The most common demographics included Caucasian race (65%, n=119), Iranian/Persian ethnicity (85%, n=160), Iranian birthplace (68%, n=124), Muslim religion (41%, n=74), and Farsi as a primary language (73%, n=133). More detailed demographics are available in Table 1. No significant demographic differences were found between non-disordered gamblers, problem gamblers, and pathological gamblers. --- Gambling Behavior and Pathology Of the 182 respondents, 77% (n=140) reported gambling at least once per year, 20% (n=37) screened positive for problem gambling, and 7% (n=13) screened positive for pathological gambling (see Table 1). The most common types of gambling reported were the lottery, slot machines, and table games at casinos (see Table 2). Of the 140 respondents who reported gambling at least once per year, 26% (n=36) and 9% (n=13) met criteria for problem and pathological gambling, respectively. --- Gambling Motives According to the GMQ, enhancement was the most common motive among respondents who reported gambling at least once per year (Table 3). There was a small but statistically significant difference (0.96-2.00, p<unk>0.001) in scores for all subscales (see Table 4). For pathological gamblers, the difference was considerably elevated between enhancement and coping subscales (mean difference=3.85, p=0.001) and between enhancement and social motive subscales (mean difference=4.69, p=0.001). For problem gamblers, the mean difference between enhancement and coping subscales was also significant (mean difference=3.00, p<unk>0.001). For non-disordered gamblers, however, no single motive subscale played a larger role than others: the mean differences between the motive subscales ranged from 0.32 to 1.34. --- Discussion Using a convenience sample of Iranian-Americans at a festival in Southern California, this study found prevalence rates of disordered gambling in this population to be greatly elevated over that previously found in the general population. The 2006 California Problem Gambling Prevalence Study, using questionnaires similar to the ones in this study, found 12% and 2% lifetime prevalence rates for problem and pathological gambling, respectively (Volberg et al. 2006). Among the Iranian-American community in this convenience sample, 20% and 7% were found to be afflicted by problem and pathological gambling respectively. These rates were even higher when examined exclusively among those who gambled once per year or more. Approximately one in ten respondents who reported gambling at least once per year screened positive for pathological gambling, and approximately one in every four screened positive for problem gambling. Socially, this population may have an elevated risk of developing problematic gambling for a variety of reasons. With the cultural acceptance of gambling as a pastime, Iranian-American children may become indoctrinated to view gambling less negatively than those in other cultures. Gambling is sometimes introduced to children in "Dowrehs" (Farsi word translated as "circles"), or informal periodical social gatherings between Persian families (Mostofi 2003). Iranians organize Dowrehs to hold discussions (e.g., on politics, religion, economics, or finances) and for leisurely activity, which can include gambling (Milani 2008;Mostofi 2003;Naficy 1993;Osanloo 2009). This may also explain why social gambling outside of gambling establishments (poker games outside of casinos, non-poker home card games for money, and board games such as backgammon for money) were commonly reported by the sample participants (see Table 2). Dowrehs take place in the members' homes, and adults who gamble may be viewed by children as social role models in favor of gambling; children may learn and maintain observed gambling behaviors, finding them appealing and reinforcing (Gupta and Derevensky 1997;Gupta and Derevensky 1998). Research suggests that adolescents who describe family gambling behavior report higher rates of gambling themselves (Dickson et al. 2008;Jacobs 2000), and early age onset gamblers display heavier patterns of gambling in adulthood than do adult onset gamblers (Lynch et al. 2004). Psychological factors may also play in the increased risk of Iranian-Americans developing gambling disorders. Many Iranian-Americans are conscious of their culture's aggregate financial, educational, and professional success (Bozorgmehr and Douglas 2011). They may feel nostalgia about previous success or envy of the success of their peers (Daha 2011;Maghbouleh 2010;Mobasher 2006;Mostofi 2003;Naficy 1993). The potentially quick and substantial earnings from gambling wins are appealing to this class, and provide an opportunity for gamblers to experience success and accomplishment. In addition to the financial rewards of winning, feelings of excitement, achievement, triumph, and enhancement also reinforce gambling behavior (Rosenthal and Rugle 1994). As gamblers experience these sensations, they may become behaviorally conditioned to gamble in an effort to re-create this response, despite any negative consequences (Blaszczynski and Nower 2002). Because of their environment and upbringing, some Iranian-Americans may be more prone than members of other cultures to desire this "success fantasy" and use gambling as a means to achieve it. This in turn may predispose them to the development of gambling disorders. The notion that these psychological factors influence gambling behavior in this population may be supported by the results of this study, in particular that enhancement was the most common motive for gambling among respondents (see Table 3). Enhancement in gambling involves internal positive reinforcements that increase positive emotions associated with gambling, such as a gambling high or feeling of excitement and euphoria (Stewart and Zack 2008). After winning, enhancement directly motivates the gambler via the rise in one's ego and the feeling of empowerment and accomplishment. Later, enhancement is indirectly experienced through the purchasing of goods with monetary gains. Moreover, our analysis of motive subscales revealed that enhancement was a significantly more common gambling motivator in this population than either coping or social mechanisms. This may be suggestive that those who are motivated by enhancement to gamble may be at increased risk for gambling disorders relative to those who gamble for coping or social reasons. Notably, the majority of participants reported incomes over $50,000 ($50,000-$100,000, 22%; >$100,000, 35%) and had attained at least a bachelor's degree (bachelor's degree, 39%; master's degree or higher, 36%). Dissimilar to other epidemiological studies which found a significant decline in the presence of disordered gambling as socioeconomic class trended upward (Welte et al. 2011;Welte et al. 2002), no statistical difference was found between the prevalence of gambling disorders and socioeconomic class in this population. Petry and colleagues (2003), in a sample of Southeast Asian refugees, also did not find a relationship between socioeconomic variables and disordered gambling. That study demonstrated an exceptionally high rate of gambling disorders, but their sample consisted of individuals with low mean education and income relative to national averages. Thus, the relationship between cultural influences and gambling pathology may be stronger than that between socioeconomic factors and gambling pathology. This study has three main limitations. The first limitation is the sampling design (convenience sample at one location), which may make these findings difficult to generalize to other Iranian-Americans. The festival may have attracted individuals who are more connected to their ethnic roots. Furthermore, the small sample size may have contributed to the non-significance in demographic characteristics (e.g., age, sex, religion, year of immigration) between gambling severity groups. However, since this was the first study to examine gambling behavior in this population, the data generated was sufficient to make preliminary conclusions and to invite further research into the association of this population with gambling disorders. The second limitation of this study is the bias produced by the selfreported nature of the questionnaires. In this population, some participants may be hesitant to disclose their gambling behavior because of the stigma associated with addictive disorders and pathological gambling. Additionally, gambling is considered unacceptable and sinful according to the majority's Islamic religious doctrine. This submits that the prevalence rates of gambling disorders in this study may even be underestimated, and the actual prevalence may be higher. The third limitation is that the participants' responses were constrained by the questionnaire, which especially pertains to the questions involving motives for gambling. Future research may opt to explore gambling motives with openended questions that would allow participants to elaborate and discuss their motivation for gambling more comprehensively. Despite these limitations, this study has important implications. The greatly elevated prevalence of gambling disorders found in this sample proposes that underlying historical and cultural factors may play an abundant role in the development of gambling disorders. We hope that this study will help raise awareness of the extent of gambling disorders in the Iranian-American community, spawn culturally geared preventative as well as treatment programs, and provide justification for further exploration into the ramifications of disordered gambling in this high-risk community. Other (Armenian, Turkish, Kurdish, Spanish, Tajiri) 5% ( 6 Data from the gambling motives questionnaire was used to explore gambling motives for current gamblers (participants who gambled once per year or more). Scores range from 5 to 20 for each category. Non-disordered gamblers were defined as not having met any DSM-IV criterion, while problem and pathological gamblers met one to four criteria and five or more criteria, respectively SEM standard error of the mean; CI confidence interval Data from the gambling motives questionnaire was used to explore gambling motives for current gamblers (participants who gambled once per month or more). Scores range from 5 to 20 for each category. Non-disordered gamblers were defined as not having met any DSM-IV criterion, while problem and pathological gamblers met one to four criteria and five or more criteria, respectively. Paired-samples T tests were conducted to examine the difference between different gambling motives SE standard error; df degrees of freedom; --- Sample characteristics based on gambling disorder CI confidence interval

Nearly half a million United States residents identify themselves as being of Iranian origin, and many in this population are of high socioeconomic status. Although games of chance have been a notable part of Iranian culture for thousands of years, there is almost no research exploring gambling in this population. The objective of this case study is to explore gambling pathology, gambling behavior, and gambling motives among Iranian-Americans using a convenience sample (N=182) at a September 2010 Iranian festival in Southern California. Of this sample, 20% (n=37) and 7% (n=13) screened positive for problem and pathological gambling, respectively. According to the Gambling Motives Questionnaire, enhancement was the preferred motive for gambling ("because you like the feeling, because it's exciting, to get a high feeling, because it's fun, because it makes you feel good"). Pathological gamblers showed a considerable difference in subscale scores between enhancement and either coping or social motives, and problem gamblers showed a considerable difference between enhancement and coping motives. Possible explanations for the higher prevalence of gambling disorders in this sample are discussed. Our results support the notion that underlying cultural factors play a role in the development of gambling disorders.

Introduction The 'Global Strategy for Women's, Children's and Adolescents' Health 2016-2030' is a global strategy aimed at eradicating all preventable maternal, newborn, and child-related deaths, including stillbirths, by 2030, as well as improving women's and children's overall health and well-being [1]. However, as the world grapples with the devastation caused by the SARS-CoV-2 pandemic, this global commitment to ensuring women's health and well-being is jeopardized. Women may be more vulnerable to the pandemic's ensuing implications in several low-income countries, where social safety nets may be limited. Furthermore, we know little about the relationship between pre-existing socioeconomic circumstances and vulnerability to the pandemic and whether female-headed families fare worse than their male-headed counterparts. We investigate these issues in the context of Ethiopia and Kenya using longitudinal data collected by the World Bank to track the impact of the pandemic on livelihoods. Female-headed families are likely to fare poorly during pandemics as women may have to balance their time between childcare and employment. To contain the virus and prevent further harm to the populace, governments around the world implemented various measures, including lockdowns, curfews, social seclusion, and wearing masks, among others. Even though such policies are implemented with the overall good in mind, they inevitably have unfavourable effects, especially when poor social safety net programmes are present. The emergence of the pandemic disrupted global food markets and, consequently, supply chains. Markets are essential in ensuring the availability and accessibility of food, which is one of the pillars of food security. In Ethiopia, value chain agents noted declining demand, clientele, and turnover, higher losses, less competition, higher transportation expenses, and modifications to the procurement landscape [2]. The pandemic had a negative impact on farmers in Kenya as well, disrupting supply chains, increasing the inflation of commodity prices, and reducing trade volumes [3,4]. Rising commodity prices limit the amount of food that can be purchased because they lower people's purchasing power or indirectly erode their disposable income. Recent research has emerged showing that restrictions imposed by governments to minimize the spread of the virus, such as lockdowns, are more likely to impact women in many respects disproportionately [5]. For example, women, particularly those living in lowincome countries, are more likely to be impacted economically since a larger fraction is in precarious or unstable employment situations, including the informal sector, entertainment industry, arts sector, and domestic services [6]. A great number of other women are also employed in poorly paid front-line positions, for example, community health workers who continue to tackle risky tasks such as COVID-19 surveillance, contact tracing, vaccinations, and monitoring quarantine and isolation centres, and yet these essential workers are often lowly or irregularly paid with inadequate protective equipment [7][8][9]. Given the erratic nature of their income sources, people in precarious employment are more likely to experience food insecurity. In this paper, we examine the aggregate impact of the pandemic on livelihood outcomes (income, consumption, and food insecurity) in Kenya and Ethiopia during the SARS-CoV-2 pandemic. Our empirical analysis relies on longitudinal data collected by the World Bank and its partners through high-frequency phone surveys. We contribute to the current discussions on the effect of the SARS-CoV-2 pandemic in low-income countries. First, we use high quality longitudinal data to provide insights on the gendered impact of the SARS-CoV-2 pandemic on livelihoods in Kenya and Ethiopia. Second, we show that pre-existing socioeconomic inequalities exacerbated the impact of the SARS-CoV-2 pandemic on livelihoods in Kenya and Ethiopia. These results underscore the need for governments to develop suitable gender-sensitive measures to mitigate the impact of future pandemics in Kenya, Ethiopia, and other economies with comparable structures, thus fostering an equitable and more inclusive response to future pandemics. --- SARS-CoV-2 Pandemic, Gender, and Outcomes The SARS-CoV-2 pandemic is more likely to exert social, psychological, health, and economic repercussions on communities, making some people more vulnerable to its negative effects on their livelihoods and well-being outcomes than others. Insecure housing, restricted access to health care, poverty, gender disparities, racial segregation, food insecurity, changing patterns of consumption, and loss of income and employment are among the factors influencing susceptibility to the SARS-CoV-2 pandemic and the impact of health and well-being outcomes [10][11][12]. The social determinants of health are a concept that encompasses all these aspects. The social determinants of health are defined by the World Health Organisation (WHO) as the "conditions in which people are born, grow, live, work and age" and "the fundamental drivers of these conditions" [11]. These life circumstances are, in turn, influenced by the distribution of income, power, and resources at both the local, national, and global levels. Premature death and disease are greatly influenced by social determinants of health, particularly among vulnerable groups such as women, children, the elderly, and minorities. Moreover, the impact of COVID-19 is less likely to be uniform across countries and even within the same country. Women, children, and the elderly are amongst the most vulnerable groups of the population. Recent evidence has also shown that women living in female-headed families are at an elevated risk of experiencing poor food security and well-being outcomes [13]. Using data from India, Bau, Khanna [13] show that women from vulnerable positions within the household, including those with children, are more likely to experience poor food security and mental well-being outcomes. One of the global measures recommended to minimize the spread of the SARS-CoV-2 virus was the implementation of lockdowns. Research has emerged showing that such measures have disproportionately impacted women and girls in several aspects [5]. Women, particularly those living in low-income countries, are more likely to be impacted economically since a larger fraction are in precarious or unstable employment, which includes the informal sector, entertainment industry, arts sector, and domestic services, among others [6]. A great number of other women are employed in poorly paid front-line positions, for example, community health workers who continue to tackle risky tasks such as COVID-19 surveillance, contact tracing, vaccinations and monitoring quarantine and isolation centres, and yet these important workers are often lowly or irregularly paid with inadequate protective equipment [7][8][9]. --- Materials and Methods --- Data This study uses longitudinal data from high-frequency phone surveys conducted in Kenya and Ethiopia by the World Bank in partnership with the local governments. In Kenya, the survey is called the "rapid response phone survey (RRPS)" [14]. Survey data for Ethiopia is available up to 12 rounds, with the 12th round ending in June 2021 [15]. For our analysis, we used the first six rounds of the Ethiopian household high-frequency phone survey (HFPS). The Ethiopian household HFPS began towards the end of April 2020, and households were then called back every three to four weeks. The survey was carried out using Computer-Assisted Telephone Interviewing (CATI) in a modular fashion. The sample of Ethiopian HFPS households was drawn from the sample of Ethiopian Socioeconomic Survey (ESS) households interviewed in the 2018/2019 round. The extensive information gathered in the ESS a few months before the pandemic provides rich information and context for investigating the pandemic's impact on livelihoods. The ESS is based on a nationally and regionally representative sample of Ethiopian households, with 6770 households interviewed in urban and rural areas. Households were asked in the survey to provide phone numbers, either their own or those of a reference household, including friends or neighbours, so that they could be contacted in the follow-up ESS surveys if they moved away from the sampled location. The sampling frame for the HFPS was made up of 5374 households that had at least one valid phone. The target sample size for the HFPS was 3300 households to obtain a representative stratum at the national, urban, and rural levels (1300 in rural and 2000 in urban areas). A detailed description of the sampling methodology and questionnaires used to collect all the data is available here [15]. The World Bank also conducted the RRPS survey in Kenya in collaboration with the Kenya Bureau of Statistics and the University of California, Berkeley. The dataset we accessed contains information from eight waves of the COVID-19 RRPS, which is part of a panel survey that began in May 2020, and the eighth wave was completed in July 2022. Even though the dataset we accessed contains eight waves of the COVID-19 RRPS, we use the first six waves for both countries to harmonise outcome variables and other key variables for the analysis. CATI techniques were used to conduct the Kenyan RRPS, just like in the Ethiopian survey. The Kenyan dataset consists of two randomly chosen samples of households that participated in the computer-assisted personal interviewing (CAPI) trial for the 2015-2016 Kenya Integrated Household Budget Survey (KIHBS) and provided a phone number [16]. The second sample of households was gathered using Random Digit Dialling technique, in which active phone numbers generated by the Kenya Communications Authority's 2020 Numbering Frame were chosen at random. The sample, which includes both urban and rural areas, was created to be an accurate representation of Kenya's cell phone-using population. A detailed sampling procedure including survey questionnaires is available here [14]. Kenya's final sampling frame included 4075 households with active phones. The purpose of these telephone surveys was to interview a nationally representative sample of households excluding those in prisons, hospitals, military barracks, and school dormitories to gauge the socioeconomic impact of the SARS-CoV-2 pandemic on livelihoods and thereby inform a targeted response. The high-frequency phone surveys collect an array of information including household background, access to basic services, employment, food security, income loss, transfers, agriculture, health, education and childcaring, and COVID-19-related knowledge among several other topics. Both surveys include a set of weights that we incorporated in our analysis to obtain unbiased estimates. --- Measures --- Food Insecurity during the SARS-CoV-2 Pandemic The high-frequency phone surveys asked several questions concerning the food security situation of individual households during the pandemic. The reference period for these questions was 30 days. However, questions were also asked on the number of adults and children that had either gone to bed hungry, skipped meals, or gone for entire days without food in the household and over the past seven days. Specifically, the questions asked were of the following form: "In the past seven days, how many days have adults in your household... (1) gone to bed hungry? (2) skipped meals or cut the amount of meals? (3) gone entire days without food?". We focus on the questions that aimed to know whether the individual respondent or any other adult member from their household: (i) worried about the possibility that the household would not have enough food to eat because of a lack of money or other resources, (ii) had to go to bed hungry because of lack of money or other resources, (iii) had to skip a meal or reduce the amount of meals due to lack of money or other resources, and (iv) had run out of food because of lack of money or other resources. In Kenya, the surveys also asked whether any children in the household had skipped a meal or reduced the number of meals due to lack of money or other resources. We use the responses to these questions to create individual dummy variables that equals one if a respondent had answered yes to a question (or had indicated having one or more days they had gone hungry or skipped a meal, or they had gone entire days with no food) and zero otherwise, thereby creating outcomes reflecting hunger, food running out, adults skipping meals, children skipping meals, and worrying over the prospect of inadequate food in the household. These indicators are also consistent with those commonly used in the literature [17]. --- Income and Consumption Losses during the SARS-CoV-2 Pandemic The high-frequency phone surveys include a separate section on income loss. In Kenya, each respondent was asked whether, during the past 14 days, his/her household had to sell livestock or other household assets to generate income (such as vehicles, furniture, kitchen or electronic equipment, and tools), took a loan for use on household consumption, as well as the kind of loan they had taken (i.e., whether from a friend/relative or commercial bank, among others), had their business closed, had to rely on credit purchases, or reduced food consumption in a bid to cope with the effects of the pandemic. These questions indicate an immediate response to the impact of the SARS-CoV-2 pandemic in Kenya. In Ethiopia, the questions in round one asked respondents whether income from several sources including overall household income, business, farm, other sources, and remittances had increased, stayed the same, reduced, or was completely lost 100% following the COVID-19 outbreak. Questions in subsequent surveys (round 2 onwards), ask respondents whether, since the last phone call, income from the same sources noted earlier had increased, stayed the same, reduced, or was completely lost 100%. These questions capture the aggregate impact of the SARS-CoV-2 pandemic on income losses in Ethiopia. We use the answers to these questions to generate several dummy variables that represent income or/consumption losses. Specifically, we created a series of dummy variables to represent income or consumption loss, and these dummy indicator variables equalled one if the respondent indicated that their income or consumption had decreased or was completely lost because of the pandemic, and zero otherwise. For consumption losses and income losses, separate variables are generated. --- Empirical Strategy The study evaluates the gendered impact of the SARS-CoV-2 pandemic on livelihoods (income, consumption, and food insecurity) and inequity in access to health services and necessities in Kenya and Ethiopia. In this study, we use the gender of the head of household as an indicator for gender. We test whether individuals living in female-headed families fare worse when compared to those living in male-headed families. To fulfil the aims of the study, the empirical analysis proceeds in two steps. First, we explore the relationship between household headship and livelihood outcomes accounting for variables that reflect potential vulnerability within households in Kenya and Ethiopia. To this end, we assess whether women from female-headed households experience worse outcomes during the pandemic compared to their counterparts in male-headed families. In this instance, vulnerability relates to those individuals from families with children, low levels of education, and lower pre-COVID-19 socioeconomic status level. The model we estimate takes the following form: Y iarct = <unk> 1 children i + <unk> 2 f emale_head i + a a + <unk> r + w t + <unk>X i + <unk> iarct(1) where i represents the individual respondent, a represents her age, r represents her region of residence, c is her county of residence, Y iarct measures the outcome variable (income losses, lost consumption, and food insecurity outcomes), children i is a categorical variable with four categories and representing the number of children in the household (1 = no children, 2 = 1-2 children, 3 = 3-4 children, and 4 = 5 or more children), and f emale_head i denotes whether the respondent resides in a family or household where the head is female. Previous research suggests that households with children are more likely to experience food insecurity [18,19]. In all the estimated specifications, we include age-fixed effects a a, region-fixed effects <unk> r, and survey-fixed effects w t. The vector of additional control variables X i incorporates the respondent's level of education (no education, primary education, and secondary or higher education) since a higher level of schooling is associated with a lower prospect of food insecurity [20]. We also include controls for whether the respondent had lost a job during COVID-19, and lives in a rural or urban area. To minimize omitted variable bias, we include pre-pandemic socioeconomic status represented by an asset index with five quintiles (poorest (quintile 1), poorer, average, richer, and richest (quintile 5)). We follow Filmer and Pritchett [21] in generating a household asset index using principal components analysis (PCA). Survey respondents were asked questions regarding ownership of several assets prior to the pandemic, including radio, mattress, charcoal jiko, refrigerator, television, landline, or computer/laptop before March 2020. In further analysis, we use this prepandemic household wealth measure to assess whether the pandemic had a greater impact on households who were already in more vulnerable positions. The standard errors, <unk> iarct, are clustered at the county or enumeration area level. --- Identifying the Impact of the SARS-CoV-2 Pandemic The study evaluates the impact of the SARS-CoV-2 pandemic on livelihood outcomes in Kenya and Ethiopia. We specifically test whether the pandemic had a disproportionately larger impact among female-headed families when compared to male-headed families. To identify the impact of the SARS-CoV-2 pandemic, we rely on the nature of the questions asked in the high-frequency phone surveys that allow us to measure the aggregate impact of the pandemic as these relate to the changed circumstances following the onset of the pandemic. For Kenya, since we were unable to identify or match the pre-COVID-19 data to the COVID-19 high-frequency phone survey data, identifying the impact of the pandemic is somewhat complicated for some of our outcomes particularly those relating to food insecurity. The questions asked in the surveys refer to the prevailing situation or changed circumstances of the household in the past one week or two weeks within the last 30 days. This is a much shorter reference window and suggestive of the changed circumstances of the household following the emergence of the pandemic. We interpret these findings on the impact of the pandemic on food insecurity bearing in mind that what we document are mere associations indicating the prospect of experiencing poor outcomes among individuals from female-headed families during the SARS-CoV-2 pandemic. However, in some of the food insecurity outcomes, the survey question allows us to clearly identify the aggregate impact of the pandemic. For example, respondents were asked the following question: "Compared to before March 2020, before the lockdown/pandemic, are you more/less/equally worried about your household not having enough food?". For outcomes linked to income losses and difficulty with access to necessities, we can clearly identify the aggregate impact of the pandemic. In Kenya, the survey questions asked respondents on the specific coping strategies that the household adopted in response to the income losses prompted by the SARS-CoV-2 pandemic (as described earlier). The coping strategies such as the selling of personal assets, borrowing from friends or relatives, taking loans from financial institutions, delaying payment obligations, credited purchases, business closures, and reducing food consumptions all clearly reflect the aggregate income losses following the onset of the pandemic. In Ethiopia, the questions on income losses clearly capture the aggregate impact of the pandemic as these reflect the changes in income from the onset of the pandemic and measured at a specific point in time. --- Results --- Summary Statistics Table 1 reports the weighted summary statistics for selected variables and using only the first wave of each country. Using the data for the first wave of the high-frequency surveys for either county, we observed that the proportion of female-headed families was 24.38% and 30.09% in Ethiopia and Kenya, respectively. The average age of respondents was higher in Ethiopia 39.01 vs. 35.30 years in Kenya. The proportion of households with no children was 22.24% in Ethiopia compared to 31.55% in Kenya. Most of the households in both countries had at least 1-2 children, with 45.33% in Ethiopia compared to 41.33% in Kenya. We observed a smaller fraction of households reporting to have five or more children, 8.07% in Ethiopia compared to 5.22% in Kenya. In terms of socioeconomic status of households prior to the SARS-CoV-2 pandemic, 16.69% compared to 26.88% of the households were classified as the poorest (asset quintile 1) in Ethiopia and Kenya, respectively. In Kenya, only 6.47% of the households were classified as the richest (asset quintile 5) compared to 22.96% in Ethiopia. The fraction of households living in rural area was comparable at 66.89% in Ethiopia and 66.52% in Kenya. Note: Data comes from the first wave of the high-frequency phone surveys for Kenya and Ethiopia. --- SARS-CoV-2 Pandemic and the Vulnerability of Women within the Household The results in Tables 2 and3, estimated using Equation ( 1), speak to the vulnerability of women during the SARS-CoV-2 pandemic in Kenya. In this instance, we explore the relationship between family structure and food insecurity outcomes including outcomes linked to income and consumption losses during the pandemic in Kenya. We show that women who are in more vulnerable positions within the household were more likely to fare worse compared to their counterparts from male-led families during the pandemic. When the head of household is female, the probability of going hungry during the pandemic increased by an estimated 3.96 percentage points (pp) and is statistically significant at the 1% level. Given that the mean of the outcome variable in our analysis sample was 39.40%, the 3.96 pp effect represents an approximate 10% 0.0396 0.394 <unk> 100 increase in the probability that the respondent or an adult from a female-headed family would go to bed hungry in the past seven days and during the pandemic in Kenya. Households with five or more children were 14.33 pp more likely to have any adult go hungry when compared to households with no children and statistically significant at the 1% level. There is a clear positive gradient suggesting that the prospect of hunger increases with the number of children in the household. The probability of an adult going hungry is exacerbated if an adult member from the household had been laid off or lost their job involuntarily since January of 2020. Having lost a job due to COVID-19 was associated with a 10.92 pp increase in the probability of going hungry during the pandemic period. Additionally, the pre-COVID-19 socioeconomic status of the household, as measured by the household wealth index, is an important determinant of the probability of going hungry. We observed that, compared to families that were classified as richest (asset quintile 5), families in the bottom poorest quintiles are more likely to experience hunger during the pandemic. Being in the poorest asset wealth group (quintile 1) before the pandemic was associated with a 16.33 pp increase in the probability of going hungry during the pandemic. The effects on hunger are also compounded when family resided in a rural area as compared to an urban locality (4.18 pp). Table 2 also indicates that when the respondent lives in a female-headed family, she or another adult from the same household is 3.42 pp more likely to skip a meal or cut the number of meals eaten in the past seven days prior to the survey and during the pandemic. The 3.42 pp effect represents an imprecise 9.88% increase in the probability that an adult from a female-headed family skips a meal. The effects on the prospect of skipping meals or cutting the number of meals eaten are also exacerbated when the family has children compared to when there are no children, when a household member had lost a job during the pandemic, the family lives in a rural area, and the family is of relatively low wealth compared to other families classified as rich prior to the pandemic. When the respondent lived in a female-led family, a child is 3.33 pp more likely to skip a meal. Given that the average probability that a child skipped a meal in our Kenyan sample was 19.40%, the 3.33 pp represents an approximate 17.16% increase in the probability that a child had skipped a meal in the seven days prior to the phone survey and during the pandemic. The probability that a child skips a meal is further exacerbated when the family has children in the household, someone in the family had lost a job, the family lives in a rural area, and the family was relatively poor prior to the pandemic. The results also show that, when the respondent lives in a female-headed family, she is 3.41 pp more likely to worry about not having enough food compared to the period before March 2020 and before the lockdown or pandemic. Since 53.50% of the respondents in the analysis sample indicated that they were increasingly more worried that their household would not have enough food, the 3.41 pp effect represents an imprecise 6.37% increase in anxiety over the prospect of the family not having enough food compared to the pre-COVID-19 period. Table 3 reports the results from estimating Equation (1) using the analysis sample for Kenya. Here, we are interested in examining whether families headed by a female respondent fare poorly in terms of income losses during the pandemic in Kenya. We include the same set of controls as in Table 2. The results show that when the respondent lives in a female-headed family, there is a 2.4 pp decline in the probability that the household would have sold livestock to cope with the effects of the pandemic. This result is statistically significant at the 1% level. The point estimates in Table 3 also indicate that being from a female-led household was associated with a 1.20 pp increase in the probability of taking a loan for use on household consumption in the past 14 days prior to the phone survey. Given that the mean of the dependent variable was 8.3% in our sample, the 1.20 pp effect represents an imprecise 14.46% increase in the probability of taking a loan for use on household consumption. When the respondent lived in female-led family, the prospect of borrowing from friends increased by 2.49 pp and is statistically significant at the 1% level. Female-headed families were 0.46 pp less likely to sell other assets during the pandemic in Kenya. The point estimates also show that female-led families were 6.51 pp more likely to report having closed a business due to effects of lockdowns or curfews and 3.88 pp more likely to rely on credit purchases to cope with the effects of the pandemic. A female-led family was 3.03 pp more likely to report reducing food consumption during the pandemic and is statistically significant at the 1% level. Given that the mean of the dependent variable was 45.6% in our sample, the 3.03 pp effect represents an approximate 6.64% reduction in food consumption among female-headed families in Kenya during the SARS-CoV-2 pandemic. Table 4 reports the point estimates from estimating Equation (1) using the data for Ethiopia. Living in a female-headed family is associated with a 2.07 pp increase in the probability that the respondent or an adult from their household went hungry and did not eat because there was not enough money or other resources for food in the past 30 days prior to the survey. Given that the mean of the dependent variable in our sample was 8.5%, the 2.07 pp effect represents an imprecise 24.35% increase in the prospect of going hungry during the pandemic in Ethiopia. The effects of the likelihood of hunger are also exacerbated when there are children in the household, the respondent or other household member had lost a job due to the pandemic, and the respondent was from a relatively poor family prior to the pandemic. These effects are statistically significant. Being able to read and write, reside in rural area, and being single appear to be negatively associated with the prospect of going hungry. The next column reports the point estimates exploring the likelihood that an adult would skip a meal during the pandemic. When the respondent lives in a female-headed family, she or another adult from their household is 4.42 pp more likely to have skipped a meal in the past 30 days prior to the telephone survey and during the pandemic in Ethiopia. This 4.42 pp effect represents an imprecise 18.89% increase in the probability of skipping a meal during the pandemic. The prospect of worrying that the household would not have enough food to eat because of lack of money was 7.2 pp higher among female-headed families compared to their counterparts. This 7.2 pp effect represents an imprecise 17.85% increase in the probability of anxiety over not having enough money or other food resources in the household in Ethiopia and during the pandemic. When the respondent comes from a female-headed household, the probability that the household had run out of food because of lack of money in the 30 days prior to the survey was 5.36 pp higher when compared to male-headed families. This effect represents an approximate 26.67% increase in the probability that the household would run out of food due to lack of money in the 30 days prior to the phone survey. The prospect of not eating for a day was 1.53 pp higher among female-headed households, representing an approximate 21.86% increase in the likelihood of not eating for the entire day. The observed effects were also exacerbated when the respondent came from a household with children, had lost a job due to SARS-CoV-2 crisis, and was from a relatively poor family before the pandemic started. Notes: *** Significant at 1% level; * significant at 10% level. In Table 5, we report the aggregate impact of the pandemic on income in Ethiopia and assess whether female-headed families fare worse compared to their male-headed counterparts. Each respondent in the Ethiopian phone survey was asked the following question: "Since [LAST CALL], has income from [SOURCE] increased, stayed the same, reduced, or total 100 loss?". We use responses to this question to indicate loss (reduced or 100% loss) of total household income, business income, farm income, lost income from other sources, and lost remittances. The results show that being from a female-headed family was associated with a 2.73 pp increase in the probability of losing overall or total household income during the pandemic in Ethiopia. Given that the average fraction of households reporting losses in total household income was 38.8%, the 2.73 pp effect represents an approximate 7.04% decline in total household incomes during the pandemic in Ethiopia. The effect on lost business income is also exacerbated when the head of the family is female when we control for pre-pandemic socioeconomic status. When the respondent lives in a female-led household, the probability of reporting reduced business income was 1.29 pp, representing an approximate 53.75% decline in the probability of reporting reduced business incomes among female-headed families in Ethiopia and during the pandemic. Female-headed families also reported having lost remittances, a 1.41 pp increase, representing an approximate 60.04% decline in the probability of remittance income. This study has shown that in Kenya and Ethiopia, the pandemic disproportionately negatively affected people who lived in female-headed households. We explored several pre-pandemic socioeconomic characteristics of households in these countries that could make female-headed families more vulnerable than their male-led counterparts. We consider factors such as education level, access to health insurance, literacy, marital status, and ownership of assets such as radio, television, and kitchen appliances. We present these results in a Supplementary Materials (Tables S1 andS2). In Table S1, we noted that a female-headed family in Kenya was less likely to report owning a radio (6.60 pp), television (12.80 pp), or a computer/laptop/tablet (2.10 pp). The results also indicate that being from a female-headed family was associated with a 3.30 pp decrease in the probability of having completed secondary school but a 3.90 pp higher likelihood of having internet access at home. All the results were statistically significant. For Ethiopia, we noted that the likelihood of owning a radio, television, gas stove, electric stove, and refrigerator declined significantly. Additionally, the prospect of being divorced was much higher (44.90 pp) among individuals from female-headed families. The probability of being married, literate, ever attending school, and of having health insurance was also significantly lower among female-headed families. For example, being from a female-headed family in Ethiopia was associated with a 14.20 pp decline in the probability of being able to read and write. The findings suggest that the SARS-CoV-2 pandemic's impact was exacerbated by the pre-existing inequality in socioeconomic circumstances of female-headed families in Kenya and Ethiopia. --- Discussion The primary objective of this paper was to investigate the impact of the SARS-CoV-2 pandemic on livelihood outcomes and ascertain whether female-headed families fare worse than their male-headed counterparts in Ethiopia and Kenya. The pandemic was associated with increased food insecurity and decreased incomes and consumption, particularly among female-headed families. Our results are consistent with emerging literature in lowincome countries (see, for example, [22,23]). Given the unequal livelihood implications, the findings highlight the need for governments to adequately prepare for future pandemics by developing strategies that are gender sensitive. Our results indicate that individuals in more vulnerable positions within the household, those with children, and those of low socioeconomic status before the SARS-CoV-2 pandemic were likelier to experience worse food insecurity outcomes, lose incomes, and reduce food consumption during the pandemic. Indeed, previous evidence across several countries, including the United States, suggests that households with children are more likely to experience food insecurity during pandemics than those without children [24]. More children, thus more mouths to feed, increases intra-household competition for food resources [25]. Given that several governments across the globe implemented policies to contain the pandemic, such as lockdowns and curfews, several households bore undesirable consequences through food insecurity, lost incomes, and lower consumption. In Kenya and Ethiopia, where a significant fraction of women was in precarious employment, losing incomes and wage employment was inevitable and had far-reaching implications on food insecurity [26][27][28][29]. The finding that female-headed families from low socioeconomic status were more likely to experience worse food insecurity outcomes reflects the vulnerability of femaleheaded families in low-income countries. The observation that female-headed families were generally poorer before the pandemic is consistent with economics studies contending that female-headed families in low-income countries are, on average, poorer [30,31]. The additional analysis further supports this finding we conducted that shows that these families are less likely to own even the very basic of assets such as a radio, television, cooking stove, and refrigerator, among others. Household wealth indicates a family's material well-being distinct from expenditure and may reflect the vulnerable households [32]. The lack of asset wealth increases a household's vulnerability to shocks such as the SARS-CoV-2 pandemic in that they will have no assets to draw on or at least sell for their own consumption smoothing needs. The latter was confirmed in Kenya (though the result was not statistically significant). We found that female-headed families were more likely to have had no assets before the pandemic when could then explain their heightened vulnerability during the pandemic. Overall, the finding that female-headed families from low-socioeconomic positions are more likely to be vulnerable to the pandemic suggests that governments in low-income countries should prioritise targeting impoverished female-headed families living with children for a more inclusive and equitable response to future pandemics. Our analysis also reveals the negative implications of the pandemic on consumption and income. In both countries, to cope with the adverse effects of the pandemic, most households lost consumption by skipping meals, while others responded through credit purchases. The impacts on consumption are more likely to be explained by the fact that many families in our sample reported losing employment, closing businesses, and having no remittances, contributing to reduced household income. We also show that pre-existing socioeconomic ine

The SARS-CoV-2 pandemic has revolutionised our lives, bringing with it the twin crises of illness and the need for an optimal mix of policies to alleviate its impact on the population. There needs to be more evidence on the effects of the pandemic on livelihood outcomes, including an understanding of whether female-headed families in low-income countries fare worse than their male-headed counterparts during pandemics. Using high-frequency phone surveys conducted in Ethiopia and Kenya, we examine the aggregate impact of the pandemic on income and consumption losses, as well as food insecurity. The empirical analysis estimates linear probability models that relate livelihood outcomes with household headship and other socioeconomic characteristics as controls. Overall, the pandemic increased the likelihood of food insecurity while decreasing income and consumption, particularly among female-headed households. In Kenya, living in a female-headed home increased the possibility of an adult going without food by about 10%, an adult skipping a meal by about 9.9%, and a child missing a meal by about 17% in the seven days preceding the telephone survey. In Ethiopia, living in a female-headed household increased the likelihood of an adult going hungry, skipping a meal, and running out of food by about 24.35%, 18.9%, and 26.7%, respectively. Salient pre-existing socioeconomic inequalities further exacerbated the effects of the pandemic on livelihoods. The findings have important implications for public policy and preparations by governments and other organisations interested in developing suitable gender-sensitive measures to lessen the impact of future pandemics in low-and middle-income countries.

shocks such as the SARS-CoV-2 pandemic in that they will have no assets to draw on or at least sell for their own consumption smoothing needs. The latter was confirmed in Kenya (though the result was not statistically significant). We found that female-headed families were more likely to have had no assets before the pandemic when could then explain their heightened vulnerability during the pandemic. Overall, the finding that female-headed families from low-socioeconomic positions are more likely to be vulnerable to the pandemic suggests that governments in low-income countries should prioritise targeting impoverished female-headed families living with children for a more inclusive and equitable response to future pandemics. Our analysis also reveals the negative implications of the pandemic on consumption and income. In both countries, to cope with the adverse effects of the pandemic, most households lost consumption by skipping meals, while others responded through credit purchases. The impacts on consumption are more likely to be explained by the fact that many families in our sample reported losing employment, closing businesses, and having no remittances, contributing to reduced household income. We also show that pre-existing socioeconomic inequalities have exposed families in terms of food insecurity, income, and consumption. Additionally, families from lower socioeconomic quintiles had a higher likelihood of experiencing worse livelihood outcomes. These results underscore the need for governments to realise that a one-size fits all approach will not be an effective way to deal with future pandemics. These results also assist the government in prioritising public resources and focusing on deprived areas and vulnerable groups of the population, such as women and children, so as not to leave anyone behind and hence focus on a more inclusive response [25,33]. Female-headed families in Kenya and Ethiopia were also more likely to have reported reduced household income, closed their businesses, and lost remittances. One possibility is that female-headed households are likely to lose their primary source of income during the pandemic since most women in low-income countries, including Ethiopia and Kenya, work in precarious employment arrangements and, hence, are more likely to lose their jobs-the primary source of their income [34]. The informal sector is relatively large in Kenya, employing at least 70% of workers, with women comprising at least 66% [35]. In Ethiopia, women are also overrepresented in the informal sector [36]. The precarious nature of such employment arrangements in the absence of social protection or where social protection is weak or slow to react makes this group of women even more vulnerable. Studies in Kenya's informal settlements demonstrated that an estimated 43% of people lost all their income as the pandemic unfolded [34]. These results are also consistent with early research in developing countries showing that the SARS-CoV-2 pandemic had an unequal impact among individuals, with women bearing the more significant brunt [37]. In several countries across the globe, women were reported to have had difficulties coping with the pandemic [31]. This finding demonstrates that pandemics such as that of SARS-CoV-2 are more likely to generate and exacerbate economic inequality. Given the lower state capacity to deal with impending shocks in many low-income countries, female-headed families and other vulnerable groups are at increased risk of experiencing food insecurity [37]. There are a few limitations to our analysis. First, our analysis does not uncover a causal relationship between the SARS-CoV-2 pandemic and livelihood outcomes. Our results could be subject to selection bias since our data is only nationally representative of the population using telephones or mobile phones in Kenya and Ethiopia. Second, many data sets, including high-frequency phone surveys, are still being collected only at the household level, and this constitutes a limitation for proper gender analysis. In this instance, we cannot know the intra-household food allocations, nor which specific family members in the household are going hungry or skipping a meal or their gender. Nevertheless, our analysis provides valuable insights into the gendered impact of the SARS-CoV-2 pandemic in Kenya and Ethiopia. --- Recommendations This study has provided new empirical evidence for Kenya and Ethiopia regarding the impact of the SARS-CoV-2 pandemic on livelihood outcomes. The results point towards the vulnerability of female-headed families during the pandemic. Notably, we show that pre-existing socioeconomic inequalities further exacerbate the impact of the pandemic, with female-headed families bearing the brunt of its effects. There are several policy options available that governments could use to better prepare for future pandemics. The ideas we suggest here are not short-term interventions, but rather medium to longer term options to address pre-existing socioeconomic vulnerabilities. Considering our results, there is a need for governments in Kenya and Ethiopia to craft policies that deliberately target the most vulnerable groups of the population, especially female-headed families living with children and families from low socioeconomic positions to promote an equitable response to future pandemics and maximise social protection. Additionally, governments with a forwardlooking perspective could prioritise policies that promote growth and development across low-income communities to reduce poverty and, as a result, socioeconomic inequalities over time, thus building resilience to future pandemics. Fundamentally, policy responses to the crisis must incorporate a gender lens and consider women's particular needs, obligations, and perspectives. --- Conclusions In this paper, we studied the implications of the SARS-CoV-2 pandemic on food insecurity, incomes, and consumption using high-frequency telephone survey data from Ethiopia and Kenya. We show that female-headed families, including those with children and from lower socioeconomic status before the pandemic, were disproportionately impacted by the pandemic. From a policy standpoint, our findings highlight the importance of clearly understanding pre-existing socioeconomic circumstances to formulate an inclusive and equitable response to impending pandemics. Future research should investigate the impact of pre-existing socioeconomic inequalities on income-related disparities in access to healthcare services and necessities during pandemics in low-and middle-income countries. --- Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph20065048/s1. Table S1: Relationship between being a female-headed family and the probability of having selected characteristics before the COVID-19 pandemic in Kenya; Table S2: Relationship between being a female-headed family and the probability of having selected characteristics before the COVID-19 pandemic in Ethiopia. --- Data Availability Statement: The data used for this study are publicly available upon a formal request to the World Bank. Data for Kenya can be requested here: https://microdata.worldbank.org/ index.php/catalog/3774/get-microdata (8 March 2023). Data for Ethiopia can be requested here: https://microdata.worldbank.org/index.php/catalog/3716/data-dictionary (8 March 2023). --- Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

The SARS-CoV-2 pandemic has revolutionised our lives, bringing with it the twin crises of illness and the need for an optimal mix of policies to alleviate its impact on the population. There needs to be more evidence on the effects of the pandemic on livelihood outcomes, including an understanding of whether female-headed families in low-income countries fare worse than their male-headed counterparts during pandemics. Using high-frequency phone surveys conducted in Ethiopia and Kenya, we examine the aggregate impact of the pandemic on income and consumption losses, as well as food insecurity. The empirical analysis estimates linear probability models that relate livelihood outcomes with household headship and other socioeconomic characteristics as controls. Overall, the pandemic increased the likelihood of food insecurity while decreasing income and consumption, particularly among female-headed households. In Kenya, living in a female-headed home increased the possibility of an adult going without food by about 10%, an adult skipping a meal by about 9.9%, and a child missing a meal by about 17% in the seven days preceding the telephone survey. In Ethiopia, living in a female-headed household increased the likelihood of an adult going hungry, skipping a meal, and running out of food by about 24.35%, 18.9%, and 26.7%, respectively. Salient pre-existing socioeconomic inequalities further exacerbated the effects of the pandemic on livelihoods. The findings have important implications for public policy and preparations by governments and other organisations interested in developing suitable gender-sensitive measures to lessen the impact of future pandemics in low-and middle-income countries.

Background With improving medical advances and the predicted increase in life expectancy of many individuals living with chronic and terminal illnesses, palliative care as an emerging field is poised for unprecedented expansion. This recognized need for expansion has led to the evaluation of palliative care as a key community health resource, particularly with recent clear evidence suggesting that early palliative care involvement can even lead to an improvement of both quality of life, and, at times, its longevity [1,2]. With mounting evidence of such benefits for early access to palliative care, policy attention is slowly being focused on the issue of existing disparities pertaining to access for this important service. This clarion call for better access to palliative care has received additional endorsement based on the recent use of rights language to describe the need for palliative care globally. This concept of palliative care as a right was first proposed by Brennen in 2007 and more recently proclaimed in the 2014 WHO Global Atlas of Palliative Care at End of Life [3,4]. Unfortunately, in spite of calls for improved resources, few advances have been made with the translation of access of these resources to vulnerable populations even in the developed world, particularly those living in the shadows of society: the homeless. Not surprisingly a review of the palliative care literature reveals that individuals with poor socio-economic status (SES) continue to have poorer access to palliative care services with subsequently poorer outcomes [5][6][7][8][9][10][11]. Why does being marginalized by socio-economic standards continue to beget such poor outcomes; particularly in countries with universal health care? While not intended to be a systematic review, this article attempts to summarize the existing literature exemplifying the current inequality in palliative care services, present a structural rationale to explain the current state of affairs, and advance a call to action on this issue. --- Discussion The Issue -The homeless: socio-economically disadvantaged and lacking access to healthcare services, including Palliative Care Poverty and homelessness are growing socio-economic and ultimately political issues in Canada. According to Statistics Canada, approximately 13 % of the Canadian population earns a 'low income' and it is estimated that annually, 150 000-300 000 individuals in Canada experience homelessness per year [12,13]. The problem of homelessness is much more than just a socioeconomic or political one as it directly affects health and access to healthcare for an already marginalized subset of our population. Individuals with a low SES are known to have poorer treatment access and health outcomes and these consequences are well exemplified in the homeless. Compared to the general population, those who are homeless have higher morbidity and mortality rates with an average life expectancy of 42-52 years (compared with 79-83 years for the general population), an increased prevalence of hepatitis of 29 times, heart disease by 5 times, cancer by 4 times, and diabetes by 2 times [5]. From a cancer point of view, a retrospective cohort study from a US national cancer registry 2008 revealed that low SES was associated with a more advanced disease stage and less aggressive treatment for individuals with breast, prostate, and colon cancer. This same study found a higher mortality rate for individuals of low SES with breast and prostate cancer [6]. Several barriers related to access to care have been identified and include: negative past experiences with healthcare workers, a lack of trust with healthcare workers, and a lack of comfort with systems not employing harm reduction strategies [5]. In addition to worse treatment access and health outcomes, individuals of low SES have been found to also have less access to community palliative care resources and hospice care. A literature review by Walshe et al. revealed that individuals who are older, male, from an ethnic minority, single, without a home caregiver, and who are socioeconomically disadvantaged (including lower education, not owning a home) are less likely to access community palliative care services [7]. A more recent review by Lewis et al. yielded similar conclusions [8]. These reviews primarily included literature from the UK, Canada, US, and Australia and pertain to patients with both malignant and non-malignant diagnoses. In the UK a recent retrospective cohort study revealed that referral rates to hospice and home care are significantly lower in regions of lower SES in defiance of a higher cancer incidence and mortality rate [9]. A Canadian study by Daneault et al. found that HIV positive individuals of low SES had poorer access to a multidisciplinary palliative care healthcare team and were less likely to die at home [10]. Similarly, a recent US study found that, individuals of a lower SES were less likely to receive continuous care and to achieve a home death, with higher rates of transfer to institutions near the end of life [11]. In spite of known poorer health status and outcomes which would warrant comprehensive management through quality palliative care, it is clear from the emerging literature that individuals from a low SES do not have access to this care. --- How are Canadian organizations addressing the issue? Notwithstanding abundant literature showing palliative care access and need discrepancy, it appears that little effort has been made to mend this gap in healthcare access. This is evidenced by the paucity of literature on programs with aims to improve palliative care service delivery and outcomes to those of low SES and in particular, to the homeless. In Canada, though considerable efforts have been done in the primary care setting to identify poverty and poor SES as determinants of health, little such efforts exist in the palliative care community. The scarce available literature related to this issue points to isolated and independent efforts existing across the country, primarily located in major urban centres. Examples of such efforts include the establishment of a social determinant of health screening tool at the Toronto Sick Kids hospital for palliative care patients, a hospice aimed to provide care to HIV patients in Toronto and in Calgary, and shelter based palliative care for the terminally ill homeless in Ottawa, Toronto and Vancouver [14][15][16][17]. More recently, an ambulatory palliative care group was established in Toronto to provide palliative care to terminally ill homeless individuals in mobile and shelter settings. Given the independent nature of the aforementioned initiatives, very little has been published by way of program evaluation and feasibility. The literature available however, suggests that palliative care programs for homeless individuals are effective in helping the homeless die comfortably and with dignity, are appreciated by the homeless population, and are cost-effective with an estimated health-care cost saving of over $1.3 Million in an Ottawa evaluative study published in 2006 [16]. In spite of the current clear need for improved palliative care delivery for Canadian populations of low SES and the successes of present individual initiatives, there continues to be no national unified strategy for providing palliative care to this group. Current systems-level palliative care campaigns including the national Speak Up campaign on advance care planning and the Canadian Hospice and Palliative Care Association's Quality End of Life Care Coalition place emphasis on home-based care as well as resources and support for family members. Though helpful for the typical Canadian population, this projected model will not address the needs of the homeless who are often without a home and estranged from family and friends. Indeed, a current Canadian qualitative study echoed that these intrinsic aspects of current end-of-life care programs are barriers to allowing the homeless access quality palliative care [17]. The review moreover demonstrated that operational policies of many palliative care programs (e.g. anti-drug, conduct and behavioural codes) inherently prohibit the homeless and most vulnerable from being included [17]. To address this issue, McNeil et al. have called for the creation of initiatives that cater to the needs of the socio-economically disadvantaged by adopting flexible, low-threshold strategies. Participants from their recent qualitative study point out the importance of adopting a harm reduction strategy to foster trust with the homeless population by demonstrating a commitment to serve them and an awareness of their life circumbstances [17]. In addition, partnering with local community agencies, and strengthening the training on the complex end-of-life care issues related to the dying homeless emerged as additional strategies to foster trust with the homeless population and thereby increase their access to end of life care services [17]. --- Summary and conclusion The history of palliative care is one of a movement that grew from an initial call to advocate on behalf of dying patients whose needs were not being met by the conventional medical system of its day. It was a response to the abandonment patients and families experienced and a recognized need for dignity to be infused into this most sacred and final act of our existence. This value was and remains its core mission. However, palliative care has become more recognized by mainstream medicine and perhaps more medicalized as a result of that encounter. In truth, our current palliative care delivery system requires a degree of monetary, in-person, and infrastructure support that often creates barriers for the homeless population; barriers they simply cannot overcome. This reality affects access to quality palliative care, not only at the end of life but also along the trajectory of a disease-where recent evidence supports its use. Current palliative care services are glaringly unequipped to properly meet the needs of the homeless. McNeil et al. have called for the creation of initiatives that cater to the needs of the poor by adopting flexible, low-threshold strategies, partnering with local community agencies, and by strengthening the training on end-of-life care issues amongst those who care for the homeless population [17]. If good quality end-of-life care is a right, it is one for all members of society and we need to re-envision our services to better support those who remain vulnerable. --- Abbreviations HIV: human immunodeficiency virus; SES: socio-economic status. --- Competing interests There are no financial or non-financial competing interests. --- Authors' contributions All authors were actively involved in the conception of and the writing of this manuscript. The idea for this paper came from a unique seminar session entitled: Palliative care and the Social Determinants of Health, as part of the Palliative Care Residency Program at the University of Toronto. --- Authors' information

Background: With an ever increasing number of individuals living with chronic and terminal illnesses, palliative care as an emerging field is poised for unprecedented expansion. Today's rising recognition of its key role in patients' illnesses has led to increased interest in access to palliative care. It is known that homelessness as a social determinant of health has been associated with decreased access to health resources in spite of poorer health outcomes and some would argue, higher need. This article aims to discuss the current state of affairs with regards to accessing palliative care for the homeless in Canada. Discussion: Recent review of the literature reveals differential access to palliative care services and outcomes with differing socio-economic status (SES). Notably, individuals of lower SES and in particular, those who are homeless have poorer health outcomes in addition to poor access to quality palliative care. Current palliative care services are ill equipped to care for this vulnerable population and most programs are built upon an infrastructure that is prohibitive for the homeless to access its services. A preliminary review of existing Canadian programs in place to address this gap in access identified a paucity of sporadic palliative care programs across the country with a focus on homeless and vulnerably-housed individuals. It is apparent that there is no unified national strategy to address this gap in access. Summary: The changing landscape of the Canadian population calls for an expansion of palliative care as a field and as many have put it, as a right. The right to access quality palliative and end of life care should not be confined to particular population groups. This article calls for the development of a unified national strategy to address this glaring gap in our healthcare provision and advocates for attention to and adoption of policy and processes that would support the homeless populations' right to quality palliative care.

Introduction With the rapid development of the economy and society, the family structure is changing sharply, and the proportion of single-parent families is increasing daily. In 2020, a total of 4.339 million couples divorced in China, an increase of 1.19 per thousand from the divorce rate in 2000 (Ministry of Civil Affairs of the People's Republic of China, 2020). In Western countries, the proportion of single-parent families has increased by nearly 20% in the past 20 years (Smock & Schwartz, 2020), and the increase in single-parent families worldwide has become more universal. The influence brought by the family structure of a single-parent family cannot be underestimated. A single-parent family refers to a family in which only one parent lives with the child due to widowhood, divorce, separation, or being unmarried (Xiong et al., 2021). Most studies on single-parent families focus on divorced families, exploring the correlation between marital conflict and children's development in divorced families, which indirectly reflects the impact of changes in family structure on children's development (Morris et al., 2007;Tiwari, 2022). However, studies of family structural changes on children's development have focused on academic achievement (Lee & Kushner, 2008), behavioral problems (Dufur et al., 2018;Mack et al., 2015), mental health (Motti & Aaron, 2016;Wang et al., 2017), and social adjustment (Chen et al., 2016). Studies have found that children living in single-parent households are more involved in crime and drug use, have lower grades, and are less well-adjusted overall than children living in two-parent households (Amato, 2005). Some studies have also found that changes in family structure affect the development of children's gender traits. For example, children from single-parent families may have fewer traditional gender schemas (Leve & Fagot, 1997), and independent growth space increases their masculine traits (Slavkin & Stright, 2000). The gender socialization of children is an important part of their personality development. Although less attention has been given to the structural changes in single-parent families and the development of their children's gender roles, their importance cannot be ignored. The development of individual gender roles is influenced not only by biological factors but also by the interaction of multilevel environmental systems such as family, school, and society. Related theories such as the challenge hypothesis clarify that parental sex hormone levels are associated with gender role parenting attitudes (Bell, 2020;Endendijk et al., 2018). Gender schema theory emphasizes the process by which individuals in the family selectively absorb environmental information to construct schemas that guide their gender behavior (Bem, 1981;Starr & Zurbriggen, 2017). Social learning theory focuses on individuals actively observing and learning the gender role behaviors of people around them (Miller et al., 1966;Morawska, 2020). Social role theory attributes gender behavioral differences to the division of social roles in the context of the social environment (Bosak et al., 2012;Koenig & Eagly, 2014). These theoretical constructs provide a strong theoretical basis for exploring the biological, familial, and social factors of I-J. Chen (Auyeung et al., 2009). At the family level, it is found that family environment factors such as family structure, parental gender parenting attitudes, and sibling relationships are pivotal in the development of children's gender roles (Mandara et al., 2005;McHale et al., 2004). Parents may influence children's gender role development through gender dialog, differential parenting, and field. Another study also found that the gender value system in the social environment affects the individual's perception and acceptance of self-gender (Cai et al., 2008). The results of the research on family structure and children's gender role development show a binary phenomenon of "influence" and "no influence". That is, most studies show that a single-parent family structure has a significant impact on children's gender role development (Jiang et al., 2007;Yang et al., 2021). However, some studies have found that changes in family structure have little effect on children's gender role development (Boothroyd & Cross, 2017;Stevens et al., 2002). In addition, the gender traits of children in single-parent families are no longer universally undifferentiated, and their gender role types are developing toward androgynous (Chen et al., 2019). Therefore, it is of practical significance to break the deep-rooted "negative influence theory of single-parent families" and pay attention to the gender role development of children in single-parent families. Therefore, the purpose of this study is 1) to clarify the factors influencing gender role development; 2) to analyze the impact of changes in family structure on the development of children's gender roles based on the research results of single-and two-parent families; and 3) to summarize the significance of the existing results and the direction for future study. --- Acquired and Innate Gender Role Development There are two major theories of gender role development: acquired and innate. The acquired influence theory began in the 1960s. The innate influence theory emerged in the 1980s. To date, these two theories have become the major trends in the investigation of gender role development. In the mid-20th century, psychologists believed that gender behavior acquisition was mainly derived from gender socialization and individuals' active observation of gender roles in the learning environment (Mischel, 1966), which laid the foundation for acquired influence theory. By the 1980s, evolutionary psychologists were applying the neural structure of the brain and sex hormones to the study of sex (Hines, 1982). It was not until the 1990s that biogenetic factors, such as sex hormones, were widely recognized as influencing the development of sex traits, reversing the dominance of the acquired theory. However, the rise of innate factors has not stopped the exploration of acquired factors. Numerous studies have found the influence of family, school, social and other acquired factors on gender roles (Wylie et al., 2013). Since the 1980s, these two theories seem to have developed I-J. --- The Foundation of Biological Genetic Factors The biological study of sex differentiation began in the 1940s with animal populations and by the 1980s had extended to humans. The role of testosterone and the genetic effects of genes are important discoveries in the study of biological sex roles. During the emerging period of biological inquiry between 1940 and 1960, most theories assumed that sex differences in organ development were caused by testicular secretions. The presence of testosterone stimulates the development of masculine characteristics and vice versa for female features. However, it is increasingly noted that the gonads are not differentiated by testosterone action but are directly controlled by genetic mechanisms and not regulated by the secretion of steroid hormones (Vigier et al., 1973). In the 1960s and 1980s, most researchers suggested that genetics is the main mechanism controlling the hormonal regulation of sex differentiation in the brain and gonads. At the end of the 20th century, sex hormones and genetics played a similar role in gender differentiation. However, some researchers have identified the one-sidedness of these theories, synthesized the two mainstream theories, and proposed new ideas: both "sex hormone regulation" and "genomic inheritance" may contribute to differences in human gender behavior (Auyeung et al., 2009;Gettler et al., 2011). In the 21st century, biological studies of human sex differentiation have reached a mature stage. Some studies also sampled T levels at multiple points and found that prenatal, postnatal, and pubertal T levels affect human sex development (Hines et al., 2016). However, experimental manipulation to obtain T levels is challenging. Most studies use atypical samples for T levels in association with sex type. For example, children with congenital adrenocortical hyperplasia are exposed to high levels of androgens early in gestation. This lacks comparison with samples with a normal sex hormone environment. In contrast, recent studies have proposed biological models linking biological factors to sex differentiation and parenting behavior. Children's biological characteristics may indirectly influence parental gender socialization through their gender behavior. For example, sex hormones or genes that lead to differences in children's sex behavior will cause parents to treat them differently, which also strengthens the gender behavior differences in children's biological factors (Gettler et al., 2011). The family ecosystem is a "petri dish" for individual gender role development. From the parents' innate heredity to the acquired attitude of gender socialization, the interaction affects the establishment of individual cognition and behavior of gender traits. To date, biological studies of sex differentiation seem to have found an interaction between innate and acquired differentiation. --- Dominance of Family Environment Factors The family is the earliest "unit" that individuals grow up in contact with. The atmosphere it creates directly affects the development of children's personality and gender socialization. Research on family environment factors has been widely explored since the rise of gender role research in the 1960s. For example, social learning theory holds that parents' gender behaviors, occupations and interests provide observational learning models for their children (Mischel, 1966). Other studies have found that the gender role of the dominant parent in the family has a greater impact on the development of children's gender roles (Bandura, 1962;Hetherington, 1965). In the past 60 years, research on the influence of the family environment on the formation of children's gender roles can be classified into three core elements: "parent-child interaction", "sibling relationship" and "family structure". One of the earliest studies on "parent-child interaction" was conducted by Freud and Hall in the early 20th century (Sears, 1940). The study of parent-child relationships in China is nearly a century later than in Western countries (Zhou, 2007). Then, came the study of "family structure", which sprang up in the 1950s (Brim, 1958) and became widespread at the end of the 20th century (Leve & Fagot, 1997). The study of "sibling relationships" was the latest to develop, starting in the 1960s and gaining more attention in the 1990s. To date, this field of research seems to have taken multiple factors into account, and the proportions of research on various factors are similar. First, the "parent-child interaction" study found that parents' selection of the type of games, gender talk, and differential parenting styles may affect their children's gender socialization. For example, parents express explicit gender stereotypes by restricting their child's preferred gaming and reinforce traditional gender behavior by using gender labels (Endendijk et al., 2014). Other studies have found that parents are more likely to use supportive parenting strategies for girls than boys (Acar-Bayraktar et al., 2019;Grazyna et al., 2009). When parents respond differentially to the same behaviors in boys and girls, children will understand the differences between boys and girls and know that certain behaviors are only suitable for specific genders (Endendijk et al., 2018). Moreover, many studies have found that family SES is also closely related to parental gender role upbringing. Children of high-education level parents and high SES families have better gender role development than children of low SES families (Yang et al., 2021). Although the research on "parent-child interaction" is fruitful, there is a lack of research on "comparative analysis of the influence of each gender in parenting". Too much of the research is cross-sectional and does not interpret the influence of other members of the family. Second, as an important source of children's social comparison, sibling relationships may also lead to the development of children's gender role differences. For example, studies have found that in families with mixed-sex children, parents have the opportunity for gender-differentiated parenting and exhibit more gender stereotypes than families with same-sex children (Leve & Fagot, 1997). However, some studies have also found that parental gender talk may be more (Endendijk et al., 2013). In addition to gender-differentiated parenting, sibling relationships may also influence children's gender role development by their gender, age, and intimacy (Bigner, 2012). I-J. For example, it was found that the number of older brothers was related to androgynous sex role development for girls, and perceived closeness to older sisters was related to androgynous sex role development for boys (Lamke et al., 2010). Sibling relationships provide children with opportunities to observe and learn the roles of the opposite sex or to reinforce the characteristics of the same sex. This is more beneficial for children's androgynous sex role development. However, there is still a lack of strong evidence on whether siblings bring about a positive effect of differential parenting on children's gender role development. Other studies point to changes in family structure as an important influence on the development of children's gender roles (Boothroyd & Cross, 2017). Research on family structure has focused on both single-parent and two-parent families. Previous studies have mostly focused on the process of gender role reduction from "two-parent to single-parent" (Tatyana et al., 2016), and few have explored the process of increasing gender roles from "single parent to reconstituted family" (Danielle, 2014). However, due to family disintegration, most children in blended families are older, and their gender role development has stabilized. Therefore, most studies focus on analyzing the structural changes from two-parent to single-parent families. A comprehensive study of the last 40 years found that changes in single-parent family structure not only created a freer context for children to develop gender roles but also reduced the formation of traditional gender roles (Slavkin & Stright, 2000). However, it may also lead to the absence of gender roles in the family, which hinders the comprehensive development of gender roles in children due to the lack of opposite-sex imitation learning objects (Yang et al., 2021). It is evident that there is a contradiction between the research results of family structure change and children's gender role development. --- The Infiltration of Social and Environmental Factors In addition to the dominant role of the family environment, the social environment cannot be underestimated. The study of social environmental factors also began in the 1960s, a period when the rise of the feminist movement in the West brought about fundamental changes in social conventionality (Freeman, 1973). At that time, research focused on the shift from "unisexuality to androgyny" under the cultural movement (Huston, 1985). The influence of the social environment on the development of individual gender roles has been widely studied (Levy, 1989). Overall, research in this area emphasizes the sociocultural factors that shape the meaning of the masculine and feminine roles of individuals. That is, individuals form gender behavioral differences in adaptation to the social environment. A comprehensive analysis of sociocultural influences can be divided into two broad categories: regional culture and media culture. Both factors emerged in the 1970s (Friedman, 1978). Because the craze of acquired influence theory has Open Journal of Social Sciences given scholars the opportunity to study gender roles from a cross-cultural perspective, the related research has been fruitful. First, the "regional culture" study found that socioeconomic and regional-ethnic culture are the main sources of the social environment influencing the development of gender roles (Minsun et al., 2022). For example, in the late 20th century, it was found that parents in high socioeconomic areas had more positive gender role parenting, which promoted the full development of children's gender roles and mathematical skills. In the 21st century, cross-cultural gender studies have developed rapidly. Relevant studies are limited to exploring gender role differences between developed and lagging regions within countries (Moreira et al., 2016), and the development of cross-cultural gender roles internationally has attracted more attention (Minsun et al., 2022). For example, the study found that Eastern collectivism and Western individualism define masculinity and femininity differently. Additionally, individuals develop different types of gender roles to accommodate cultural differences (Cai et al., 2008). However, among Mexico-born adolescents, females exhibited significant declines in traditional attitudes from early to late adolescence, but males' attitudes were stable over time. U.S.-born females and males, in contrast, did not differ in their gender attitude trajectories (Updegraff et al., 2014). Although there is a large body of cross-cultural research on gender roles, most studies use gender roles as moderating variables to predict individual abilities or developmental levels, and there is a lack of independent research on individual gender role types. Second, the "media culture" study found that mass media and schooling may influence individual gender role development. For example, in the 1970s, most studies concluded that television gradually became the main information medium, and there was a clear gender division in TV programs for preschool and primary school children (Sternglanz & Serbin, 1974). The number of females leading in television casting was much lower than that of males, and women were underrepresented (Gerbner, 1970). Other studies have found that textbooks and teaching aids are also important tools for the permeation of students' gender roles (Bernard, 1979). By the 21st century, the media culture of the Internet era has added more channels for the dissemination of gender role norms, such as advertisements, games, video software, and electronic publications, which quickly infiltrate individuals' gender cognition (Scharrer & Warren, 2022). For example, most of the spokespersons for clothing, beauty and home appliances are women, which implicitly assigns women the gender roles of being beautiful, slim, and doing housework (Xu, 2010). Technological advances have also led to an increase in the exploration of acquired factors, but there is still a lack of longitudinal data to track how the social environment affects gender role development over time. --- The Effects of Changes in Family Structure on Children's Gender Roles In conclusion, gender roles are the result of both innate and acquired interac- One group of studies argues that children's gender socialization is influenced by changes in family structure (Chen et al., 2019;Tatyana et al., 2016), while another group of studies indicates that changes in family structure have no significant effect on children's gender role development (Boothroyd & Cross, 2017;Stevens et al., 2002). --- The Universal Truth: Changes in Family Structure Significantly Affect Children's Gender Roles This study found approximately 90 studies supporting the idea that family structure has a significant effect on the type of gender roles of children. It seems to be a common "truth" in the history of gender role research that changing from a two-parent to a single-parent family structure can have a negative impact. The findings of the study can be summarized into three arguments: First, children in single-parent families have high masculinity. Most studies have found that children in single-parent families have higher masculinity than children in two-parent families (Leve & Fagot, 1997;Tatyana et al., 2016). Parents' gender-role parenting attitudes play a critical role in their children's perception of gender roles and gender identity (Morawska et al., 2021). Changes in the structure of single-parent families can lead to changes in the gender parenting style of parents, which in turn affects the gender socialization of children (Heath & Cavanaugh, 1993;Mandara et al., 2005). Moreover, single-parent families are more likely to develop unisex gender roles for children due to the absence of one parent's parenting role (Chen et al., 2016). Single parents, especially single mothers, are more likely to expect their children to behave in traditionally masculine roles (Slavkin & Stright, 2000). Another study found that most of the children from single-parent families are characterized by a high level of masculine-feminine identity (48% of the respondents), and their peers from nuclear families showed a high level of masculine-feminine identity only in 18% of the cases (Tatyana et al., 2016). It is evident that children in single-parent families are prone to unilateral attachment in imitating their parents' gender roles and exhibit relatively homogeneous masculine gender traits. Second, the proportion of undifferentiated children in single-parent families is high. It was found that the proportion of undifferentiated children in single-parent families was higher than that in two-parent families (Tatyana et al., 2016). For example, an intergenerational study found that single parents had significantly higher undifferentiated gender roles with their children than two-parent families. Single parents' undifferentiated gender roles can be passed on to their children in an intergenerational manner (Yang et al., 2021). Undifferentiated gender roles of parents can easily cause conflict and changes in marital relationships (Chen et al., 2019). Changes in family structure have forced single parents to take on the gender responsibilities of absent parents, making it easier for them to confuse their own gender roles and pass on undifferentiated gender roles to their children (Carlson, 2011). The lack of gender awareness in most single-parent families aggravates the imbalance in the development of children's gender roles (Mandara et al., 2005). In single-parent families, children may misinterpret gender images when observing parental gender behavior due to the lack of intact intersex role models (Chen et al., 2016), contributing to the emergence of children's undifferentiated roles. Changes in family structure affect both parents' and children's gender socialization. The highly undifferentiated roles of single-parent children are the result of intergenerational interactions between family systems. Third, the gender composition of single-parent families affects the development of children's gender roles. Some scholars have explored the gender differences of single parents on children's gender role development and found that single-father families have a developmental disadvantage, with a lower proportion of undifferentiated children in single-mother families than in single-father families (Boothroyd & Cross, 2017). It is also inferred that the traditional mascu-I-J. Chen et al. DOI: 10.4236/jss.2023.116024 385 Open Journal of Social Sciences line role of single fathers is deeply entrenched. Single fathers are more likely to disrupt their children's gender roles than single mothers due to changes in family structure that lead to "two roles in one" (Beaty, 1995). Other studies have further analyzed the gender pairing of single parents and their children. Among the four categories of "mother-son, mother-daughter, father-son, and father-daughter" studies, it was found that children in single-parent families with same-sex parenting of "father-son and mother-daughter" are more likely to develop masculine or feminine traits corresponding to their parents, which weakens the possibility of gender role confusion in single-parent families (Jim & Yesilernis, 2003). Specifically, boys raised by single mothers have fewer masculine traits and more feminine traits than boys from other structural families, whereas girls do not have such differences (Stevenson & Black, 1988). It has also been suggested that boys who live with their fathers are more likely to acquire stable gender role perceptions and engage in play appropriate to their gender identity. Moreover, the warmth of fathers facilitates the formation of androgynous traits in their sons but has no connection with the development of daughters' gender roles (Boothroyd & Cross, 2017). Other studies have also revealed that the negative parenting attitudes of single parents affect the gender socialization of heterosexual children. For example, when single mothers express hatred for their ex-husbands in parenting, they not only fail to provide behavioral norms for masculine roles but also aggravate boys' avoidance of masculine traits, resist masculine behaviors, and develop feminine tendencies (Ding, 2008). Alternatively, transgendered parents show excessive spoiling of their children, resulting in the extreme display of their children's masculine or feminine traits, which is not conducive to the development of gender roles (Wang, 2008). Despite the overwhelming evidence supporting family structural changes and children's gender role development, a few current studies questions this finding and predict that children in single-and two-parent families will have increasingly similar gender role types. --- The Status of the Few: Family Structure Changes Have Little Impact on Children's Gender Roles Although only 22 of the 112 articles supported the "no significant effect of family structure change on child gender role type", they also provided novel interpretations of the gender role development of children in single-parent families. In the "no effect theory" studies, nearly 50% noted no significant differences in the gender roles of children in single father families versus two-parent families, with family structure changes having a much weaker effect on daughters' gender roles (Stevens et al., 2002). For example, Boothroyd and Cross (2017) analyzed samples of European and American countries through the Bem scale and found no evidence of correlation between father absence and gender behavior changes in daughters. They also suggest that the impact of changes in family structure on children's development is due to their reactive behavior in response to changes in the family environment rather than to changes in children's gender role types. Open Journal of Social Sciences In addition, the increasing similar proportion of children in single-and two-parent families may also lead to the "no effect theory". For example, studies have found that the gender role types of single parents and their children are no longer traditionally unisex but are more androgynous (Chen et al., 2019). The study of the gender roles of children in two-parent families in China over the past 20 years shows that children in two-parent families have moved from traditional unisexuality to "androgynous and undifferentiated predominance" (Huang et al., 2010). Children of two-parent families have a similar trend to children of single-parent families in terms of "substantial increase in the proportion of undifferentiated" and "transition from unisexual to androgynous". This also reveals that although there are complete gender roles in two-parent families, parents do not regulate their gender role behaviors or do not actively participate in parenting, which may lead to delayed development of children's gender roles. With the advent of the information age, children in single-parent families have more access to gender information, and their gender role cognitive constructs tend to be more complete. At the same time, the reference effect of other male figures in the lives of single-parent children also alleviates the disadvantage brought by the absence of parental gender roles (Stevens et al., 2002). Overall, although the "no effect theory" research is weak and suffers from "subjective bias in data sources due to single parent reporting", some of the findings do highlight a holistic perspective on gender role development. --- Pushing Back to the Source: Exploring the Reasons for the Contradictory Conclusions of Family Structure Changes The research in this area has grown richer and richer, with "effect theory" and "no effect theory" competing. Among the 112 studies that have been conducted, there are more findings that "family structure has an effect on children's gender role development" than "no significant effect of changes in family structure". Therefore, this study suggests that there are three reasons to interpret the differential effects of family structure changes on children's gender role development: First, the age of the sample varies widely, and individual development interferes with the effects of changes in family structure. Kohlberg first studied gender roles from a cognitive developmental perspective and found that gender constancy is gradually formed between the ages of 2 and 7. After the age of 7, children observe and learn the social norms of gender roles (Kohlberg & Ullian, 1974). In high school, there is steady development (Jiang et al., 2007). However, current research on gender role development involves children from preschool to college age. For example, Chen et al. (2016) investigated children aged 8-18 years, andJiang et al. (2009) used 434 high school students as respondents. Therefore, it is not possible to exclude the effect of age differences in the sample group on their gender role development. Moreover, children of different ages experience single parenting at different times. Therefore, to investigate the influence of family structure changes, we cannot ignore the length of time children have experienced single parenthood and their own gender role development. Open Journal of Social Sciences Second, the fit of multiple measurement instruments affects the source of scientific data. The current approach to assessing gender role traits is based on questionnaires and interviews. At present, questionnaire surveys and interviews are the main ways to evaluate gender role traits. The most famous measurement tool is the Sex Role Scale (BSRI) developed by Bem in 1974(Bem, 1974). BSRI is the originator of psychological scales for measuring androgynous traits, and its first mention of androgynous gender role typology created a boom in gender role research after the 1980s (Myers & Gonda, 1982). Several studies have tested the reliability and validity of this scale (Boothroyd & Cross, 2017). Despite this, the BSRI also revealed shortcomings in areas such as "older definitions of androgyny" and "scales developed with roots in American culture may not be appropriate for other country samples". For example, studies have pointed out that compared with the 1990s, American college students in recent years are less likely to agree with the feminine characteristics presented by the scale. BSRI concepts of masculine and feminine traits need to be updated to better reflect current gender role norms (Donnelly & Twenge, 2017). The first self-administered gender role scale in China was created in 2000 when the Chinese College Student Sex Role Inventory (CSRI) was developed based on socially accepted gender role norms and social stereotypes in Chinese society (Qian et al., 2000). This scale broke the single-quadrant positive gender schema of the BSRI and innovatively treats the traditional gender schema as a four-quadrant distribution with "one dimension and two endpoints". Its high reliability and validity have been tested in many studies (Zhang et al., 2012). However, there are still shortcomings, such as the lengthy scale and the thin test study of the multidimensional theory of gender schema. Subsequently, the BSRI was revised in 2003 to consider the cross-cultural differences of the BSRI scale due to its general diffusion in China (Lu & Su, 2003). However, the revision was limited to the deletion of low-load vocabulary from the BSRI entries, and no replacement vocabulary was considered. Because of the inadequacy of the existing scales, Liu et al. (2011) developed the CSRI-50 based on the BSRI and CSRI, which has a significant change in word selection compared with the Bem and CSRI and retains 50 moderate items with good reliability and validity. However, the CSRI-50 was developed nearly 10 years ago, and entries that fit the gender culture at that time may not be applicable to current gender roles. Moreover, most of the test groups were college students, so we need to be cautious about whether the scale can be extended to other age groups. Thus, gender role research requires both culturally appropriate and sensitive measurement tools, as well as an up-to-date examination of changes in sample groups. Third, a single measure is prone to subjective bias in data reporting. Although gender role studies can obtain data through scales and interviews, most studies tend to use scales to measure children's gender types (Russell & Denise, 1991). The tests only considered unilateral parental reports (Stevens et al., 2002) or child reports (Zhang et al., 2012). Few studies have used paired parent-child co-participation tests to examine differences in parent and child subjective re-I-J. (Hillary & Maureen, 2016). Gender role parenting in families is formed gradually during the life interactions between parents and children, so the unilateral reports by parents and children are prone to subjective bias. In conclusion, it was seen that out of 112 included studies, nearly 90 studies agree with the theory and showed that the changes in family structure led to a considerable impact on children's gender roles. Studies have indicated a higher prevalence of masculine traits and undifferentiated types in children of single-parent families, which further found that the gender configuration of single-parent family members influences children's gender role development. In contrast, 22 studies supported the theory of no effect of family structure, providing a dialectical perspective on the relevant studies. In exploring the differences in the results of these studies, it can be concluded that the sample source, measurement method, and instrument selection in the process of gender role research may have an impact on the whole body and affect the differences in research results. --- Summary and Prospects The study of gender roles has developed for nearly a century. Despite some not- (Wang, 2005). Studies also found that the father's role has a greater impact on children's gender roles (Jacklin et al., 1984). Therefore, it is necessary to further compare the influence of single parent gender types. Second, multiple research methods should be integrated, and measurement instruments should be scientifically selected. To prevent subjective reporting bias of questionnaires, the survey can be conducted by using "questionnaire-interview" and "parent-child", which are collected from multiple perspectives. In the past, single-parent comparative studies were mostly conducted with crosssectional data. However, the journey of a family from intact to broken up requires a longitudinal study to deeply analyze the impact of changes in family structure on children's gender roles. Third, to control for multiple factors that influence gender role development, we need to draw precise conclusions about family structure and children's gender role types. Related studies have shown that socioeconomic status (Marks et al., 2009), parental education (Yang et al., 2021), sibling relationships (Bigner, 2012), and parent-child relationship quality (Zhou, 2007) In summary, focusing on the gender role development of children in single-parent families not only emphasizes parenting for parents but also requires continuous exploration of the gender types of single parents and children by experts and scholars to offer more valuable and constructive scientific conclusions for single parents. --- Data Availability Statement This paper is a critical review, so there are no shared data. --- Conflicts of Interest The authors declare no conflicts of interest regarding the publication of this paper.

As the number of single-parent families is increasing, the issue of structural changes in single-parent families is gaining more attention. The impact of single-parent families on children's gender role development due to the absence of one parent is worth exploring. Gender roles are shaped by the interaction of innate and acquired factors. Innate factors such as sex hormones and genome lay the basis for gender socialization. In contrast, acquired factors are dominated by the family environment (e.g., parent-child interactions, sibling relationships, family structure, etc.) and permeated by the social environment (e.g., regional culture, media culture, etc.), which together influence gender role development. To explore the in-depth role of family factors in gender role development, especially the impact of changes in family structure, this study explored five databases (four English and one Chinese) and identified 1573 studies; 112 studies were included in a systematic analysis. To summarize relevant studies, there are two main arguments for the "effect theory" and "no effect theory" on family structure changes and children's gender role development. Moreover, the results that support the effect theory account for more. Studies holding significant effects indicated a higher prevalence of masculine traits and undifferentiated types in children of single-parent families and found that the gender configuration of single-parent family members influences children's gender role development. However, a few studies have also concluded that changes in family structure have little effect on children's gender roles. For this reason, the article proposes three reasons for the discrepancy and highlights future research breakthroughs.

Background Empowering service users and caregivers (hereafter'service users' unless otherwise specified) to be involved at all levels of the mental healthcare system (advocacy, policy, planning, service provision, monitoring, research and education) is at the centre of the international mental health policy agenda [1]. Involvement has been increasingly acknowledged as fundamental to the scale-up of quality mental health services in low-and middle-income countries (LMICs) [2][3][4]. However, actual involvement remains largely aspirational, with little guidance on how to achieve involvement in practice [2,3,5]. In many LMICs, service users are still largely excluded from involvement in mental health systems [2,6]. Multiple factors hinder service user involvement, including the limited attention given to empowerment and mobilization of service users, and the lack of preparation of other stakeholders to work with service users [2,6,7]. There is a strong argument for interventions to build the capacity of service users and health professionals (including service managers) to facilitate active participation and to overcome stigma and exclusion [2,8,9]. Studies from high-income countries indicate that empowerment and training of key stakeholders is helpful and can lead to successful working partnerships [10][11][12]. However, the evidence base for effective models of capacity-building is weak, limited to descriptions of training initiatives, and there are no programmes to equip key stakeholders to work together within LMIC mental health systems [13][14][15]. In Ethiopia, there have been no training programs designed to equip service users and health professionals for collaborative working in mental health system strengthening. Indeed the key stakeholders themselves reported being ill-equipped and expressed a strong need for training on how to work together to effect change [16,17]. As a part of an intervention to pilot a new model of service user involvement in mental health system strengthening in Ethiopia, the authors developed and delivered interactive training to service users and health professionals in the Sodo district of southern Ethiopia. In this paper, we describe the development of the pilot training program and evaluate the feasibility, acceptability, and perceived outcomes. --- Methods --- Study design We used an uncontrolled, convergent mixed-method study design. This study formed part of a larger research project aiming to develop service-user involvement in mental health system strengthening in Ethiopia. This was initiated as part of the 'Emerging mental health systems in low-and-middle-income countries' (Emerald) project including six LMICs (Ethiopia, India, Nepal, Nigeria, South Africa and Uganda) [18]. In addition, this specific study was part of the OPAL (Optimizing Provider Attitudes and competence in Learning mental health systems) project [19], and was one component of an intervention to pilot a Theory of Change model for service user involvement in mental health system strengthening in rural Ethiopia, described elsewhere [17]. --- Setting The study was conducted in Sodo District of the Gurage Zone of the Southern Nations, Nationalities and Peoples' Region, Ethiopia. The capital of the district, Buie, is located about 100 km south of Addis Ababa. At the time of the study, the district had 58 kebeles (sub-districts); four urban and 54 rural. The population of the district was about 170,000. Health care was provided by one primary hospital, eight primary health centres and 58 health posts. --- Participants The participants of the study were service users, caregivers, and generic health professionals, purposively selected based on pre-set inclusion criteria (See Table 1). The service users had confirmed diagnoses of psychosis or epilepsy or alcohol use disorder or depression; and they all had accessed integrated mental healthcare in primary healthcare as part of a previous study implementing the World Health Organization mental health Gap Action Program (mhGAP) [20]. The clinician involved in the person's care (a psychiatric nurse based in Buei Primary Hospital) recommended the service user participants based on his professional judgement. For participants who were not under the clinician's immediate care, a project psychiatric nurse assessed their capacity to consent using a semi-structured approach used in the setting previously. empower service users to be involved in mental health system strengthening in this setting. Further study to assess the impact on health systems strengthening is warranted. --- Keywords: Training, PhotoVoice, Service user, Mental health, Sub-Saharan Africa The health professionals were recruited from the Sodo district health office (the administrative office co-ordinating all non-hospital health care in the district) and three health facilities (Buie Hospital, Kela and Tiya primary centres) in the district. All participants were invited by formal letter to participate by the Sodo district health office and in collaboration with PRIME project field coordinators. --- Overview of the training program --- Theoretical lens The training program design and interactive delivery was informed by principles of participatory action research (PAR) [21] and experiential learning theories [22,23], suitable for adult learning and for people with low literacy. Key characteristics of this approach were: (i) the attention paid to empowerment, building on strengths and resources inherent in participants, and being defined by mutual trust, co-learning, active participation, and respect to promote social action; (ii) the value placed on life experiences as a central and necessary part of the learning process, and (iii) acknowledgement that participants bring a wealth of experiential knowledge on the topics discussed and are well-positioned to speak about their experiences [21][22][23]. The REducing Stigma among HealthcAreProvidErs (RESHAPE) program is an initiative to engage service users and their caregivers to improve training of primary care workers in mental health services [15,19]. RESHAPE was developed based on stigma reduction theories from social psychology, social neuroscience and medical anthropology using a 'What Matters Most' framework of motivation [15]. The core methodology is PhotoVoice, which is a participatory photographic narrative technique. PhotoVoiceis a PAR methodology, involving creation of a visual reality of a person's lived experiences with a particular topic and their reflections. The PhotoVoice approach is rooted in participatory empowerment education/critical consciousness theory and other participatory approaches [24,25]. These theoretical underpinnings promote power sharing, understanding of lived experiences and encourage the participants to construct and share their reality to take collective action for positive change. PAR PhotoVoiceh as been employed within the field of mental health to: (i) explore and understand service user experiences of treatment of mental illness and the recovery process, (ii) engage in psycho-education programmes, (iii) dispel issues such as shame, social exclusion and stigma associated with mental illnesses, and (iv) advocate for more inclusive, co-produced and participant-centric ways of uncovering and discovering the lived experiences of service users [24,25]. Photovoice combines photography, interviews, and group discussions mainly guided by SHOWED questioning techniques to elicit discussion about participants photographs [24,25]: What do you See here?;What is really Happening here?;How does this relate to Our lives?;Why does this condition Exist?;and What can we Do about it? For RESHAPE, PhotoVoice was used to help health workers understand the experience of mental health conditions through the eyes of persons with lived experience. RESHAPE was developed in Nepal, where service users and caregivers were trained in PhotoVoice and then presented the photography narratives to primary care workers participating in mhGAP training, local community groups and policy makers. Proof-of-concept testing has demonstrated the feasibility and acceptability of the approach in Nepal, as well as suggestive benefits for reducing stigma among primary care workers in mhGAP training [15]. The current study in Ethiopia is the first test of the RESHAPE approach outside of Nepal. --- Development of the training program The development of the training program was informed by identifying needs through in-depth qualitative interviews with service users, health professionals, planners and policy makers [16]. Potential approaches to address these needs were identified by mapping out existing materials, consulting with experienced researchers in LMICs, particularly informed by a training manual developed by a Nepal research team [15], and a review of the international literature on similar initiatives [10,11]. Based on these inputs, co-authors (SA, CH, and HL) developed two training manuals for service users, and health professionals. The appropriateness, contextual relevance, basic delivery and assessment modalities for the translated manuals (Amharic) were reviewed by six professionals who were either university faculty staff members or researchers from diverse disciplines (psychiatry, health education, clinical psychology, public health, and mental health epidemiology). The core content of the training manual was similar for both target groups, but the service users training included additional content on basic knowledge about mental health and illness, treatment options and their rights to receive care. The training manuals are freely available [26]. The training program for service users had additional content on the development of recovery narratives using PAR PhotoVoice techniques based on RESHAPE [14,15,19]. The Ethiopian team had a close dialogue with the OPAL Nepal team who had first-hand practical experiences of providing training and applying PhotoVoice techniques. --- Training delivery The interactive training was delivered face-to-face separately for (i) health professionals, and (ii) service users. The training for health professionals was conducted for a full day, which was divided into seven sessions. The training for service-users consisted of two parts, including an initial three day training course that focused on core topics, and eight PAR photovoice sessions, each of 3-4 hours duration conducted two times per week. The photovoice training had two components. First, field activities whereby service-users were supplied with digital cameras, a notebook and pen to take photographs of people, places or objectives/mementos that depicted their own recovery journey, followed by preparation of service user narratives closely linked to the self-selected photos. Participants were first trained in the use of the digital cameras, including about the ethical aspects to consider when taking photographs of people. The participants then took photos for each session on three themes about the time before receipt of professional treatment, during treatment and after completing treatment or continuing to receive treatment. Second, there were interactive classroom learning and reflection sessions. During the classroom sessions participants were asked to choose photographs that were most relevant to the theme and that they felt comfortable discussing in a group. The photographs were transferred onto laptops to display the photos for discussion with all participants. Accordingly, every service user presented and was involved in discussions about the photographs using the SHOWED techniques. A range of facilitated discussions were conducted on several topics, including identifying strengths and resources, combating stigma and myths about mental illness, advocacy, communication skills, understanding and managing crisis. In line with principles of PAR and experiential learning theories, the learning goals, content, and activities for both target groups were designed and delivered consisting of interactive techniques, including didactic, small group discussions and presentations, case scenarios analysis, audio-visual material showing the recovery narratives from lived experiences in the local area, and own recovery narratives using PAR photovoice, and action plan development exercises. The training for service user and health professionals groups was held in a private setting in Buie town in Amharic (the official language in the study district). However, there are diverse local languages spoken in the study site. Therefore, during the training, the psychosocial workers involved in the training assisted us by translating key concepts into the local languages for the specific training sites (Kistane Guragena). The first author (SA) with two research assistants led the training for health professionals. The training for service users was facilitated by SA and MG. A senior professor of psychiatry from Addis Ababa University (AA) taught the mental health related topics for service users. The trainers were assisted by three psychosocial service providers who lived in the study area. SA and MG had received a fiveday intensive face-to-face training specific to facilitate the planned photovoice sessions by three professionals from Nepal. The OPAL Nepal team also provided ongoing supervision and support throughout the training. --- Evaluation of the training program We evaluated the training program using multiple sources of quantitative and qualitative data [27]. --- Quantitative assessment The quantitative data collection consisted of selfreported or interviewer-administered (for those with no or low literacy) rating scales and training process data; which included demographic data, program feasibility and acceptability, and change in understandings of training content. --- Demographic data Socio-demographic data including gender, age, education, and experiences (work/caregiving or living with mental health condition) were gathered at the beginning of the first session. --- Feasibilityand acceptability of the training program To assess the feasibility of the training program, we collected process information that consisted of the number of participants who enrolled, attended and completed each training session. Acceptability of the program was operationalized as participants' satisfaction with the training sessions, standard of the training and fulfilment of their expectations. This was assessed immediately after the completion of training using a questionnaire comprising both structured and unstructured questions [9]. Aspects of the training sessions were rated on a 4-point Likert scale from 1(excellent) to 4(poor), and questions about the standard of training and fulfilment of expectations were rated from1(strongly agree) to 4(disagree). --- Evaluation of understanding We employed a retrospective pre-post-test questionnaire to investigate whether the training produced changes in participants' self-reported understanding related to the objectives of the training program. The concept of service user involvement is relatively new within the Ethiopian mental health system, and prior to this training participants had no experience of involvement and working collaboratively in this area [16]. We anticipated that in most cases participants lacked sufficient knowledge to provide pre-training ratings, as has been reported previously [28]. With the retrospective pre-post-test approach, participants estimate their preintervention levels of understanding training content after receiving the training and can therefore situate themselves in relation to what they have since learned. This avoids response-shift bias, which occurs when participants may not fully understand how to assess the target understandings before the training, therefore leading to inaccuracies in pre-test ratings [28,29]. We employed a bespoke questionnaire developed for Emerald [9] in line with the training goals. The instrument consisted of 17 items for health professionals and 15 items for service-users, with 13 items common to both groups. Participants reported their response to the questions via a five-point scale of 'not very well' (1),'somewhat well' (2), 'don't know' (3),'moderately well' (4), or'very well' (5). The participants completed the questionnaire immediately at the end of the first day (health professionals) and after the three-day training (service users). For each item, participants provided two ratings (i) their current level understanding after the training, and (ii) retrospective estimate of their level before the training. Five interviewers wrote the responses for participants who had low literacy (n = 15) with their agreement. --- Qualitative assessments --- Exit interviews Face-to-face in-depth interviews were conducted in Amharic individually with the service users and caregivers (n = 24) immediately at the end of the last Photo-Voice session. The interviews explored why participants attended the training, the aspects of the program that they found most helpful and enjoyed, aspects that made them feel uncomfortable, how the program could be improved, and perceived benefits/impacts of the training. The interviews were conducted by five (2 female and 3 male) project field research assistants. The interviews were audio-recorded and lasted between 6 and 46 minutes (with an average of 13 minutes). In addition, qualitative responses to open-ended questions nested in the quantitative survey related to the perceived challenges or comforts about the training process were collected. --- Follow-up in-depth interviews Individual in-depth interviews were conducted in Amharic with subsets of service user (n = 4) and caregivers (n = 4) approximately 6 months after the end of the training sessions. In between the training and the interviews, each of the participants had gained practical experience of giving testimonies to health professionals on a five-day training programme (each participant had attended at least two training sessions). The topic guides were adapted from similar research used by the OPAL team. In addition to the exit interview questions, the topic guide aimed to explore participants' impressions of the training, knowledge and skill they had acquired, and how they had applied the knowledge and the skills since the past training sessions. All interviews were conducted by a single experienced interviewer (female), who had not been engaged in the training programme. The interviews were conducted in Amharic and were audio-recorded. Interviews lasted between 16 and 58 minutes (with an average of 30 minutes). --- Data analysis We utilized a triangulated convergent mixed-methods approach [27,30], whereby quantitative and qualitative data were collected concurrently or at follow-up, analyzed separately, and the results were triangulated to develop comprehensive understanding of the training program for the key stakeholders in relation to feasibility, acceptability and perceived benefits/impact. --- Quantitative data Descriptive statistics were used to analyze feasibility and acceptability assessments and improvements in understanding. Individual group scores/absolute numbers were calculated/ summarized, session attendance, completion of the program, and frequency of agreement (agree or strongly agree) for survey items. We did not perform inferential statistical testing on retrospective pre-test assessment, because of the small number of participants involved; rather we compared descriptive data. --- Qualitative data SA typed the responses to the open-ended questions and the data from process documentation to develop a dataset, and then open-coded and categorized the dataset into themes using inductive thematic analysis [31]. This analysis was augmented with the themes from interviews of the qualitative data. Qualitative data from exit and follow-up interviews were transcribed verbatim by independent transcribers and translated into English by research assistants from Addis Ababa University (n = 7) and SA (N = 21). Audio records of three exit interviews were not clearly audible and were excluded. The remaining transcripts (n = 19) of the exit interviews were included to develop codes, which ensured data saturation [32]. Transcripts were analyzed by SA using thematic inductive coding to uncover meaning in participants'accounts of their involvement in the training process [31]. Details of narratives of service users' recovery journey and challenges of living with mental illness will be reported separately. --- Rigor To enhance trustworthiness of the findings we employed different strategies, including data and methods triangulation, use of field observation notes, and prolonged engagement with participants [33]. We applied triangulation of multiple data sources that helped to include various points of data in the analysis. Through prolonged engagement in the field with participants, SA took observational field notes throughout the study period that helped to enrich data from the interview transcripts and compare personal experiences during the subsequently bracketing [34] and data analysis. Information for the photovoice narratives were participants' own observations and interpretation included, which is considered as member checking [35]. --- Results The results of the training are presented in three sections, comprising participant characteristics, training program feasibility and acceptability, and perceived outcomes. --- Participant characteristics Twelve service users, 12 caregivers and 18 health professionals attended and completed the training and provided response to the program assessments. Sociodemographic data for participants is presented in Table 2. To maintain anonymity, the participants were represented with identification numbers to describe their qualitative accounts in the results, for example service user (e.g. SU1, SU2), caregiver (e.g. CG1, CG2) or health professionals (HP1, HP2). --- Training program feasibility and acceptability Feasibility Feasibility of the training program was supported by high enrolment, training completion and response to the training assessment by all eligible service users (12/12), caregivers (12/12) and health professionals (18/20) (See Table 3). Moreover, all participants who enrolled attended all sessions, except one who missed one session because of a competing commitment. --- Educational attainment Non-literate The analysis of the qualitative datasets (open-ended questions, interviews: exit and follow-up) strongly supported the feasibility of the training programme. The participants reported only a few challenges to attending the training program. These logistical issues included challenges with transportation, accommodation, overlap with their regular duties, and health challenges. --- I attended all the training attentively with interest. The time schedule was good and it did not waste our time. (SU20, Exit interview) I attended all the 10 days, previously I used to feel health problems all the days, but there was no problem during the training. I saw changes [health] in this regard. (SU3, Exit interview) Only one participant (coming from a rural area) reported challenges related to transportation and weather conditions that affected attendance. --- Sometimes I was delayed getting transportation and when there was rain we did not get transport, but there was nothing else. (CG20, follow-up interview) Some participants coming from rural areas reported challenges related to accessing a power supply to charge the digital cameras. Notwithstanding the affirmative feedback about the training content and values, almost all health professionals indicated areas where the training programme could be improved in terms of time and supplementary reading materials. --- The training was very nice, but the time was very short. If the time was longer we could have discussed many points. The trainer covered a lot of key points in [a] very short time. (HP13, Response to open-ended questions) --- Acceptability The quantitative analysis indicated that the majority of participants reported strong agreement (31/42) or agreement (11/42) that the training was of a high standard and that their expectations had been fulfilled (strongly agree = 33/42; agree = 9/42) (See Table 3). This positive feedback was captured in the qualitative data in participants' accounts about the adequacy of the training content, recommendations about how the training could be improved, and the most useful aspects of the training. Almost all participants mentioned that the training included many useful topics, and their expectations were largely met. They described the training program using words and phrases like 'interesting','very nice','very relevant','very useful/helpful','very happy', and one participant said "This lesson cannot be gained even by paying for it, " (CG12-Follow-up interview). Many of the participants recommended no change or additional topics to the training program, except three participants who wanted to additionally receive training about income generation (SU11, SU21,CG4). --- I am very happy about the training; the issues discussed were very important. The recommendations and solutions raised during the training should be changed into practice involving all stakeholders (HP6, Response to open-ended questions) --- The training contained many useful topics. It helps to learn how to properly use medication, and how to teach other people. It helps to clearly learn from others' stories, for example about how to avoid alcohol use and factors that worsen mental illness. (CG 22, Exit interview) Almost all participants recommended the need for expansion of the training programme to reach more people, including involving more people with lived experiences and engaging community stakeholders. The service users also expressed their readiness to share what they had gained from the training and their lived experiences. --- This training was provided for 12 service users and 12 caregivers. They should share the knowledge to many people. The training program should be supported by governmental and non-government organizations to reach many people. For example, if teachers participate in the training, the lesson can reach many people. (CG12. Follow-up interviews) The participants appreciated the different activities and interactive training techniques, including case scenarios analysis and PAR photovoice recovery narratives, which were considered to encourage collaborative learning, share experiences and made training understandable. --- The service users actively participated by capturing and describing photos and this made the training very clear to understand. This training included many things and was unusual. There were photographs, videos, and people with the mental health conditions shared their experiences throughout the training. I liked it very much. (CG 24, Exit interview) The PhotoVoice was helpful to express idea easily. It nicely helped me to express my ideas. I am very happy talking about my experiences to others. (SU21, Exit interview) --- Perceived outcomes of the training programme --- Improved understanding/knowledge The descriptive analysis of the retrospective pre-test responses showed that increased numbers of participants reported higher understanding levels after the training than before (see Additional file 1 and Table 4). Consistent with the quantitative findings, the qualitative data supported those participants perceived this training improved their understanding. The participants reported that the PAR photovoice enhanced their active involvement and understanding of the training. The participants reported learning new skills that they could use or are using in various areas. --- The training gave us sufficient knowledge about mental illness, medication use and stigma and discrimination. I got good knowledge when I heard their advice (participants sharing lived experiences) and the way they talk. (SU16, Exit interview) The training had a lot of benefit. When participants presented their experiences of living with mental illness, this becomes a good experience and advice for us [ --- participants]. When they [service users] talk we [participants] were happy. Just when they speak about their past experiences and how well they are doing now it gives hope service user. (CG15, Followup interviews) Besides the improved understanding and skills, the qualitative data highlighted numerous perceived outcomes/benefits of the training programme, including feelings of empowerment, social benefits, and perceived therapeutic benefits. --- Sense of empowerment The participants mentioned the valuable contribution of their participation in the interactive PAR photovoice group activities and having the opportunity to give testimonies to health professionals. For many participants, the social space for sharing their recovery journey, having their views heard, valued and acknowledged by health professionals had led to improved self-confidence, a sense of valued contribution and increased self-worth. ---...they (health professionals) were very happy, clapped their hands for us, and they told us to help us [service users and caregivers] if we faced any problem even at night. (CG20, Follow-up interview) I am very happy having a freedom to discuss with others and sharing my experience about my previous situation. I feel like I rebirth and feel like my age is just like a child. Previously I have suffered a lot... when that time passes and I become to this stage, this is rebirth for me. I am happy. (SU 17, Exit interview) --- Perceived social benefits Almost all participants valued the social benefits of the training programme in terms of improved social acceptance. The participants mentioned the PAR photovoice process helped them to be valued and acknowledged for their lived experiences in the family and community. We were stigmatized and discriminated by community, no one used to visit us, including our neighbours, and we did not go to other peoples' homes. --- After we started participation in this training many people started greeting us. My relatives who used to reject me are now accepting me. People started treating us as human beings. We have hope. God knows the future. (CG7, Exit interview) In my family no one was looking for me because of my illness. They were wishing my death; no one used to give me even free rain water and considered me as use- --- less. My husband is very nice person and he encouraged me and after my involvement in the training my family started visiting me. (SU11, Exit interview) The training showed us many good things. I am communicating with my family very well now and they are also asking me to share my ideas with them. They are also accepting my ideas. (SU1, Exit interview) Although there was improved acceptance of service users in the community, the participants mentioned that more work needed to be done in the community and institutions through awareness creation and involving service-users. There are many people who stigmatise and discriminate against people with mental illness. They lack understanding about mental illness and people with mental illness. We should provide awareness to all people. (SU 11, Exit interview) At beginning of the training process, we observed that some participants felt uneasy about sharing their own recovery stories because of being upset due to past unpleasant experiences of discrimination. This problem was anticipated by the trainers before the training started and managed by probing participants to share their personal strengths rather than focus on their personal mental illness. The participants liked the opportunity to meet, spend time and interact with people with similar mental health conditions. The participants welcomed the opportunity to have the time and space to exchange their lived experiences and learn with or from each other. They reported that this created a sense of belonging, strengthened their relationships and stimulated action to establish long-term relationships through getting organized into a grass roots service user association. The participants reported that being organized into an association would support their collective efforts to tackle problems facing people with mental health conditions. The training process brings people to deeply connect with each other. Our (service users) coming and discussing together enabled us to establish a service users association, and hope that it [association] will be the best association in the Sodo district and the entire country. (CG12, Exit interview) --- Perceived therapeutic benefits Many of the participants reported making changes in their lifestyles and behaviour with perceived therapeutic value because of the knowledge and skills gained from the training. Some reported making changes to unhealthy aspects of their lives, including reducing/stopping alcohol drinking, improved medication use, attending health facilities, improvements in the way they experienced their family relationships and the wider community. The man (service user) used to be in bed and feeling pain all day; and I was not able to attend meeting even at my neighbourhood. I am very happy for attending this training. The man is effectively taking his medication. The man was a difficult person, but after the training [he] has showed improvement. He started meeting people and participation in social activities. He is started looking for and feeding cattle. The training gives hope. (CG2,Exit interview) This training was more than taking medication, helpful to change the mind and giving hope about the future. Thanks to God we are here to get involved here. We used to chain her all the days just like a sheep and goat. I remember many people visited and took her photo while she was chained. (CG7, Exit interview) Many of the service users mentioned involvement in peer support, advocating for human-rights and supportive relationships with health service providers that could help to improve health service quality. --- When both of us (service user and caregiver) go to health facility, we don't feel shameful and [can] freely talk about her (service user) mental health condition without fear. I ask for timely service without waiting for long. They (health professionals) are also serving us with good manner and respect. (CG10, Exit interview) --- Discussion In this paper we have described the development, delivery and evaluation of the feasibility, acceptability, and perceived impacts of a participatory training programme for mental health service users and health professionals. This was one component of an intervention to support service user involvement in mental health system strengthening in Ethiopia [17]. There was high participant enrolment, completion, response to assessments, and few challenges or discomforts during the training process. The training program content, delivery process, standard, and relevance were well received. The participants recommended continuing and expanding the training program to reach many people. The training had positive impacts, including improved understanding of issues related to service user involvement, mental illness, increased confidence, social impacts and perceived therapeutic benefits. Our program showed that training service users and health professionals for greater involvement in mental health system strengthening is feasible, acceptable, and can have several benefits at small scale in rural Ethiopia. The preliminary findings of feasibility and acceptability of the training programme indicate that this is a promising approach to equip key stakeholders to support greater service user involvement [17,36]. This step was identified as a necessary precondition to service user involvement in our Theory of Change model for mental health system strengthening in Ethiopia [17]. This study also contributes to the small evidence base on training programs designed to empower mental health service users in LMICs [5,8,9,13]. Our findings indicated that participants valued the wide range of interactive training strategies utilized during the training (e.g., group exercises and reflections, photovoice recovery narratives). The findings reinforce studies from high-income countries that reported similar active involvement of service-users [11,12]. These activities are key strengths of PAR and experiential learning approaches that we applied, which integrate participation, promote supportive relationships, and recognise knowledge and lived experiences as vital to the knowledge generation process [22,23,37]. Many publications report that the PAR photovoice method values the expertise of participants and their direct involvement in all aspects of research is positive, including the visual data analysis about the meaning and importance of self-selected of photos and their narratives [24,25,38]. PhotoVoice is inclusive and particularly suitable for those with low literacy, because this method does not presuppose any working knowledge of traditional research methods; and participants mainly use photography as the medium through which they communicate their lived experiences [24,25]. Combined use of visual data (photos), together with individual and group reflective sessions (voice), can facilitate an authentic, active participation, co-creating/co-production of knowledge, and provides a powerful means for marginalized groups to communicate their experiences [25,38]. Our findings show that the training program positively improved participants' understandings related to service user involvement, and mental health recovery journeys. This is in line with evidence that show that PhotoVoice approaches support participants to take an active role and improve understanding by conveying the why, how, and what of living with mental illness, the complexities of their recovery process, and their experiences of dealing with mental health challenges in the private and public domains [25,39]. In addition, through the active involvement of participants in the PhotoVoice process by learning new skills (e.g. photography, collaborative group working) can promote increased self-confidence, self-worth, and sense of contribution [25,39]. Moreover, we identified that service users reported the adoption of more healthy lifestyles in a number of ways, for example, reduced alcohol consumption, effective use of medication, and improved social interaction and healthcare attendance. These outcomes are in line with evidence that greater participation in PhotoVoice can have therapeutic benefits for participants to improve selfunderstanding, a sense of empowerment, self-esteem, and assistance to overcome stigma [25,40]. Similarly, case studies from diverse countries reported that programmes that engage service users to share their recovery narratives can reduce stigma, facilitate healing, and promote engagement in mental healthcare, while also assuring that service users have autonomy in the content and form of the narratives [41]. Our results show that the participants reported improved social acceptance in their community. This is consistent with the evidence that active participation and reflection during photovoice can achieve social acceptance through addressing multi-level stereotypes, stigma and feeling of shame and isolation that is often associated with mental illness [25,38]. This'social contact' aspect of PhotoVoice in challenging stigma is likely to be crucial in overcoming the main barrier that service users face to meaningful civic participation and steers away from the past systemic disadvantages in most facets of their daily lives in LMICs [42,43]. The training process had social benefits, including opportunities to meet and engage with people with similar experiences, share experiences and learn from each other, and building social networks. Others studies also reported that such sharing of stories can instil a sense of connectedness, and that participants benefit from group support due to the shared experiences [25,39,44]. In our study, service users who participated in the training decided to establish a service user association in Sodo district. The association comprised of 24 founding members (12 service users and 12 caregivers) to support each other to expand their networks in an organized way for involvement in mental health systems strengthening and to advocate for their human rights. We also identified areas of the training programme that need further attention. Health professionals suggested future training might be improved by extending the one-day program to facilitate in-depth coverage of the training content. Our training process observations show that some service users were highly emotional when narrating their recovery journey. This implies the importance of taking necessary precautions and preparation of participants for strategic disclosure of recovery narrative during the PAR photovoice process. Other studies also recommended the need for preparatory training (e.g. "Coming Out Proud") to support participants to affirm their own readiness to disclose their mental illness and combat self-stigma and reframe the meaning of mental illness [45]. --- Strengths and limitations The program development and delivery were informed by participatory action research and an experiential learning approach that promoted active involvement of participants. We employed a prospective convergent mixed methods design, which utilized multiple data sources and allowed us to triangulate findings. Nonetheless, there were several limitations of this pilot study that warrant consideration. (i) The study sample was small, (ii) there was no control group and (iii) participants were selected purposively, with the potential for selection bias that may limit generalizability of the findings. There was also concern that an (iv) interview

Background: Increased service user involvement is recommended to improve weak mental health systems in lowand middle-income countries (LMICs). However, involvement is rarely implemented and interventions to support involvement are sparse. In this study we evaluated the acceptability, feasibility and perceived outcomes of an empowerment and training program for service users and health professionals to facilitate service user involvement in mental health system strengthening in rural Ethiopia. Methods: REducing Stigma among HealthcAreProvidErs (RESHAPE) is a training curriculum for service users, their caregivers and aspirational health workers, which uses PhotoVoice methodology, to prepare them in participation of mental health systems strengthening in LMICs. We delivered the RESHAPE training augmented with empowerment content developed in Ethiopia. The interactive face-to-face training was delivered to service users and caregivers (over 10 days), and health professionals (1 day) separately. The study was an uncontrolled, convergent mixed-methods design. The quantitative data consisted of process data, satisfaction questionnaire, and a retrospective pre-test survey. Qualitative data included exit and follow-up in-depth interviews with the service users. Descriptive statistics were performed for quantitative data, and qualitative data were thematically analysed. The findings were integrated through triangulation for convergent themes following analysis. Results: Twelve service users, 12 caregivers and 18 health professionals were enrolled, and completed the training. Participants valued the content and delivery process; the standard of the training program met their expectations and participation led to positive gains in understanding about mental illness, stigma, service-user involvement and human rights. The qualitative findings identified positive impacts, including increased self-confidence, sense of empowerment, social -and perceived therapeutic benefits.We found that the RESHAPE training with added content for Ethiopia, delivered using the Photo-Voice methodology, is feasible, acceptable and of value to develop and implement training programmes which can

might be improved by extending the one-day program to facilitate in-depth coverage of the training content. Our training process observations show that some service users were highly emotional when narrating their recovery journey. This implies the importance of taking necessary precautions and preparation of participants for strategic disclosure of recovery narrative during the PAR photovoice process. Other studies also recommended the need for preparatory training (e.g. "Coming Out Proud") to support participants to affirm their own readiness to disclose their mental illness and combat self-stigma and reframe the meaning of mental illness [45]. --- Strengths and limitations The program development and delivery were informed by participatory action research and an experiential learning approach that promoted active involvement of participants. We employed a prospective convergent mixed methods design, which utilized multiple data sources and allowed us to triangulate findings. Nonetheless, there were several limitations of this pilot study that warrant consideration. (i) The study sample was small, (ii) there was no control group and (iii) participants were selected purposively, with the potential for selection bias that may limit generalizability of the findings. There was also concern that an (iv) interviewer administered assessment could lead to social desirability bias. The quantitative measures used in this study, although based on other studies of a similar nature, have undergone limited psychometric testing or were unstandardized measures with unknown psychometric properties. We employed a PAR real-time engagement approach and several triangulation techniques to obtain in-depth assessment of the perceived understanding changes, however, use of validated measures in future evaluation of this training will be useful in benchmarking improvements against other training approaches. Moreover, this study did not evaluate the impact on health workers attitudes and behaviors, which are the end targets of the RESHAPE intervention. --- Conclusions Our study has provided preliminary evidence of the acceptability, feasibility and perceived benefits of a training programme to empower and equip service users and health workers to collaborate to strengthen mental health systems in rural Ethiopia. Future work needs to focus on evaluation of the programme at a larger scale and to further investigate the ongoing support required to achieve sustainable service user involvement. --- Availability of data and materials The datasets generated during and/or analysed during the current study is available from the corresponding author on reasonable request. --- Abbreviations --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12913-022-08290-x. --- Additional file 1. --- Authors' contributions --- Declarations Ethics approval and consent to participate All methods were carried out in accordance with guidelines and regulations of the Declaration of Helsinki. This study was approved by the Institutional Review Board of Addis Ababa University College of Health Sciences (027/16/Psy). All participants were fully informed about the study. Informed consent for study participation was obtained from all subjects (if subjects are under 16/illiterate/Non-literate/dead, from parent or legal guardian/next). --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Increased service user involvement is recommended to improve weak mental health systems in lowand middle-income countries (LMICs). However, involvement is rarely implemented and interventions to support involvement are sparse. In this study we evaluated the acceptability, feasibility and perceived outcomes of an empowerment and training program for service users and health professionals to facilitate service user involvement in mental health system strengthening in rural Ethiopia. Methods: REducing Stigma among HealthcAreProvidErs (RESHAPE) is a training curriculum for service users, their caregivers and aspirational health workers, which uses PhotoVoice methodology, to prepare them in participation of mental health systems strengthening in LMICs. We delivered the RESHAPE training augmented with empowerment content developed in Ethiopia. The interactive face-to-face training was delivered to service users and caregivers (over 10 days), and health professionals (1 day) separately. The study was an uncontrolled, convergent mixed-methods design. The quantitative data consisted of process data, satisfaction questionnaire, and a retrospective pre-test survey. Qualitative data included exit and follow-up in-depth interviews with the service users. Descriptive statistics were performed for quantitative data, and qualitative data were thematically analysed. The findings were integrated through triangulation for convergent themes following analysis. Results: Twelve service users, 12 caregivers and 18 health professionals were enrolled, and completed the training. Participants valued the content and delivery process; the standard of the training program met their expectations and participation led to positive gains in understanding about mental illness, stigma, service-user involvement and human rights. The qualitative findings identified positive impacts, including increased self-confidence, sense of empowerment, social -and perceived therapeutic benefits.We found that the RESHAPE training with added content for Ethiopia, delivered using the Photo-Voice methodology, is feasible, acceptable and of value to develop and implement training programmes which can

Introduction Being overweight or obese is defined as abnormal or excessive fat accumulation that may impair health. In 2016, over 340 million children and adolescents aged between 5 and 19 were overweight or obese [1]. Over the last fifty years, the obesity rate in adolescents has increased by approximately 5% per decade, and approximately one-quarter of children are now either overweight or obese [2]. Rapidly developing countries are showing a significant increase in childhood obesity [3]. In China, the overweight rate among children and adolescents aged between 6 and 17 was 9.6%, and the obesity rate was 6.4% in 2015, increasing from 2002 by 5.1% and 4.3%, respectively, demonstrating a rising tendency [4]. Obesity in children and adolescents harms both physical and mental health. Consequences mainly include: (1) continued obesity in adulthood; (2) increased risk of dis-eases, such as type 2 diabetes, cardiovascular disease, chronic kidney disease, and cancer; (3) increased death and premature mortality [3]. A growing body of literature links obesity to negative psycho-social consequences in adolescents, including poor emotional adjustment and stigmatisation, distorted perceptions of body image, antipathy from peers and adults, social withdrawal, and aggression, which are even more distinct in girls [5][6][7][8]. Currently, potential mediators of the association between being overweight and emotional well-being have become the focus. For example, one study showed that when body dissatisfaction was statistically controlled, the association between obesity and low self-esteem was no longer evident, nor was the relationship between obesity and depression [5,9]. These findings suggest that psycho-social factors should be taken into account when designing research or intervention programmes that deal with obesity [10][11][12][13]. In light of the above findings, interventions aiming to alleviate adolescent obesity should be considered. First, researchers have to identify the factors related to adolescent obesity. In general, physical activity (PA), dietary behaviour (DB), and family functioning (FF) are usually considered risk factors for adolescents being overweight and obese, despite several opposite conclusions. Physical activity has been a popular focus in this regard. It is well-established that regular PA benefits health and helps adolescents improve cardiorespiratory fitness, control weight, improve psychological well-being, and reduce the risk for chronic diseases [14,15]. Moreover, the relationship between PA and an epidemic of childhood obesity has been well-documented, in which PA is commonly regarded as a protective factor against adolescent obesity [16,17]. According to the Amsterdam Healthy Weight program, policies that encourage active transport and safe walking or cycling effectively reduce adolescent obesity [18,19]. The fundamental to reducing adolescent obesity is improving weight-related health behaviours, which basically balance energy intake and consumption [20]. Currently, the main perspective is still that high consumption of fried food or snacks high in sugar and fat leads to poor nutritional status and childhood obesity [21][22][23][24]. A meta-analysis concluded that sufficient consumption of fruits/vegetables, snacks, or fried food had a nonsignificant effect on childhood obesity, but consuming excessive sugar-sweetened beverages increased the risk of obesity by 24% [14]. According to the data from the China Health and Nutrition Survey (CHNS), the Chinese modern dietary pattern is characterised by high consumption of red meat and processed meat, as well as the use of the unhealthy cooking method of deep-frying, and such a dietary pattern abundant in saturated fat and cholesterol is reported to be positively associated with a higher risk of obesity [25]. Moreover, some studies have found that Chinese parents and grandparents are more inclined to push their children to eat a set amount of food at each meal, despite the child's hunger and satiety signals, which causes common overeating among children and adolescents [26]. In addition to PA and dietary behaviours, environmental, genetic, biological, and socioenvironmental factors, especially family factors, are also crucial [3,13]. For example, research reported a positive correlation between household income and obesity, in which household income may alter the total calorie intake, DB, or PA of school children [27]. Among the studies on the interrelationship between family and children's health behaviours, the prevailing conclusion is that parents influence their children directly through parental practices and role modelling, which affects children's weight status [13,28,29]. In recent years, the family context has become an important factor in the formation of healthy behaviour in adolescents [13,[28][29][30][31]. One of the key contextual factors of concern is family functioning (FF) [32]. In short, FF is how families cope with their day-to-day affairs by fulfilling their roles in the family [33]. Most studies showed a strong association between FF and adolescent obesity [31,34,35], but some studies concluded that there was no significant relationship between the two [36]. To our knowledge, no study has yet addressed the temporal association between FF and being overweight. FF was regarded as a moderator between health behaviours and parenting practices or parental modelling, but few existing studies have reached a consensus on its effects [29,37,38]. Multiple previous studies with small and accessible samples focusing on the relationship between FF and obesity-related behaviours have shown that healthier FF is associated with healthier dietary intake, less sedentary behaviour, and more physical activity, which is consistent with the finding of a national study among adolescents in the USA [29,31,33,39]. Children in dysfunctional families resort to effective but eventually dysfunctional eating habits, and they often consume energy-dense junk food to relieve emotional and stress-related pressure. All of these factors lead to poor self-rated health and severe obesity [8,40]. Therefore, a supportive and harmonious family environment is important for children's growth and development [20,32,34]. In conclusion, PA, DB, and FF are all considered to be associated with adolescent obesity. Moreover, associations between FF, exercise, and diet have been well-documented in several studies, including those conducted among overweight adolescents [41]. Given these studies, health behaviours and FF may jointly influence childhood obesity and being overweight. However, few studies have examined the interaction role of PA, DB, and FF in exerting their influence on adolescent obesity and being overweight and, therefore, how they work together is still unclear [32,41]. This study provides updated and innovative evidence of the multiplication effects and marginal effects between FF and weight-related health behaviours, including physical activity and dietary behaviour, that occur daily on adolescent BMI so that more efficient early screening and prevention can be taken into practice to manage the global adolescent obesity epidemic. --- Materials and Methods --- Participants and Procedures This study, conducted from September to December 2015, was a school-based crosssectional health survey utilising a random cluster sampling design in Shanghai, China. The research team reached out to five junior high schools in Shanghai and introduced this study to them, with the aim of establishing collaborations. Among these, three schools in Jing'an District showed interest. All students in the 6th, 7th, and 8th (10 to 16 years old) grades of these schools were invited to take part in the study on a voluntary basis. The sample size calculation principle was adopted in this study, which required that the sample size statistically cover ten times the number of items [42,43]. An 87-item questionnaire was synthesised to respond to the research question, with 9 items for sociodemographics, 3 items for physical activity, 15 items for dietary behaviours, and 60 items for family functioning. Hence, a sample size of 870 was the minimum. Considering a nonresponse rate of 20%, the target sample size was increased to 870/0.8 = 1088. The consent forms with descriptions of the study and the questionnaires were distributed to students and their parents by headteachers at the beginning of the semester. By returning the consent forms with both the parent's and the student's signatures, along with the completed questionnaires, the students indicated participation in the study. Out of 1088 students from three schools, 937 (86.12%) effective responses were received. A schematic of the study design is shown in Figure 1. --- Measures The data collection consisted of two parts: (1) anthropometric measurements conducted by trained researchers and (2) a self-report questionnaire completed by students that covered information about sociodemographics, physical activity (PA), dietary behaviours (DB), and family functioning (FF). --- Outcome Variables Body weight (to the nearest 100 g) and height (to the nearest 0.5 cm) were obtained using a portable height and weight machine by the researchers. Respondents wore only light clothing and did not wear shoes during the measurement. For each student, two readings were taken for body weight and height, the averages of which were adopted for analysis. Body mass index (BMI) was computed following the formula BMI = kg/m 2, where kg refers to the weight in kilograms, while m2 stands for height in metres squared. Age-and sex-specific BMI z scores were calculated based on a Shanghai local reference. BMI status was classified into three categories: non-overweight (underweight and normal weight) (BMI <unk> 84th percentile), overweight (BMI <unk> 85th and <unk>95th percentile), and obese (BMI <unk>95th percentile) [44]. --- Exposure Variables Physical Activity (PA) Selected items from the CDC Youth Risk Behaviour Survey were translated and reviewed by the research team experts to evaluate physical activity levels [45]. Indicators included days of exercise and the number of PE classes per week, which were rated on a scale from 1 to 7; in addition, participation in sports teams in the last year was rated from 1 to 4. The total score ranged from 3 to 9 after coding. A higher score indicated a higher level of physical activity. A score less than 4 was regarded as not active; otherwise, it was regarded as active. --- Measures The data collection consisted of two parts: (1) anthropometric measurements conducted by trained researchers and (2) a self-report questionnaire completed by students that covered information about sociodemographics, physical activity (PA), dietary behaviours (DB), and family functioning (FF). --- Outcome Variables Body weight (to the nearest 100 g) and height (to the nearest 0.5 cm) were obtained using a portable height and weight machine by the researchers. Respondents wore only light clothing and did not wear shoes during the measurement. For each student, two readings were taken for body weight and height, the averages of which were adopted for analysis. Body mass index (BMI) was computed following the formula BMI = kg/m 2, where kg refers to the weight in kilograms, while m2 stands for height in metres squared. Age-and sex-specific BMI z scores were calculated based on a Shanghai local reference. BMI status was classified into three categories: non-overweight (underweight and normal weight) (BMI <unk> 84th percentile), overweight (BMI <unk> 85th and <unk>95th percentile), and obese (BMI <unk> 95th percentile) [44]. --- Exposure Variables Physical Activity (PA) Selected items from the CDC Youth Risk Behaviour Survey were translated and reviewed by the research team experts to evaluate physical activity levels [45]. Indicators included days of exercise and the number of PE classes per week, which were rated on a scale from 1 to 7; in addition, participation in sports teams in the last year was rated from 1 to 4. The total score ranged from 3 to 9 after coding. A higher score indicated a higher level of physical activity. A score less than 4 was regarded as not active; otherwise, it was regarded as active. --- Dietary Behaviours (DB) The questionnaire for dietary behaviours was constructed with reference to the Dietary Practices Questionnaire used in a qualitative study examining Hong Kong residents' dietary practices [46]. There were 13 questions exploring dietary behaviours concerning (a) fruit and vegetable consumption (e.g., "What is the amount of your average daily intake of fruits?"), (b) protein and dairy consumption (e.g., "What is the amount of your daily intake of meat, fish, and egg?"), (c) high-fat food consumption (e.g., "What is the amount of your daily intake of fried food"), and (d) dietary habits (e.g., "Apart from meals, how often do you have snacks within a day?"). Respondents were asked to recall the frequency/amount of food consumption and their dietary habits for the past 7 days. A total score was calculated for each of the four categories. The total score ranges from 4 to 10 for fruit and vegetable consumption, 3 to 7 for protein and dairy consumption, 2 to 6 for high-fat food consumption, and 6 to 17 for dietary habits. A higher score indicates healthier behaviours. All scores were divided into two classes: unhealthy and healthy. --- Family Functioning (FF) The McMaster Family Assessment Device (FAD) was employed to study family functioning [47] in our study. To our knowledge, the FAD is the only measurement that has a subscale that specifically assesses the general FF. There are seven subscales of the FAD: (1) problem solving (i.e., the ability of family members to solve problems to achieve good family functioning), ( 2) communication (i.e., the exchange of information between family members), ( 3) roles (i.e., if the family has regular patterns of behaviour to deal with family functioning), (4) affective responsiveness (i.e., the ability of family members to respond emotionally appropriately to environmental stimuli), (5) affective involvement (i.e., the degree of warmth among household members), (6) behavioural control (i.e., whether the family has a norm or standard to guide individuals' reactions to emergencies), and (7) general functioning (i.e., the overall level of family functioning). Respondents were required to answer 60 items using a 4-point Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). Sample items are "We are able to make decisions about how to solve problems" (positive items) and "We cannot talk to each other about sadness we feel" (negative item); in addition, there are 35 positive questions and 25 negative questions. A mean score was computed by averaging the item scores after reverse scoring negatively worded items. For analytic purposes, the score was dichotomized. Participants were classified into the healthy FF group if their average scores were 3 or 4, while those with scores of 1 or 2 were classified as unhealthy FF [29,48]. --- Adjusting Variables Demographic variables were also collected, including sex, grade, age, length of stay in Shanghai, birth weight, primary caretaker, and monthly household income. According to the law on mandatory education in China [49], students of the 6th, 7th, and 8th percentiles were divided into three categories: <unk>11, 12, and <unk>13 years old. Length of stay in Shanghai was classified as <unk>5, 5-10, and <unk>10 years. Birth weight was grouped as <unk>2.5 kg, 2.5-4 kg, and <unk>4 kg [50]. Primary caretakers were classified into four groups: parents, grandparents, grandparents, and others. Monthly family income was divided into <unk>CNY 3000, CNY 3001<unk>6000, CNY 6001<unk>9000, CNY 9001<unk>12,000, and <unk> CNY 12,000. In addition, parental BMI was classified into non-overweight (underweight and normal), overweight, or obese (<unk>24.0 kg/m 2, 24.0-27.9 kg/m 2, <unk> 28.0 kg/m 2, respectively) [51]. --- Statistics The data were analysed with Stata version 17.0. Descriptive data were calculated to indicate the prevalence of different BMI statuses. Frequencies and percentages were calculated for the remaining categorical variables. Chi-square tests were conducted to compare the difference in BMI status within the identified groups. Multinomial logistic regression analyses were used to calculate the odds ratio (OR) for "non-overweight", "overweight" and "obese". Multiplication interaction models were used to analyse the interaction of risk factors, including PA, HF, and FF, on BMI status. We then computed marginal effects at means following the logistic regression models. Statistical significance was defined as p <unk> 0.05 (two-tailed). --- Results --- Sample Characteristics and BMI Status Among the 937 participants (54.6% male), 36.71% were overweight or obese, with a mean BMI of 23.20 for overweight participants and 28.16 for obese participants (Table 1). The characteristics of the participants and the results of the chi-square tests are shown in Table 2. The average age of the sample was 12.47 <unk> 1.14 (M <unk> SD). The number of students from Grade 6, Grade 7, and Grade 8 was 362, 290, and 285, respectively. The average birth weight was 3.62 <unk> 1.04 kg, and 9% of students had macrosomia. Regarding the primary caretakers, 89.3% of students were cared for by parents, and others were usually raised by grandparents. In addition, approximately half of the students' family incomes were more than CNY 9000 per month. A total of 53.7% and 24.1% of students' fathers and mothers were overweight or obese, respectively. Among the sociodemographic variables, sex (p <unk> 0.001), length of stay (p <unk> 0.05), birth weight (p <unk> 0.05), monthly family income (p <unk> 0.05), and father's BMI status (p <unk> 0.001) were found to be significantly related to student BMI status (Table 2). In addition, the results showed that 27% of participants were physically active. There was a significant relationship (p <unk> 0.05) between students' physical activity and BMI status. Regarding dietary behaviours, only high-fat food consumption was related to BMI status (p <unk> 0.05), with 9% of the students reporting having an excessive amount of HF. In addition, BMI status was found to be significantly dependent on family functioning (p <unk> 0.05). Most of the students' families (87%) had healthy family functioning. The primary caretakers and monthly household income index of 42 and 66 people were unknown, respectively. * Chinese yuan (currency unit). The average exchange rate between USD and the CNY in 2016 was 6.642. --- Multinomial Logistic Regression Analysis of BMI Status Multinomial logistic regression analysis was conducted to identify predictors of BMI status. In this model, the dependent variable was BMI status; variables that were found to be related to BMI status based on chi-square tests were entered as independent variables. The results are displayed in Table 3. Male students were more likely to be obese vs non-overweight (OR = 0.229, 95% CI [0.142, 0.369]) and overweight (OR = 0.538, 95% CI [0.316, 0.915]) than female students. The odds of having the BMI status of non-overweight (OR = 2.503, 95% CI [1.149, 5.542]) and overweight (OR = 2.672, 95% CI [1.075, 6.641]) were both twice as high as obesity in students whose birth weight was between 2.5 and 4.0 kg relative to those who were heavier than 4.0 kg. Students whose family monthly income was between CNY 3001 and CNY 6000 compared to more than CNY 12,000 had 2.4 times higher odds of being non-overweight than overweight (OR = 2.432, 95% CI [1.289, 4.589]). Students with non-overweight fathers had 2.8 higher odds of being non-overweight than obese in contrast to students with obese fathers (OR = 2.873, 95% CI [1.496, 5.517]). Physically active students had higher rates of being non-overweight than overweight (OR = 2.260, 95% CI [1.318, 3.874]), as well as better odds of being overweight than obese (OR = 2.096, 95% CI [1.167, 3.766]) in comparison with physically inactive students, indicating that a healthy level of PA was a protective factor against adolescent obesity. Students who did not consume excess HF were less likely to become non-overweight as opposed to obese (OR = 2.048, 95% CI [1.105, 3.796]). Healthy FF reduced the odds of being obese compared to overweight compared to unhealthy family functioning (OR = 2.084, 95% CI [1.099, 3.952]). --- The Multiplicative Interaction among PA, HF, and FF To investigate how PA, HF, and FF work jointly to influence BMI status, we examined the multiplicative interaction among the three indicated variables. As shown in Table 4, there was a multiplicative interaction between FF and PA. Comparing the BMI status of non-overweight and obesity, students with healthy FF, whether their PA was active or not, had a lower risk of obesity, and those whose PA was active were more likely to stay as non-overweight (inactive: OR = 1.794, 95% CI [1.062, 3.030]; active: OR = 3.44, 95% CI [1.760, 6.722]). Overweight students who lived in families with healthy FF were less likely to become obese regardless of whether they exercised (inactive: OR = 2.181, 95% CI [1.114, 4.272]; active: OR = 3.870, 95% CI [1.719, 8.713]). A similar trend was observed between FF and HF. Adolescents who maintained healthy high-fat food consumption in families with healthy FF were less likely to be obese (OR = 4.989, 95% CI [1.930, 12.899] for non-overweight). Those with healthy FF were at less risk of obesity even though they consumed superfluous HF (unhealthy: OR = 4.615, 95% CI [1.049, 20.306]; healthy: OR = 5.116, 95% CI [1.352, 19.362]). --- The Marginal Effects between FF and PA or HF Based on logistic regression, we tested the marginal effects of FF, PA, and HF on BMI status. As shown in Figure 2a, the MEs of inactive PA and unhealthy FF were -0.071, 0.035, and 0.036 for non-overweight, overweight, and obese individuals, respectively (p <unk> 0.05). Figure 2b shows that the ME of unhealthy HF and healthy FF was -0.267 to non-overweight and 0.198 to obese (p <unk> 0.05). --- Discussion The study set out to examine the multiplicative interaction among physical activity, high-fat food consumption, and family function with respect to their connections with adolescents' BMI status. In this study, 36.71% of the students were overweight or obese, with a mean BMI of 23.20 kg/m 2 for overweight and 28.16 kg/m 2 for obese, which was much higher than the prevalence of being overweight and obese among children aged --- Discussion The study set out to examine the multiplicative interaction among physical activity, high-fat food consumption, and family function with respect to their connections with adolescents' BMI status. In this study, 36.71% of the students were overweight or obese, with a mean BMI of 23.20 kg/m 2 for overweight and 28.16 kg/m 2 for obese, which was much higher than the prevalence of being overweight and obese among children aged from 7-18 in China of 19.4% in 2014 (41,608/214,354) [52]. This may be because the study was conducted in Shanghai, which is a well-developed urban city. It is well documented that obesity rates among urban adolescents are higher than those in rural areas, which is associated with nutritional intake, physical activity, and air pollution due to traffic [53]. Connections between BMI status and several sociodemographic factors were identified in this study. First, consistent with a national investigation from China [52], boy students were more likely to be obese than girls. Possible explanations for this result concerns both children's and parents' health perceptions and behaviours. For example, Chinese parents tend to believe that physical development is different between boys and girls. This may lead to variance in feeding behaviours, mainly in the form of overfeeding boys, resulting in higher rates of boys being overweight and obese [54]. In addition, girls and boys have different perceptions of weight and diet, with girls feeling more pressure to be thinner and having lower confidence regarding their weight [53][54][55]. It also found that students with relatively low household incomes were more likely to be obese than non-overweight students compared to those with high monthly incomes. This finding was consistent with previous studies that revealed a negative association between income and obesity in China [56,57], and the burden of disease may shift to those of lower socioeconomic status in the future [56,57]. Likewise, studies from Europe and the USA also reported that decreasing family income and socioeconomic indicators were related to increasing the risk of childhood obesity and being overweight [58]; however, in low-and middleincome countries, affluence is positively associated with obesity [59]. Dietary behaviours could be a possible explanation for the difference. The high rate of being overweight among adolescents from economically affluent households may be due to a significantly higher weekly intake of meat or fish, eggs, dairy products, legumes, fruits, and vegetables than those from lower-income populations [56]. The high rate of being overweight in comparatively low-income households could be a characteristic of the disease burden of obesity in comparatively high-income countries. This is mainly because children from low-income families are more likely to eat unhealthy foods and develop unhealthy eating habits due to budget constraints [60]. Finally, the overweight and obesity rates in this study were associated with birth weight and paternal BMI, genetics, and nutrition during pregnancy, which may account for these connections [39,61,62]. The results concluded that unhealthy FF, inactive PA, and unhealthy HF were all risk factors for obesity. This is consistent with the findings of several previous studies [14,16,23,24,34,39]. Furthermore, we confirm our hypothesis that unhealthy FF and poor health behaviours have a synergistic influence on obesity in adolescents, together with increasing the risk to a large extent. There were MEs of inactive PA on being overweight and obese when family functioning was unhealthy and an ME of excessive HF when family functioning was healthy. Our findings are complemented by interaction effects that confirm the dynamic relationships among PA, HF, FF, and obesity. In particular, our conclusions add to the extensive literature on the influence of FF on children's health behaviour, including those in obese populations [28,29,31,33,37,39,41]. A Canadian study demonstrated that FF was not an intermediate factor between parental practices and role modelling and healthy behaviour and, thus, it should be studied separately with regard to health behaviours [41]. However, similar to this study, a few studies are revealing that FF interacts with health behaviours. For example, it was shown that unhealthy FF was associated with less parental monitoring of their children's diet and exercise [39]. Families with more conflict tend to pacify their children with tasty but high-fat food [63]. Moreover, previous evidence from intervention trials suggests that improved FF precedes weight-related changes, and that unhealthy FF has been shown to contribute to obesity [41,64]. These findings have shed light on designing obesity intervention programmes, where not only healthy behaviours but also FF should be included. In summary, previous evidence suggested that DB, PA, and FF were all independently associated with BMI status, which was in line with our findings. Furthermore, our study is the first to prove that health behaviours and FF jointly influence adolescent obesity. However, the temporal order is still unclear [32,41]. With known findings, adolescents with health-related behaviour problems and obese students may be at greater risk of a poor family environment. Therefore, we suggest that clinicians assess the FF of these adolescents' families. Both society and schools should seek opportunities to intervene in interfamilial problems [34,64]. Additionally, family members are expected to be aware of the importance of FF and strive to enhance family relationships through efficient ways, such as family health conversations [63][64][65]. Because this study is the first of its kind to investigate the interaction relationships among the indicated factors, one of the future directions could be replicating this study in more locations to gain sufficient data to verify the found associations. Moreover, cohort intervention studies that target improving family functioning in families with overweight/obese adolescents should be conducted to further clarify the temporal association between FF and health behaviours. Currently, there is a paucity of literature on how the seven subscales of FF influence specific behaviours [39,66]. Future research may be interested in linking FF and healthy behaviours in a more detailed manner to provide more precise directions for interventions. The limitations were as follows: (1) the study adopted a cross-sectional design, so we could not establish causal or temporal relationships between FF and adolescent BMI and health behaviours. Additionally, there might be residual and unmeasured confounding. However, our adjustment for potential demographic confounders had no substantial impact on our association estimates. (2) The study was based on an economically developed urban city, where the economic level was higher than the national general average. Therefore, replications of this research in different socioeconomic backgrounds are recommended for future generalization. (3) Only fried food and desserts were included as high-fat foods. Future research may benefit from including more types of high-fat food, such as takeaway food. --- Conclusions Unhealthy HF, inactive PA, and unhealthy FF will increase the risk of being overweight and of obesity among adolescents. Unhealthy FF is a regulatory variable of either inactive PA or unhealthy HF increasing the risk of obesity in youths, which leads to a much higher risk, and healthy FF is also a regulatory variable of active PA or healthy HF helping maintain being non-overweight, which intensifies the effect. Institutional Review Board Statement: The studies involving human participants were reviewed and approved by the Ethics Committees of Tongji University (ref: LL-2016-ZRKX-017). The participants provided their written informed consent to participate in this study. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. --- Data Availability Statement: The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding authors. --- Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. --- Abbreviations

Objectives: Family functioning (FF), physical activity (PA), and high-fat food consumption (HF) are associated with adolescents being overweight and obese; however, little is known about their interactions. Therefore, this study aimed to examine how they work jointly on adolescent obesity with BMI as the outcome variable. Methods: A cross-sectional survey utilizing a cluster sampling design was conducted. Multinomial logistic regressions, multiplication interaction (MI), and marginal effects (MEs) were tested. Results: Active PA (non-overweight vs. obesity: OR = 2.260, 95% CI [1.318, 3.874]; overweight vs. obesity: OR = 2.096, 95% CI [1.167, 3.766]), healthy HF (non-overweight vs. obesity: OR = 2.048, 95% CI [1.105, 3.796]) and healthy FF (overweight vs. obesity: OR = 2.084, 95% CI [1.099, 3.952]) reduced obesity risk. Overweight students with healthy FF were less likely to become obese regardless of PA (inactive: OR = 2.181, 95% CI [1.114, 4.272]; active: OR = 3.870, 95% CI [1.719, 8.713]) or HF (unhealthy: OR = 4.615, 95% CI [1.049, 20.306]; healthy: OR = 5.116, 95% CI [1.352, 19.362]). The MEs of inactive PA and unhealthy FF were -0.071, 0.035, and 0.036 for non-overweight, overweight, and obese individuals, respectively (p < 0.05); the MEs of HF and healthy FF individuals were -0.267 and 0.198 for non-overweight and obese individuals, respectively (p < 0.05). Conclusions: Unhealthy FF regulated the influence of inactive PA or unhealthy HF on adolescent obesity, altogether leading to a higher risk of obesity.

INTRODUCTION Social scientists have increasingly invested in research to understand how to improve data quality and measurement in household surveys. In particular, there is ongoing innovation to better capture measurement of sensitive topics, including indicators of intimate partner violence (IPV), which are traditionally thought to be underreported. Survey respondents may under-report or answer untruthfully due to factors related to stigma, shame, social desirability or social pressures, and fear of retaliation or legal action, among others (Blair 2015;Palermo et al. 2014). Because of compounding influences and differences by respondent subpopulations, it is often difficult to understand in a meaningful way the directions and magnitude of bias in a particular sensitive indicator. Additionally, motivation for measurement innovation may stem from the desire to measure sensitive behaviors within the context of a multi-topic evaluation, where little time and resources are available to incorporate ethically challenging topics which are not the primary objective of the data collection. List randomization (or list experiments), a technique originally used by political scientists in political polling and opinion surveys (Kuklinski et al. 1997a(Kuklinski et al. /1997b)), has recently been applied to sexual behavior and health. In basic form, the method aggregates a response to a question of sensitive nature alongside responses to non-sensitive questions, thereby masking the specific sensitive response to the interviewer. By randomizing lists containing a sensitive item and lists containing non-sensitive items, in aggregate, researchers can then identify prevalence within a given population. If respondents' believe their answer remains undisclosed to the interviewer, they may be more likely to report behaviors which are private in nature. List randomization has been shown to increase disclosure to sensitive items, however has also been critiqued for complexity and implementation challenges (Blair 2015). List randomization has recently been applied to solicit attitudes around female genital mutilation/cutting (FGM/C) in Ethiopia (De Cao & Lutz 2015), and sexual behavior including condom use, number of partners and unfaithfulness in Colombia and Uganda (Chong et al. 2013;Jamison et al. 2013). To our knowledge, this technique has never been applied to collect IPV. We utilize list randomization to collect a measure of IPV in an experimental impact evaluation of the Government of Zambia's Child Grant Program (CGP), an unconditional cash transfer. The CGP was initiated in 2010 with the overall goal of reducing extreme poverty and curbing the intergenerational transfer of poverty. The CGP was implemented by the Ministry of Community Development, Mother and Child Health (MCDMCH) and operated in Kalabo, Shangombo and Kaputa districts which represent the districts with the highest rates of mortality, morbidity, stunting and wasting among children aged zero to five years. In eligible areas, the CGP transfers a fixed cash sum of approximately US $24 transferred on a bi-monthly basis (US $12 per month equivalent, or approximately 27% of baseline household consumption), to the primary female adult in households with a child under the age of five. While there is evidence that the CGP met its basic poverty-reduction targets, as well as increased women's financial standing and intra-household decisionmaking power specifically (Bonilla et al. 2017;Handa et al. 2015;Natali et al. 2016), there was specific interest in understanding potential for impacts on IPV. Existing literature on cash transfers, primarily from Latin American suggests that cash can play a role in reducing IPV experienced by women in beneficiary households, through reductions in poverty-related stress and increases in women's status within the household (Buller et al. 2016;Hidrobo et al. 2016;Hidrobo & Fernald 2013). Despite promising evidence, the relationship between cash transfers and IPV is not unambiguously negativein particular there is a possibility of an increase in IPV in cases where men may seek to gain control over resources or re-assert power (Bloch & Rao 2002;Gibbs et al. 2017). However, to date there are only two quantitative impact evaluations from Africa examining impacts of cash transfers on IPV outcomes: in Kenya, a 12-month unconditional cash transfer is found to decrease 6-month experience of physical and sexual IPV across villages, however has mixed findings comparing within villages (Haushofer & Shapiro 2016); in South Africa a four-year cash transfer conditional on schooling is found to decrease physical IPV among females aged 13 to 20 years at baseline (Pettifor et al. 2016). One reason for lack of evidence is that implementers of already overstretched multi-topic surveys do not have the time or resources to devote to collecting additional indicators-particularly those of a sensitive nature which require additional ethical considerations and referral processes (WHO 2001). --- METHODOLOGY The evaluation of the CGP was designed as a cluster-randomized control trial (cRCT) with 90 clusters dived into treatment (45) and control (45). A baseline survey among a random sample of 2,515 beneficiary and non-beneficiary households was carried out in October-November of 2010. Four longitudinal follow-up surveys were subsequently collected after 24, 30, 36, and 48 months. The IPV list randomization was conducted in the 48-month wave which contained a module on women's empowerment asked of the primary female (CGP beneficiary) in each household. We include in our sample 2,171 women age 15 to 49 among the full sample and 1,617 among the currently married or co-habiting sub-sample who answered the list randomization. Further, we restrict our analysis to females who have data at 48-month follow-up and who were living in study households since the start of the evaluation, for an attrition rate of 13% (p = 0.14 difference between treatment and control groups, calculated at the household level among a sample of baseline households with responses to the women's empowerment module). Therefore, our estimation of program impacts relies on the successful randomization into treatment and control at baseline, which has been demonstrated elsewhere (AIR 2015). We show overall and differential attrition rates in Appendix Tables 1 and2 and similar to other publications which have interrogated attrition for the sample of women answering the empowerment module, do not found attrition to be problematic for internal validity (Bonilla et al. 2017;Natali et al. 2016). The list randomization was implemented as follows: Enumerators were instructed to ask either Panel A or Panel B based on the even/odd numbering of the last digit of the household identification number. Panel A instructed to read: "Here is a list of four things that some people have done, and some people have not. Please listen to them and then tell me HOW MANY of them you have done or experienced in the last 12 months. Do not tell me which you have done, just tell me how many total. Wait until I have read the entire list to respond." The list in Panel A included the following (non-sensitive) items: --- 1. Taken care of a sick relative who is unable to care for themselves --- 2. Gone to visit one of your child's teachers to talk about their progress in school --- 3. Traveled to the district center to buy goods or sell items you produced --- 4. Attended the wedding celebration of a friend Panel B was identical to Panel A, however included a fifth item: --- 5. Been slapped, punched, kicked, or physically harmed by your partner Thus, the list comparison is between the group of women randomized to answer the 4-item non-sensitive list, and the group randomized to answer the 5-item list containing the sensitive question about IPV. Subtracting the number of affirmative reported statements in the 5-item group from the 4-item group gives us an estimate of the percentage of women in our sample who report IPV. It is important to note that we ask about physical IPV only, and ask only three behaviorally specific outcomes: (slapped, punched and kicked), in contrast to the typical modified Conflict Tactics Scale implemented in the WHO Violence Against Women surveys and the Demographic and Health Surveys (DHS) (Ellsberg & Heise 2005;Hindin et al. 2008). We estimate the impact of the CGP on IPV (total number of list outcomes reported), given by the coefficient on the interaction of the CGP treatment indicator and a binary indicator of whether the woman was asked the 5-item list containing the violence question. We run multivariate adjusted and unadjusted models, controlling for age of woman, highest grade completed, marital status, log of household size, and district of residence (baseline values of all covariates) and cluster standard errors at the primary sampling unit level. --- RESULTS Table 1 shows the descriptive results of outcome indicators and background characteristics by randomization to the "violence set (5-item list)" or "control set (4-item list)." Descriptive statistics show that the mean age at baseline of the women interviewed is 28.2 with an average of approximately grade four education completion. The majority of women (nearly 3/4 of the sample) were married or co-habiting at baseline, while 14% were divorced/ separated or widowed and remaining 12% had never been married or had never cohabited. All covariates were balanced between those who were asked the violence set and control set, with the exception of household size among the married or co-habiting sample (Panel B). In total, respondents identified 1.9 list items which were true (affirmative answers), with means of 1.978 (SE=0.039) for the 5-item group (violence set) and 1.832 (SE 0.034) for the 4-item group (control set). This indicates that approximately 14.6% (1.978-1.832; SE=0.052) women experienced physical IPV in the last 12-months. This same percentage is 14.95% in the married or cohabiting sample (1.959-1.809; SE=0.0598). The percentage of women asked the 5-item scale was 50.5%, and paired with linkages to household identification numbers, shows enumerators were able to follow instructions regarding the random assignment to groups A and B, and thus the list randomization functioned as intended. However, the statistical validity of list randomization measures depends on a number of assumptions, including absence of design effects. We formally test for this possibility and discuss in the Appendix alongside Table A3, initial results suggest we cannot reject the null hypothesis of no design effects, and thus proceed with analysis under the assumption of no design effects (Blair & Imai 2012). The prevalence of 12-month physical IPV reported in the Zambian DHS 2013-14 among ever-married women age 15-49 is 21.2% (95% confidence interval: 20.0-22. 5, CSO et al. 2014), indicating that our estimates are in the range expected according to national statistics. The lower incidence likely stems from the fact that DHS asks seven behaviorally specific questions related to physical IPV [1) pushed, shook, threw something at you, 2) Slapped, 3) Twisted your arm or pulled your hair, 4) Punched with fist or with something that could hurt you, 5) Kicked or dragged or beaten, 6) Strangled or burned on purpose, 7) Threatened you or attacked you with a knife, gun or other weapon]. Table 2 shows the result of the regression analysis with unadjusted (1 and 3) and adjusted models (2 and 4) predicting the number of list randomization questions women answered affirmative for in the full women's sample and the married/cohabiting sub-group. We find no measurable significant impact of the CGP on IPV in any of the models as shown by the interaction term (Treatment*Asked violence set). Following Holbrook and Krosnick (2010) we run additional analysis with interactions between asked violence set and demographic indicators, and find no change in results (Appendix Table A4). --- DISCUSSION As interest grows in the potential of other economic-based interventions to reduce IPV, innovations in methodology are needed to include and test hypotheses in evaluations not directly set up to measure violence (World Bank, 2015). This could include instances were evaluations are overstretched, and resources cannot be dedicated to logistical necessities (e.g. same sex enumerators) or ethical standards (e.g. ensuring safe referral mechanisms, additional enumerator training). This could also include instances where there is no buy-in from stakeholders for fear of discovering adverse effects, or in situations where IPV is taboo and where low women's status prevents or threatens credible responses to direct questioning. We demonstrate that list randomization is a credible method to implement in cases were direct elicitation is not possible. There are several limitations to our analysis. First, the question we implemented to capture IPV was basic, and it is likely that we captured a lower bound of 12-month physical IPV. In addition, although we find no significant impacts on IPV-we cannot reject the possibility that our study lacked power to detect differences in this measure. Further, we were not able to implement a more complex survey design where we collected full standard IPV modules to validate the list experiment, for precisely the same reasons which motivate this paper. Finally, Glynn (2013) suggests methods such as the double list randomization in order to boost the power of the list experiment. These variations and validation efforts are needed to advance our understanding of potential for this method to capture IPV. Future research efforts should expand on methodology available to measure IPV in noninvasive ways. For example, there are other techniques being developed for reducing bias, such as using non-verbal response cards to minimize stigma (Lindstrom et al. 2015) or validating sensitive measures using qualitative methods (Blattman et al. 2016). Although much more research is needed to improve the quality and reliability of measures, this effort has shown that it is possible to implement a list randomization for IPV in a large multi-topic survey-and that results can fill evidence gaps in understanding program impacts which would not have been possible without this methodological innovation. --- Supplementary Material Refer to Web version on PubMed Central for supplementary material. Notes: Bivariate t-tests were run to test for differences in covariates at baseline between women asked the violence set (5-item group) and control question set (4-item group);

Social scientists have increasingly invested in understanding how to improve data quality and measurement of sensitive topics in household surveys. We utilize the technique of list randomization to collect measures of physical intimate partner violence in an experimental impact evaluation of the Government of Zambia's Child Grant Program. The Child Grant Program is an unconditional cash transfer, which targeted female caregivers of children under the age of five in rural areas to receive the equivalent of $24 USD as a bi-monthly stipend. The implementation results show that the list randomization methodology functioned as planned, with approximately 15% of the sample identifying 12-month prevalence of physical intimate partner violence. According to this measure, after four years, the program had no measurable effect on partner violence. List randomization is a promising approach to incorporate sensitive measures into multitopic evaluations, however more research is needed to improve upon methodology for application to measurement of violence.

Background In 2015, there were 19 million people living with HIV in eastern and southern Africa, including nearly 1 million people newly infected by the disease [1]. In rural sub-Saharan Africa, people living with HIV frequently seek allopathic clinical services while they concurrently receive treatment from traditional healers [2][3][4][5][6]. In the rural Bushbuckridge sub-district of Mpumalanga Province, South Africa, 69% of HIV infected patients report routinely visiting both types of practitioners to treat their HIV disease, opportunistic infections (Tuberculosis, Kaposi's sarcoma), and side effects from antiretroviral medication (ART) [2]. As knowledge about HIV transmission and allopathic treatment practices becomes more widely understood, healers may be incorporating clinical information about CD4 cell counts, virial load, or the development of drug resistance into their own diagnostic and treatment strategies. Traditional healers in sub-Saharan Africa are thought to be able to diagnose and treat physical and emotional ailments resulting from sources such as social transgressions, spirits, curses, and sorcery, that allopathic health care workers are unable to address [7][8][9][10]. Despite allopathic practitioner's concerns with traditional medicine use, patients still seek, and often prefer, alternative treatments. Healers generally speak the same local languages as patients, spend more time with patients explaining the source of the illness and the necessary treatments, offer greater perceived compassion than clinicians, and are oftentimes more easily accessible [11]. When experiencing illness, it is common for patients to go back and forth between the allopathic health system and the traditional health system [2,[11][12][13]. Medical pluralism is common among patients dealing with a variety of chronic and acute ailments, including HIV, leading to poorer health outcomes among individuals who delay or abandon allopathic treatment for biomedical conditions [10,[14][15][16][17][18]. Although there is no legal framework to control healer treatment of HIV, with those claiming the ability to cure the disease censured by provincial or district health offices and by their own healer organizations. Little work has been done to document HIV-related treatment provided by the traditional health system currently functioning within the rural Bushbuckridge area where the MRC/Wits Agincourt Research Unit has been running a Health and socio-Demographic Surveillance System (HDSS) since 1992. While we know some people employ traditional services for their HIV care and treatment, we know little of healer perspectives about which HIV patients they choose to treat, the type of treatment offered, outcome expected, and the cost of treatment. This study was undertaken to address this paucity. --- Methods --- Study location The study was conducted in the Agincourt subdistrict, Mpumalanga province, northeastern South Africa. The MRC/Wits Agincourt Research Unit oversees the maintenance and operation of the Agincourt HDSS. Roughly 500 km northeast of Johannesburg, the unit has been engaged in population-based health and demographic research since 1992 [19][20][21]. The Agincourt HDSS population is comprised of roughly 115,000 individuals -mainly xiTsonga-speaking-spread throughout 21,000 households in 31 research villages. Roughly one-third of the permanent HDSS population is comprised of former Mozambican refugees, who immigrated to South Africa during the 1980s. The study area contains eight government run clinics, a public-private community health center, and one large public health center. Patients are referred to three district hospitals, each located at 25-55 km from the site and relying primarily on public means of transport. HIV testing and treatment is available at all public health care facilities in the Agincourt sub-district. Therapy initially became available at district hospitals in 2007, with a complete roll-out of ART to satellite clinics by 2010. --- Study population Traditional healers who are registered with Kukula organization [22] were randomly selected using the random selection function in Stata® (StataCorp, LP, College Station, Texas, USA). The Kukula organization consists of more than 300 traditional healers who have organized to champion for healer rights and responsibilities in the Bushbuckridge region. The majority of healers living in the area are members. Of the 169 participants (from a population of 280) selected to participate, four were no longer living in the area and five had died. Of the 160 healers who were located and approached to participate in the study, all accepted (100% acceptance rate). --- Ethics, consent and permissions This study was approved by the Vanderbilt Institutional Review Board (IRB # 140646), the University of Witwatersrand Institutional Review Board (IRB #140547), and the Mpumalanga Department of Health's Research Ethics Committee. All participants provided written informed consent. --- Data collection We conducted 27 in-depth interviews followed by 133 quantitative surveys between July 2014 and August 2015. We used the in-depth interviews to inform and further refine the development of the questionnaire used for the survey. Open, semi-structured interviews were conducted in xiTsonga by a trained qualitative field supervisor (SN; the other co-authors participated in several of the interviews) and were focused on generating data on how healers diagnosed illness, how they learned to treat HIV, how they decided which HIV-infected patients to accept for treatment, and whether they believed they could cure the HIV virus from infected patients. Traditional healers were also asked about the ability of allopathic medicine to treat HIV. Interviews lasted an average of 61 min. Quantitative surveys, borne from the findings of our qualitative interviews, asked healers how they diagnosed HIV, preferred treatment strategies, anticipated outcomes, and costs charged to patients. Data was also collected on the number and type of patient diagnoses in the past seven and 30 days. Given limitations in healer literacy, survey data was collected via face-to-face interviews with healers at their preferred location. --- Data analysis Survey participant characteristics and service uptake were presented as frequencies with percentages or medians with interquartile ranges (IQR). Multivariable log-binomial regression was used to model the odds of treating HIV. The model was adjusted for age and education level. Qualitative data were transcribed within two weeks of the interview into xiTsonga and were subsequently translated into English. Two researchers participated in the thematic analysis, both with extensive experience conducting qualitative research. Agreement in coding was reviewed and discrepancies resolved by consensus. A comparison of coding agreement on the first six interviews found 90% agreement using Cohen's Kappa in MAXQDA® (VERBI GmbH, Berlin, Germany) software. --- Results --- Traditional healer demographics Among healers who participated in the quantitative surveys, healers in the rural Bushbuckridge were mostly female (77%), older (58 years [SD: 14]), had been practicing traditional medicine for an average of21 years (SD: 14)), and had low levels of education (mean3.7 years [SD: 3]). Ninety-two healers (69%) self-identified primarily as an Inyanga (herbalist), with the remaining healers identifying as a Sangoma [divine treatment; 34 (26%)], herbalist [2 (1%)], or "other" [5 (4%)]. Among all healers, the mean number of patients treated in the past month was 6; however, among active healers (those seeing a least one patient per month), an average of 12 patients. The majority of healers did not identify with any particular religion (n = 68, 52%), but the most common religious affiliations were Zion (n = 19, 15%), Apostolic (n = 11, 9%), and Bandla Lama Nazaretha (n = 10, 8%). Immigration from Mozambique (either themselves or their parents) was reported by 94 (71%) of healers; however, only 4 (3%) reported speaking fluent Portuguese. The most commonly reported languages spoken included: xi-Tsonga (n = 130, 98%), isiZulu (n = 40, 30%), English (n = 12, 9%), and Xhosa (n = 8, 6%). --- Diagnosis and treatment When initiating the diagnosis process, healers described the legal rationale of accompanying the patient to the allopathic health care facility for HIV and TB testing. "Yes, it is a law that if a patient comes to us we have to refer them to the hospital to be tested. We are not allowed to treat a patient without being tested..." (female, unknown age). Despite the health regulation requiring them to refer all patients suspected of having HIV to clinical sites for HIV counseling and testing, healers acknowledged that they might unknowingly treat HIV if a patient lies about their status or refuses to test. As such, traditional healers often offer treatment for symptoms suggestive of HIV infection, without knowing whether or not a patient is HIV-positive. One healer noted, "The problem with HIV is that you don't know if you are treating it or not; because when a patient has diarrhea we give her treatment, when she is coughing or she had sores or anything, we just give her treatment to take. Then we end up not knowing if we are treating it [HIV] or not" (female, 54 years). Healers who claimed to cure HIV reported diagnosing the illness through traditional practices (throwing the bones, for example), while others discussed accompanying suspected HIV cases to the clinic for testing. Healers who brought patients to the health facility waited with them, and, if they tested HIV positive, ascertained their CD4 cell count from the patient. Qualitative interviews revealed healers differentiating between two categories of known HIV-infected patients: (1) those with CD4 <unk> 350 cells/mm 3 and (2) those with >350 cells/mm 3. Healers were less likely to treat patients with a low CD4 for fear that they will die in their care: "If the person is not bedridden I do treat her but when she comes to me being critical ill, I will not cure her because her CD4 count will be very low and they [her immune system] are failing to fight the virus" (female, 50 years). Among those surveyed, only 18 (14%) reported successfully curing HIVinfected adult patients with CD4 counts <unk>350 cells/mm 3, while 39 (30%) reported successfully treating HIVinfected adult patients with CD4 cell >350 cells/mm 3. Treatment strategies included providing herbal remedies to be ingested and/or injected. One healer describes the treatment process: "I will give him two liters of the treatment [containing a mixture of six herbs]. He will take it and after two or three days he will start to have diarrhea and when he tells me that he has started to have diarrhea I will give him a due date; and I will ask him if they have the tree that I want him to dig its roots and when he says they have it I will tell him to dig or to eat its leaves... his diarrhea will stop and I will also tell him to stop taking the medication that I have given him. He has to come back and collect another treatment that will cure him. All the powers of things that will create AIDS will be gone after taking the treatment and having diarrhea." (Male, 78 years). Healers who reported treating HIV-infected patients with >350 cells/mm 3 (versus those who did not) treated more patients (8.7 vs. 4.8 per month; p = 0.03), had been practicing for less time (16.9 vs. 22.8 years; p = 0.03), and had lower levels of education (2.8 vs 4.1 years; p = 0.017) (Table 1). Inyangas had higher odds of treating someone with HIV versus Sangomas (OR: 2.49), but this was not statistically significant (p = 0.068). Modified Poisson regression using robust errors revealed Inyangas were more likely to treat (RR: 2.20 [CI: 1.00-4.83] p = 0.049) (Table 2). Being above the median age and increased educational attainment reduced the risk of treating HIV-infected patients but fell short of the 0.05 threshold for significance. Qualitative interviews revealed that leaves and roots from the Ximuwana, Gashu, and Nkompa trees were most commonly used to treat HIV symptoms. Healers charged a median of 92 USD (IQR: 43-123) to treat HIV patients with high CD4. --- Referral to and from allopathic care providers Healers were generally confident in their ability to treat certain conditions, but they also frequently referred patients to the health facility for HIV and TB testing, to treat dehydration/blood loss, and to assist patients they judged themselves as unable to cure. Among the healers in this study, 113 (85%) had referred a patient for dehydration or for a blood transfusion and 47 (36%) referred a patient to a clinic because they could not successfully treat them. While these transfers were common, they also led to patients shuttling back-and-forth between the two systems. One healer explains, "When looking at the patient, I can see that this one doesn't have water in his body because he will be very weak... What I do is to tell him to go to the hospital and get tested and after testing and have the results you will come back to me for treatment" (Male, 67 years). --- Discussion Traditional healers in Bushbuckridge, where ART is available at public hospitals and clinics, continue to treat HIV-infected patients for both HIV and other opportunistic infections resulting from HIV. Inyangas, or herbalists, were at higher risk of believing they could cure an HIV-infected patient than a Sangoma, an unexpected finding given the history of Sangomas treating HIV in South Africa [23]. Our data revealed the increasing sophistication of healer practices: healers selected which HIV-infected patients to treat based on CD4 counts. Healers were avoiding the treatment of any perceived end-stage terminal conditions, as the death of a patient is bad for business. Furthermore, treating HIV-infected individuals seemingly resulted in increased business. Healers claimed that they could cure HIV, leaving patients with no negative health implications after treatment, a potentially more tempting offer to patients than ART for life. Engagement with healers may need to focus on healer's ability to both appreciate and subsequently advertise their own survival rates to gain entry into the traditional system. Traditional healers were universally hesitant to disclose their strategies and sources of medication to treat, highlighted their concern with the theft of their intellectual property [24]. They discussed the use of unspecified herbs (ingested and injected), baths, and ceremony to cure the patient from HIV. The use of razors to inject herbs under the skin of patients is concerning, given the risk of HIV transmission from the patient to the healer [25]. The cost of traditional treatment is high, considering both median income and that primary health care, including antiretroviral therapy, [21] is free to the patient in South Africa. A recent study from South Africa exploring the cost of traditional healer treatment among people with epilepsy found costs to be somewhat lower than HIV treatment, but still substantially higher than allopathic health care costs [26]. While we cannot identify the number or proportion of HIV-infected patients paying these high fees, qualitative interviews and previous research in Bushbuckridge suggest a large number of patients do accept their treatment [2]. Traditional healers do refer patients to the allopathic health care system to ensure that patients suspected of being HIV-infected are tested and, in some cases, treated for HIV. In this study, at least 85% of healers had previously referred patients to allopathic services. This interaction highlights, at least, some current level of engagement between the two systems and while this study suggests that this engagement is possibly a result of fear of legal prosecution, further research on whether voluntary engagement is warranted given the strong relationship between community members and traditional healers, and the large number of registered healers in the Bushbuckridge area. Further engagement could facilitate the incorporation of their work into the allopathic health system. If trained and effectively engaged, traditional healers could help increase early diagnosis and therapy uptake with prompt referrals, adherence support, and avoidance of herb-drug interactions and toxicity, [6,9,10,[27][28][29][30][31][32][33] ultimately improving health outcomes for people with HIV in an region beset with extremely high levels of HIV [19]. Efforts to increase healer knowledge about specific diseases, improve relationships with allopathic practitioners, and reduce delays to allopathic care have been piloted with traditional healers in diverse countries, including Brazil, Uganda, Kenya, Ghana, Cameroon, Lesotho, Gambia, Nepal, and India [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48] with mixed results. Our random sample of traditional healers living in northeastern rural South Africa allowed us to generate data that reflects the source population. However, several factors may limit our results. Recall bias may constrain a healer's ability to remember the number of patients (and their diagnosis) treated in the past 30 days. Social desirability bias may have resulted in fewer healers reporting the treatment of HIV given healer's knowledge that treating HIV through traditional means is viewed negatively in South Africa. The interviewer attempted to reduce this bias by assuring healers that the data would not be reported to local authorities nor was she associated with any of the local health facilities. --- Conclusion Traditional healers in rural South Africa are continuing to treat HIV symptoms in patients, often at prices much higher than the free primary allopathic healthcare available in the area. Further understanding of commonalities and differences between traditional and allopathic health care systems, ways of ensuring risk reduction among the traditional healers, and promotion of honest and targeted dialogue between the two systems may allow for greater coordination between the two systems and an ultimate improvement in HIV patient care in rural South Africa. --- Abbreviations ART: Anti-retroviral therapy; CD4 cells: Cluster of differentiation 4; HDSS: Health and socio-Demographic Surveillance System; HIV: Human immunodeficiency virus; IQR: Interquartile range Authors' contributions CMA guided the conception of the study, analyzed and interpreted the data, and drafted and revised the manuscript. SN revised data collection tools, collected the data, interpreted the data and helped revise the manuscript. RGW refined the study design, interpreted the data, revised and reviewed the manuscript critically for important intellectual content. All authors read and approved the final manuscript. --- Ethics approval and consent to participate This study was approved by the Vanderbilt Institutional Review Board (IRB # 140646), the University of Witwatersrand Institutional Review Board (IRB #140547), and the Mpumalanga Department of Health's Research Ethics Committee. All participants provided written informed consent. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Human immunodeficiency virus (HIV) substantially contributes to the burden of disease and health care provision in sub-Saharan Africa, where traditional healers play a major role in care, due to both their accessibility and acceptability. In rural, northeastern South Africa, people living with HIV often ping-pong between traditional healers and allopathic providers. Methods: We conducted 27 in-depth interviews and 133 surveys with a random sample of traditional healers living in Bushbuckridge, South Africa, where anti-retroviral therapy (ART) is publicly available, to learn: (1) healer perspectives about which HIV patients they choose to treat; (2) the type of treatment offered; (3) outcomes expected, and; (4) the cost of delivering treatment. Results: Healers were mostly female (77%), older (median: 58.0 years; interquartile range [IQR]: 50-67), with low levels of formal education (median: 3.7 years; IQR: 3.2-4.2). Thirty-nine healers (30%) reported being able to cure HIV in an adult patients whose (CD4) count was >350cells/mm 3 . If an HIV-infected patient preferred traditional treatment, healers differentiated two categories of known HIV-infected patients, CD4+ cell counts <350 or ≥350 cells/mm 3 . Patients with low CD4 counts were routinely referred back to the health facility. Healers who reported offering/performing a traditional cure for HIV had practiced for less time (mean = 16.9 vs. 22.8 years; p = 0.03), treated more patients (mean 8.7 vs. 4.8 per month; p = 0.03), and had lower levels of education (mean = 2.8 vs. 4.1 years; p = 0.017) when compared to healers who reported not treating HIVinfected patients. Healers charged a median of 92 USD to treat patients with HIV. Conclusion: Traditional healers referred suspected HIV-infected patients to standard allopathic care, yet continued to treat HIV-infected patients with higher CD4 counts. A greater emphasis on patient education and healer engagement is warranted.

Introduction Rural places are continuingly shaped and re-shaped by the economic activities taking place within them, be they economic, recreational or residential (Antrop, 2005;Claval, 2005). Rural places and local landscapes are highly intertwined as assemblages of heterogeneous elements: farmlands, woods, gardens, grassland, mountains, hiking trails, lakes, roads, villages, animals and people. Relating to community and landscape assets, this paper seeks to offer a deeper understanding of the different practices of different types of rural businesses, the barriers and opportunities that they face and the implications for local rural development. A series of business interviews in the UK and Denmark are used as a means of interrogating an earlier categorisation of rural businesses where location alone was considered an inadequate parameter for defining a "rural business" (Bosworth 2012). Additional characteristics such as their embeddedness within local areas and the diversity of connections with rural landscapes are considered to be important for understanding how rural places are re-shaped by, and influence the performance of, rural business activities. This perspective has recently been reinforced in a policy note from the CLA (2016), a membership organisation for owners of land, property and business in rural England and Wales, which asserts that there is "something intrinsically different" about rural businesses. Similar trends are identified in Denmark, where a large number of municipal strategic plans for rural areas focus on the facilitation of alternative economic initiatives supported by various landscape values such as accessibility to farmland, openness, health issues and remoteness. Thus, our working research question is "how do rural businesses draw on community-based and landscape assets to sustain their everyday practices and performance?" In order to enrich our understanding of how landscapes and rural places are enacted in the interplay with new business practices, we apply tools from Actor-Network Theory (ANT). This approach, first manifested through the book We have never been modern by Latour (1993), (and further developed into many different strands within human and economic geography) allows us to unfold how not only humans but also non-humans (eg. landscapes, villages, policies) can be assigned with agency. This is crucial to describe and discuss the sometimes unexpected effects of rural business developments as well as the material or discursive relations, that make them stable or make them fall apart. Within the existing policy and research paradigm, landscape values are most often perceived as amenities, ecosystem services or public goods, which leaves them outside the focus of the way they a produced or maintained, namely by the economic activities taking place in the countryside. Understanding the nature of a "rural business" is equally contested. Figure 1 illustrates an earlier attempt to stimulate debate in this area (Bosworth, 2012), suggesting that a rural business must satisfy at least two of the criteria illustrated. Additionally, the connection with the local landscape could be integrated into a deeper understanding of what makes a business "rural" in character (Finke, 2014) while other criteria such as the value of rural social relations and networks (Moyes et al., 2015) or the employment of local people might also be relevant. --- INSERT FIGURE 1 NEAR HERE --- Interrogating Rural-ness Although small businesses were found by some to have suffered disproportionately during the recession (Chow andDunkelberg, 2011, cited in Moyes et al., 2015), other research has shown that, despite, or perhaps because of, the predominance of smaller businesses in rural areas, rural businesses were more resilient during the economic crisis (Anderson et al., 2010). The prevailing rhetoric of rural disadvantage in economic and business literature, is characterised by sparse networks, peripherality, remoteness and a general lack of service provision which constitutes a "rural penalty" (Malecki, 2003;Bell, 2010). Instead, we argue that a deeper understanding of the effects of "rural-ness" for small businesses, particular those exhibited through rural landscape values, can inform new research agendas, practices and policy in this field. Definitions of rurality and how to support the rural economy have been challenging themes among scholars from different academic fields as well within European policy and planning for at least the last two decades (Bosworth and Somerville 2013;Cloke, 2006Cloke, & 2015;;Halfacree, 1993). As Woods (2011) notes, the term 'rural' encompasses everything outside the urban sphere and therefore attracts attention across many different research disciplines. Complex rural-urban interrelations increase the challenges of understanding the contemporary rural economy, particularly where rural businesses rely on these interconnections for many new demands for nature experiences, health services, tourism activities and products with rural brands (Ploeg and Marsden, 2008). This is reflected in new business activities emerging from both agricultural and rural business diversification as well as from urbanites attracted or returning to rural environments where they can employ their entrepreneurial capabilities to draw on rural landscape assets in new businesses (Finke, 2014). Such changes in the rural economy raise questions about what distinguishes the ex-urbanite, the diversifying farmer or the indigenous village shopkeeper in terms of their business activities and their rural-ness or urban-ness. Questions also emerge about their engagement and interaction with the local community and countryside as a process in the creation of new possibilities through which the re-valorisation of local landscapes emerge. This calls for a re-interpretation of the rural, where unforeseen phenomena are identified without any a priori assumptions of <unk>authenticity <unk> or aspects of the <unk>rural idyll <unk> as the basic ideas of what constitute the assets of a rural space. The meaning of local landscape heritage is also constantly re-produced (Braaksma et al., 2016) through different appreciations of landscape aspects and new rural practices. Hence, our key contention here is, that questions of the <unk>authenticity <unk> as well as the notion of <unk>identity <unk> are not constructive in capturing the ongoing changes, as those concepts tend to assign the rural places with a certain stable image of rurality excluding the factors that produce the changes. This blurring of what is 'rural' and what is <unk>urban <unk> divides opinions between scholars; some see newcomers' aesthetically-motivated and entrepreneurial actions as "violating the rural" (Vasiliki, 2013) while other, arguably more progressive voices, focus on the rural power related to intangible and symbolic values as part of the interdependency that shapes today's rural-urban reality (Cook et al., 2004). Structural definitions of rurality form the basis for economic and broader rural development policies under the New Rural Paradigm (OECD, 2006) where the countryside has shifted from a space of pure production to a multifunctional space (Mitchell, 2013). Drawing on Schumpeter's notion of creative destruction, Mitchell (2013) sees the new functions of landscape within the multifunctional paradigm as part of a "creative enhancement" that builds upon previous landscape uses rather than over-riding them completely. Compared to the urban lifestyle, landscape values form a more predominant part of the rural environment, but both natural and built environments play a role in shaping local cultural identities (Antrop, 2005;Claval, 2005). Despite the symbiotic association between rural communities and rural economies, cultural factors related to local behaviour and place-based development are often absent in the analysis of economic change and development (Huggins and Thompson, 2015). It could be argued that the increasingly consumption-led drivers of the rural economy (Slee, 2005) include desires to consume features of rural communities and rural landscapes -whether directly through being in these landscapes or indirectly through the consumption of products represented of or bounded to symbolic values of rural places (Bell et al., 2010). Rural landscapes carry romantic or subjective memories (Cloke and Jones, 2001;Antrop, 2005) of the past, yet they are simultaneously spaces of continuous development. Recent contributions to the development of conceptual frameworks for practice driven rural research, suggest the landscape as the point of departure when studying the rural transitions taking place all over Europe (Pinto-Correia & Kristensen, 2013). Thus, the role of the landscape as a local community asset, and the impacts upon the physical landscape relating to rural economic changes demand greater attention, in particular how the landscape asset forms can enhance rural entrepreneurial opportunities. With the composition of rural economies increasingly mirroring those of more urban areas (DEFRA, 2015), traditional sectoral distinctions are less useful. Instead, literature has focused on rural businesses as being smaller (Galloway and Mochrie, 2006) and more sparsely located, thus emphasising problems such as the need for technology to overcome peripherality (Salemink et al., 2015;North and Smallbone, 2006), the thinness of local labour markets (Anderson et al., 2010;North and Smallbone, 2006) and the fragility of rural networks and supply chains (Newbery et al., 2013). In the light of some rural regions being more economically resilient in the past recession (McCann et al., 2014), a more positive categorisation that seeks to identify how rural businesses generate social and cultural as well as economic value for their local communities through the employment of local assets could spark new approaches towards rural development strategies on different levels. Furthermore, as suggested by Korsgaard et al. (2015) there should be a differentiation of what is <unk>rural entrepreneurship <unk> and <unk>entrepreneurship in the rural<unk>, as to support and acknowledge the latter form with the nature of being smaller non-growth-businesses as they engage with local resources in a broad sense in re-enacting local places. --- Methodology For a better understanding of the ways in which rural businesses engage with and interact with landscape assets, six cases of rural entrepreneurs in England and Denmark form the basis of this study. Three business cases demonstrating particular connections with their local landscapes were drawn from a larger study on Funen (Denmark) and these were matched with three further cases in Lincolnshire (UK) due to the prima facie similarity of their business activities. Semi-structured interviews were carried out along with walking tours through the physical environment of the business. The six interviews were transcribed and further sketched into visual forms, identifying the actor-relations involved in each case. This enabled analysis of the networks of relations and the role of community and landscape assets as co-producers within the business practices. Lincolnshire and Funen are both considered rural regions within their respective countries, each having just one major city: Lincoln is 120 kilometres from London while Odense (Funen) is 175 km from Copenhagen. Funen is often referred to as "Denmark's Garden", while Lincolnshire has been described as the "bread basket of England". Hence, the rural representation of the contexts mirrors strongly the production landscapes, and for that reason the cases are chosen in order to identify how landscape values beyond the production aspects associated with agribusinesses co-exist with emerging business types. Some striking similarities between the landscapes of Funen and Lincolnshire, namely the spatial dividing of fields with hedgerows, few woodlands, small villages and highly cultivated hilly landscapes form an immediate impression of being in a "rural" setting. We apply Actor-Network Theory (ANT) as a method to investigate human and non-human relations within the business-pairs. ANT assumes an ontological worldview without any distinction between nature and society; "For each state of Nature, there exists a corresponding state of Society" (Latour, 1993;94-95). Thus, trying to unfold connections between non-humans like landscapes, the notion of rurality, business development and humans (communities, new entrepreneurs) ANT is a powerful tool to grasp this complexity (Allen, 2011). ANT is thus a method that helps us to unravel, unfold and describe the detailed webs of relations through empirical case studies that follow. Inspired by the ANT proposition that non-humans as well as humans have agency, we have applied the perspective that landscapes (and other non-humans) can be actors, whose effects operate in conjunction with actions of other social and economic actors. As Mol (2010;255) explains," An actor acts. It, she, he does something, makes a difference. If the actor were eliminated from its setting, it would take others a lot of work to replace these actions". Following this concept, the role and impact of the rural landscapes are the scope of analysis. Hence, if we accept, that the landscapes can make a change, they are productive and therefore shouldn't be seen as merely externalities. One of the founders of ANT explain this as "Economists invented the notion of externality to denote all the connections, relations and effects which agents do not take into account in their calculations when entering into a market transaction" (Callon, 1999;187). This is useful in order to explore the agency of the landscapes and communities beyond definitions of public goods or 'community capitals <unk> as these can contribute to valuable difference for the business development and rural economy more generally. Essentially, this means, that if neither the landscapes nor the <unk>community capital <unk> is enacted within the business constructs, they do not make a difference. As a method more than a theory, this requires a sensitivity in pursuing the landscape relations (eg. plant materiality, cultivation methods, desires, activities, land use policies) -involved in a business construct, and to outline the differences in the rural perceptions and to pay attention to effects of their ways of engaging with the local community-landscape itself. Hence, in the following analysis we investigate how landscapes are enacted, created and re-reshaped through the entrepreneurial activities. Specifically, we explore how it is possible to unfold the notion of the rural character of a business and consider the effects of actor-relations that might make them 'rural'. --- Cider and wine production: Re-configuration of farm traditions The physical interaction with plants, the growing and cropping of apple trees and vines, the pressing of fruits into beverages and the management of fermentation processes describe the daily activities of the cases of cider and wine production in Lincolnshire and on Funen. Together, these afford a rural character to the businesses. In each case, the entrepreneurs also purchased additional land from other local famers, thus integrating themselves into local land-based community networks. Skidbrooke Cider is managed by one person with a young assistant, producing 1,000 litres of apple juice and 20-30,000 litres of cider per week for wholesale and supplies to local pubs. It is housed in a row of traditional farm barns attached to the family house and, besides the eye-catching stainless steel fermentation tanks, the structure of the building fulfils the idyllic image of rural life with its local brick construction. The start-up in 2004 was inspired by the owner seeing wasted apples from an old orchard and then beginning a local community initiative to gather the apples in return for some of the cider. Explaining his motivation, he said: "You can't really get decent cider in Lincolnshire and there was such a lot of wasted fruit -there were three orchards with more than 100 trees and it was just rotting so we were just taking something that people didn't want and turning it into something that they wanted". The entrepreneurial opportunity arose from a combination of the desire for the use of waste apples and the rural community's will to support the enterprise. Further emphasising the <unk>rural-ness <unk> of the venture, pictures of the annual wassailing festival (a celebratory ritual giving thanks to the apple trees) are displayed in the barn-shop. Of course, as Lincolnshire does not have a heritage of cider production, one could argue that this is a newly invented "tradition" for the local area. Nevertheless, due to the broader cultural heritage and dominant representations of the <unk>rural idyll <unk>, it sits easily in the predominantly agricultural landscape and community. Originating from Somerset, a traditional cider producing area of England, the owner was familiar with the process but still relied on extensive advice networks, including online forums with contacts in the USA and mainland Europe. Combining these external connections with his local embeddedness, he acts as a conduit for the integration of external knowledge into local networks. "....just on the internet we talk to each other and exchange ideas and things. That's people in the US, Europe all over but mainly for me its people in the UK, Somerset, Dorset and Southampton. For example, I'll say so-and-so has happened, what do you think it is? Or "where can you get 500 or so bottles?" it's a good group. You have to be a bit more careful about telling people how much your wholesaling litres for though" Nyholmgaard vineyard was founded by a Danish couple in 2007 whose investment was inspired by the wish for a new lifestyle, spending as much of their daily work life outside as possible and choosing not to have employees, as they had in their former city life. They explained: "We didn't know anything about wine growing. We haven't been to wine tastings and we never travelled around in the world to visit wineries. It was the call for doing something else. We saw a programme on television about a Danish winery that got a silver medal... and we thought, if they can... so can we!" As they began to establish the enterprise, with such limited experience in the physical, rural aspects of the business, they had to seek out information and training -and with wine-growing such a small niche in Denmark, this relied on external contacts. They commented; "We have seen a lot of YouTube movies. When we started growing vines, we participated in a two-day course in a winery association". Despite a self-taught, amateurish approach, the first harvests yielded silver and bronze medals in international food contests in London, followed by a broad attention on their wines from television and gourmet restaurants. This led to an immediate stabilising of the business network, as the medals acted as mediators of quality attracting local chefs and tourists to their products. Inspired by this early success, the entrepreneurs established wine-tastings, vineyard tours and a farm-shop to complement their product offering. --- Reinventing traditional farm landscapes -between global knowledge and local practices The cider and wine producers are changing the local rural landscapes from fields covered by cereal crops into apple-orchards and straight lines of trellised vines. In the case of cider production this is importing an old English rural heritage to a new locality, whereas wine production in the Danish climate is regarded as something quite exotic and challenging. Changing wind directions and temperatures throughout the year are common features of both regions leading to further treeplanting to create windbreaks in the landscape. Within these new sheltered enclosures 'half-standard' apple trees and approved varieties of vines are imported and grown, developed over many years of breeding processes funded by national agricultural ministries to produce varieties adjusted to efficient harvesting and EU production standards. Although cider and wine are labelled as local products, it is apparent that a series of non-local and non-human actors are participating in their production (See Figure 2). Furthermore, the local communities are engaged in the business constructs as consumers of the products and as beneficiaries of the enhanced local reputation and representational attractiveness, for example appearing in tourist brochures. Both cider and wine producers are self-taught, picking up knowledge through the internet, social media, television and their professional networks and due to the motto <unk>learning by doing <unk>. Such an approach to entrepreneurial learning places an emphasis on critical moments and the need for valuable networks to support entrepreneurs' business development (Cope and Watts, 2000). Arguably we see rural skills and former local practices transformed into new versions of rurality as a result of this learning within diverse rural-urban and global-local network interactions, emphasising the complexity of categorising rural-ness. --- INSERT FIGURE 2 NEAR HEAR (the wine bottle-actors) Text to Figure 2: The complexity of the urban and rural, global and local networks related to the production of a bottle of wine, illustrated by the many actors -human and non-human -involved in the business of Nyholmgaard vineayrd. --- Craft gallery and Micro Folk High School: Different perceptions of landscape and cultural values The copywriter/gallery and the Micro Folk High School are located in relatively isolated rural areas surrounded by manorial/estate landscapes featuring open irregular fields following the contours of the land and divided by hedgerows, areas of woodland and impressive solitary trees several hundred years old. The landscapes, nominated as 'areas of outstanding natural beauty' (UK) or regulated by conservations with specific guidelines for the cultivation (Denmark) are part of their business concepts, enacting aesthetic landscape values in their activities. --- The copywriter and gallery This conjoint business is based in a converted barn with a retail arts and crafts gallery on the ground floor and private office space upstairs for the owner's second copywriting business. The copywriting business employs two staff locally as well as other free-lancers in London and Leeds while the gallery employs 3-4 young people to cover weekends, providing good opportunities for work experience in the local community. Copywriting for urban customers is the basic activity creating the economic stability, but the passion for arts and crafts and the landscape surrounding them had led to the idea of running a workshop, presented on the website as: --- "A lifelong passion for arts and crafts, salvage and retail has been realised by the building of [the premises], our rural business idyll in the <unk>Area of Outstanding Natural Beauty <unk>in the Lincolnshire Wolds. A unique blend of working yard, antiques and salvage, offices, art gallery with workshops and boutique accommodation, it <unk>s the sort of place I would travel miles to shop, visit and stay. Let <unk>s hope others do too!" Attached to the gallery is a small cottage with a panoramic view over the open cultivated fields of The Wolds. The first floor guest-suite is designed with an old-fashioned bathtub standing directly in front of a large window as a mediation between the visitor and the landscape. Demonstrating her awareness of the role of economic activity in creating this landscape, the owner explained: --- "It's an agri-landscape, it's not like your looking out onto something that's untouched, it's a working landscape -and this to us is very important." The entrepreneur also encountered difficulties resulting from the different perceptions of local people. For example, she commented that other local rural businesses are less familiar with paying for services like PR and copywriting: "None of my clients were from Lincolnshire, they [the local people] don't want to pay... they don't value branding and marketing I'm afraid." Instead ten jobs have been created in other cities, and social media is her business lifeline. However, as she becomes more integrated into the local area she now employs a local apprentice, recognising that compromises may be needed for her to respond to the different demands and expectations of local businesses. Moreover, she hopes that building new networks locally can also encourage local businesses to become more aspirational. The Micro High School emerged by the encounter between a photographer and an art historian/psychotherapist creating a new school-concept for artists and professionals situated in a clearing in the woods. With many years of professional academic experience they share a common aim of promoting the existential aspects of life and human-nature relations. Through courses in the Micro Folk High School, they offer participants new personal insights improving their abilities to perform in their daily professional lives. The courses contain humanistic leadership, photography and a mix of both activities attracting business leaders, judges, professional and amateur photographers from all over Denmark. An old and carefully renovated forest cottage is rented from a nearby manor-house and constitutes the physical frame around the courses, as well as the surrounding manorial landscape which is directly implemented in the course activities. The couple have extensive experience and professional capacities and networks across Denmark and internationally but they have no aim of expansion: "The micro is essential for the activities in this school. The intimacy is very important, to provide the being on a micro-level, because that <unk>s where you find new possibilities. That reflects, that we don <unk>t want to be <unk>macro <unk>. We want to grow through new insights -not in the common sense of growth. In my point of view, the notion of growth is being abused politically these years... nothing can ever be big enough!" The photographer uses the physical landscapes, both actively during the courses and as an environment for artistic inspiration. During self-development courses the participants are walking around in the woods "reaching a state of concentration that wouldn <unk>t be possible within a traditional business centre". The photographer explained, "The landscape is not necessarily something that is finished out there; it is something one can create. It can be varied infinitely... and this is something I will discuss with the participants." Besides course activities taking place in the small cottage the photographer also arranges photo exhibitions in the nearby town of Kerteminde, featuring contributions from amateur photographers alongside well known artists. Landscape development and changes brought about through human impacts like agriculture or pollution are the central aspects of these exhibitions which facilitate a public awareness of the different attitudes that exist between people with various appreciations of the aesthetical and economic dimensions of landscape uses. --- Enacting rural aesthetics The gallery and the Micro folk high school provide clear examples of different ways of engaging with landscape heritage values, and their neighbours and as they assign their local environment with different forms of appreciations and through that they are co-producers of the local heritage (Braaksma et al., 2016). The creative and philosophical interactions with rural landscapes are forms of relations enacted by business activities such as workshops and course activities that draw upon the landscapes they are located in. Professional academic skills form the basis of the business concepts in a co-production with the landscape context owned, maintained and regulated by national policies, agricultural institutions and landed estates. As the Micro folk high school depends on the surrounding landscape for its inspirational capacity whilst also renting from the owner of the manor-house who sustains the landscape, it can be argued, that the school is intertwined in the business concept of the manor-house and that the landscapes and not merely an externality of agricultural production (Callon, 1999). The gallery owner and copywriter, however, is more focused on the economic potential of the rural landscape. Understanding non-local perceptions of rurality, she recognises a landscape value that is appreciated by outsiders who visit and are drawn to her enthusiastic portrayal of an idyllic rural. These diverse networks allow her to generate capital from the landscape in ways that local people might not be so quick to identify. contrast, trying to sell copywriting and marketing services from her urban world to a rural audience encountered very different barriers where the norms of rural cultural traditions were more difficult to assimilate compared to the hegemonic norms associated with natural or landscape assets. --- Chapel Studios and Film Funen: combining landscape and cultural production The "Film Funen" company and "Chapel Studio" are examples of business activities facilitating artists within film and music production drawing on both landscape assets and rural culture and communities. The entrepreneurs behind the two companies have careers rooted in the metropolises of Copenhagen and London, but each has subsequently moved to a quieter rural lifestyle with their families outside the stressful and noisy city life, described by the studio-owner as "the life you have, when you are young!" Chapel studios has been running for over 20 years in a former chapel on the edge of a very small Lincolnshire village. With no close neighbours to disturb and excellent acoustics it is perfect for music recording. The studios are rented out to musicians fully equipped with recording equipment and a technical assistant for a few days or, as with the band present at the interview, for residential periods of up to five weeks. There is also a house next to the studios where bands can stay as part of the package. The technician explained that, "Bands are more relaxed and more focused... the ability to go out and take in the scenery and things... it's better that just going out from a dingy city studios for a Greggs or a Subway!" [Referring to the ubiquitous national sandwich shop chains] The local pub was also an essential link in the local business networks, especially collaborating with the farmer who cultivates the surrounding landscapes and the walking trails with famous musicians enjoying the opportunity to relax, relatively incognito, over an evening beer while a few of the other local residents were also able to enjoy a few rare moments of star-spotting. Sadly, the closure of the pub has taken away a valuable link here, reiterating the vulnerability of these sparse rural business networks. He was disappointed that pub had closed as this has taken away a key asset for his businesses. He felt that he also contributed to the local community having interesting bands visiting the pub! --- Insert Figure 3 here (photo from Chapel studio) Film Funen is a regional funded company, supported by the Funen municipalities promoting filmmaking outside the dominant cultural centre of Copenhagen. In the last two decades there has been increasing interest among Danish film-writers for using rural landscapes and villages as a contrast to urban environments. One of the pioneers of this cultural decentralisation is a former radio host and skilled journalist, who moved with his family to Funen and manages Film Funen from a rural office, drawing on his networks locally, nationally and internationally. The film productions take advantage of lower cost in the countryside and the cooperative locals. A central philosophy requires workers and contractors associated with the set to have a Funen address in order to support local artisans, producers, hotels and restaurants serving the film set and actors. The manager explained: "The film writers and especially the producers love to shoot their films on Funen. That <unk>s because the whole set moves to Funen during the working process and it enables a certain focus and efficiency for the production. They are fascinated by the landscape, and harsh scenes are often compensated by soft yellow wheat fields and green forests". Film premieres are always staged in the location (e.g. manor, barn, village, street or fisherman's house) where a film has been shot or in a local village cinema, exclusively for the local community who have the chance to meet the famous actors on the red carpet -enacting a proud spirit in the local environment. "In the countryside, people are very interested and proud about the film productions in their environments, and they collaborate nicely -which is so different in the cities, where it <unk>s hard to manage a free setting within the busy streets. We have experienced a whole island funding the last money for a film production -as the really saw the possibilities in their island being promoted through the film." --- Valuing rural remoteness Cultural values are re-invented and re-configured through processes that enrol local actors and actresses in the performing of film productions. Aside from creating jobs and supporting other local businesses, this affects local cultural self-understanding and raises awareness of the cultural values of rural places, their landscapes and buildings. In the case of music and film productions these activities are made possible by entrepreneurial personalities with wide networks rooted in the cultural urban elite. Where local communities are engaged in co-production or informally linked to these new networks, the role of businesses in providing social fulfilment alongside economic opportunities is further revealed. Located in or drawing on rural assets the new businesses work on the basis of commodifying (Perkins 2006) aspects of remoteness, the open fields and the tranquility as profoundly contrasting features to those of the major urban markets and urban business models. This is an example of how entrepreneurs are becoming aware of how it is possible to activate the landscapes they operate within by valuing these features of the remote rural areas. In other words, as Allen (2011) describes, the landscapes as make a difference, because they are involved in the actor-network of the business concepts. --- Discussion -Supporting rural businesses as agents of change and continuity Our six cases of rural entrepreneurship are summarised in Table 1 (see below) with the implications for understanding their relationship with the rural landscapes and communities. What the cases demonstrate are, that the businesses adapt and enact landscapes in various ways, and that these landscape features are transformed into new ideas and appreciations of their rural context. Along with the business development, controversies also occur when non-traditional activities, like wine production or copywriting, are challenged by divergent views on what a rural business is or should be. Furthermore, the new businesses receive only a fraction of the over-arching rural policy support which continues to be dominated by productivity-driven conventional farming practices that are detached from the innovative activities of new rural entrepreneurs. As well as revealing the hegemony of agribusiness regimes, both in rural Lincolnshire and Funen, this highlights the complex real and perceived divides between production and consumption, rural and urban, global and local, as well as innovation and tradition. Together, these distinctions challenge the structural definitions of rural and urban, as well as the perspective of what rural growth should constitute. --- (Insert Table 1 near here) Understanding the roles that local community and landscape assets play in rural business development in order to better characterize what makes what makes these businesses <unk>rural <unk> were the key objectives for this article. As seen in Table 1 our cases have identified how different forms of rural assets are enacted within the entrepreneurial initiatives to generate value and to sustain their businesses. Although each of the businesses is located in a rural area, none of them focus principally on the local rural market for their trade and therefore, it is the third dimension -"sell a rural product" from the Venn diagram presented in the introduction that appears most critical in determining the rural character of these businesses. From our cases, while some of the products (e.g. local crafts, cider and wine) might be deemed typically rural, a film producer of recording studio might not. This is, however, an insufficient way of understanding the rural business activities as the outcome of artistic practices are equally enacting rural features -both symbolical and material -thus revealing the agency or "the power of rural" (Bell et al., 2010) of the landscapes in a new perspective. We argue that the mutual re-configuration between the urban and rural implies an inevitable cultural and economical exchange, but that the enactment of landscape materialities make a difference between urban and rural. Scattered settlements, walking alone, waste apples, the changing images of farmland, quietness, bird life, plants, physical work with trellised vines are all crucial element in facilitating a diverse and inspiring world for rural entrepreneurs with vibrant rural places. Rural landscapes are places of inspiration, and as stated by Bell et al. (2010) "the rural also pleases us, soothing our worries through book and film and song, and rewarding our ambitions through walks, weeding and woodcutting", thus offering more than simply production or consumption spaces. Remoteness of rural life itself is found to be a quality and a positive driver for new business practices and not perceived as places to mourn about. Furthermore, the often highlighted distance parameter is turned to be a positive feature of the rural as tranquility and less populated areas are essential assets of the business concepts shown in this study. Technological devices bridge the spatial distances for knowledge exchange and customer relations and translate information into innovations of practices at a local scale, as seen in the case of invention of hot-steaming in the Danish vineyard, to overcome the cold climate in the springtime (See Table 1). Therefore, to return to the earlier typology, the conceptualisation of a rural business must recognise a wider range of rural assets that contribute to the reshaping of the concept of a rural business. Rather than the vague "selling a rural product", we therefore propose that this should be widened to be "drawing value from rural assets" which can include local employment, social and cultural resources and aesthetics all captured within the concept of rural landscapes. This might also encompass the rural setting as the profound motivation for the entrepreneur her/himself. The title of our paper indicates that new business activities perform roles that bring about

The purpose of this paper is to use a series of business interviews in the UK and Denmark as a means of interrogating the categorisation of rural businesses presented in an earlier publication which sought to understand the rural character of different businesses (Bosworth 2012). Rather than defining a rural business purely on its geographical location, the engagement of a business within a rural economy and its relations with both rural and urban environments are more complex issues. A wider conceptualisation how rural social relations, cultural influences and landscape values, forms the basis for analysing how these assets can create value for rural businesses. Understanding these issues can better inform organisations that are seeking to support the rural economy and rural communities. It can also guide business owners themselves as to how they might benefit from being a part of, or associated with, the rural economy in capturing the possibilities linked to a growing urban population. The findings reveal how landscape assets are intertwined in the business concepts as positive drivers despite many voices within rural and economic research pointing at disadvantages relating to distance, sparseness of infrastructure or service provision. The research analysis draws on Actor-network theory and this tool reveals promising prospects for future placebased studies in how landscape assets can be enacted in the development of the rural economy.

found to be a quality and a positive driver for new business practices and not perceived as places to mourn about. Furthermore, the often highlighted distance parameter is turned to be a positive feature of the rural as tranquility and less populated areas are essential assets of the business concepts shown in this study. Technological devices bridge the spatial distances for knowledge exchange and customer relations and translate information into innovations of practices at a local scale, as seen in the case of invention of hot-steaming in the Danish vineyard, to overcome the cold climate in the springtime (See Table 1). Therefore, to return to the earlier typology, the conceptualisation of a rural business must recognise a wider range of rural assets that contribute to the reshaping of the concept of a rural business. Rather than the vague "selling a rural product", we therefore propose that this should be widened to be "drawing value from rural assets" which can include local employment, social and cultural resources and aesthetics all captured within the concept of rural landscapes. This might also encompass the rural setting as the profound motivation for the entrepreneur her/himself. The title of our paper indicates that new business activities perform roles that bring about change in rural areas, but we also recognise that through the provision of employment, the enhancement and maintenance of landscapes and build environments, the delivery of services and the upkeep of meeting places they play key roles in the continuity of rural communities too. The maintenance of open landscapes and continuance of rural places can also become a value in itself for many rural businesses, such as the "timeless" agrarian views and peaceful settings of Lincolnshire and the inspirational manor landscapes preserved as national cultural values of Funen. Of course, these are not really stable landscapes but are being continually renegotiated and reshaped by businesses and other actors (land use policies, farming practices, nature conservation) participating in rural activities It is here that the rural entrepreneur treads a fine line in presenting a positive image of the rural while simultaneously enacting development within that rural setting. Therefore, there is a need to understand the processes that drive changes in rural landscapes (Antrop, 2005;Pinto and Kristensen, 2013;Mitchell, 2013;Korsgaard et al., 2015) and the inner logic of entrepreneurial initiatives lead us to identify new modes of ordering the rural economy. In other words, we argue, that landscapes are more than merely amenities or externalities deriving from production systems (Callon, 1999). Neither are they solely framed within eco-system services without any capability to exhibit agency in a rural economy. As seen in all our cases, rural assets need to be activated in order to create values, and here the ANT tool is especially constructive in detecting the unexpected pathways of creativity enrolling the rural assets when studying these changes (Mol, 2010;Cook, 2004). Entrepreneurs themselves also need to recognise the potential landscape values they can enact. Approaching the rural character by applying ANT as method to open up the <unk>black-boxes" of the business practices has provided a viable tool to understand the interactions between heterogeneous components of both the rural and urban world as well as how global knowledge exchange can create local opportunities. Further studies to sustain these new types of <unk>landscape-businesses <unk> are needed to explore how rural landscape assets can activate the rural economy and develop policies to sustain them. Especially there is a need to study the business activities diversifying from conventional agricultural production as they manage larger land areas and thus are changing the rural images and practices in more dramatically way than seen in the six cases here. --- Conclusions The cases in this study show potential for both entrepreneurs and rural communities to reveal new rural economic agendas. The open landscape becomes a central actor and mediator for emerging businesses as it embodies the very difference to the urban locations. This creates a new interplay between agricultural and non-agricultural values attached to rural landscapes and these landscapes can influence new business innovations related to the multifunctionality of rural resources (Lefebvre et al., 2015;Korsgaard et al., 2015;Pinto-Correia and Kristensen, 2013). The six business models analysed in this paper are expressly different to the anything that the highpressured, high price, urban setting can offer and therefore rural businesses can avoid competing on the unfavourable, urban-economy terms but instead provide unique services, tailored to their rural context. Through the use of technology, they are able to achieve the best of both worlds -maintaining daily communication across diverse networks of customers and suppliers whilst recognising and communicating the values that await to be activated within their rural communities and landscapes. Translating this into relative stable rural business models requires rural entrepreneurs to become better at recognising these forms of value and incorporating them into the development of highervalue products and services for people living and enjoying recreation in rural areas as well as through wider-reaching supply chains. As rural areas continue to attract more counterurbanisers with extensive urban networks and experiences, and as urban-rural interdependencies grow, the opportunities to draw on these values can be expected to continue to increase. These processes are changing rural areas in many ways, socially, culturally, physically and economically, but at the same time they are building on the patterns of perceived continuity that underpin the attractiveness of rural places; namely that of a sparsely populated and relatively "natural" environment. Before turning these findings into policy recommendations, however, we must also remember that a number of rural entrepreneurs are pursuing lifestyle goals too. This may be the maximum carrying capacity of some local rural economies and as such, we must acknowledge that the non-economic values that are generated might be just as important as employment or growth indicators. In this light, we advocate policies -and research -that seek to promote the <unk>rural power <unk> (Cloke, 2015;Bell et al., 2010) in a wider sense, not narrow employment and growth-based objectives. This will look very different across different rural areas and among different types of rural businesses which makes any prescriptive policy approach almost impossible. Few rural policy strategies could have incorporated vineyards in Denmark or recording studios in remotest Lincolnshire so we need to identify the forms of value that are active agents of change and continuity through diverse business activities and formulate polices to enrich these. As demands for tranquillity and landscape values are increasingly demanded in an urbanised world, new economic perspectives on externalities is a first step towards renewed acknowledgement of a qualitative rural economy. In conclusion, improved digital infrastructure and rural business support policies combined with new ways of thinking about rural planning that enable rural entrepreneurs to take advantage of new opportunities could also yield positive outcomes. Where "rural businesses", as defined herein, are drawing value from different forms of rural assets, it will be in their interests to develop sympathetically to those values that they, and their customers, cherish. Furthermore, without development permissions, a number of traditional rural businesses might become so constrained that they are forced to leave, thereby endangering an integral component of the perceived continuity of rural life. In this respect, the management of the landscape values at both a national and community level seems to be one key element in the rejuvenation of rural places by further stimulation of a multifunctional rural economy that can meet growing urban demands for rural experiences and products alongside continuing local needs.

The purpose of this paper is to use a series of business interviews in the UK and Denmark as a means of interrogating the categorisation of rural businesses presented in an earlier publication which sought to understand the rural character of different businesses (Bosworth 2012). Rather than defining a rural business purely on its geographical location, the engagement of a business within a rural economy and its relations with both rural and urban environments are more complex issues. A wider conceptualisation how rural social relations, cultural influences and landscape values, forms the basis for analysing how these assets can create value for rural businesses. Understanding these issues can better inform organisations that are seeking to support the rural economy and rural communities. It can also guide business owners themselves as to how they might benefit from being a part of, or associated with, the rural economy in capturing the possibilities linked to a growing urban population. The findings reveal how landscape assets are intertwined in the business concepts as positive drivers despite many voices within rural and economic research pointing at disadvantages relating to distance, sparseness of infrastructure or service provision. The research analysis draws on Actor-network theory and this tool reveals promising prospects for future placebased studies in how landscape assets can be enacted in the development of the rural economy.

Introduction The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the causative agent of the transmissible coronavirus disease (COVID- 19), which was declared a pandemic by the World Health Organization (WHO) on March 11, 2020 [1]. As of September 28, 2020, 7,059,087 cases were reported in the United States of America (USA) [2]. The pandemic has resulted in the introduction of drastic changes and restrictions to prevent further spread of the virus such as social distancing, quarantine when necessary, mask-wearing, and deliberate hand hygiene [3]. Implementing these individual preventive behaviors requires a clear understanding of the actual risk of COVID-19 and the will to put in the effort and time to adapt to these changes [4]. This subject is of great importance for the Latino population, which is at increased risk of acquiring SARS-CoV-19 and having unfavorable outcomes [5]. Available data from the Centers for Disease Control and Prevention (CDC) has demonstrated that the highest rates of hospitalization and death from COVID-19 in the USA are among racial and ethnic minority groups when compared to non-Hispanic white groups [5]. This high risk is increased explicitly among non-Hispanic Black people, Hispanics, and Latinos, and American Indians/Alaska Natives [5]. In a CDC analysis of the data from 5,187,853 cases, race/ethnicity was available for 2,669,792 (51%) cases. From those, 29.2% (779,596) were Hispanic/Latinos, whereas 18.4% (490,378) were non-Hispanic/Black. This was further supported by a recent study in the Baltimore-Washington DC area that showed that Latinos had a significantly higher prevalence of COVID-19 compared with other ethnic groups [6]. The difference in the prevalence of COVID-19 among Latinos and other ethnic groups can be attributed to many given factors such as low socioeconomic status, overcrowded living conditions, inequalities, and limited healthcare access [7][8][9]. The vast majority of Latino people work on the front-line job sectors such grocery stores, waste management, cleaning/sanitation services, and food delivery, making them constantly exposed to people or surfaces potentially infected with the virus [7]. In addition, Latino communities are more prone to be living in a multi-generational household with overcrowded conditions, making it challenging to take safety measures to safeguard senior family members or isolate those who are sick if space in the house is inadequate. Studies have shown that an adequate perception of COVID-19 risk is associated with better engagement in recommended preventive behavior such as social distancing [3,10]. On the contrary, populations with inadequate COVID-19 risk perception are less likely to commit to precautionary actions to avoid COVID-19 and are associated with higher infection rates, which have deleterious effects on the individual's health, community health, and global value by worsening the pandemic [3,11]. This is of great importance, especially in vulnerable populations and high-risk ethnic groups such as the Latinos, where a higher prevalence of the infection will be associated with higher complications and mortality. To date, there is limited knowledge of the level of COVID-19 risk perception faced by Latinos in the USA and their level of engagement in preventive measurements recommended by the CDC to avoid SARS-CoV-2 infection. This could be a significant factor influencing the high prevalence of SARS-CoV-2 infection is this population. Therefore, our study aims to assess the prevalence of COVID-19 and evaluate COVID-19 risk perception and preventive behavior implementation in an asymptomatic Latino population in the USA. This will provide insight into the gaps in knowledge faced by Latinos and efforts needed to be taken to provide adequate education and social support to decrease the virus spread, morbidity, and mortality. --- Materials And Methods --- Population The study population consisted of 410 Latino individuals recruited within a religious community in Maryland, USA, between July 25 and August 25, 2020. Patients aged 18 years or older with no known current or previous history of COVID-19 infection were included in the study. This study was approved by the Institutional Review Board, and informed consent was obtained. --- Questionnaire Participants who were found to be eligible to participate in the study sat for a semi-structured questionnaire/interview. Participants were briefed about the study's objective, risks, benefits, and voluntariness, after which written informed consent was signed. Each participant's interview took approximately 20 minutes and was conducted in Spanish or English, according to the participant's preference. The questionnaire was implemented to assess four main components: patients' demographics and socioeconomic status, COVID-19 risk perception, precautionary behavior, and presence of COVID-19 symptoms. Patients' demographics and socioeconomic status questions assessed variables such as age, sex, chronic diseases, marital status, educational level, household size, and annual income (quantified as lower or higher than the 125% federal poverty guidelines -the measure of income used by the U.S. government to determine who is eligible for subsidies, programs, and benefits) [12]. The COVID-19 risk perception component measuring the subjective understanding of COVID-19 threat perceived by each individual was assessed using a scale of 1 to 3, where 1 indicated low risk, 2 indicated moderate risk, and 3 indicated high risk. The precautionary behavior evaluating patients' behavioral aspects during the pandemic was measured using a scale of 1 to 3, with 1 indicating never, 2 indicating sometimes, and 3 indicating always. --- Physical examination and COVID-19 testing We conducted a focused history and physical examination to evaluate the asymptomatic status of patients. Patients' blood pressure, heart rate respiratory rate, pulse oximetry (SpO 2 ), and the presence or absence of pulmonary auscultatory findings were evaluated. Additionally, patients were referred for COVID-19 polymerase chain reaction (PCR) testing in a nearby testing center. --- Follow-up After receiving COVID-19 test results, positive patients were contacted to be informed of their results and oriented to healthcare's next steps. Patients were referred to their primary care physician or the local health department for follow-up. Subsequently, we assessed for the development of related symptoms. --- Statistical analysis Analyses were conducted using SPSS Version 24 for Windows (IBM Corp., Armonk, NY, USA). All data measurements were compared employing an independent t-test for numerical variables and Pearson's correlation for categorical variables. Categorical variables are presented as frequencies with percentages, and continuous variables are presented as means <unk> standard deviations. Results were considered statistically significant if p-value was <unk>0.05. --- Results Of the 410 participants included in our study, 47.6% of the participants were aged 35-44 years followed by 45-54 years (19.3%) and <unk>55 years (12.9%). Females accounted for 65.9% of the population, and 32.9% of total participants had a history of chronic disease, with hypertension (78.5%) and diabetes mellitus ([DM] 12.6%) being the most common. Almost 90% (89.3%) of participants were under 125% for poverty category. Also, 48.1% lived in a crowded house of six members. Of the participants, 93.2% stated to lose their job due to the COVID-19 pandemic. In addition, 74.1% had a close family member or friend infected with the virus (Table 1) When comparing COVID-19 positive and negative patients, we noticed that positive patients were significantly younger than negative patients (p<unk>0.001). Interestingly, COVID-19 positive patients were more likely to be asthmatic (31.8% vs. 0% for negative), diabetic (13.6 vs. 12.4% for negative), and both diabetic and hypertensive (18.2% vs. 0.9% for negative) than negative patients (p<unk>0.001). COVID-19 positive participants (69.7%) lived in overcrowded houses with six members significantly more than negative participants (43.1%) (p<unk>0.001). Unpredictably, COVID-19 negative patients had a lower education level compared with positive patients, where 81.3% had no education or only accomplished primary school studies in negative patients versus 29% in positive patients (p<unk>0.001). Total (n=410) COVID- Table 2 represents the descriptive statistics of the risk perception of COVID-19 infection. Almost all participants (97.3%) believe that their risk of getting infected by COVID-19 is low, and 80% stated not to be worried at all. However, when asked about this risk to their neighbors, to the state, and the USA, 61% and 81% replied "moderate risk" and 75.6% replied "high risk", respectively. When comparing the different levels of COVID-19 risk perceptions of COVID-19 positive and negative patients, we noticed that negative patients (99.1%) thought their risk of having COVID-19 to be low significantly more than positive patients (89.5%) (p<unk>0.001). Nevertheless, they had a better perception of the risk of COVID-19 at the level of their neighborhood and the USA compared with positive patients (p<unk>0.05). --- Discussion COVID-19 is extensively being studied among minor ethnic communities in the USA and worldwide [13][14][15]. To our knowledge, this is the first study to determine the prevalence, level of risk perception, and preventative measures practice of COVID-19 in asymptomatic Latinos in Maryland and in the USA. Our study of 410 Latino adults compared the different baseline characteristics, level of risk perception, and preventive behavior between COVID-19 positive and negative patients. We established an inadequate level of risk perception in this minority group, which is significant in individuals testing positive for COVID-19 (p<unk>0.001). Our data also demonstrated that a significant number of the COVID-19 positive participants had adopted only some preventive measures such as facemask wearing, with mostly abandoning other protective behaviors (i.e., hand washing and social distancing). These findings support that Latino minority groups are being disproportionately affected by COVID-19 in the setting of pre-existing unfavorable social determinants of health, such as poverty, low education, and healthcare access. Around 80% of our study participants were 35 years and older and considerably healthy, with only about a third reporting history of chronic disease (<unk>80% hypertension), the majority of which were females (<unk>66%). Our data highlight a population of low socioeconomic and educational statuses, with 90% living in poverty, with only <unk>6% who having made it to college. Almost half of the participants of our participating community were living in densely populated households of six family members and more. When asked about their likelihood of acquiring an infection with COVID-19, nearly all individuals (97.3%) stated that they personally have a low chance of getting infected. However, as we slowly stretched the diameter to surrounding regions (i.e., neighborhood, state, and country) and asked about their impression of the risk on other individuals (i.e., neighbors, work partners, friends, and family), the majority of participants responded that others were at a moderate-to-high risk of getting infected with COVID-19. Also, 74% said they are very likely to come across someone infected. These findings are interesting and may translate to our population having a good understanding of the impact of the COVID-19 threat on the country and the general USA population; however, they somehow seemed to believe that they were more immune or at lower risk than other people. This could be due to optimism bias in which individuals tend to underestimate their risk of developing a disease and complications compared with other similar individuals [16]. A recent study performed in three different countries (the USA, United Kingdom, and Germany) suggested that optimism bias could affect the way people respond to the pandemic and the frequency of assumed preventive behaviors against COVID-19 [17]. In addition, our findings are consistent with a study conducted in the USA that compared COVID-19 awareness and concerns between different genders, races, and ethnicities. The study showed that women (27.9%), Black (36.15%), and Hispanic persons (32.8%), especially the ones living below the poverty level, were significantly more likely to state that they "were not at all likely" to get COVID-19 compared with males (19.7%), whites (17.5%), other races (29.45%), and non-Hispanics (22.3%) [18]. Another study contradicted our findings, where it showed high awareness of the possibility of being asymptomatically infected by SARS-Cov-2 in Hispanic and non-Hispanic Black individuals [19]. This difference could be explained by the fact that almost all of our population was under the same discrepancy with low socioeconomic and educational statuses. This low socioeconomic status is prevalent in the Latino population in the state of Maryland, and it is also very common that people with low socioeconomic status attend the religious groups looking for help and support. Gaps in the knowledge and awareness of COVID-19 in socioeconomically disadvantaged groups have been shown to affect the way individuals perceive and respond to the pandemic [18]. In the study by Jones et al., almost 85% of the population had at least some college education and an annual income of $30,000 or more [19]. Another possible explanation for this difference could be the fact that our population comes from a religious community that could result in giving a sense of protection toward COVID-19. In a late study, Qaiser et al. described how it may be challenging for healthcare professionals to achieve successful management of COVID-19 among different religious groups that would potentially replace preventative measures with their faithful beliefs, i.e., being protected by God and prayers would be enough for preventing and treating the infection [20]. In another study, spiritual leaders have been called to participate in the fight against COVID-19 to prioritize pandemic guidelines and assist in overcoming the stigma [21]. Participants were tested for COVID-19 infection, and, interestingly, 18.5% tested positive, 55.3% of which were 35 years and older, and almost 30% had a history of chronic disease (36.4% had hypertension and 31.8% had asthma). Minority ethnic groups are at higher risk of disease (particularly cardiovascular disease), and recent studies have shown that severe cases of COVID-19 infection in the elderly have been associated with multiple different underlying chronic health conditions including hypertension [22,23]. In addition to that, almost 70% of our positive COVID-19 cases were living in a crowded household. This may explain the high rates of COVID-19 positive patients in the group of participants living in overcrowded houses since it is challenging to maintain proper social distancing, face mask-wearing, and handwashing practices among family members [24]. These findings are alarming since these conditions make the situation difficult for COVID-19 positive participants to self-isolate and hence may explain our results where 94.7% of COVID-19 patients said that they will never be able to self-isolate if recommended, resulting in an additional risk on healthy household members. There was a noticeable low perception of the risk of COVID-19 and inconsistent compliance with the government-recommended preventive measures in our study. This was noticed explicitly in COVID-19 positive patients. Although there was a high prevalence of participants who were consistent in wearing face masks, they stated they were less likely to wash/sanitize their hands and avoid contact/crowded places and comply with social distancing [25]. An interesting point here is that a significant number of COVID-19 positive participants were ones that reported always wearing their face masks (<unk>50%). Based on these findings, our data may emphasize the importance of fully (vs. partially) adhering to the WHO-and CDCrecommended COVID-19 mitigation guidelines, including social distancing and proper hygiene [26]. Historically, a strong relationship has been witnessed between risk perception and precautionary behavior, which also applies to the COVID-19 pandemic, as shown in recent studies [17]. Individuals with better knowledge, awareness, and guidance on the risks and severity of consequences associated with the novel virus are more likely to adopt proper preventive measures [14]. Another interesting aspect of our study was the significantly higher incidence of positive cases in the female population when compared with males. It is common to have a higher female population among religious communities; however, it is uncommon to have higher positive cases when compared with the male population. This is a contrast with the findings in a recent study, where the COVID-19 cases were equally distributed among females and males, but the severe disease was most frequent among males. Additionally, when analyzing the population that required admission to ICU, all of them were males. It is demonstrated in the literature that males are more at risk of adverse outcomes regardless of age or comorbidities [27]. On follow-up of the 76 positive COVID-19 cases, we found that <unk>85%, <unk>75%, 14%, and <unk>5% were symptomatic, hospitalized, ICU admitted, and died. Is important to note that all the hospitalized and death patients were older and presented comorbidities such DM and asthma. These outcomes further warrant increased attention to the underprivileged Latino communities in the USA during the COVID-19 pandemic era. In general, nationwide Latino populations are at high risk for morbidities and mortalities due to the socioeconomic discrepancies and racial inequities that result in their lack of appropriate education, employment, housing, and inadequate access to healthcare and medical insurance [28]. Besides the evident socioeconomic impacts of COVID-19 and increased number of jobs lost, the persisting structural health disparities have shown increased numbers in COVID-19 diagnoses and deaths among the Latino populations and other ethnic minority groups across the USA (e.g., Black communities) [13]. This has been exacerbated by the multifactorial vulnerability of the underserved Latinos, putting them at a particular disadvantage during the COVID-19 pandemic. Our study has some limitations. First, our data may contain selection bias since it focuses on a proportion of Latinos living significantly in poverty, which does not reflect the actual socioeconomic status of all Latinos across the USA. Another limiting factor is that all the collected data for our study were self-reported, which could have led to recall bias. Third, our sample was taken from a small-scale region and is based on observational and descriptive analyses. It would be useful to carry out a wide-ranging, heterogeneous data collection with more extensive analytical methods in the future. --- Conclusions Our study identified concerning inadequate COVID-19 threat perception and lack of engagement in acceptable preventive behavior among a group of Latinos living in the USA. This provides a snapshot of the actual miscommunication and disadvantage present in this vulnerable group, which is under low socioeconomic status with limited educational levels. Therefore, we believe that Latino communities across the USA are at especially high risk of acquiring, spreading, and dying of COVID-19. We call for governments to come together with social service organizations and local media outlets to directly deliver COVID-19 information to their local populations in several non-English languages to overcome linguistic barriers. We argue that focused efforts should be put toward increased awareness, trusted educational resources, guideline implementation, financial and social support, and collaboration with faith-based organizations, and greater efforts should be taken to prioritize COVID-19 testing in at-risk populations. In our study, patients were asymptomatic, they did not have a clear knowledge about accessing healthcare in the case of COVID-19, and most of them were uninsured and were hesitant to access to healthcare due to lack of migratory status. This could contribute to the fact that most of them would be going late to receive medical care, resulting in higher mortality among the population. Due to all these factors, further research on COVID-19 among minority ethnic groups is needed. --- Additional Information Disclosures Human subjects: Consent was obtained by all participants in this study. Solutions IRB issued approval Protocol #2020/07/18. The protocol COVID-19 Prevalence, Perceptions and Preventive Behavior in a 410 Asymptomatic Latino Population attending El Calvario Church, Silver Spring, Maryland, United States: Asymptomatic Latino Population: A Cross-Sectional Study. has been approved by the IRB Chair under the rules for expedited review. The expedited review categories applicable to the study approval are: (6) Collection of data from voice, video, digital, or image recordings made for research purposes. This project should be conducted in full accordance with all applicable laws and regulations, and must adhere to the project plan that has been approved by Solutions IRB. You must notify Solutions IRB immediately of any proposed changes that may affect the status of your approval or the risk to participants. You must also report any unanticipated problems involving risk to participants or others to Solutions IRB immediately.. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.

To determine the prevalence, level of coronavirus disease 2019 (COVID-19) risk perception attitude and preventive behavior implemented by the Latino population in the United States of America (USA).This cross-sectional study was conducted between July 25 and August 25, 2020, and included asymptomatic Latino participants (n=410) with no current/previous COVID-19 within a religious community in Maryland. Participants answered a questionnaire consisting of three components: patient demographics/socioeconomic status, COVID-19 risk perception, and precautionary behavior. Additionally, a focused history taking and physical examination were performed, and nasal swabs for COVID-19 testing were obtained.Around 80% of our study population was 35 years and older, considerably healthy, with only about a third reporting history of chronic disease (~80%); most were females (~66%). Of our participants, 90% lived under poverty; only ~6% had made it to college. When asked about the likelihood of acquiring COVID-19, 97.3% stated they have a low risk of getting infected. However, as we asked about the risk of individuals living in their neighborhood, state, and country, the rates changed to moderate to high (78.4%, 86.3%, and 86.6%, respectively). When asked about preventive behavior, 71.2% stated they never wear masks and 85.4% mentioned they never keep social distance. Additionally, 76 (18.5%) tested positive for COVID-19, whereas 64 (84.2%) developed symptoms at follow-up, 57 (75%) were hospitalized, and 4 (5.2%) died.Our study identified inadequate COVID-19 threat perception and lack of engagement in preventive behavior among a group of Latinos living in the USA. We believe that Latino communities across the USA are at markedly high risk of acquiring, spreading, and dying of COVID-19.

Introduction The human immunodeficiency virus (HIV) affects approximately 1.2 million persons in the US [1] and an estimated 35 million worldwide [2]. The introduction of combination antiretroviral therapy (cART) in 1996 has resulted in dramatic improvements in the quality of life, CD4 cell counts, and survival rates of those with HIV [3]. However, despite these dramatic improvements in major health and wellbeing outcomes, HIV-associated neurocognitive disorders remain a substantial problem, and some reports suggest rates of cognitive impairment have increased among those who are not immunosuppressed [4,5]. The success of cART therapies has resulted in an increasing number of persons with HIV living into old age. In 2013, over a quarter of all persons with HIV were over the age of 50, and the Centers for Disease Control estimate that by 2015, more than half of all persons with HIV will be over 50 years old [1]. Adults living with HIV into old age are at greater risk of cognitive impairment [6] ; however, the factors associated with this are not well understood. One potentially important factor is loneliness. Feelings of loneliness generally increase with age among older adults, and this has been associated with a number of negative health and psychological outcomes [7]. We previously demonstrated that loneliness is associated with cognitive decline and the development of Alzheimer dementia [8]. Loneliness may be a particular problem in the context of aging with HIV, given the well-documented social stigma and prejudice that HIV infection confers [9], and this seems particularly deleterious among older persons [10,11]. The net effect of increased feelings of loneliness due to aging and long-term HIV infection may place neurocognitive abilities at heightened risk of decline or impairment in old age. Black Americans have the most severe burden of HIV of all racial/ethnic groups in the US, specifically accounting for higher proportions of those with new HIV infections, of those living with HIV, and of those ever diagnosed with AIDS [12]. Black adults with HIV report a particularly high level of social stigma [13], which has been associated with social isolation [14], and in general, older Black adults have fewer social resources, namely smaller social networks and less social engagement [15], than older White adults. For these reasons, older Black adults might experience loneliness differentially and more drastically than older White adults. A theoretical model of racial/ethnic HIV health disparities has been proposed [16], suggesting that minorities with HIV often possess multiple intersecting stigmas at the structural level (e.g., segregation) as well as the individual level (e.g., perceived discrimination). Older Black adults with HIV not only have to contend with the negative historical consequences of institutionalized racism, but also with the social stigma associated with HIV infection, and these may interact to produce downstream effects upon health outcomes such as cognition. Since loneliness has been associated with poorer cognitive functioning in old age [8], this in turn may have an effect of exacerbating any negative effects of the HIV virus on cognitive functioning in old age among Black adults. We tested the hypotheses that (1) older Black adults with HIV would experience greater loneliness than older White adults with HIV, and (2) the association of greater loneliness with lower cognitive function would be stronger in Black adults with HIV than in White adults with HIV. Participants came from the Research Core of the Rush Center of Excellence on Disparities in HIV and Aging (CEDHA), a community-based cohort study of older Black and White adults aging with and at risk of HIV, based in the Chicago metropolitan area. In regression models that adjusted for the effects of age, education, income, and sex, we additionally explored whether our findings were specific to those older adults with HIV by including control participants who were HIV negative. --- Methods --- Participants Participants were recruited as part of the Research Core of the CEDHA, which is a collaboration among the Rush Alzheimer's Disease Center of Rush University Medical Center, the Ruth M. Rothstein Core Center of Stroger Hospital, and the Community Outreach Intervention Projects within the School of Public Health at the University of Illinois Chicago. This study, which began in 2012 and has ongoing longitudinal data collection, enrolls participants who are 50 years of age and older, identify as either Black or White, and are HIV seropositive or seronegative. HIV-seropositive participants are eligible for the study if they have a CD4+ count <unk> 200 cells/mm 3 on cART or a CD4+ count <unk> 500 cART na<unk>ve, and viral loads ranging from undetectable up to 50,000 copies. All participants undergo assessment of risk factors and annual clinical evaluations. The clinical evaluation is comprised of a medical history, cognitive and motor function testing, and assessment of risk factors, including a questionnaire to assess loneliness. To more fully capture variation in socioeconomic status, income was measured using a show card methodology. Participants were shown a card with the following 10 possible categories and asked to choose the option that represented their household annual income: (1) USD 0-4,999, (2) USD 5,000-9,999, (3) USD 10,000-14,999, (4) USD 15,000-19,999, (5) USD 20,000-24,999, (6) USD 25,000-29,999, (7) USD 30,000-34,999, (8) USD 35,000-49,999, (9) USD Loneliness, Race, and HIV In order to further test the robustness of our loneliness findings in the CEDHA cohort, we also conducted additional loneliness analyses on a separate group of 1,180 participants without dementia or HIV, including 590 White participants of the Memory and Aging Project (MAP) [17] and 590 Black participants of the Minority Aging Research Study (MARS) [18]. The MAP and the MARS are both community-based longitudinal epidemiologic studies of aging and dementia administered through the Rush Alzheimer's Disease Center. Both the MAP and the MARS focus on adults over the age of 65, but the MARS is exclusively focused on older Blacks, while MAP includes all racial and ethnic groups. The CEDHA, in contrast, includes only Blacks and Whites, and has a focus on midlife adults (age 50 and older) who are either infected with HIV or at risk of HIV infection. Importantly, all 3 cohort studies have a common core of data collection, allowing the studies to be compared on the key variables of interest in this paper. Demographic and other information on the samples used from these 3 cohorts is presented in Table 1. --- Loneliness Assessment We assessed loneliness at each evaluation with a modified version of the de Jong-Gierveld Loneliness Scale [19,20]. Three modifications were made to the original scale. First, we eliminated 5 items assessing social loneliness because emotional loneliness was the intended focus of our interests. Second, we combined 2 similar items and made minor wording changes to another item to improve the clarity of the scale for older adults. Third, we asked participants to rate agreement with each item on a 5-point scale (rather than dichotomously) to improve the sensitivity of the scale. The following 5 items were employed: "I experience a general sense of emptiness," "I miss having people around," "I feel like I don't have enough friends," "I often feel abandoned," and "I miss having a really good friend." An average of the item scores yielded a total score that ranged from 1 to 5, with higher values indicating greater loneliness [8]. --- Assessment of Cognition A battery of 21 cognitive measures was administered by technicians trained by a board-certified clinical neuropsychologist. Measures of cognitive function assessed a wide and diverse range of cognitive functions and were identical in all essential details with the MARS [18]. Two of the 21 tests, the Complex Ideational Material and the Mini-Mental State Examination (MMSE), are only used for diagnosis of cognitive impairment and description. Raw scores on the remaining 19 tests were converted to z scores using the mean and standard deviation (SD). A global cognition score was calculated by averaging the z scores across these 19 measures of cognitive function as previously described [21]. --- Statistical Analyses Descriptive and bivariate statistics were calculated to characterize the sample. <unk> 2 tests were used for categorical variables and t tests for continuous variables. Linear regression models were then performed to examine the associations between race and loneliness in persons with HIV. All models included terms to control for the potentially confounding effects of age, education, and sex. Subsequent models investigating the association of race with loneliness also controlled for the potentially confounding effects of global cognition and income. Next, models investigating the interaction between race and loneliness were conducted with global cognition as the outcome in persons with HIV. Finally, models investigating the interaction between race and loneliness were conducted in participants who were HIV negative. Analyses were programmed in SAS version 9.3. --- Results --- Descriptive Statistics of Participants with HIV The CEDHA HIV-positive participants ( n = 177) included 124 Black and 53 White adults. The mean age of the total sample was 58.71 years (SD = 5.46), the mean education was 13.19 years (SD = 2.83); 134 identified as male, 42 as female, and 1 as transgender. The descriptive statistics characterizing the sample are presented in Table 2. Black and White participants differed with respect to age, education, income, MMSE total score, and global cognition. Black participants were older, had fewer years of education, had lower income, scored slightly lower on the MMSE, and performed lower on the global measure of cognition than White participants. No racial difference in CD4 count was observed. --- Racial Differences in Loneliness among Participants with HIV We hypothesized that older Black adults might endorse higher levels of loneliness than older White adults. In models adjusted for the effects of age, education, sex, income, and global cognition, we observed the opposite finding in the CEDHA participants. In this model, older Black participants reported significantly lower levels of loneliness than older White participants ( Table 3 ), though the difference was notably small (Cohen's d = -0.312). In additional post hoc analyses (online suppl. Table 3; see www.karger.com/doi/10.1159/000455253 for all online suppl. material), older Whites endorsed not having enough friends more than older Blacks, suggesting that perceptions regarding friendships might play a role in this racial difference. --- Loneliness and Cognitive Function among Participants with HIV In order to investigate the association of loneliness with global cognition, we conducted linear regression models adjusting for the effects of age, education, sex, income, and race in the CEDHA participants. There was no significant association between loneliness and global cognition. However, the results revealed an interaction between race and loneliness such that in older Black adults, greater loneliness was associated with lower global cognition ( Table 4 ). This interaction is visualized in Figure 1. Notably, we calculated Pearson and Spearman correlation coefficients for loneliness by global cognition stratified by race (Black, White) among HIV-positive persons. For Blacks, loneliness was significantly correlated with global cognition (Pearson r = -0.24128, p = 0.0069; Spearman <unk> = -0.20833, p = 0.0202); however, for Whites, loneliness was not significantly correlated with global cognition (Pearson r = 0.19706, p = 0.1573; Spearman <unk> = 0.03300, p = 0.8146). --- Race, Loneliness, and Cognitive Function in Those without HIV Although we demonstrated that older Black adults with HIV who indicated greater loneliness also had lower cognitive function, we wanted to investigate whether this pattern was due to race or to HIV infection by considering Black adults without HIV in the CEDHA cohort. We hypothesized that HIV status in particular would be as-sociated with negative health consequences, given its well-documented associations with neurological impairment, poorer psychological wellbeing, and mortality [e.g., 4-6 ], among others. We therefore investigated whether older Black adults showed greater loneliness than older White adults in subsequent analyses of HIV-negative participants who were recruited as part of the same study as a control group. The differences in sample characteristics between HIV-positive and -negative participants are described in Table 5. By design, the HIV-positive and -negative groups did not differ with respect to age, education, sex, or race. Neither did they differ on scores of loneliness or income in bivariate comparisons. Interestingly, differences in MMSE and global cognition were observed such that HIV-positive persons exhibited higher scores than HIV-negative persons; the differences were about 1.5 points on the MMSE and 0.20 z score on the measure of global cognition. These differences, although significant, are notably small and may reflect the fact that the HIVpositive participants were more likely to be linked to medical care than the HIV-negative participants. This greater access to medical care may have addressed comorbidities that could have an effect upon cognition. In a regression model that adjusted for the effects of age, education, sex, income, and global cognitive function, older Black CEDHA participants who were HIV negative also indicated less loneliness than older White adults (<unk> = -0.4652, SE = 0.1265, p <unk> 0.001; online suppl. Table 6). However, in regression models investigating the association between loneliness, race, and cognitive function, the interaction was not significant in HIV-negative persons (online suppl. Table 7). The 3-way interaction between race, loneliness, and HIV status in the combined group of HIV-negative and -positive participants was significant (<unk> = -0.3381, SE = 0.1495, p = 0.0244). This suggests that higher loneliness among older Black adults with HIV is associated with lower cognitive function ( Table 6 ). We found that older Black adults reported less loneliness than older White adults in the CEDHA cohort, regardless of HIV status, and this was contrary to our initial hypothesis. In order to test the generalizability of this finding, we conducted additional analyses on a separate group of 1,180 older participants without dementia and without HIV, including 590 White participants of the MAP [17] and 590 Black participants of the MARS [18]. The participants of this subsequent analysis were different from those in the previous CEDHA analyses and were demographically matched according to age, education, and sex using propensity score matching methods. In regression models controlling for the effects of age, education, and sex, older Black adults reported less loneliness than older White adults in this separate larger group. In additional models further adjusting for global cognition and income, this finding remained significant ( Table 7 ). --- Discussion Contrary to our first hypothesis, we observed that older Black adults with HIV reported less overall loneliness than older White adults with HIV in the CEDHA cohort. However, consistent with our second hypothesis, the association between greater loneliness and lower cognitive function was stronger in Black adults with HIV than in White adults with HIV. As a comparison reference, we tested the generalizability of this finding in 2 separate populations -in the control CEDHA subjects (similar in demographics but without HIV) and in a group of adults 15-20 years older without dementia or HIV. We observed that older Black adults without HIV indicated less overall 259 loneliness than older White adults without HIV in the CEDHA cohort, but there was no racial difference in the association of loneliness and cognitive function among those without HIV. However, in a 3-way model of the combined cohort of positives and negatives, we observed a racial difference in the association of loneliness and cognition by HIV status. We also found in the cohort of older adults without dementia or HIV that older Blacks Over a quarter of persons with HIV are over the age of 50, and since the advent of cART, a rapidly increasing number of adults with HIV are living into old age [1]. Black adults account for a higher proportion of new and current HIV infections than other racial groups [12]. Our results support a difference by race in overall loneliness experienced among older adults living with HIV; however, the exact mechanisms for this are unclear. Research has suggested that HIV status among older Black adults is associated with significant stigma and prejudice, and this may consequently result in depression and social isolation [22]. For these reasons, we hypothesized that older Black adults might report greater overall rates of loneliness than older White adults; however, we found the opposite to be true. Previous reports investigating persons of different cultures using this loneliness measure have found greater loneliness endorsed among minority and immigrant groups [23,24]. Our results are surprising as they are discrepant from these reports. First, it should be noted that many studies of loneliness among older adults have not focused on older Blacks, and it is possible that Black adults as a group have different experiences from other minority or immigrant populations. Second, our measure of loneliness was slightly modified in that the items reflecting social loneliness were dropped and only the items reflecting emotional loneliness were retained. This difference could also be a possible explanation for our discrepant findings, as social loneliness may reflect the objective absence of others, whereas emotional loneliness may reflect a subjective perception of loneliness. Third, although racial differences in loneliness were significant, the overall differences were small. However, in order to test the direction and strength of the finding, we conducted additional analyses on a separate group of 1,180 nondemented participants from the MAP and MARS cohorts. As stated previously, older Black adults in this separate larger group also reported less loneliness than older White adults, supporting our loneliness by race finding. Some work has suggested that Black individuals may show greater resilience to psychosocial stressors [25]. Consistent with this notion, previous work has suggested that among older Black and White persons with HIV experiencing comparable levels of distress, older Black adults engage in more adaptive coping strategies [26,27]. The engagement of more adaptive coping strategies might explain our overall observation of less reported feelings of loneliness among older Black adults. It is also possible that older Black adults may show greater resilience to loneliness due to the necessary development of social or emotional coping strategies to deal with well-documented and long-standing institutionalized racism and systemic injustice [28]. More work is needed to clarify the mechanisms involved in this finding. Although older Black adults might report less loneliness overall, it is possible that they might experience loneliness differentially and more drastically than older White adults for multiple reasons. For example, older Black adults, regardless of HIV status, have been found to have fewer social resources (smaller social networks and less engagement in social activities) than older White adults [15], and therefore may be more prone to the negative health effects of loneliness after learning of an HIV infection. Older Black adults with HIV not only have to contend with the negative historical consequences of institu- tionalized racism, but also with the social stigma associated with HIV infection, and these may interact to produce downstream effects upon health outcomes such as cognition. For these reasons, we secondarily hypothesized that loneliness experienced among older Black adults with HIV would be associated with poorer cognitive functioning in the context of an infection believed to result in cognitive impairment. Our finding of higher loneliness associated with lower cognitive function among older Black adults with HIV is striking and suggests that race and loneliness may be important linked factors to consider for assessing the risk of cognitive impairment in the context of aging with HIV. However, the exact mechanisms for this are again unclear. From a neurobiological perspective, loneliness has been associated with decreased dendritic arborization in the hippocampal and prefrontal cortices in animal models [29], and dendritic injury has been described as a pathological substrate for HIV-related cognitive impairment [30], but it is unclear how these associations may be influenced by race. Again, one possible explanation may lie in the association between loneliness and stress in old age. Older adults who report experiencing a greater number of chronic social stressors also report greater loneliness [31]. Chronic stress induces the release of glucocorticoids, which in turn induces dysfunctional changes in glutamate neurotransmission in the hippocampus and prefrontal cortex [32]. Older Black adults with HIV may experience greater stress from multiple intersecting stigmas compared to White adults, and the association of loneliness with lower cognition among Black adults may be an indicator of increased race-specific stress from multiple sources acting upon neurobiological functioning. Older Black adults may also be more resilient to loneliness than White adults due to the development of necessary coping strategies in response to long-standing institutionalized racism. If this is true, then this may explain why Blacks report less loneliness overall. Furthermore, any loneliness endorsed among older Black adults may represent a more serious breakdown in social or other resources than among White adults, and this in turn may be associated with worse cognitive function. Future research is needed to examine the neurobiological and psychosocial mechanisms underlying racial differences in the association of loneliness with cognitive functioning. Future research is also needed to explore the effects of stigma upon neurobiological functions, and to determine what resources might be useful as intervention strategies to increase resilience. The present study had some limitations. First, temporal ordering and causal inferences cannot be based on cross-sectional data. It is currently unknown how the present findings represent the possible time courses of loneliness, cognition, and HIV status. It is plausible that Black adults with HIV who have poorer cognition might experience more loneliness. Longitudinal studies would greatly assist in investigating causal links in our results. A final limitation is the relatively small number of Whites in the study, which may have reduced the power to detect an association between loneliness and cognition in this population. This study also has multiple strengths. These include the use of participants from a community-based sample, a large sample size, the controlling of multiple demographic characteristics that could confound the results, and the use of an HIV-negative control group that was similar in age, sex, education, income, and race to elucidate the specificity of results to HIV. Our findings suggest that loneliness among older Black adults with HIV may be associated with worse cognitive functioning relative to White adults with HIV. This suggests that the development of interventions aimed at reducing the loneliness of older Black adults with HIV might have a consequent protective benefit upon cognitive functioning. Future work is needed to examine this and the neurobiological or psychosocial mechanisms underlying these associations.

study of the Rush Center of Excellence on Disparities in HIV and Aging (CEDHA) completed a 5-item self-report scale used to measure emotional loneliness and a battery of cognitive measures. Results: Contrary to our expectations, older Black adults indicated less overall loneliness than White adults (β = -0.3893, SE = 0.1466, p = 0.0087) in models controlling for the effects of age, education, sex, global cognition, and income. However, in models with cognitive function as the outcome, an interaction between race and loneliness was observed, such that older Black adults who indicated greater loneliness showed poorer cognitive function relative to White adults (β = -0.2736, SE = 0.1138, p = 0.0174). Conclusion: Older Black adults with HIV reported less loneliness than older White adults; however, the inverse association between loneliness and cognitive function was stronger in Black than White older adults. Additional work is needed to elucidate the mechanisms underlying this interaction.

Introduction Welfare state theory has often categorized states in terms of the pattern of interplay between social policies and the structure of the labour market (cf. Esping-Andersen 1990). Moreover, it has frequently offered an exclusive focus on state policy (Fraser 2009). In both cases, it has failed to fully recognize the role of the formally organized third sector. Added to this, such theory has given insufficient attention to the global trend of state restructuring and the rise of meso-governance. This study offers a corrective view. In 1998/99, government third sector policy in Scotland, Wales and Northern Ireland ceased to be decided by territorial ministries of central government. Instead, it is determined by administrations elected in three new political systems (re-)created by devolution (Chaney 2013b). This has transformed the way party politics influence government policy on the third sector in the UK. In the case of Scotland, the Scottish Labour Party and the Scottish Liberal Democrats formed successive executives until the 2007 elections; subsequently the Scottish National Party has held office. In Wales, the Welsh Labour Party has been in government since 1999 (including periods of coalition with the Welsh Liberal Democrats 2000-03 andPlaid Cymru 2007-11). In contrast, under the singular arrangements in Northern Ireland the exercise of executive functions has been done on the basis of power-sharing between parties. Accordingly, this article makes a distinctive contribution by focusing on policy discourse and how this transformation in territorial politics is impacting on the way that the third sector is envisioned as a welfare provider. It, therefore, addresses a key lacuna in contemporary understanding, namely how 'welfare pluralism' is shaped by the process of state decentralization. In the following discussion, 'welfare pluralism' is a descriptive label which refers to the situation whereby service contributions by the voluntary and private sectors complement to state welfare delivery (Beresford and Croft 1983). The involvement of the third sector has long-standing links with political attempts to recast public service provision, yet emphasis on encouraging voluntarism and harnessing the contribution of the sector has heightened in recent decades (Hanlon et al. 2007). In definitional terms, we are mindful of Brenton's (1985: 57) rejoinder that 'the voluntary sector's pluriformity and lack of clear boundaries do not lend themselves to the definitions and classifications upon which statistical methods are based'. In response, this article follows existing research practice (Casey 2004) by using the umbrella term 'third sector' to refer to the principal collective signifiers associated with nongovernment advocacy and service organizations; namely, 'voluntarism', 'voluntary sector', 'third sector', 'civil society' and 'non-profit sector'. The UK presents a propitious research context because the process of devolution initiated in 1998/99 has recast the territorial governance of the third sector with social policy responsibilities being transferred to newly (re-)established legislatures in Scotland, Wales and Northern Ireland. Such state restructuring can be seen as part of a wider 'devolutionary trend [that] has swept the world [... involving widespread] transference of power, authority, and resources to subnational levels of government' (Rodriguez-Pose and Gill 2003: 334). In exploring the interplay between state restructuring and public policy discourse, we make an original methodological contribution to understanding the impact of devolution. Comparative discourse analysis is used to examine the key policy texts published by government in the constituent polities of the union state. 1 The aim is to focus on the formative phase of policy-making and examine the contrasting political visions for the third sector as set out by devolved and central government through an emphasis on policy discourse. Accordingly, the article's principal aims are as follows: 1. To examine the framing of policy on the third sector, including its welfare role, in the four polities of the UK. 2. To examine the nature and development of social policy narratives related to the third sector in each territory. 3. To consider whether the data provide evidence of the territorialization of third sector policy in the wake of devolution. 2 The remainder of this study is structured as follows. Following an outline of the research methodology the findings section consists of four parts: 1. The policy framework prior to devolution: analysis of policy framing in the state-third sector formal agreements or 'Compacts' of 1998. 2. The policy framework following devolution: exploration of the framing practices in the principal third sector policy documents in each territory 1999-2012. 3. Detailed examination of the visions of the third sector's welfare role in each polity (in the section 'Welfare pluralism and the territorialization of third sector policy'). 4. Examination of the territorial policy narratives in each polity, reflection on their underlying causes and consideration of the application of our methodology to other liberal democratic regimes. The main findings and their implications are considered in the concluding discussion. --- Methodology The present method combines content and critical discourse analysis. In the former case, by recording the number of incidences of key words, ideas or meanings in a text, content analysis can further an understanding of the nature and salience of issues in policy formulation. In the current study, this is complemented by critical discourse analysis, operationalized here by frame analysis (Yanow 1999) of how, as key texts, policy documents enable policymakers to construct (or 'frame') measures to address social issues and effect change. The documents analyzed constitute the principal government policy documents on the third sector published in each of the four polities 1998-2012 (see 'Policy Documents Included in Analysis'). Framing here refers to 'collections of idea elements tied together by a unifying concept that serve to punctuate, elaborate, and motivate action on a given topic' (Creed et al. 2002: 37). Our focus on salience and policy framing enables exploration of political narratives associated with the development of the third sector and its social policy role across polities. This methodology was applied as follows. Electronic versions of each government's core policies on the third sector were analyzed using appropriate software. 3 It should be noted that all of the texts analyzed were territorially discrete and referred to either: England, Scotland, Wales or Northern Ireland (as opposed to Great Britain or the UK). The policy texts were divided into 'quasi-sentences' (or an argument that represents the verbal expression of a single political idea or issue, Volkens 2001: 96) centred on the incidence of a key term. 4 Thus each quasi-sentence was classified using an inductive coding frame based on key frames derived from the policy literature on the third sector (see 'References'). In order to offer a sophisticated exploration of policy discourse, this study adopted a tiered, or sequential, approach to the frame analysis in order to examine what Minsky (1975: 223) describes as'sub-frames'. These are explanatory or descriptive signifiers which attach to primary issue frames (in the case of the 'welfare pluralism' frame, they comprise the different underlying motives for cross-sectoral working -efficiency, effectiveness, choice, etc.). In terms of the temporal comparison in the analysis, the periods 1998-2003, 2003-07 and 2007-12 were used (each period is discrete, divided before and after the elections in a given year). This was a 'best-fit' approach aimed at capturing the first three terms of devolved government whilst acknowledging the fact that Westminster and meso-government operate on different electoral cycles. 5 We apply the foregoing method in order to first compare policy framing in the state-third sector formal agreements or Compacts of 1998, and then to analyze the principal third sector policy documents in each territory, 1999-2012. --- Comparative Analysis of Policy Framing on the Third Sector 1998-2013 The state-third sector compacts of 1998 From an international perspective, the state-voluntary sector agreements or Compacts introduced in the UK in 1998 (see figure 1) were an innovation in contemporary governance; one that has subsequently been emulated in other countries. They are formal statements that set out mutual obligations and define each sector's role. As Kendall (2003: 2) observes, the 'Compact idea is completely without precedent, representing an unparalleled step in the positioning of the third sector in public policy'. With a separate Compact for each UK nation they also prefigured devolution by adopting an explicitly territorial approach to state-third sector relations. A key question here is the extent to which there is continuity in the policy frameworks applying in each territory before and after devolution. In answer, critical discourse analysis reveals that whilst there are inter-polity differences in individual policy frames, overall there is broad consistency in the territorial framing profiles. The variance of the four data sets is not statistically significant (P = 0.432). 6 The reason for this continuity lies in state-wide electoral politics. The Compacts have shared roots; each stems from New Labour's initial policy document, Building the Future Together -Labour's Policies for Partnership between Government and the Voluntary Sector (HM Government 1997). Thus, on the eve of devolved governance, notwithstanding some local variations, the framing in the four documents amounts to a general state-wide agenda on the values and priorities of state-third sector relations as set by a single party governing at Westminster. Thus, frames such as 'partnership' (where there is marked similarity in the discourse across territories) provide evidence of this shared framework. For example, in England reference is made to 'working in partnership towards common aims and objectives. [... this] improves policy development and enhances the design and delivery of services and programmes' (Home Office 1998: 8). At the same time the Scottish, Welsh and Northern Irish counterparts make similar assertions, 'the government is concerned with promoting partnerships between public and voluntary sectors through its policies' (Welsh Office 1998: 3); 'the government will meet with the sector to develop policy and practice and promote effective dialogue' (Scottish Office 1998: 10); and, the government will 'involve the voluntary and community sector in partnership working and the process of developing and monitoring public policies' (Northern Ireland Office 1998: 14). However, the beginnings of distinctive territorial approaches to framing policy may also be detected in the Compacts. Examples include the disproportionately high level of attention paid to community and local level working in the English Compact (it accounts for 44.3 per cent of the overall policy discourse compared to a mean of 25.2 per cent for the other polities), typified by statements like, 'it is important that the distinctive needs and interests of community groups are taken into account. A code of good practice will be developed to facilitate and reflect this' (Home Office 1998: 12). A further example is the emphasis placed on third sector organizations' role in promoting equality in the Northern Ireland Compact; for example, 'equality of access to resources and decision-making processes for all the people of Northern Ireland' (Northern Ireland Office 1998: 11). Framing in the principal third sector policy documents in each territory, 1999-2012 An initial indication of the level of contrast or continuity in the framing practices pre-and post-devolution can be gained by examining the distribution of quasi-sentences made under each frame across the four polities 1999-2012. In other words, this is an aggregate measure of the total number of incidences of each frame in all key policy documents analyzed over the period broken down by territory. The result is empirical confirmation of the territorialization of policy. Statistically significant differences emerge in framing practices when the polities are compared in this way (P = <unk>0.001) (see table 1). 7 From a comparative perspective, the greater attention afforded to a number of frames in the English policy documents underlines how the 'postdevolution' policy discourse became territorialized. For a series of key frames the incidence of quasi-sentences is more than double the mean for the devolved polities. These include 'local/community working', 'funding issues' and 'partnership'. Crucially, reflecting what our later analysis shows to be a greater emphasis on welfare pluralism in England (see below), the 'public service delivery' frame is given significantly more attention in the English public policy discourse than elsewhere (accounting for 38.1 per cent of all references under this frame, compared to 19.4, 11.4 and 31.1 in Scotland, Wales and Northern Ireland, respectively). However, territorial distinctiveness in framing practices is not restricted to England. It is also evident in the discourse relating to the devolved polities. Thus, policy in Northern Ireland accounted for most references to 'capacity building' (47.6 per cent),'strategic aims and vision' (40.8 per cent) and, reflecting attempts to engage voluntary groups in civil conflict resolution (cf. Chaney 2012a), 'leadership and political commitment' (40.5 per cent). In contrast, Scottish policy predominated on 'promoting volunteering', a prominent trope in post-1999 debates (Fyfe et al. 2006) (32.9 per cent), as well as 'communication' (31.9 per cent) and 'best practice/effectiveness' (35.6 per cent). The Welsh policy framework placed particular emphasis on 'participation' (54.8 per cent of all references) and 'citizenship' (36.6 per cent); both are tropes in the inclusive governance discourse promoted by parties across the political spectrum to bolster initially fragile support for devolution (Chaney and Fevre 2001a). --- Welfare pluralism and the territorialization of third sector policy A key welfare governance issue which attaches to state decentralization is whether there is policy continuity on third sector involvement in public service delivery spanning the pre-and post-devolution eras, or whether 'devolved' governance is fostering contrasting approaches in the constituent polities of the union state. In other words, we need to understand what happens to the way that the third sector's welfare role is envisioned when state-wide practices are replaced by four territorial political systems. To explore this issue in depth, a two-tier methodology was employed. Following the initial coding process (see 'Methodology'), all quasi-sentences associated with the public services frame in the key third sector policy documents covering the period 1999-2012 were (re-)coded according to subframes detailing the motive underpinning each reference to third sector involvement in service delivery (see table 2). The results are striking and reveal statistically significant differences between polities (P = <unk>0.001). 8 This is significant because it confirms the rise of territorially distinctive approaches to welfare pluralism. Most notably, the policy framework covering the third sector in England stands out as the most'market-oriented'. In other words, it places greatest emphasis on the three elements defining this frame; namely,'securing better efficiency over state provision', 'added value' and'marketization' (P = <unk>0.001). 9 Just over a quarter (25.5 per cent) of references to third sector involvement in public service delivery in England related to the aforementioned motives, compared to 15.2 per cent in Scotland, and just 6.2 per cent and 5.9 per cent in Wales and Northern Ireland, respectively. Further insight can be gained by examining of the development of substantive government policy on third sector welfare service delivery in England. This gained momentum in the wake of the UK government's 2002 and 2004 cross-cutting policy reviews (applying to England) (HM Government 2002Government, 2004)). These set out the governing Labour Party leadership's aspiration for 'a transformation of the third sector to rival the market and the state' (Brown 2004). It was an agenda that was discursively packaged in a way that linked Examples of the market-oriented discourse include, 'encouraging the growth of a diverse and competitive market in which the third sector is expected to play a growing role' (Cabinet Office 2006: 17); and 'we believe in the power of competition to increase standards and deliver better value' (HM Government 2012: 23). It is a discourse that spans the neo-liberalism of New Labour (1999-2010) and the subsequent Conservative-Liberal Democrat Coalition government (from 2010). Compared to the other polities, the English policy texts have a more assertive tone. This is particularly noticeable under the Coalition government, where it is often wrapped-up under its 'localism' agenda. 10 For example, the self-stated aim of the latest iteration of the state-third sector Compact is: Promoting contestability... these rights will give local community and third sector organisations the opportunity to challenge their local authority where they believe services or facilities would be better run by alternative providers. It will... give people a voice over local issuesalso it will open up more contracts to third sector providers (HM Government 2010: 6). A further notable aspect of the discourse is the way it is articulated in terms of the market discipline of competition, 'government's public service reforms will enable charities, social enterprises, private companies and employeeowned co-operatives to compete to offer people high quality services' (HM Government 2012: 3) and, 'we will help the sector become more competitive in this emerging landscape, in particular through our new plans to run a series of commercial skills "masterclasses" in 2013' (HM Government 2012: 5). Compared to the other devolved polities, Scotland's greater embracing of the'market-oriented' discourse is, in part, explained by the policy dynamic of the Labour Party simultaneously holding government office in Westminster and Edinburgh (cf. Laffin et al. 2007). 11 Thus in the early-to-mid-2000s, the Scottish Executive followed key aspects of Westminster policy (Hassan and Shaw 2012). This is evident in the policy texts. For example, 'the UK review [covering England] identified several obstacles and challenges that need to be overcome to enable the sector to develop its public service delivery role. These included the following development needs which apply equally within a Scottish context' (Scottish Executive 2004: 8, emphasis added). Examples of the Scottish market-oriented discourse include, 'the social economy in Scotland is becoming much more business-like in its approach to service delivery -and this is helping some organisations to generate significant profits on some of their services. Some might call them not-for-profit businesses. But we see them as more-than-profit organisations' (Scottish Executive 2004: 5). 12 Here the stated aim was 'to make Scotland a world leader in the development of an enterprising third sector... [to] develop new services and [for them to] market themselves effectively'. A further common trope crosscutting the market-oriented discourse in the English and Scottish policy frameworks is the pervasive -yet empirically contested -notion of non-state organisations' superior ability to innovate in service delivery (cf. Borins 2001). For example, 'the best organisations [have]... an ability to be flexible, offer joined-up service delivery and... the experience to innovate' (HM Treasury and Cabinet Office 2007: 49) and, 'we recognise the added value that the sector brings to the delivery of public services. Social economy organisations have a real understanding of the area in which they operate. They are flexible and able to innovate' (Scottish Executive 2005a: 2). In Wales, at a rhetorical level at least, Labour has been keen to espouse the existence of putative 'clear red water' 13 between itself and the Party at Westminster. Instead, it has preferred to style itself as 'Classic' rather than 'New' Labour (Chaney and Drakeford 2004), thereby signalling its opposition to the latter's neo-liberal agenda. Such a standpoint is evidenced by the Welsh party affording less attention than its English counterpart to the market-oriented sub-frame. Nevertheless, analysis shows welfare pluralism is still a feature of the third sector policy framework in Wales. Examples include, 'the model, which we have opted for, seeks to maximise efficiency gains through the scale economies of more effective co-operation and coordination between agencies across the whole of the public sector, not excluding the independent, voluntary and private sectors' (WAG 2004: 3) and, there must be 'a willingness to consider new ways of providing services, including an increasingly mixed economy of provision, with the potential for a greater role for the third sector in delivery' (WAG 2006: 21). Compared to the English policy documents, the present analysis reveals that the policy frameworks in the devolved nations place greater emphasis on extolling the community benefits of third sector involvement in welfare service delivery. According to the devolved administrations' policy discourse, third sector involvement is generally more responsive than services provided by the state. Thus the 'community benefits' of service delivery by third sector organizations, or welfare pluralism, is a sub-frame that constitutes 17.4 per cent, 21.4 per cent and 16.8 per cent of all references in Scotland, Wales and Northern Ireland, respectively -compared to 10.5 per cent in England. It is typified by the assertion by the Northern Ireland Executive that 'effective partnership between Government and the Voluntary and Community Sector can make a valuable contribution to more responsive and people-centred public services' (DSD 2011: 2). The idea that third sector organizations deliver more effective services than the state receives greatest attention in the Scottish policy discourse (it accounts for 17.8 per cent of references, double the mean for the other polities). Examples include, 'cross-boundary solutions... are increasingly being used by local authorities to generate efficiencies and to ensure more effective delivery of services' (COSLA/SCVO 2009: 5) and,'recent years have seen increasing numbers of voluntary sector organizations playing an effective part in delivering on key agendas in terms of service delivery' (Scottish Executive 2004: 6). In contrast, and reflecting the more comprehensive equalities law applying in the province (Chaney 2011), the involvement of third sector organizations in service delivery in order to meet the needs of disadvantaged groups receives three times the level of attention in Northern Ireland compared to the other polities (19.7 per cent of references compared to a mean of 7.3 per cent in the other territories). Examples include, 'voluntary and community organisations have a track record of tackling social need and deprivation and are well placed to develop and deliver improved frontline services, particularly to the most disadvantaged people in society' (DSD 2005a: 18; see also DSD 2005b). Lastly, the analysis shows the policy discourse in Wales to have given greatest attention to the idea that third sector organizations are subject to higher levels of trust than state service providers (see Taylor 2002) (7.4 per cent of references compared to a mean of 2.1 per cent elsewhere). For example, 'we believe the third sector is in a particularly strong position to provide frontline services when users have multiple disadvantages... the service is targeted at users who are likely to mistrust businesses or state providers' (WAG 2006: 27; see also WAG 2009). Before summarizing the implications of the present findings we first consider the territorial policy narratives in each polity, and reflect upon the apparent drivers as well as the transferability of this study's methodology to other liberal democratic regimes. --- Territorial policy narratives The following examination of policy framing alongside the principal policy developments in each polity (see table 3) reveals the structural narrative of the third sector in each territory. It is a technique that offers a temporal perspective of how frames as narrative devices, develop and become more or less prominent and persuasive over time (Petersen and McCabe 1983). In this way it enables an 'understanding of the dynamics and frameworks of decision makers that supports and articulates policy choices and the claims underlying them' (Roe 2011: 541). --- Scotland. A key aspect of the narrative in Scotland is far-reaching structural change affecting the third sector (e.g. Scottish Executive 2003b, 2005b). During the early years of devolution the Scottish Executive supported a tripartite system of Volunteer Centres, Councils for Voluntary Service and Social Enterprise networks (for a discussion, see Fyfe et al. 2006). Following its election in 2007, the Scottish National Party government introduced a SOCIAL POLICY & ADMINISTRATION, VOL. 48, NO. 7, DECEMBER 2014 Concordat between Scottish government and local authorities; a development which has recast state-third sector relations. Its effect has been to devolve power and give local authorities greater control in dealings with the sector. On one level this is nothing new. It is part of the wider international trend of promoting localism evident over recent decades (cf. Page 1991). Yet the means by which it has been pursued in Scotland are distinctive. Since 2011, new third sector 'interfaces' have been developed in each community planning area. The aim is to align the sector with the Community Planning Partnerships and Single Outcome Agreements. In turn, this is designed to support to local third sector organizations, boost volunteering and develop social enterprise. Crucially, its proponents claim that it allows a more strategic approach to the third sector's welfare role. Framing data provide further details of the developing policy narrative (see figure 2a). Over the period 1999-2012, the greatest increase has been in the 'promoting volunteering' frame (+23.7 percentage points); for example, the 'Scottish Executive will... ensure that volunteers are supported and encouraged in every possible way' (Scottish Executive 2004: 2). The underlying concern here is to increase the capacity of the sector to deliver services as well as to boost active citizenship. Recent work (Asenova et al. 2012: 17) reveals how, in the face of growing austerity, local authorities' principal strategic policy response is'service transfer' to the third sector. Such a trend is supported by the framing data which show an increase in proposals under the 'public services' frame (+14.7 percentage points) underlining the growing emphasis on welfare pluralism as a response to the current recession and cuts in welfare spending (estimated by the Scottish government to be £4.5 billion in the five years to 2014-15). 14 In contrast, the frames'monitoring and evaluation' (-11.5 percentage points), 'communication' (-7.2 percentage points) and 'best practice' (-5.9 percentage points) have all declined in salience since 1999. --- Wales. A core aim at the outset of constitutional reform in Wales was 'establishing a new, more inclusive and participative democracy' based on devolved government 'working in partnership with the voluntary sector' (Welsh Office 1998: 3). Uniquely, this commitment was enshrined in the subsequent devolution Acts such that statute requires Ministers publish a Scheme setting out how they propose to engage with and promote the interests of voluntary organizations (Chaney and Fevre 2001b). Accordingly, successive Welsh administrations have sought to develop third sector capacity and mainstream the sector into the conduct of public business (see figure 2b). As a result, the policy narrative in Wales is one of increasing attention to 'participation and engagement' (+26.6 percentage points), 'community development' (+13.3 percentage points) and 'promoting volunteering' (+8.3 percentage points [Chaney 2013a]). Examples of this discourse include, 'we are also keen to promote greater partnership working between third sector organisations themselves, not only to ensure a stronger voice for citizens locally, but also to improve efficiency through the sharing and pooling of capacity' (WAG 2008: 47). A seemingly counterintuitive finding is the decline in the salience of 'funding issues' (-12.2 percentage points) and 'third sector involvement in public service delivery' (-13.0 percentage points). This is a key difference compared to Scotland. There are two main explanations. As noted, successive Labour Party-led administrations in Wales have been keen portray themselves as champions of state provision of social welfare; thereby distinguishing themselves from the mixed economy approaches espoused in England and Scotland. In addition, the most recent key policy document in the Welsh dataset is from 2010 (as opposed to 2012 in the Scottish case). At that juncture, the full extent of government austerity measures was not appreciated. The latest data reveal the full gravity of the situation and point to the likelihood of revised framing practices in future rounds of policy-making and an attendant shift towards welfare pluralism. Underpinning this scenario Northern Ireland. In Northern Ireland, the structural narrative of the third sector forms part of the wider project to secure inclusive governance in the wake of the civil conflict (Chaney and Rees 2004;DSD 2007aDSD, 2007bDSD, 2008DSD, 2012)). Thus the Compact agreed by the government and the sector in 1998 (NIO 1998) espoused the themes of accountability, active citizenship and the participation of the sector in the development of public policy. It is an agenda facilitated by the Joint Government -Voluntary and Community Sector Forum which was established to promote discussion of general issues of common concern. The re-imposition of direct rule and successive crossgovernment policy reviews (DSD 2006(DSD, 2011;;PWC 2004) underline the many challenges facing the sector. Notably, government's second strategy (DSD 2006) prioritized 'building communities' and 'targeting disadvantage'. In 2008 (and following a period of direct rule), a resolution of the Assembly called on the Executive to set out further measures to strengthen co-working and engagement with the sector. 15 After protracted negotiations a successor to the Compact was published in 2011 (DSD 2011). This singular policy history has resulted in a post-1998 structural narrative characterized by an increase in policy framed in terms of 'local/community co-working' (+19.6 percentage point) and, significantly, third sector involvement in public service provision (+10.8 percentage points). For example 'we will encourage and support more effective and wider-ranging involvement of voluntary and community organisations in the delivery of public services' (DSD 2005a: 4). As in the case of Scotland, the growing emphasis on welfare pluralism is driven by austerity measures. This is evident in the province's Budget settlement 2011-15 which sets out 'a real terms reduction in resource DEL [Departmental Expenditure Limit] of 8 per cent, and a real terms reduction in capital DEL of 40 per cent by the end of the Spending Review period' (NIE 2011: 21). The role of the recession in promoting service transfer from state to third sector is underlined by the fact that over a half (55.6 per cent) of post-1998 references to service delivery have been made since 2007 (see figure 2d). England. The structural narrative in England is characterized by a strong post-1998 emphasis on welfare pluralism (Home Office 2003, 2004, 2005). In contrast to the other polities -and reflecting the neo-liberalism of the Blair administrations -the mixed economy discourse is evident from the outset and is a core feature of the three successive Compacts between the sector and government (cf. Cabinet Office 1998a, 1998b, 1998c, 1998d, 2009;HM Government 2010). It has been subject to renewed emphasis by the current Coalition government such that policy is concerned to make it 'easier for civil society to work with the state'. For example, 'we are making it easier for civil society to... access public service contracts... help the sector become more competitive... [and] identify opportunities' (HM Government 2010: 4). Once again reflecting the role of austerity in driving a mixed economy of welfare approach, almost two-thirds (63.7 per cent) of the post-1998 total of references to funding and 39.5 per cent references to public services have been made since 2007. Summary. The foregoing territorial narratives highlight the power dimension to the third sector's changing welfare role. This, and the discursive processes which drive the policy process in territorial political systems with multi-party elections, are not unique to the UK. Thus the present methodology is applicable to other liberal democratic regimes where it may complement existing structural analyses. Examples of the power dimension to the envisioning of the sector's welfare role in the present study include national administrations' framing of third sector policy in terms of autonomy, independence and localism. Yet, as the foregoing analysis suggests, the underlying power dynamic is one of central authorities retaining power in the allocation of resources as well as 'high level' policy and lawmaking functions, whilst transferring to local government and third sector bodies day-to-day service delivery responsibility along with the attendant political risks (policy examples include the Scottish Community Planning Partnerships and the Localism Act 2011 in England). The power dimension is also evident in the discourse on welfare pluralism. Notably, the foregoing analysis reveals this to be framed in terms of empowerment of the sector and the communities that they serve. Yet, here austerity is a key under-acknowledged driver; one that spans polities. In this regard, the third sector provides the political elite with a viable means to deliver welfare outside of state provision whilst avoiding exclusive reliance on the private sector. Again, the power dynamic is one of government retaining control but transferring political risk in the form of delivery responsibility (as well as answerability to regulators and budgetary oversight). In an era of multi-level governance, the present comparative structural narrative methodology also reveals how welfare pluralism is both a 'devolved' and shared construct. In other words, economic imperative underpins its commonality across polities, yet distinctive framing practices apply in each (in turn, reflecting territorial party politics). This means that it is advanced and applied in different ways (e.g. as the contradictory elements in the Welsh public policy discourse attest -at once underlining the need to maintain and support state provision yet embracing non-state provision). The structural narrative approach also underlines the way in which policy change is presented in order to appeal to, and meet the expectations of, specific audiences. This is captured in the literature on strategic framing (cf. Pan and Kosicki 2007) whereby common policy objectives are advanced by use of contrasting policy frames. In the present case, it is again typified by the discourse on welfare pluralism. Thus compared to the Welsh and Northern Irish -and to a lesser extent the Scottish policy discourses (which all underline co-working, partnership, local engagement, and are pitched to the support base of the Left-of-centre parties which have predominated in devolved government) -the neo-liberal UK coalition government covering England places greater emphasis upon efficiency, competition, accountability and innovation in relation to non-state provision of welfare (tropes that 'play well' with traditional Conservative Party supporters). --- Discussion By focusing on the changing ways in which territorial politics envisions the third sector's role as a provider of welfare, the present study complements a vast body of macro-comparative studies of welfare regimes (classically, Esping-Andersen 1990). Yet whilst this literature has placed increasing, if inconsistent, attention on the place of voluntary associations and informal care as pillars of welfare governance alongside the state and (labour) market (Arts and Gelissen 2010), it has given insufficient attention to the welfare role of the formally constituted third sector. Moreover, it has largely ignored the way that shifting sectoral welfare roles play out in the context of devolution and the global rise of multi-level governance. The evidence from the UK shows these to be significant factors shaping the nature of social welfare in the 21st century. In methodological terms this study marks an initial step in investigating the nexus between policy discourse, state restructuring and welfare. However, the limitations of the findings also need to be acknowledged in that they are derived from examination of a specific type of policy discourse; albeit the principal statements of government policy in each territory. To this end it is suggested that future study needs to explore the policy

Welfare state theory has struggled to come to terms with the role of the third sector. It has often categorized welfare states in terms of the pattern of interplay between state social policies and the structure of the labour market. Moreover, it has frequently offered an exclusive focus on state policy -thereby failing to substantially recognize the role of the formally organized third sector. This study offers a corrective view. Against the backdrop of the international shift to multi-level governance, it analyses the policy discourse of third sector involvement in welfare governance following devolution in the UK. It reveals the changing and contrasting ways in which post-devolution territorial politics envisions the sector's role as a welfare provider. The mixed methods analysis compares policy framing and the structural narratives associated with the development of the third sector across the four constituent polities of the UK since 1998. The findings reveal how devolution has introduced a new spatial policy dynamic. Whilst there are elements of continuity between polities -such as the increasing salience of the third sector in welfare provision -policy narratives also provide evidence of the territorialization of third sector policy. From a methodological standpoint, this underlines the distinctive and complementary role discourse-based analysis can play in understanding contemporary patterns and processes shaping welfare governance.

which territorial politics envisions the third sector's role as a provider of welfare, the present study complements a vast body of macro-comparative studies of welfare regimes (classically, Esping-Andersen 1990). Yet whilst this literature has placed increasing, if inconsistent, attention on the place of voluntary associations and informal care as pillars of welfare governance alongside the state and (labour) market (Arts and Gelissen 2010), it has given insufficient attention to the welfare role of the formally constituted third sector. Moreover, it has largely ignored the way that shifting sectoral welfare roles play out in the context of devolution and the global rise of multi-level governance. The evidence from the UK shows these to be significant factors shaping the nature of social welfare in the 21st century. In methodological terms this study marks an initial step in investigating the nexus between policy discourse, state restructuring and welfare. However, the limitations of the findings also need to be acknowledged in that they are derived from examination of a specific type of policy discourse; albeit the principal statements of government policy in each territory. To this end it is suggested that future study needs to explore the policy discourse and deliberations of exogenous interests, most notably civil society organizations and policy networks, as they set out their vision for the sector. It also needs to examine the role of state-wide versus regionalist parties in shaping the nature and extent of policy convergence/divergence across territories -as well as the extent to which civic nationalist parties' influence on third sector policy is shaped by the twin imperatives of (sub-)national autonomy and state-building. In addition, this study's focus on executive policy documents needs to be complemented with analysis of legislative proceedings, including backbenchers' and opposition parties' discourse as they seek to (re-)define the sector's welfare role. Earlier work shows that, in policy terms, the impact of devolution on the third sector in the UK was not immediately apparent. Thus, one study of the early 2000s concluded that, 'although there is evidence of significant structural change in the forums for engagement flowing from devolution, there has been less to suggest significant policy divergence' (Alcock 2009: 9). In contrast, we have been concerned to explore the new formative processes which now shape third sector policy. This has explanatory power to complement traditional analysis of policy outputs, and, as the current discourse data show, reveal the ongoing territorialization of social policy following the redrawing of political boundaries in the union state. The latter is a function of the fact that the interlocutors in the four post-1998/99 state policy-making systems in Scotland, Wales, Northern Ireland and England are markedly different in number, political complexion and influence to those found at Westminster before 1998. Crucially, 'postdevolution' policy-making is grounded in starkly contrasting notions of identity, culture, as well as political and constitutional ambitions for the nature and functioning of the modern state (Chaney 2013c). The present methodology reveals this, yet it also underlines that the changes affecting the third sector in the UK are more than just structural in nature; they are also about the situated internal party politics of state-wide parties as they seek to manage tensions between their UK role and presence in the devolved nations. This was notable, for example, in the discourse on the role of the private sector and market-based approaches to welfare (with strong continuity between Scottish Labour and 'New' Labour at Westminster, yet clear contrasts with Welsh Labour). Accordingly, this study's process-oriented approach underlines the complex and contingent ways in which the territorial politics of devolution shapes the third sector's welfare role. The rise of'sub-state' welfare pluralism also needs to be viewed within the context of the third sector's importance in the current economic climate. The latter is a product of the development of state-third sector relations over the past two centuries. A period which spans 'the development of the "new philanthropy" in the late 19th and early 20th centuries [and...] the expansion of state welfare provision after 1945' (Harris 2010: 25;Wincott 2011Wincott, 2012)). Over the past century, governments from both Right and Left have advocated co-working with the sector, albeit in contrasting ways and with varying motives. The post-2008 global recession has added fresh impetus to this agenda. As Wilding (2010: 97) cogently observes, it 'has shaken the confidence of and prospects for the UK voluntary and community sector... Looking forward, managing relationships with government in a period of substantive public expenditure cuts is likely to be the biggest test of the sector's ability to survive and even thrive, in a recession'. Indeed, a growing body of empirical work reveals the impact of austerity and the ways in which financial constraints are limiting organizational effectiveness. According to Brown et al. (2013: 58), this underlines 'a contradiction between the current [UK] coalition government's "Big Society" ideas 16 and the reality as it unfolds in... the third sector'. The present, therefore, marks a critical juncture: if ever there was momentum to roll back the welfare state, it is the (aftermath) of the financial crisis of 2008-09. All theoretical perspectives within comparative welfare state research predict radical reform in this circumstance... budgetary constraints are forcing political actors to make tough choices and introduce austerity policies. As a result, the question of who pays what, when, and how will likely give rise to increasingly sharp distributional conflicts (Vis et al. 2011: 338) Secondary data underline this point and explain why parties from across the political spectrum view the third sector as an appealing option to make good any shortfall in welfare provision arising from public sector cuts (notwithstanding sharp divisions on how this should be operationalized). They reveal that the efforts of the 10.6 million people in the UK who volunteer once a month contributes the equivalent to the work of 1.3 million full-time employees at a potential cost to the state of £23.1/US$35.9 billion (based on median hourly wage) (NCVO 2011: 21). Thus the present study underlines how the contrasting policy frameworks in the four UK polities continue to be tempered by the recession. Prior to 2007/08, the devolved governments' policy discourse on mixed economy approaches to welfare was often driven by more expansive visions of welfare (compared to Westminster). Post-recession, they are increasingly shaped by the administrations' differing responses to austerity, thereby giving added impetus to policy divergence. This economic dimension to welfare pluralism is certain to accelerate in future years owing to: 1. the greater size (compared to England) of the public sector(s) in the devolved nations (with inherent vulnerability to downsizing as part of ongoing austerity measures); and 2. the recent devolution of significant taxation and borrowing powers to Scotland (with Wales likely to follow). The latter transition to a quasi-federal taxation system post-dates the current dataset. Yet it will mean that meso-governments' embracing of third sector welfare delivery will no longer be informed by the general necessities of block grant transfers from the UK Treasury. Instead, it will be shaped by a new economic imperative as devolved administrations take on the political risks associated with raising direct taxation, determining the level of welfare spending 17 and, crucially, its allocation between sectors. Overall, the foregoing analysis reveals the key significance of devolution to the third sector. The new territorial politics associated with the creation of four distinct political systems in the UK means that policy is now shaped by the contrasting ways that parties in the constituent polities envision the sector's role. This is (re-)defining the nature of contemporary welfare and driving third sector policy divergence within the union state. Sub-state electoral politics and ongoing constitutional change mean that, in the new millennium, the direction of travel appears to be firmly away from predominantly state-wide policymaking which characterized social welfare for much of the 20th century. 1. 'Union state' refers to the United Kingdom -a state founded on the union of England and Wales, followed by unions with Scotland and Ireland (subsequently, post-1920, Northern Ireland). 2. The territorialization of policy is a spatial process whereby, compared to past practices, policy decisions are made at a more local level and tailored to local circumstances (see e.g. Chaney 2012b). 3. Adobe Reader XI. 4. For example, the single statement, 'our approach will amend the schools curriculum to encourage young people to undertake voluntary work in order to help others and benefit the area in which they live' consists of two quasisentences: (a) 'encourage young people to undertake voluntary work', a quasisentence coded under the 'volunteering' frame; and (b) 'benefit the areas in which they live', a quasi-sentence coded under the 'community development frame'. 5. It is an inexact match because devolution in Northern Ireland pre-dated devolution elsewhere by a year (and has been characterized by the re-imposition of periods of direct rule). Moreover, 2012 was added to the last period in order to capture the latest policy documents. 6. F-test, df = 3, F = 0.93. 7. <unk> 2 = 2704.213, df = 27, P = <unk>0.001; F-test df = 1, F = 6.54, P = 0.015489. 8. <unk> 2 = 362.573, df = 27, P = <unk>0.001. 9. <unk> 2 = 477.571, df = 3, P = <unk>0.001. 10. An agenda axiomatically to devolve decision-making to local communities, given legislative effect in the Localism Act 2011. 11. The Scottish case is explained by internal Labour politics and a close alignment between the Edinburgh and Westminster leaderships (see Laffin et al. 2007). This was initially true of Wales, yet superseded by a political determination to be seen as different to New Labour following the departure of Alun Michael AM as Welsh First Minister (see Flynn 1999). 12. It should be noted that this policy document seems to be using a rather unusual definition of what is in the 'third sector' and also that this very confusion in the text is itself an important finding from the data. It reflects the policy standpoint which says that what is being sought is any alternative to state/ governmental service provision and there is no particular preference for the third sector. 13. 'Clear red water' is the term used by the then Welsh First Minster, Rhodri Morgan AM, in a keynote speech in 2002 to signal the difference in approach to welfare between the Welsh Labour Party and Tony Blair's New Labour Party in England. The speech is available at http://www.sochealth.co.uk/the-socialist -health-association/sha-country-and-branch-organisation/sha-wales/clear-red -water/ (accessed 31 October 2013). 14. http://www.scotland.gov.uk/News/Releases/2013/03/benefit-cuts-24032013# (accessed 26 July 2013). 15. Northern Ireland Assembly Official Report, 25 November 2008. 16. This is a narrative in UK Coalition government policies. It is variously concerned with redefining the role of the market, voluntary sector and state and addressing anti-social behaviour ('Broken Britain'), etc. For a discussion, see Smith 2010. 17. Excluding, for the present, income maintenance benefits or'social security' (apart from limited competence in the case of Northern Ireland).

Welfare state theory has struggled to come to terms with the role of the third sector. It has often categorized welfare states in terms of the pattern of interplay between state social policies and the structure of the labour market. Moreover, it has frequently offered an exclusive focus on state policy -thereby failing to substantially recognize the role of the formally organized third sector. This study offers a corrective view. Against the backdrop of the international shift to multi-level governance, it analyses the policy discourse of third sector involvement in welfare governance following devolution in the UK. It reveals the changing and contrasting ways in which post-devolution territorial politics envisions the sector's role as a welfare provider. The mixed methods analysis compares policy framing and the structural narratives associated with the development of the third sector across the four constituent polities of the UK since 1998. The findings reveal how devolution has introduced a new spatial policy dynamic. Whilst there are elements of continuity between polities -such as the increasing salience of the third sector in welfare provision -policy narratives also provide evidence of the territorialization of third sector policy. From a methodological standpoint, this underlines the distinctive and complementary role discourse-based analysis can play in understanding contemporary patterns and processes shaping welfare governance.

Introduction The global youth unemployment rate is 14.9%, whereas in Africa, the rate is 12.7% (Statistics South Africa, 2022). South Africa is currently facing a challenge, especially youth unemployment. Despite a drop of 34.5% in the first quarter of 2022 from 35.3%, challenges still exist (Statistics South Africa, 2021b). Statistics reveal that 31% of graduates under the age of 24 are not working in South Africa (Statistics South Africa, 2021a). The majority of these are graduates (Bangani, 2019). Studies reveal some reasons for this high state of unemployment. Saliently, these can be reduced to (1) a lack of skills that enhance employability; (2) the qualification type attained, including the field of study; (3) the superiority of the tertiary education acquired based on the institution; (4) higher prospects from graduates; (5) a lack of knowledge about the employment hunt and limited work experience (Chiwara et al., 2017;Harry et al., 2018;Mseleku, 2019). Proviso graduate joblessness rates continue growing, resulting in continued idleness, poverty and frustration among the graduates (Graham & De Lannoy, 2016). This is exacerbated by the growing number of graduates exiting the tertiary sector (Harry & Chinyamurindi, 2020;Rogan & Reynolds, 2016). There is a need for strategies that enhance the integration of the unemployed, especially graduates, into the labour market (Harry & Chinyamurindi, 2019). This can be in the form of tools such as technology that remain popular, especially among young people in countries like South Africa. There is a noted growing popularity of smartphone usage amongst students (Shava et al., 2016). Popular sites amongst students include Facebook, LinkedIn and Twitter, all for communication purposes (Cilliers et al., 2017;Shava et al., 2016). Changqing et al. (2016) pointed out that graduates are the most active mobile phone operators as they utilise text messaging and the Internet. Robinson (2011) proclaimed that with the receipt of mobile communication among young adults, there is a perception of social media use to increase youth employability in South Africa. Technology is a practical tool to help youth to search for employment and thereby increase the employability of graduates from tertiary education in South Africa (Albal, 2018;Benson et al., 2014). Technology plays a pivotal role in displaying skills required by employers and thus has the potential to increase graduate transition into the labour markets (Benson et al., 2014;Tanser, 2017). Social media has been identified as a solution to employment challenges faced by graduates. Students actively use social media for personal reasons (Beavon, 2019;Garcia, 2011). Joosten (2012, p. 4) defined social media as 'different methods of electronic communication employed by (employers; recruitment companies as well as graduates) to build online communities so that information, ideas as well as other content can be shared'. Changqing et al. (2016, p. 3) described social media as 'websites and applications that allow endusers (graduates, managers and people) to make a profile, facilitating collaboration and sharing of information in the virtual social world'. Tess (2013) added that social media transforms how youth converse and cooperate. Conversely, there are concerns associated with social media use by graduates. These take in online security concerns, the cost of data and the social media experience. Therefore, graduates have scant knowledge necessary to effectively use social media as an employability tool (Albal, 2018). This study aimed to establish the role of social media in addressing the employability challenges graduates face. The article discusses the literature review and is then followed by a discussion of the methodology applied in the study. This is followed by a section which presents the study results and after that, a discussion of the findings. Lastly, the paper presents the study's limitations and the conclusion. The following section presents the underpinning literature. --- Literature review --- Theoretical lens The study has employed two theoretical frameworks. Firstly, the CareerEDGE framework defines the concepts that influence the employability of the youth cohort. The power of CareerEDGE rests on its effortlessness because it can be explained to graduates with ease (Dacre Pool & Qualter, 2013). On the other hand, this strength could also be observed as a flaw in that it does not help in clarifying to nonexperts in the field, mainly the youth themselves and their parents, precisely what employability means. The CareerEDGE model offers information on what should be considered when developing programmes and interventions on employability. Moreover, the CareerEDGE model is critical in knowledge transfer activities among the youth cohort (Dacre Pool, 2020). The CareerEDGE model has been used in several articles on employability within a South African context (Coetzee & Esterhuizen, 2010;Harry et al., 2018). A second theoretical model is the social media integration model (SMIM). Garcia (2011) alluded that social media integration is an interactive process. It facilitates information exchange among employees and graduates. It creates longlasting communication and enhances graduate-employer interaction. The SMIM has been used in various studies (Rou<unk>kova, 2015;Valos et al., 2016;Webb & Roberts, 2016). The SMIM was used in the business setting to examine how small organisations are communicating externally with their customers and clients, using social media to connect with their stakeholders and updating their social media platforms to ensure connectivity, sharing and exchange, exposure and feedback through activities such as blogging, social networking, social bookmarking and social recommendations and reviews. The following section discusses the empirical literature. --- Empirical literature --- Graduate challenges Today's graduates lack job-search skills. This lack of knowledge about the job search among graduates may hinder employment opportunities (Nassiriyar et al., 2015). Molek (2010, p. 93) defines job search as a 'normal occurrence which Job seekers use as a form of investment to enhance their position in the labour market'. The job search needs time and determination to achieve an objective, which in this case is being successful in attaining employment. The advent of technology has affected every sector of society, together with the process of looking for employment (Piveca & Ma<unk>ek, 2019). Graduates can now use social media and job sites to look for job opportunities (Becton et al., 2019). Social media has also made it easier for youth seeking jobs to put themselves out there, displaying their Curriculum vitae and skills. With Internet connectivity and a mobile device, youths can look for employment wherever in the world, from anywhere in the world (Ibrahim et al., 2016). New graduates seeking employment opportunities need to gain the competence in skills, capabilities and experience needed to find job opportunities and settle themselves in the workplace (Lord et al., 2019). In most cases, graduates still need to make it past the interview stage. Many graduates need to gain interview and technical skills necessary to get employment (Krishnan et al., 2017). Steyn et al. (2014) add that communication skill is critical for a graduate to pass the interview stage. Similarly, Krishnan et al. (2017) found that many youths lack a good command of English and find it challenging to converse effectually with potential employers http://www.sajhrm.co.za Open Access in interviews. Graduates could use webinars to help them prepare for the interviews, increasing their chances of being hired (Abdu & Siudikiene, 2016;Albal, 2018). Graduates today also find it challenging to be employed; they need to gain knowledge of applying principles of what they have learned at tertiary institutions. Hooley (2017) stated that university youth find it challenging to convert the theoretical knowledge they have learnt into practical ones while in the workplace. Another factor is unrealistic expectations from graduates. Graduates think obtaining the top qualifications will guarantee high incomes and management positions (Shyju & Chandra, 2018). In the same way, Kalufya and Mwakajinga (2016) stated that graduates assume they begin with high incomes; conversely, some do not have any experience and are never exposed to experiential learning by their programmes. Graduates find it challenging to commence from a lower position with such expectations. Consequently, some decide to stay unemployed, awaiting their dream occupation (Krishnan et al., 2017). If unrealistic expectations among youth persevere, it will additionally escalate youth joblessness. Graduates must have an employability aptitude and must cope with emerging challenges arising every time to endure their work environment after attaining their employment opportunities (Ismail, 2017). Kalufya and Mwakajinga (2016) pointed out that the youth should show their graduate employability capacities, including combining personal qualities, understanding and practices and applying these skills at their workplaces. --- Social media and employment Emerging technologies for employment are novel in South Africa (Maltby, 2011). Social media started being used around early 2000. Initially, it was used to carry out a background check of employees. A growing number of independent companies are emerging to assist graduates in connecting to employers (Maltby, 2011). Through such connections, young people are often trained in how they can acquire the necessary employability skills. This can also include using emerging technologies to increase employability (Careerbuilder, 2011). The labour market is evolving, demanding that graduates be technologically literate in industry environments (Abdu & Siudikiene, 2016). The learning and teaching methodologies at tertiary institutions should evolve and adapt to offer learners technical and social media skills. Social media is evolving how individuals, as well as firms, collaborate and function. Graduates must know how to use social media to find employment opportunities and shape their online presence (Koch et al., 2018). Through social media, graduates can express the skills they have mastered in tertiary education with ease. Social media has progressively changed from a tool for individual activities to a tool for professional activities. Benson and Morgan (2013) stated that emerging technologies are part of daily life for everyone. Further, social media provides a plinth for self-presentation and self-disclosure (Kaplan & Haenlein, 2010). Research studies by Benson et al. (2013) and Albal (2018) found that young graduates use social media for employability. Additionally, Albal (2018) stated that social media tools provide a platform for graduates to effectively access extensive information about organisations of their interest. Social media allows graduates to find helpful information about their potential employers (Benson et al., 2014). Social media (for example, a blog) permits graduates to contribute to an online discussion (Kaplan & Haenlein, 2010). Graduates could display their critical thinking and decision-making skills by participating in the debate. Employers could use social media to respond to graduate students who have displayed their skills and talents in the company and discussion forums (Ibrahim et al., 2016). For that reason, social media offers structures that stimulate graduateemployer interaction. Fletcher ( 2016) stated that social media has to become prominent, with employers utilising social media to carry out background checks of potential personnel and conduct talent searches. Suen (2018) stated that graduates must behave well when using social media, as it could affect their online presence. In developed states, more than 75% of employers do background checks on the social profiles of potential employees during the recruitment process. However, graduates' profanity and poor selling on social media decrease their probability of being hired (Salm, 2017). Krishnan et al. (2017) revealed similar results. Therefore, graduates should make use of social media to sell their talents. Koch et al. (2018) look at the present state of the hiring processes in South Africa. The study established that employment agencies in South Africa are following their global counterparts. LinkedIn was found to be the primary platform for their recruitment processes. Some higher education institutions have started designing special modules and training to assist learners in developing a professional online identity (Rienties et al., 2012). The University of Central Lancashire in the United Kingdom developed a social media initiative called 'Twitter Critters', used by the learner to engage with employers by sharing their work sketches and receiving expert feedback and advice (Swan, 2012). Emerging technologies such as LinkedIn and Facebook have been recognised for their valuable role in assisting graduates in obtaining employment. Nevertheless, the youth residing in parts of the country with low Internet access rates and those from rural backgrounds sometimes need to be aware of these priceless resources. Thus, it becomes an obligation for policymakers in the education sector to spread awareness about these resources. Awareness could be achieved by http://www.sajhrm.co.za Open Access including enough information about social media in their curriculum and teaching to educate students on the best practices to successfully use social media to acquire the skills needed to get their first job (Mashigo, 2014). --- Research methodology This article adopted an interpretivist research philosophy. A qualitative research method was employed (Brynard & Hanekom, 2006) to establish social media tools' role in addressing employability challenges faced by graduates. This technique was selected because it allows gathering data directly linked to subjective experiences; thus, it is investigative in its application (Flick, 2011). A qualitative method was considered suitable as it provides valuable insights into gaining an in-depth understanding of employability challenges faced by graduates. The study employed online interviews that fall under interviews. The study was conducted during the coronavirus disease 2019 (COVID-19) period when the authors needed to practise social distancing. --- Sampling and research participants The article used a nonprobability sample method and a purposive sampling procedure. The research respondents were considered 'information-rich' and could offer helpful information regarding the problem under investigation. The objective was to select participants with characteristics required in the sample, that is, professionals within the human resources (HR) field. Participants were identified through known contacts and the following roles in the organisation: (1) deputy director of human resources, (2) HR officer or administrator and (3) HR practitioner. Fifteen HR practitioners participated in this study. These HR practitioners were from the Eastern Cape region of South Africa. The justification for selecting the populace in the Eastern Cape was to reduce the study cost. Seven were male, and six participants had more than 8 years of experience in the HR field. The data were collected over 2 months. The study applied inclusion criteria to select its participants: a participant should have at least 3 years of practice in their discipline area to take part in the study and also had to be an expert in any field of HR. This ensured dependability in the information collected from well-experienced participants. --- Data collection and data recording The study was conducted during the COVID-19 pandemic; the authors used any convenient method to gather the data. The researchers were required to observe social distancing and could not see participants in person. Therefore, the researchers contacted participants via telephone and e-mail with a link to a qualitative survey on Google Forms (Alphabet, Inc., Mountain View, California, United States). The study objectives and research question made using a qualitative research approach possible. The study used structured interviews, which require that all the research participants answer predetermined questions in the same order. --- Strategies to ensure data integrity The researchers paid attention to confirmability, transferability, credibility and dependability to ensure the quality of qualitative data (Korstjens & Moser, 2018). Research participants participated online; their responses were recorded over Google Forms and enabled each participant's recording (with permission). Human resource practitioners who participated in the study were involved to ensure the study's credibility. Credibility was achieved as the data were collected through the actual HR practitioners. The trust was built based on the professionalism between the parties. The study focused on understanding labour market behaviours and ascertaining the role that social media plays, which allowed the researchers to gather opinions of the HR practitioners concerning the study. --- Data analysis The data were analysed through a thematic narrative analysis (Long, 2014). Thematic narrative analysis was employed to understand HR practitioners' opinions on the social media role in addressing employability challenges graduates face. The thematic analysis allowed the researchers to identify and examine shared patterns within the data. The researchers extracted themes and subthemes from the collected data. Gibbs (2007) pointed out that extracting themes and subthemes from the participant's responses is essential as it refines interpretations of the interviews and then encapsulates and integrates the gathered data. Through thematic analysis, the scholars understood participants' understandings through individual analysis, followed by evaluations of the participants' experiences to generate themes (Neuendorf, 2019). The themes were generated using the six steps advocated by Braun et al. --- Ethical considerations This study guaranteed participants' confidentiality, anonymity and privacy (Ferreira, 2018). The researchers observed all ethical considerations, including informed consent (Saunders & Lewis, 2018). In this study, participants had the right to pull out from participating at any time (Neuman, 2014). The study observed ethical guidelines. Ethical clearance was obtained from the participating university (ref. no. CIL021SMUR01). --- Findings This study aimed to establish a social media role in addressing employability challenges graduates face after completing their tertiary education. Two primary outcomes emerged from the research study. Firstly, a need exists for unemployed graduates to use social media as a marketing utility for their skills and abilities. Secondly, social media tool usefulness as an information utility to enhance awareness about labour market issues. Table 3 illustrates themes and subthemes. These results are presented next. --- Theme 1: Social media playing a utility role in marketing the prospective candidate The participating HR practitioners expressed that social media plays a utility role in marketing the prospective candidate. Two subthemes emerged: firstly, social media is helpful in applicant screening, and secondly, it increases and broadens the number of applicants. Emerging technologies and media have transformed the way companies connect with graduates. Companies investigate prospective employees' social media accounts before they hire them. A participant pointed out that social media can be used to identify'red flags' as they disclose the nature and personality of potential personnel. Youth seeking job opportunities must behave in this digital industry when using social media, as it influences their online profile and employability. Participant 3 noted the following: 'Social networks should not be a qualifying factor. They should be handy in picking out possible red flags (such as racism and extremist behaviour).' (Male) Additionally, Participant 11 also noted that: 'Information on social media is not always accurate; some graduates forget to update their profile. Therefore, employers do not need to depend on the information found on a graduate's profile. Graduates must be careful about what they post and comment on social media as it affects their employability.' (Female) Participant 14 cautions especially the younger generation regarding the content they post on Facebook and Twitter, as this can affect their probabilities in a competitive labour market: 'Some people can lie about their information, and some do not update it. HR cannot rely on that information. People need to be careful of what they post and comment on.' (Female) Social media helps spread information between youth concerning job opportunities or internships available. Consequently, Subtheme 2 (social media usefulness as an information utility to enhance awareness about labour market issues) is convenient for attracting new talents through sharing information about employment opportunities available transversely over geographic borders and broadening the pool of applicants. Participant 1 had this to say: 'Social media platforms are only relevant for attraction purposes only. However, other administrative requirements of the --- Phase Description --- Familiarising yourself with your data Actively reading and rereading data to obtain an overall understanding. --- General initial codes Noting important aspects of data. --- Searching for themes Identify codes and form codes into themes. --- Reviewing themes Relating the themes to codes and the entire data set. 5. Defining and naming themes Producing clear definitions and names for themes. 6. Producing the report Final analysis of themes relating to the initial research question. Source: Braun, V., Clarke, V., Hayfield, N., & Terry, G. ( 2019). Thematic analysis. Springer, New York TABLE 3: Social media playing a utility role of marketing the prospective candidate: Illustrating quotes. --- Social media is helpful in candidate selection Social media enhances and widens the pool of applicants 'Social media is public, so private thoughts shouldn't be aired on it. If it is, then it is fair to use this in verifying the candidate.' (Participant 2, male) 'It is fair as we are living in technology infested world.' (Participant 5, male) 'Social media is a platform to winddown for a number of people and so it may indeed be unfair to use comments therein to pass judgment on individual character.' (Participant 7, male) 'It is important in that nowadays, graduates need to understand the importance of using online resources including social media platforms for recruitment and selection purposes as these are becoming more and more popular and relevant today.' (Participant 1, male) 'Social media should be explored to a certain degree as this reveals a lot about the nature and character of the individual which they are happy to display to the public. Their consent is having posted it on a public platform, so anything they post there, they should be comfortable with knowing that a prospective or current employer will be able to view it.' (Participant 4, male) 'I think social media can be used to post available posts. One can also teach the graduates how to prepare for field work.' (Participant 8, female) 'I believe people write to entertain the reader and basically to start a healthy argument with others or the personal audience in social media. To use then that to assess my ability to work in your team is a little bit unfair because you really do not know how far i have evolved from the day of the post.' (Participant 9, female) 'Yes, social media can also be used as a platform of advertising as it can attract pool of applicants.' (Participant 13, female) 'Social media should be used for its advantages in the employment and recruitment industry. Graduates should be warned about the kind of stuff they should and should not post on social media and to ensure that their online media presence is closely monitored and updated so that they are aware of what "others" can see about them. Social media allows for quick and easy judgement-making, so it would be wise for graduates to ensure that it (their social media presence, profile and image) is a favourable one, as far as possible.' (Participant 10, female) When using social media, individuals can associate universally in the world if they have access to the Internet. Through social media, information reaches the graduates promptly and at a very inexpensive cost. Participant 5 narrated as follows: 'Information can reach the globe on time as there will be no geographical boundaries.' (Male) Table 4 illustrates additional quotes from the interviews supporting Theme 1 of the study. This leads to the study's second finding, and social media was noted (in addition to the marketing utility identified) as playing a related information utility role in enhancing awareness about labour market issues. These results are presented next. --- Theme 2: Social media's information role in enhancing graduate employability through enhancing information awareness A finding of the study reviewed that social media helps increase youth employability through enhancing information awareness. The subthemes that emerged were that social media imparts youth with employment opportunity information, inspires social connectedness on matters of mutual interest and helps with graduate skills development. Participant 4 pointed out: 'It would be important for the graduate to learn how to use social media for their benefit in their employability journey -how to search for job opportunities, advertise their skills and profiles on social media for prospective employers to view etc. Social media allows for several technological innovations to help graduates get their "name" out in the world to a larger scope of prospective employers and to help showcase their portfolio if they have developed something like this.' (Male) Emerging technologies, for instance, social media, could assist youth in creating relations with companies and others with whom they share a related interest. Youth ought to use LinkedIn, to generate strong networks with employers. Participant 15 had to say the following: 'Very useful to highlight relevant skills and network to possible employers. Makes young applicants more accountable for what is posted, if possible employers are watching.' (Male) Social media provides an opportunity for graduates to learn. Graduate skills may be established if graduates go through stories posted on social media related to their line of interest. These stories might elucidate how an individual develops his or her skills and becomes prosperous in obtaining a job opportunity. Participant 9 said the following: 'We all need social media for relevance; we need to learn online. The employability may improve with social media, especially if you read the stories on links related to your field of interest.' (Female) Furthermore, social media assists in creating social communities for graduates that enhance information access and provide an opportunity for youth to create networks with employers. Participant 6 had to say: 'It can be a good platform for good networks, and others become entrepreneurs and open businesses through experiences shared via social media.' (Female) A social media platform allows an individual to post employment opportunities and share them with youth seeking employment. Corporations may utilise emerging technologies to broadcast employment prospects available. Additionally, Participant 2 expressed the following: 'The best use is to ensure that jobs are posted and shared on the platform to potential graduates to search and apply for them. With that, a campaign to educate and advise graduates on the skills that are lacking both in graduates and in the job market could mitigate some unemployment and possibly drive graduates to further skills themselves on what is highlighted by the campaign.' (Male) Social media is a communication medium used by numerous businesses to share information regarding their prospects from potential employees. As a result, graduates must follow businesses' social media platforms to receive updates from future employers concerning employment. Participant 3 noted this view: --- Discussion of results This research study intended to establish social media's role in addressing employability challenges graduates face after completing their tertiary education. Social media transformed the way companies connect with potential employees. The study revealed that social media has become prominent, and HR departments within organisations utilise social media to carry out personnel background checks before they invite the individual for an interview and to conduct talent searches. Graduates must behave well when using social media because it could influence their online identity. Graduates' profanity on social media decreases their employment probability (Salm, 2017). Krishnan et al. (2017) revealed similar results. Therefore, graduates must utilise social media for their benefit, such as selling skills and talents, because employers are searching for them online. The results replicate comparable findings from the studies carried out by Hussain et al. (2017). Their findings pointed out the need to educate graduates on how to behave in the digital world. Furthermore, the results pointed out that policies ought to be available and govern how recruitment companies may integrate social media into recruitment as well as the selection of potential employees (Wirthman, 2016). Through social media, job information is available to graduates. These findings are comparable to the current study from Becton et al. (2019), who point out that organisations' HR departments use social media tools to attract or lure new talents because job opportunity information can reach the audience (graduates) in no time. Therefore, this study proposes that youth ought to acquire employment-hunting skills when using social media and market their graduate skills on social media (Sangeeta & Ahlawat, 2018). (Albal, 2018). Social media enables graduates to create a social community that youth may use for sharing available employment opportunities (Benson et al., 2013). The study's findings corroborate previous studies that presented ways social media could help increase youth's employability (Albal, 2018;Piveca & Ma<unk>ek, 2019;Shyju & Chandra, 2018). In addition, the study's results reveal similar findings from several researchers (Abdu & Siudikiene, 2016;Garcia, 2011;Hussain et al., 2017). Emerging technologies are a critical tool that could project individuals' positive, more constructive side. Graduates should use social media to develop employer-employee relations, construct their online identities and display their skills. --- Implications Some implications can be drawn from this study. The findings offer insights into social media roles in addressing graduates' employability challenges. This study adds to the discussion of social media's role in employability in modern societies. The study has revealed social media's potential to enhance the employability of graduates. Further, the study adds to existing knowledge on social media for employability enhancement. Higher education institutions have to present knowledge regarding the prominence of social media to graduates. This could be through sharing stories on how youth must shape their online identity, as most employers conduct a background check on the applicant on social media sites. From a law-making perspective, policymakers (legislators) in the Department of Education must find ways to implant social media in higher education programmes and teaching to teach graduates best practices to employ social media for employment purposes effectively. The finding that social media aids the employability of graduates has significant implications for HR practitioner practices. From a HR perspective, policies that underline HR practitioners' restrictions when selecting new talent should be in place. Finally, this study sheds more light on the role social media plays towards the employability of exiting students. Overall, graduates must utilise social media for their professional development. --- Limitations of the study and directions for future research The purpose of the study was to establish social media's role in addressing employability challenges graduates face. Human resource practitioners in the Eastern Cape province of South Africa were used as a sample; therefore, the results generally relate to this setting. Nevertheless, research may use a quantitative approach and examine all the HR practitioners in South Africa. Moreover, the study should be extended to Southern African Development Community (SADC) universities in the future. --- Conclusion The receipt of smartphones among the youth has led to the use of social media, such as Twitter and Facebook, to gain access to job opportunity information on these platforms in South Africa. The study's purpose was to examine the social media tools' role in addressing employability challenges graduates face after completing their tertiary education in South Africa. Two primary outcomes emerged from the research study informed by the thematic analysis conducted where a need exists for unemployed graduates to use social media as a marketing utility of their skills and abilities. Social media can also be used as an information utility to enhance awareness about labour market issues. The contribution of the study is a framework with theory-practitioner and policy ramifications. The study establishes an insight into youth employability challenges and provides reasons for what causes them. --- Data availability The data used were obtained via interviews; the data are represented in the manuscript in the presented tables. --- Competing interests The author(s) declare that they have no financial or personal relationship(s) that may have inappropriately influenced them in writing this article. --- Authors' contributions --- Disclaimer The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any affiliated agency of the authors.

Orientation: South Africa is currently going through a graduate unemployment challenge. Calls exist for research that addresses this challenge. Research purpose: This research study aimed to establish social media tools' role in addressing employability challenges faced by graduates after completing their tertiary education. Motivation for the study: Calls exist in the literature for nuanced ways of addressing challenges youths face after completing their studies. The role of social media in this remains an area of inquiry with noted importance.The study utilised an interpretivist philosophy relying on a qualitative research approach using interview data with a sample of 15 experts within the human resources field. Thematic analysis was used as a means of data analysis. Main findings: Two primary outcomes emerged from the research study. Firstly, a need exists for unemployed graduates to use social media as a marketing utility for their skills and abilities. Secondly, social media can also be used as an information utility to enhance awareness about labour market issues. Practical/managerial implications: The findings offer insights into addressing the employability challenges faced by graduates when seeking employment opportunities. The role of social media in mitigating graduate challenges is proposed. Through the findings, graduates have the potential to enhance their portfolios to improve their integration into the labour market.The study proffers insights into how social media helps increase the employability prospects of graduates entering the job market.

Factors Underlying Cross-Cultural Differences in Stigma towards Autism among College Students in Lebanon and the United States Stigma, originally defined by Goffman (1963, preface) as "an attribute that is deeply discrediting," reduces well-being among autistic people and their family members around the world by fostering self-doubt and reducing access to social support, mental health care, employment, and other opportunities (Dehnavi et al., 2011;Divan, Vajaratkar, Desai, Strik-Lievers, & Patel, 2012;Gray, 2002;Grinker and Cho, 2013;Grinker et al., 2011;Ha et al., 2014;Mak and Kwok, 2010;Robertson, 2010). The earliest emerging and largest body of research examining stigma towards autism focuses on the parents of autistic people, many of whom report experiencing "courtesy stigma" arising from their close association with autistic people (e.g., Farrugia, 2009;Gray, 2002;Mak & Kwok, 2010;Woodgate et al., 2008). More recently, autistic scholars have begun to highlight ways in which the deficit-oriented assumptions that are widespread in the research literature may contribute to stigma towards autism (e.g., Robertson, 2009) and researchers have begun to examine stigma experienced by autistic people themselves (e.g., Crane et al., 2018;Shtayermman, 2009). Emerging research suggests that stigma towards autism is associated with reduced mental health among autistic adults (Botha & Frost, 2018) and decreased help seeking among autistic college students, a disproportionate number of whom face co-morbid mental health issues (Jackson et al., 2018;McMorris et al., 2018). To better support the growing population of autistic college students, a rapidly expanding body of research has begun to examine stigma and associated attitudes towards autism in college students (e.g., Gardiner & Iarocci, 2013;Nevill & White, 2011;Sasson & Morrison, 2017;Tipton & Blacher, 2014;White, Hillier, Frye, Makrez, & 2016). Thus far, such research has focused primarily on college students in Westernized nations. However, the degree to which autism is stigmatized is believed to vary greatly across cultural contexts (e.g., Grinker et al., 2012;Tekola et al., 2016;Tilahun et al., 2017). Indeed, Goffman (1963) highlighted the relational and contextually situated nature of stigma by indicating that stigma does not reside in an attribute itself, but arises when others interpret the attribute as an "undesired difference from what we ["the normals"] had anticipated." Subsequent researchers have often emphasized the relative and collectively defined nature of stigma. For example, Stafford and Scott (1986, p. 80) defined stigma as "a characteristic of persons that is contrary to the norm of a social unit." This description of stigma as a deviation from collectively defined norms aligns with evidence that stigma towards mental health conditions more generally may be heightened in more collectivistic cultures wherein people are encouraged to subordinate their individual needs to the collective goals of their ingroups (e.g., family or tribe) in order to preserve social harmony (Abdullah & Brown, 2011;Griffiths et al., 2006;Papadopoulos et al., 2013). Indeed, the first studies to directly compare stigma towards autism across countries revealed that college students in Lebanon and Japan reported higher levels of stigma towards autism (as measured by greater desired social distance from a "person with autism1 ") than college students in the US (Obeid et al., 2015;Someki et al., 2018). Lebanon and Japan are often considered slightly more collectivistic than individualistic, although pronounced variation in cultural values within each country is also apparent (e.g., Ayyash-Abdo, 2001;Hofstede Insights, n.d.;Matsumoto, 2007). In contrast, the US is often considered a primarily individualistic culture, wherein people are encouraged to prioritize personal achievement and autonomy over conformity with group norms. Therefore, Obeid, Someki and colleagues interpreted their findings as evidence that stigma towards autism is heightened in more collectivistic cultural contexts. However, they did not assess the degree to which their participants actually endorsed collectivistic or individualistic cultural values. Thus, the primary aim of the current study was to identify cultural value orientations, and other individual differences, that contribute to differences in stigma (assessed as in prior cross-cultural comparisons using a social distance scale; Obeid et al., 2015;Someki et al., 2018) among college students in two multicultural countries: the US, a vast country (9,161,923 square kilometers) with substantial, but unequally distributed, supports for autistic people (Mandell et al., 2009), and Lebanon, a small (10,452 square kilometers) country in the Middle East with very limited autism resources (Daou, 2014). --- What Insights Can be Derived by Comparing Stigma in Lebanon and the US? Autism in Lebanon and other Arab countries has received very little research attention (Hussein & Taha, 2013). Government-based autism services are infrequently available in Lebanon and, when available, they are often of low quality (Daou, 2014). The insufficient autism services in Lebanon may contribute to lack of understanding about and heightened stigma towards autism. Indeed, Obeid and colleagues (2015) found that Lebanese college students expressed less accurate overall knowledge about autism and more stigma (i.e., desired social distance) than college students in the US. Nevertheless, Lebanese students were more likely to correctly indicate that autism is diagnosed more often in wealthier and more educated families living in countries where richer people have better access to healthcare. The US has higher levels of resources in general (the GDP of Lebanon in 2017 was $8,524 while the GDP of the US was $59,532; http://data.worldbank.org/indicator/NY.GDP.PCAP.CD), and autism resources in particular, than Lebanon. Although the US leads the world in the quantity of autism research produced (Office of Autism Research Coordination, 2012) and the US government is required to provide appropriate educational services to all public school children with disabilities, striking disparities in access to diagnosis and treatment for autism remain apparent in the US (Durkin et al., 2010). Not only do socioeconomic disparities contribute to deeply problematic inequalities in access to autism resources within and between countries (Divan et al., 2012;Grinker et al., 2011;Mirza, Tareen, Davidson, & Rahman, 2009), how people respond to inequalities may amplify the negative effects of inadequate resources by contributing to stigma. Indeed, conceptualizations of stigma increasingly highlight power inequalities as the necessary starting point for stigma. In an attempt to unify the proliferation of diverse ways of defining and measuring stigma that Goffman's research inspired, Link and Phelan (2001) defined stigma as a process that is rooted in and exacerbates power inequalities which consists of the following interrelated components: identifying and labeling differences, linking differences to negative stereotypes, and categorizing people possessing these differences as separate from the categorizers ("us" versus "them"), resulting in status loss and discrimination for those who are stigmatized. This definition suggests that an investigation of other cultural values besides collectivism, particularly how people respond to power inequalities, may be essential for understanding variations in stigma within and across cultures. --- Which Cultural Value Orientations Contribute to Stigma? The once myopic focus of cross-cultural psychology on individualism and collectivism as the primary dimension (or dimensions depending on one's perspective) to consider when examining crosscultural differences has been roundly critiqued (e.g., Voronov & Singer, 2002), with increasing emphasis placed on dynamic context-dependent relationships between varied cultural values and other factors (e.g., Schwartz, 2006;Taras, Steel, & Kirkman, 2012). For example, Triandis and Gelfand (1998) asserted that there are many different forms of individualism and collectivism that arise from variations in the degree to which horizontal (de-emphasizing hierarchy) or vertical (emphasizing hierarchy) social relationships are valued. They developed a scale to assess these cultural value orientations, the Culture Orientation Scale, which we used in the current study. The Culture Orientation Scale assesses four domains that may vary independently of one another (Triandis & Gelfand, 1998;Triandis, Chen, & Chan, 1998): vertical individualism (i.e., valuing competitive attempts toward the top of hierarchies), horizontal individualism (i.e., valuing doing one's own thing without comparing oneself to others), vertical collectivism (i.e., valuing being part of a collective [typically the family] where inequality is accepted), and horizontal collectivism (i.e., valuing cooperative union with one's in-groups). The cultural value verticality, or "acceptance of inequalities among people" (Singelis, Triandis, Bhawuk, & Gelfand, 1995, p. 259), can be measured on either the individual or cultural level and is conceptually similar to Hofstede's cultural level factor, power distance, or "norms establishing and rewarding some forms of inequality" (Singelis et al. 1995, p. 269). The most widely known research assessing power distance and individualism on the cultural level is Hofstede's (1983) work wherein he compared the average degree to which "people in a given country" (i.e., a sample of IBM workers who filled out his survey in that country) endorsed specific values without reference to variation in values between people in the country. In his work, each country was a data point and variation within people in the country was ignored. In research examining cultural values as individual difference variables, each person's response is a data point and variability between people is a focus of analyses (Kemmelmeier et al., 2013). Hofstede and McCrae (2004, p. 65) emphasized the importance of not confusing these levels of analysis: "individuals are to societies as trees are to forests; comparing trees is not comparing forests writ large." In order to generalize from individual difference measures of cultural values (e.g., individual responses to the Culture Orientation Scale) to broader cultural values, Triandis (1996) suggested identifying cultural values that 90% of participants from a culture endorse. It is important to note that cultural value orientations as we have conceptualized them in this paper are individual difference characteristics, or values expressed by individuals. This distinction is conceptually rather opaque, partially because researchers have been inconsistent in the degree to which they follow it. Although Triandis and Gelfand (1988) described individualism versus collectivism on the individual level as a personality dimension, wherein characteristics of individualism were labeled "idiocentrism" when assessed on the individual level and characteristics of collectivism were labeled "allocentrism" on the individual level, in subsequent work using the Culture Orientation Scale, they again refer to individualism and collectivism by their more commonly known cultural level labels (Triandis & Gelfand, 1998). To more clearly link our work to prior research, we follow them in their choice of terminology but highlight that individualism and collectivism and vertical and horizontal cultural orientations refer to individual level value orientations in this paper. Although prior research has not related the Culture Orientation Scale to stigma towards autism in particular, prior research suggests that vertical cultural values may contribute to stigma towards disabilities more generally. Rao and colleagues (2010) found that individualism, specifically vertical individualism (assessed using the Culture Orientation Scale), was associated with heightened stigma towards job candidates with mental or physical health conditions (i.e., psychosis, substance dependence, HIV, cancer), among employers in China and the US. Unexpectedly, Chinese employers reported heightened vertical individualism relative to US employers, which the authors attributed to the rise of capitalism. Similarly, Ditchman and colleagues (2017) found that vertical individualism was associated with more negative attitudes towards the sexuality of people with intellectual disabilities while horizontal individualism and horizontal collectivism were associated with more positive attitudes. These findings suggest that inconsistencies in observed associations between individualism, collectivism, and stigma may be attributable to the vertical (accepting inequality as a necessary part of social life) vs. horizontal (rejecting power differences in favor of equality) dimension of the Culture Orientation Scale rather than the individualism vs. collectivism dimension. Indeed, the first study to our knowledge to examine individualism vs. collectivism as a single dimension predicting stigma contained evidence that the vertical vs. horizontal dimension is important to consider. Papadopoulos and colleagues ( 2013) conducted a survey with different ethnic groups in the UK which revealed that heightened individualism relative to collectivism (assessed with the Culture Orientation Scale) was associated with reduced stigma towards mental health conditions among people from American and Chinese backgrounds, but was not related to stigma among people from White-English or Greek backgrounds. They interpreted these findings as evidence that individualism vs. collectivism helps explain stigma towards mental health conditions among people from cultures where mental health conditions are either more (Chinese) or less (American) stigmatized but not in cultures with intermediate levels of stigma (White-English or Greek). Although they did not include the vertical vs. horizontal dimension of the Culture Orientation Scale in analyses predicting stigma from individualism vs. collectivism, they provided a graph depicting vertical and horizontal dimensions of individualism vs. collectivism separately for each group. This graph revealed that participants from both American and Chinese backgrounds reported pronounced vertical relative to horizontal cultural tendencies; American heritage participants' vertical tendencies were highly individualistic while Chinese heritage participants' vertical tendencies were highly collectivistic. In contrast, participants from White-English backgrounds reported heightened horizontal relative to vertical orientations and Greek heritage participants reported similar levels of vertical and horizontal tendencies. This pattern suggests that the vertical vs. horizontal dimension may be associated with stigma. To the best of our knowledge, no prior research has examined the independent contributions of individualism vs. collectivism and vertical vs. horizontal cultural orientations to stigma. --- Do Other Personal Characteristics and Experiences Also Contribute to Stigma? In prior cross-cultural work, reduced stigma towards autism (assessed with a social distance scale) has been associated, albeit inconsistently, with the following participant characteristics: being female, heightened autism knowledge, reduced autistic traits, prior contact with an autistic person and being in a helping major, such as education, psychology, social work or nursing (Obeid et al., 2015;Someki et al., 2018). Although prior cross-cultural research did not assess associations between prior quality of contact with autism and stigma, quality of contact is more consistently associated with autism acceptance than quantity of contact (Gardiner & Iarocci, 2013). Indeed, contact theory states that highquality contact between groups can decrease biases (Allport, 1954). Greater openness to experience (i.e., the tendency to be drawn to novelty and think independently; McCrae & Costa, 1987) among students in some "helping majors," such as psychology, compared to students in professional majors (Vedel, 2016), may also contribute to differences in stigma. Openness to experience is a more consistent a predictor of mental health stigma than prior contact with people with mental health conditions (Brown, 2012). Openness to experience has also been directly associated with prejudice against gay men and lesbians (Cullen et al., 2002) and indirectly associated with generalized prejudice (racism, sexism, prejudice toward homosexual individuals, and individuals with mental disabilities) through right-wing authoritarianism (Ekehammar et al., 2004;Sibley & Duckitt, 2008), which in turn is associated with both individual and cultural differences in vertical cultural value orientations (Kemmelmeier et al., 2003). Reduced openness is also associated with heightened acceptance of inequality on a cultural level (Hofstede and McCrae, 2004). --- Research Hypotheses We hypothesized that lower quality of prior contact with autistic people (i.e. less enjoyable interactions with autistic people), less openness to experience and less trait emotional intelligence (a constellation of self-perceptions individuals have about their emotional abilities, including empathy, that facilitates prosocial behaviors; Mavroveli and Sanchez-Ruiz, 2011;Petrides et al., 2016) would be associated with heightened stigma towards autism and underlie cross-cultural differences in stigma. In light of evidence that stigma towards mental health conditions is heightened in collectivistic cultures (Abdullah & Brown, 2011;Griffiths et al., 2006;Papadopoulos et al., 2013) and evidence that stigma towards autism is heightened among college students in Japan and Lebanon relative to the US (Obeid et al., 2015;Someki et al, 2018), we considered the possibility that stigma towards autism might be positively associated with collectivism. However, other literature suggested that stigma toward autistic individuals may be reduced in potentially more collectivistic cultural contexts such as Morocco (Haldane & Crawford, 2010), the Navajo Nation (Kapp, 2011), and Nicaragua (Kim, 2012), and that a preference for social inequality may more closely relate to prejudice (Duckit, 2016;Ekehammar et al., 2004;Sibley & Duckitt, 2008). After discussing these conflicted patterns in the prior literature, the authors of this report were unable to agree upon a unified hypothesis concerning cultural values and stigma. Therefore, analyses of potential associations between cultural values and stigma should be regarded as exploratory. Given that social desirability may influence self-reported preferences for social distance and culture value orientations (Link et al., 2004;Triandis, Chen, & Chan, 1998) and was not assessed in prior research comparing social distance towards autism across cultures (Obeid et al., 2015;Someki et al., 2018), we included susceptibility to the social desirability bias in analyses. --- Method The current study is a replication and expansion on Obeid and colleagues' (2015) original crosscultural comparison of stigma toward autism among college students in Lebanon and the US. We initially obtained approval to conduct this study from the IRBs of two urban universities in the US and two urban universities in Lebanon. However, only one university in each country was able to provide academic credit for participation. Schools that could not provide credit were not able to recruit more than a handful of students, so analyses focus on the two universities that could provide credit for participation. --- Participants College students (N=1150) in Lebanon (n=611) and the US (n=539) were recruited via online subject pools in the psychology departments of each university. All participants completed an online consent form wherein they were provided with the authors' contact information should any concerns arise, followed by an online survey that included pre-tests (assessing autism knowledge and stigma), an autism training (Gillespie-Lynch et al. 2015;Obeid et al., 2015) and post-tests (identical to pre-tests), as well as assessments of participants' cultural value orientation (individualism-collectivism and verticalhorizontal dimensions), social desirability bias, openness to experience, trait emotional intelligence, autism symptoms, and demographics, including types and quality of contact with individuals with autism. Participants were also invited to share their feedback on the survey at the end of the survey. The findings described in this report focus on pre-test measures of autism stigma in relation to individual differences2. Seventy-four participants were excluded from analyses because they did not indicate that they were fluent in English when asked at the beginning of the survey "Are you fluent in English?" The final sample consisted of 556 students from Lebanon and 520 students from the US. --- Materials Autism Awareness Survey. This measure of autism knowledge (<unk>=.62 Lebanon; <unk>=.68 US) was adapted from an original scale created by Stone (1987). Participants were asked to rate 13 statements about autism using a 5-level Likert scale (i.e., -2 for strongly disagree to 2 for strongly agree; 7 items were reverse scored). See Gillespie-Lynch and colleagues (2015) for specific items; note that "People with autism have empathy" was replaced in the current study with "People with autism care about and feel the pain of those who are suffering" to align with research suggesting that cognitive, but not affective, empathy may be impaired in autism (Jones et al., 2010). The total autism knowledge score could range from -26 to 26 with a higher score indicating more accurate knowledge of autism. Autism Social Distance Scale. The measure of stigma used in the current study was an adapted version of the Social Distance Scale (Bogardus, 1933), a commonly used measure of stigma that typically exhibits good internal-consistency and evidence of validity (Link, Yang, Phelan, & Collins, 2004). We utilized a social distance scale to be able to relate our findings to prior cross-cultural autism stigma research which also used a social distance scale (e.g., Obeid et al., 2015;Someki et al., 2018). To ensure that random responding was not driving cross-cultural differences, we expanded the 6-item scale (with no reverse-scored items) used in prior work to include 11 items (5 of which were reverse scored; see Appendix A). Stigma scores could range from -22 to 22 with higher scores indicating heightened stigma. The internal-consistency of our adapted social distance scale remained good in the current sample (<unk>=.88 Lebanon; <unk>=.89 US). Given that we added additional items to a previously used scale, we also conducted a principal component analysis, which revealed one factor with an eigenvalue greater than 1.00. This factor, with an eigenvalue of 5.24, explained 48% of the variance in the scale. All 11 items had a loading of.60 or higher on this factor. (Triandis & Gelfand, 1998). This 16-item measure consists of four 4item subscales: vertical individualism (e.g., "Winning is everything."), vertical collectivism (e.g., "It is important to me that I respect the decisions made by my groups."), horizontal individualism (e.g., "I often do'my own thing.'"), and horizontal collectivism (e.g., "If a coworker gets a prize, I would feel proud."). --- Culture Orientation Scale Following Papadopoulos and colleagues (2013), we assessed the relative degree of individualism vs. collectivism by subtracting each individual's total collectivism score from their total individualism score. Similarly, we assessed the relative degree of vertical vs. horizontal orientation by subtracting each individual's total horizontal score from their total vertical score. Exploratory factor analysis of this measure in the current sample revealed that all items loaded on a single factor, except for items from the vertical individualism subscale, which comprised their own factor. However, the internal-consistency of the overall scale was acceptable (<unk>=.75 Lebanon; <unk>=.79 US). --- Demographic survey. Participants were asked to indicate their age, gender, major (classified as helping majors if education, nursing, occupational/physical therapy, psychology, or social work were reported), country of residence (US vs Lebanon), and personal relationships with autism (''Please select as many of the following types of relationships as you have had with people with autism spectrum disorders: yourself, your child, your parent, your sibling, your spouse, your extended family member, your friend, your coworker, your student, your fellow student, your acquaintance, or other.''). Participants were also asked to rate the quality of their prior contact with autism using 4 Likert scale items from Gardiner and Iarocci (2013;e.g., "In the past, my experiences with individuals with autism have been pleasant."). (Reynolds, 1982). This measure of social desirability bias (<unk>=.57 Lebanon; <unk>=.62 US) consists of 13 binary items (5 reverse scored). Higher scores indicate greater susceptibility to the social desirability bias. (Neo Five Factor Inventory-3;NEO FFI-3;McCrae and Costa, 2007). --- Short Marlow-Crowne Social Desirability Scale --- Openness to Experience Openness to experience (<unk>=.74 Lebanon; <unk>=.74 US) was assessed using a subscale of the NEO Five Factor Inventory-3 comprising 12 items (5 reverse scored). Higher scores indicate greater intellectual curiosity and openness to experience. (Eriksson et al., 2013). This 14-item measure (1 reverse-scored item) was designed to be a rapid screener for autism risk. Higher scores indicate more autistic symptoms (<unk>=.76 Lebanon; <unk>=.84 --- RAADS-14 --- US). Trait Emotional Intelligence Questionnaire (Petrides, 2009). This 30-item measure (15 reverse scored items; <unk>=.88 Lebanon; <unk>=.89 US) consists of four subscales (well-being, self-control, emotionality and sociability). Higher overall scores indicate heightened emotional intelligence. --- Analytic Approach We evaluated kurtosis (which was excessively high for one variable: age), skew and multicollinearity. The social distance scale used in the current study was far more normally distributed than a prior version of the scale that lacked reverse-scored items (Gillespie-Lynch et al., 2015). Exploratory correlations for each country separately can be viewed in Appendices B and C. For our primary analyses, we conducted independent sample t-tests (for continuous variables) and chi-square tests (for categorical variables) to examine differences between participants from Lebanon and the US. Then, we conducted a regression with gender and country as predictors to evaluate if Obeid and colleagues' (2015) finding that social distance scores were heightened in Lebanon relative to the US replicated in this sample. Next, we conducted a regression analysis to determine which variables predicted social distance scores when other potential predictor variables were accounted for (including individualism vs. collectivism and vertical vs. horizontal orientations). Given the controversy about whether cultural values should be conceptualized as poles of one dimension or categories that co-vary, we reran this regression with the same set of variables but replaced vertical vs. horizontal and individual vs. collectivistic cultural values with the four domains of the Culture Orientation Scale: vertical individualism, horizontal individualism, vertical collectivism, and horizontal collectivism. To protect against Type 1 error, we used a Bonferonni correction for primary analyses, resulting in an alpha level of.001. --- Results --- Differences between Participants from Lebanon and the US Participants in Lebanon were younger and less likely to be helping majors than students in the US (see Table 1). They reported less frequently having had any prior relationships with as well as lower quality of prior contact with autistic people, less knowledge about autism, and a higher preference for social distance from autistic people, yet more openness to experience relative to students in the US. Unexpectedly, students in Lebanon and the US did not differ in self-reported cultural values overall. --- Predictors of Stigma towards Autism An initial regression analysis predicting social distance scores from only gender and country replicated prior research (Obeid et al., 2015); males reported greater desire for social distance from autistic people than females and Lebanese students reported greater desire for social distance than US students (see Table 2; overall R 2 =.04). When all potential predictors (with cultural values operationalized as dimensions of individualism vs. collectivism and vertical vs. horizontal orientations) were included in a regression model predicting social distance scores across countries (see Table 3), lesser autism knowledge, quality of contact, and openness to experience, and greater vertical relative to horizontal cultural value orientation were associated with higher preferred social distance (overall R 2 =.29). Heightened individualism relative to collectivism (p =.003) was marginally associated with heightened social distance scores. Country (p =.06), major (p =.051), gender (p =.11), trait emotional intelligence (p =.09), lack of prior contact with autism (p =.12 3 ), social desirability bias (p =.15), and age, (p =.66) were not significantly related to social distance scores in this model. In a follow-up regression wherein the two polar dimensions of cultural values were replaced with vertical collectivism, vertical individualism, horizontal collectivism, and horizontal individualism (Table 4), lesser autism knowledge, quality of contact, openness to experience, and horizontal collectivism, and more vertical individualism were associated with heightened preference for social distance from autistic people (overall R 2 =.30). --- Discussion Contrary to prior research focused on mental health conditions using country level (e.g., Griffiths et al., 2006) or individual level (e.g., Papadopoulos et al., 2013) analyses of cultural values, we found no evidence that greater collectivism was associated with heightened stigma towards autism, as assessed with a social distance scale. Instead, heightened acceptance of inequality was associated with higher stigma towards autism. A trend toward positive associations between greater individualism relative to collectivism and stigma towards autism was also observed. A follow-up analysis wherein cultural value orientations were conceptualized as four independent categories rather than two dimensions revealed that vertical individualism was associated with a heightened preference for social distance from autistic people while horizontal collectivism was associated with reduced social distance scores. These findings align with evidence that vertical individualism is associated with heightened stigma towards disabilities (e.g., Ditchman et al., 2017;Rao et al., 2010). Findings also align with a larger body of research linking belief 3 Although we focus in primary analyses on the presence or absence of any prior contact with autism, we also examined if a dimensional measure of contact (or the number of different types of relationships with autism participants reported) was associated with social distance. Both measures were associated with reduced social distance in correlations. Aligning with prior literature (e.g., Gardiner & Iarocci, 2013), such associations were no longer apparent once quality of contact was accounted for. systems and/or personality traits (i.e., social dominance orientation and authoritarian personality) that promote inequality as a desirable aspect of human interaction with stigma towards many ways of being different, including disability status, race, economic status, gender, sexuality, and immigration status (Craig & Richeson, 2014;Duckit, 2016;Ekehammar et al., 2004;Triandis et al., 1965). As Link and Phelan aptly (2001, p. 375) pointed out, "Stigma is entirely dependent on social, economic and political power-it takes power to stigmatize." Prior evidence that stigma (assessed with a social distance scale) is heightened among college students in Lebanon relative to the US (Obeid et al., 2015) was replicated in the current study. Contrary to expectations, no differences in cultural value orientations were observed between participants in each country. Although this study was designed to elucidate underlying factors that may contribute to international differences in stigma, it is important to note that gender and country explained very little of the variance in stigma towards autism even before other factors were accounted for. Autism knowledge and high quality of contact with autistic people, greater openness to experience, and the belief that people should be equal (more horizontal cultural value orientations) were more strongly associated with stigma towards autism than what country participants lived in. --- Limitations and Future Directions The small magnitude of differences in preferred social distance between Lebanese versus US students in the current study might be comparable to variations in stigma between different racial/ethnic groups in a single multicultural country. Indeed, variations in stigma toward mental illness between different ethnic/racial groups within the US have been well documented (Abdullah & Brown, 2011). Although we asked participants to indicate their racial/ethnic background using an open-ended response, responses were not sufficiently clear to permit analysis of the degree to which variations in racial/ethnic background might contribute to stigma towards autism. For example, many students in Lebanon described themselves as White and many others described themselves as Arabic; both would be considered Caucasian according to US census categories. Other students indicated that race "does not matter" or provided their religious affiliation instead. Indeed, religious background may impact racial/ethnic self-identification in the Middle East, with Christians more likely to identify as White (e.g., Gualtieri, 2001). Future research examining individual cultural heritage in relation to cultural value orientations and stigma should use closed-ended options that separately address both race/ethnicity and sectarian/religious background with open-ended opportunities to elaborate. Our comparison of students from a single college in each country conflates potential differences between specific colleges with potential cultural differences. Although we initially tried to collect data from an additional school in each country, we were only able to provide academic credit for participating at the two schools described in this report; recruitment proved untenable when participation was not incentivized. Our requirement that Lebanese participants be fluent English speakers raises concerns that Lebanese participants in this study might be more Westernized than non-English speaking Lebanese people might be. Although the majority of Lebanese people are educated in either English or French and most majors in Lebanese universities require fluency in at least two languages (Esseili, 2017), a study examining variations in individualism vs. collectivism among college students in Lebanon revealed that students who elected to take a survey in Arabic tended to be more collectivistic than students who elected to take the same survey in either French or English (Ayyash-Abdo, 2001). However, identifying as Muslim rather than Christian also predicted collectivistic values, further supporting the importance of assessing religious orientation in future research about cultural values and stigma. Such research should also assess degree of Westernization (e.g., Stigler et al., 2010). Given that our reliance on a convenience sample of college students who might be more individualistic than the general population may have attenuated some of the cross-cultural effects which our study set out to explore, future work with more funding to support it should assess stigma across multiple settings in each country, preferably with a general population sample, using measures that are translated into the various languages people in each country speak. Such work should also compare stigma across a range of countries, in order to be able to explore associations between both cultural level and individual level cultural value orientations and stigma. The current findings and prior literature suggest that associations

is now at the Lebanese American University. We would like to thank Anna Schwartz for her insights about the distinction between cultural and individual differences.

examining variations in individualism vs. collectivism among college students in Lebanon revealed that students who elected to take a survey in Arabic tended to be more collectivistic than students who elected to take the same survey in either French or English (Ayyash-Abdo, 2001). However, identifying as Muslim rather than Christian also predicted collectivistic values, further supporting the importance of assessing religious orientation in future research about cultural values and stigma. Such research should also assess degree of Westernization (e.g., Stigler et al., 2010). Given that our reliance on a convenience sample of college students who might be more individualistic than the general population may have attenuated some of the cross-cultural effects which our study set out to explore, future work with more funding to support it should assess stigma across multiple settings in each country, preferably with a general population sample, using measures that are translated into the various languages people in each country speak. Such work should also compare stigma across a range of countries, in order to be able to explore associations between both cultural level and individual level cultural value orientations and stigma. The current findings and prior literature suggest that associations between collectivism and stigma may vary depending on the level of analyses, while highlighting that established measures of cultural level value orientations (e.g., Hofstede, 1983) may no longer be accurate (e.g., Rao et al., 2010). Differences between the current findings and prior work (e.g. Papadopoulos et al., 2013) may be attributable to differences in the conditions studied and measures used. However, in a follow-up to the current study (manuscript in preparation), we found that heightened vertical relative to horizontal cultural value orientations (and not heightened collectivism) were associated with heightened stigma towards a range of other conditions (e.g., schizophrenia, social anxiety, depression). Nevertheless, reliance on a single self-report measure of stigma, the social distance scale, is a significant limitation of this study. Although social distance scales have commonly been used in prior research examining stigma towards autism (e.g., Tilahun et al., 2017) and other mental health conditions (e.g., Griffiths et al., 2006) in diverse cultures, is possible that differences between the current findings and the findings documented by other researchers (e.g., Papadopoulos et al., 2013) might be attributable to their use of a different, and more comprehensive measure, the Community Attitudes Toward the Mentally Ill Scale (CAMI; Taylor & Dear, 1981). Although social distance scores and CAMI scores have been correlated in prior research (Lee & Lee, 2016;Smith & Cashwell, 2011), future research should use diverse measures of stigma to allow direct comparisons across studies. Given that the ways stigma and cultural value orientations are enacted are context-dependent, future research examining stigma towards autism internationally should include in-depth qualitative interviews and/or behavioral measures of actual interactions with autistic people in relation to measures of structural stigma (institutional practices and norms that disadvantage people in specific sociopolitical contexts). Given the relatively low alpha of the autism knowledge scale used in this study, future work should develop a comprehensive open-access measure of autism knowledge and stigma in collaboration with the autism community. Such a measure should highlight strengths as well as weaknesses associated with autism and ways in which stigma may be expressed and experienced differently in different cultures, depending on the cultural values that are most salient to a particular group (see Abdullah & Brown, 2011). Most importantly, future research should focus on decreasing structural stigma towards autism by building on promising innovations, including participatory autism research (e.g., Nicolaidis et al., 2011) and autism trainings for students (e.g., Gillespie-Lynch et al., 2015), health care workers in low-resource regions (Tilahun et al., 2017), and employers (e.g., https://proautism.org.uk/). --- Conclusions The findings described in this report suggest that, at least in multicultural countries like Lebanon and the US, cultural values (particularly acceptance of inequality), personal experiences, and individual differences in openness to experience exert a stronger influence on one's stigma towards autism than one's national or gender identity. Findings suggest that equalizing experiences (such as positive contact) and personality characteristics that contribute to such experiences (such as openness to experience and a commitment to fostering equality in social interactions) may reduce stigma towards autism. The evidence described in this report that heightened vertical cultural orientation is associated with more stigma towards autism is striking given that income inequality is a societal level indicator of vertical cultural orientation (Triandis & Gelfand, 1998) and recent evidence indicates that income inequality is increasing worldwide, particularly within the US (Alvaredo, Chancel, Piketty, Saez, & Zucman, 2018). Future research should focus on fostering high-quality positive contact with autistic people while identifying effective techniques to resist structural stigma and combat harmful misconceptions in order to work against societal forces that may, if unchecked, contribute to increased stigma towards people with disabilities. Note. Bolded items are significant at the alpha level of.001 selected for this study. Note. Bolded items are significant at the alpha level of.001 selected for this study. --- Appendix A Social Distance Scale The 5-point response scale ranged from strongly disagree to strongly agree.

is now at the Lebanese American University. We would like to thank Anna Schwartz for her insights about the distinction between cultural and individual differences.

INTRODUCTION Globally, there are growing inequalities concerning health and health care delivery [1][2][3], and to help reduce these inequities, nations around the world are in the pursuit of achieving universal health coverage. However, several nations emphasize their efforts in reducing national mortality and morbidity rates over resolving inherited issues caused by imbalances in health care systems [4]. For instance, every day, more than 800 women are estimated to die from complications related to pregnancy and delivery throughout the world due to low prenatal coverage, with sub-Saharan Africa accounting for the majority of deaths [5,6]. Furthermore, inequities in health include not only the unequal distribution of health but also unjust distribution of health as a result of unfair or insufficient social systems [7]. Health disparities result from a systematically uneven allocation of power, prestige, and resources among societal groups [8]. Evidence suggests that women and children in disadvantaged demographic segments have lower preventive treatment coverage and usage as well as poorer health outcomes than more advantaged populations [9,10]. These discrepancies are serious public health problems with social and economic ramifications [9][10][11]. Access to equitable quality maternal health care is critical to improving maternal health outcomes and is extensively advocated for. In low-income nations, disparities in residential location (urban/rural), household affluence, and the mother's educational status are identified as important barriers to obtaining high coverage [12,13]. Thus, health equity has become a worldwide goal in the post-2015 Sustainable Development Goals (SDGs), and it is also a focus in Ethiopia [14]. Ethiopia has implemented a pro-poor health policy through a variety of health programs, including the Health Extension Program, the Health Development Army, and the establishment of community-based health insurance schemes aiming to enhance maternal health usage services [15,16]. The health care that a woman receives throughout her pregnancy is critical for both the mother's and her baby's survival and well-being [5]. Antenatal care (ANC) provides the chance to provide appropriate therapy to reduce low birth weight and boost newborn survival [17]. In Ethiopia, ANC coverage has significantly increased over the previous two decades [17]. Despite the progress, significant disparities in health outcomes persist due to variations in economic position, education, location of residence, and gender [4]. Maternal health service consumption discrepancies among demographic groups continue to be a challenge in decreasing maternal mortality in Ethiopia. Inequities in maternal health services consumption were growing in 2016 compared to 2000, with low socioeconomic status, illiteracy, rural location, lack of employment, and lack of access to mass media being the main factors [15]. Access to maternal health care services remains a significant problem in low-income areas [18]. Even though maternal health services are provided free of charge in Ethiopia, the full attendance of ANC is about 62%, and attendance is much better in urban areas than in rural areas [19]. This study focuses on the disparities in ANC utilization among a local underprivileged population known as the "golden hands" community. In the Kembata Tembaro zone, there are outcast artisan groups known as "Fuga" who are akin to the "untouchables" in India. Kambati Menti Gezima (KMG) Ethiopia is striving to improve the social and economic situations of these villages with the support of USAID project monies using the tried-and-true tactics of community discourse and social mobilization. KMG has collaborated with both populations to uncover the root causes care (ANC) service use by the golden hand community. Age, urban residence, and wealth index were the most important independent factors with statistically significant contributions to changes owing to differences in effects (discriminated difference). --- Conclusion: Since ANC service discrimination is prevalent, the government and nongovernmental organizations should take steps to ensure that marginalized groups in society, such as golden hand women, the poor, the uneducated, and rural people, have equal access to service utilization opportunities. of prejudice and exclusion [20]. The study addressed existing equity gaps in which to intervene, and evaluating the interdependence of multiple factors of health disparities may be a significant component in understanding and acting on socially and economically disenfranchised women. As a result, the purpose of this study will be to assess differences in ANC service inequity depending on community groupings such as the golden hand and non-golden hand strata. --- METHODS AND MATERIALS STUDY DESIGN AND SETTING A community-based comparative cross-sectional was conducted in Ethiopia's Kembata Tembaro zone, SNNPR, from January 3 to February 27, 2022. --- STUDY POPULATION The study population was divided into two categories: golden hand and non-golden hand. The study population included all women of reproductive age residing in selected kebeles (small administrative units) representing non-golden hand communities across the zone, as well as women from golden hand communities throughout the zone. The study included women of reproductive age (15-49 years) who had lived in the study areas for at least six months and had at least one child. Women were excluded if they were unable to participate due to illness and had a known history of infertility at the time of data collection. --- SAMPLE SIZE AND SAMPLING PROCEDURES To estimate sample size, Epi info version 7 was used. The following hypotheses were used: The concentration index (CI) of inequity in the utilization of maternal health care was 13.2, 17.5, and 14.1 percent, respectively; the service is highly favored by the economically better off [21], 95% (confidence interval), power of 80%, and design effect of two because the sampling technique uses multistage sampling; and 10% nonrespondent rate was also included. The sample size was calculated using the same assumptions for all outcome variables and the second objective, and the maximum sample size was taken as the minimum required sample size, which is 880 (440 golden hand vs. 440 non-golden hand). However, all eligible participants in the chosen cluster were interviewed; therefore, a total of 1,210 (440 golden hand vs. 770 other non-golden hand community) study participants were included in the study. The subjects for the study were chosen using stratification (stratified sampling) based on population characteristics, followed by a multistage cluster sampling method. The study's participants were split into two groups: the golden hand community and the non-golden hand community. The study then enrolled all members of the golden hand community who met the inclusion criteria, regardless of where they lived, for a total sample size of 440. A multistage cluster sampling approach was used to sample members of the non-golden hand community. In the first stage of cluster sampling, more than 20% of the nine districts (woredas) in the Kembata Tembaro zone were chosen, resulting in the selection of four districts (woredas). In the second stage of cluster sampling, more than 20% of kebeles from each selected woreda were included in the study, totaling eight kebeles with similar sociodemographic characteristics as the golden hand community subgroup final administrative unit in the sampling process (two from each woreda). As a result, from the households in the kebeles described in the study, 770 women between the ages of 15 and 49 who had at least one child in a non-golden hand community were chosen. --- DATA COLLECTION INSTRUMENTS, PROCEDURES, AND DATA QUALITY CONTROL The survey instruments were adapted from the Demographic and Health Survey and the WHO framework for measuring health equity [8,22]. Data was collected using the open data kit (ODK) electronic data collection platform. We used a structured interviewer-administered questionnaire. Sixteen data collectors with a minimum of grade 12 and the ability to communicate in the local language, as well as eight nurses with a supervisory diploma, were recruited and assigned as data collectors and supervisors, respectively. To ensure the consistency of the questionnaire, the English version of the instrument was translated into the local language (Kembatissa) and then retranslated into English by independent translators. Before the actual data collection time, a pretest was conducted in 10% of the sample size outside of the study population with similar socioeconomic backgrounds. Significant changes and a logical flow of ideas were implemented based on the pretest results. The principal investigators trained them for two days on the study's objectives, including how to select households; conduct interviews, including consent-taking methods; and handle data. Supervisors reviewed and checked the questionnaires for completeness and relevance every day, and all necessary feedback was provided to data collectors the following morning before data collection. Participants' overall knowledge on ANC was classified as "good" if the score was between 70% and 100% (7-10 points), "moderate" if the score was between 50% and 69% (5-6 points), and "poor" if the score was less than 50% (0-4 points). Attitude toward ANC was assessed using seven questions and graded on a 5-point Likert scale. The overall level of attitude was categorized using modified Bloom's cutoff points as positive if the score is 80%-100% (<unk> 20 points), neutral if the score is 60%-79% (15-19 points), and negative if the score is less than 60% (<unk>15 points). --- MEASUREMENT OF INEQUITIES In this study, inequity is measured using concentration curve and concentration index. Concentration indices are frequently used to measure inequality in one variable over the distribution of another. To estimate equity for ANC utilization, four standard equity measures-equity gaps, equity ratios, concentration curves, and concentration indices based on community grouping-were used [3,23]. First, we examined differences in coverage for the outcome variable and explanatory variables by community subgrouping using bivariate statistics. To determine equity, we used simple comparative rates/measures of coverage for two groups [24]. Furthermore, when the wealth quantile was taken into account, the concentration index and concentration curve were used to demonstrate the inequity and difference in service utilization between the two groups. --- CONCENTRATION CURVE AND INDEX We created a concentration curve to illustrate inequality in ANC service utilization by wealth index and population subgroupings. A concentration curve that falls below the line of equality indicates that wealthy people use the ANC service more. To calibrate the degree and statistical significance of inequality, we used a concentration index, which measures differences in healthcare utilization between the two groups regardless of wealth status [23]. The concentration index ranges from 1 to +1; a value of zero indicates that the health variable is being used equally. When the concentration index is negative, it indicates that health variables are disproportionately concentrated among the poor. --- DECOMPOSITION ANALYSIS To analyze the key causes and the amount to which each component contributes to the discrimination in ANC service consumption, a nonlinear logit decomposition model was utilized. The analysis decomposed the total gap in ANC service utilization into two components: endowments and coefficients. Endowments (E) represent differences due to characteristics or endowments between the two groups, such as education, wealth, etc. Coefficients (C) represent differences due to differential effects of endowments or discrimination/bias against one group. Moreover, the raw difference (gap) in ANC service consumption was used to indicate the distinct undecomposed service utilization difference between golden hand and non-golden hand community members. All analyses were judged statistically significant with a p-value of less than 0.05 and a confidence range of 95% [25]. This suggests that the non-golden hand community used ANC services 6.31 times more than the golden hand group. Furthermore, 83.92% of ANC users were in the fifth quantile of the wealth index, 83.55% were urban residents, 86.08% were government employees, and 90.3% of ANC users had finished at least secondary education (Table 2). --- UTILIZATION OF ANTENATAL CARE SERVICES --- KNOWLEDGE AND ATTITUDE OF STUDY PARTICIPANTS TOWARD ANC SERVICES According to modified Bloom's cutoff points, the overall magnitude of good, medium, and bad knowledge among research participants was 79.87% (95% CI = 77.48 %-82.05 %), 17.44% (95% CI = 15.38%-19.71%), and 2.70% (95% CI = 1.90%-3.80%). The classification was made based on modified Bloom's cutoff points, and the overall magnitude of positive, neutral, and negative attitude toward the ANC service was 53.72% (95% CI = 50.89%-56.52%), 40.91% (95% CI = 38.16%-43.71%) and 5.37% (95% CI = 4.23%-6.80%), respectively. --- CONCENTRATION INDEX AND CONCENTRATION CURVE SHOWING SERVICE UTILIZATION INEQUALITY The rate-ratio in Table 3 shows the ratio of service utilization between the richest and poorest wealth quintiles. For ANC service utilization, the richest had 3.64 times higher utilization than the poorest. For ANC knowledge, the richest had 4.02 times higher good knowledge than the poorest. The concentration index shows the degree of socioeconomic inequality in each indicator. Positive values indicate higher utilization/knowledge among richer groups. --- SOCIOECONOMIC VARIABLES TOTAL NUMBER IN EACH CATEGORY ANC UTILIZATION --- Community subgroups Golden The attitude index was close to zero (0.04), indicating minimal wealth-related inequality in attitudes. The concentration indices for utilization and knowledge were statistically significant, as evidenced by the low p-values. But for attitude, the p-value was above 0.05, indicating the concentration index was not statistically different from zero. The difference in concentration indices between golden hand and non-golden hand groups was significant for utilization but not for attitude. This suggests wealth-related inequality in utilization was different between the two groups, but attitude inequality was similar. Therefore, there were pro-rich inequalities in ANC service utilization and knowledge but not in attitudes. The inequalities were more pronounced among the golden hand group for utilization (Table 3). The concentration curve graphs the cumulative percentage of the health indicator (y-axis) against the cumulative percentage of the population ranked from poorest to richest (x-axis). The concentration curve for the investigated indicator (ANC) services revealed a pro-rich bias (see Figures 1 and2 below). The concentration curve of ANC revealed that the services were highly utilized by a wealthy segment of the population. For ANC service utilization, the concentration curve was below the line of equality. This means that as you move from poorest to richest along the x-axis, the cumulative percentage of ANC utilization rises faster than the cumulative share of the population. Therefore, the ANC concentration curve being below the equality line graphically depicts that utilization of ANC services was concentrated or disproportionately higher among the wealthier groups compared to the poorer groups in the study population. This matches the prorich inequality found in the quantitative concentration index analysis (Figure 1). --- GENERALIZED LORENZ CURVE SHOWING INEQUALITY IN ANTENATAL CARE SERVICE UTILIZATION BETWEEN GOLDEN HAND AND NON-GOLDEN HAND COMMUNITY MEMBERS The Lorenz curve for ANC service utilization revealed that the service was lowly utilized by the golden hand community net of characteristics composition. The Lorenz curves depicted below clearly indicate that the service distribution for the golden hand community was more unequal than the service distribution for non-golden hand community members (Figure 2). --- DECOMPOSITION ANALYSIS FOR ANTENATAL CARE SERVICE UTILIZATION The total gap in ANC service utilization between the two groups was 0.68, as indicated in Table 4. This means one non-golden hand group had 0.68 higher ANC service utilization than the golden hand group. The raw difference in ANC service consumption in the table below was statistically significant, indicating that there is a distinct undecomposed service utilization difference between golden hand and non-golden hand community members. The gap of 0.689 quantifies how much higher the ANC utilization rate was for one group compared to the other. It provides a numeric measure of the absolute difference in ANC service utilization rates between the two groups. This differential can be presented separately by decomposing it into endowment and discrimination components in order to examine the percentage contribution of each component and forward a clear recommendation on specific components, allowing specific action to be taken on specific components rather than the general recommendation on both components. The coefficient (discrimination) component is clearly bigger than the endowment component (characteristics), showing that evidence for ANC service use discrimination exists in addition to individual characteristics such as educational achievement, employment, location, and so on. Discrimination explains 60.23% of the lower ANC service consumption by golden hand community members as compared to non-golden community members. Endowment accounts for just 39.47% of the reduced ANC service consumption for the golden hand community subgroups as compared to non-golden hand groupings. The tiny endowment components demonstrate that individual characteristics such as education and other endowment disparities explain a minor portion of the service use differentials among community groupings. However, the discriminating component is substantial, resulting in decreased service consumption by golden hand persons with the same domicile, employment, wealth index, and educational levels as non-golden hand individuals (Table 4). --- DISCUSSION Inequity between groups in health service utilization are common in developing countries. This study attempted to assess inequities in the utilization of ANC services in the golden hand community in Southern Ethiopia. ANC service is by far utilized more by pregnant women living in urban areas than those living in rural areas. This may be due to increased availability and accessibility of the services in urban areas and to individuals migrating to urban areas from the rural area when they have a strong economic base, which also enhances health-seeking behavior. This finding from the current study is consistent with findings from other inequity studies conducted in Ethiopia in which urban residents were found to utilize the service at a higher rate than rural residents [4], and it is also is in line with findings from a study in a different area [26][27][28][29][30]. Furthermore, ANC service is by far utilized more by educated women than women with no education according to the finding of the current study. This may also be due to the increased intention to utilize ANC service that occurs as a result of advanced educational status. This finding is supported by findings from different studies [28][29][30] and another inequity study in Ethiopia conducted using EDHS data throughout the country [4]. ANC service is utilized more by richer women than poorer women, because as financial capacity increases, health-seeking behavior increases; this is also supported by findings from another study [28]. The rate of ANC service utilization is 4.36 times higher in women in the top wealth quintile than in the bottom quintile. The concentration index was 0.28, which implies that the rich mainly utilize ANC services. This finding is higher than the one reported from Namibia [31], rate-ratio of 1.06 and CI of 0.0130. Even though they are not considered skilled providers, health extension workers (HEWs) provide preventive and promotive health services. ANC is one of the services provided by HEWs. Lesseducated pregnant women, those living in rural areas, and those at the bottom quintile mainly utilize ANC services by HEWs. This could be because HEWs are the first contact with the healthcare system in Ethiopia, especially in rural areas [32,33]. As a result, they can reach the part of the population that is unable to visit health centers or hospitals. However, the study also found evidence of inequity in the utilization of ANC, and the utilization of at least one antenatal visit was more common among women in the richest quintile [34]. While it is difficult to definitively identify the reasons, the pro-rich inequality in ANC knowledge and utilization may be related to women in the wealthiest quintile having greater access to information about ANC and greater means/opportunity to utilize ANC services. The wealth divide appears to enable richer women to gain more knowledge about ANC and translate that knowledge into higher service utilization compared to the poorest women. Pro-wealth inequality in the services provided in the antenatal program is of great concern, as it implies that the least wealthy pregnant women are not attending the program. Several studies from varying contexts and populations have reported similar findings of pro-rich inequality and lower utilization of ANC services among disadvantaged minority ethnic groups and those with lower education levels [35][36][37]. For instance, research in Brazil showed improvements in ANC coverage over time but persistent socioeconomic gaps, with women in the lowest wealth quintile having higher inadequate ANC compared to the highest quintile [35]. Several studies from high-income countries like the Unite States have also revealed pronounced racial/ethnic disparities in maternal health services, with Black, Hispanic, and minority women being more likely to face barriers to accessing optimal ANC [37,38]. Furthermore, studies have indicated that there are differences in the utilization of maternal health services between the better-off and poorest populations [24,39,40]. These findings emphasize the need to investigate and assess contextspecific causes of varying use of maternal health care if safe motherhood is to become a reality in developing countries [41]. Overall progress toward health-related goals can be reached even if health disparities across socioeconomic groups remain or worsen because certain groups may be left behind or may not benefit to the same extent as the relatively more affluent or privileged [1]. In terms of ANC --- DATA ACCESSIBILITY STATEMENT All data underlying the results are available as part of the article, and no additional source data are required. --- service knowledge, the rate of having excellent knowledge was greater among economically wellsituated women than among the poorest women. This is corroborated by a calculated CI of 0.15, indicating that the wealthier section of the population is more knowledgeable about prenatal care services than the poorest segment. Concerning attitudes about ANC services, the rate of positive attitudes toward ANC services was greater among economically well-situated women than among the poorest women. This is supported by a computed CI of 0.04, revealing that the wealthy segment of the population has a positive attitude toward ANC services compared to the poorest segment. Women's understanding of ANC is critical for using ANC services throughout pregnancy. Women's chances of having a solid understanding of ANC services are therefore a driver for increasing ANC seeking. National-level policies that guarantee targeted ANC services become an inherent component of ANC and other maternity care services and are critical to assisting women in improving their understanding, addressing misconceptions, and removing other healthcareseeking difficulties while utilizing ANC services [28,42,43]. This suggests that while personal characteristics contribute to low service consumption, discrimination based on living conditions and socioeconomic level accounts for a larger amount of the difference. As a result, the impoverished have higher rates of disease and mortality than the rich, and they typically use healthcare less, even if they have higher needs. These disparities are severe public health issues with social and economic consequences [9][10][11]. --- CONCLUSIONS Our ANC service utilization gap decomposition results indicate that a huge service utilization in Ethiopia is out of the ambit of affirmative action policies and here the government should initiate to reserve the right to equal education, employment, earnings, and ANC service utilization opportunity for marginalized sections of the society, like golden hand women and poor, uneducated, and rural people. As a result, policymakers are concerned about huge disparities in prejudice. In terms of endowment differences, improving the economic status and urbanization of marginalized people, job opportunities in government (public and private sectors), and safeguards against discriminatory practices such as norms and many invisible barriers that prevent golden hand women from receiving needed ANC services should be provided to close the huge ANC service utilization gap between golden hand and non-golden hand women. --- ETHICS AND CONSENT Wachemo University's Research Ethics Committee granted ethical approval. An official letter from Wachemo University was filed to the Kembata Tembaro zone's Health Department, followed by a letter to the designated district (Woredas), and lastly, district authorization was acquired for the selected kebeles to conduct the study. To maintain confidentiality, research participants were identified using codes rather than names. Mothers were informed that their participation was entirely optional and at their discretion. Informed consent was obtained from all subjects. Furthermore, all methods were carried out in accordance with relevant guidelines and regulations. --- COMPETING INTERESTS The authors have no competing interests to declare. --- AUTHOR CONTRIBUTIONS DE, SM, and BA: conceptualization and design, data collecting, analysis, and interpretation. AS was responsible for the article's conceptualization and design, data collection, and revision. SM, BA, TE, and DE: Data analysis and manuscript preparation. All authors have given final consent to the version to be published and agreed to accept responsibility for all aspects of the work. --- AUTHOR AFFILIATIONS

Background: Health equity has emerged as a global issue in the post-2015 Sustainable Development Goals, and Ethiopia is no exception. Despite positive improvements, inequities in maternal health service utilization among demographic groups continue to be one of Ethiopia's significant challenges in decreasing maternal mortality. This study focuses on antenatal care service discrimination among a local poor group known as the "golden hands" community in Ethiopia's Kembata Tembaro zone. The subgroup community consists of outcast artesian groups known as "golden hands," formerly known as "Fuga," who face discrimination in all aspects of life owing to their living conditions and ethnic background. Methods: A community-based comparative cross-sectional study was conducted in Ethiopia's Kembata Tembaro, zone in the Southern Nations, Nationalities, and Peoples' Region (SNNPR), from January to February 2022. The study focused on two groups, "golden hands" and "non-golden hands," consisting of women aged 15-49 years. Using stratified and multistage cluster sampling, 1,210 participants were selected, with 440 from golden hand communities and 770 from non-golden hand communities. Data was collected through translated questionnaires, and data quality was rigorously monitored. The concentration curve and index, as well as logistic-based decomposition analysis, were used to examine inequality. The statistical significance threshold was set at p < 0.05 with a 95% confidence interval. Result: This study comprised 1,210 eligible participants, 440 of whom were golden hand community members. Discrimination accounted for 60.23% of the decreased antenatal 2 Mekango et al.

Introduction Research has consistently demonstrated that sexual minority (e.g., gay, lesbian, bisexual) youth are more likely to use alcohol, marijuana, and illicit drugs than their heterosexual peers (Kann et al., 2018;Marshal et al., 2008). For example, in the 2019 Youth Risk Behavior Survey, sexual minority youth were more likely than heterosexual youth to report current (past-month) alcohol use (34% vs. 29%), marijuana use (31% vs. 21%), and prescription drug misuse (12% vs. 6%) (Jones et al., 2020), as well as lifetime illicit drug use (28% vs. 13%; inclusive of cocaine, inhalants, heroin, methamphetamines, ecstasy, and hallucinogens; CDC, 2020a). These disparities tend to increase from adolescence to young adulthood (Marshal et al., 2009), highlighting the need to understand and address them early in development. While most studies have focused on sexual minority youth in aggregate, accumulating evidence suggests that bisexual youth are especially likely to use substances. In a meta-analysis, sexual minority youth reported higher rates of substance use compared to heterosexual youth, but moderator analyses revealed that this was only true of bisexual youth and not gay/lesbian youth (Marshal et al., 2008). Despite this evidence, little is known about substance use experiences among youth who experience attractions to more than one gender or regardless of gender (e.g., those who identify as bisexual, pansexual, or queer; collectively referred to as bi+). To address this, we used qualitative data from a study of bi+ male youth to explore various aspects of substance use, including motivations for and against using substances, how they made decisions related to substance use, what consequences they experienced, and connections to experiences related to one's sexual orientation. --- Minority Stress Theory and Substance Use The predominant conceptual framework used to explain mental health problems and substance use among sexual minority youth is minority stress theory (Goldbach & Gibbs, 2017;Meyer, 2003). Minority stress theory proposes that sexual minority youth experience unique stressors related to their stigmatized social status (e.g., discrimination, internalization of negative societal attitudes) and these stressors contribute to their increased risk of mental health problems and substance use (Meyer, 2003). The associations between minority stressors and substance use are well documented among sexual minority youth. For example, in a meta-analysis, victimization, negative reactions to disclosure of one's sexual orientation, and lack of support from parents and other adults at school were significantly associated with substance use among sexual minority youth (Goldbach et al., 2014). Of note, bi+ individuals face unique stressors at the intersection of heterosexism and monosexism (e.g., being stereotyped as confused about their sexual orientation, experiencing discrimination from both heterosexual and gay/lesbian people; Feinstein & Dyar, 2017), and attitudes toward bi+ men tend to be particularly negative (Dodge et al., 2016). Although few studies have specifically examined the associations between minority stressors and substance use among bi + individuals, the available evidence suggests that bi+ stressors are indeed associated with substance use among bi+ youth (Feinstein et al., 2019) and adults (Dyar et al., 2020;Feinstein et al., 2017). --- Motivational Models of Substance Use While research on substance use among sexual minorities has emphasized the role of minority stress, research on substance use in general has identified other important contributing factors such as motives. Motivational models of substance use propose that people use substances to attain valued outcomes, such as to reduce negative emotions (coping motives), to enhance pleasure (enhancement motives), to bond with others or improve social gatherings (social motives), and to avoid social censure or gain others' approval (conformity motives), and these motives have been identified among adolescents and adults (Cooper et al., 2016). Motivational models have also distinguished between the extent to which substance use is motivated by a desire to pursue a positive outcome versus avoid a negative one, and the extent to which motives are internally/self-focused versus externally/other-focused (Cooper et al., 2016). These two dimensions can be used to create the four categories of motives described above: (1) self-focused approach motives (i.e., enhancement motives); (2) self-focused avoidance motives (i.e., coping motives); (3) other-focused approach motives (i.e., social motives); and (4) other-focused avoidance motives (i.e., conformity motives). Related to conformity motives, prior studies have demonstrated that peer pressure is associated with substance use among adolescents (Santor et al., 2000), and meta-analytic findings support the influence of peers on adolescent substance use (Allen et al., 2003). However, other studies suggest that personal choice and curiosity have greater influences on youth substance use than do peer pressure and the desire to conform (McIntosh et al., 2003). In addition, several other motives have been identified among adolescents, including experimentation motives (e.g., drinking to find out what the experience is like; Lo & Globetti, 2000;Palmqvist et al., 2003) and identity motives (e.g., drinking to assert one's independence or to feel more mature; Kloep et al., 2001). Although few studies have explored substance use motives, including peer pressure, among sexual minority youth, one study found that sexual minority youth were more likely than heterosexual youth to endorse coping and conformity motives (Bos et al., 2016). In addition, research on young gay and bisexual men has found that coping motives help to explain the associations between minority stressors and substance use problems (Feinstein & Newcomb, 2016). Given the limited research in this area, additional research is needed to understand bi+ youth's motivations for using substances, especially the extent to which their motivations are influenced by experiences related to their minoritized identities. In addition, even fewer studies have examined people's motivations for not using substances. One study of 30 young adults who did not use alcohol found diverse motivations for not drinking including wanting to be in control of one's behavior, wanting to avoid the negative health consequences, not liking the taste, knowing someone who died for reasons related to alcohol use, having parents who did not drink often or at all, and religious/cultural values (Hardcastle et al., 2019). However, we are not aware of any prior studies that have examined why some sexual minority youth choose not to use substances. It is likely that some motivations for not using substances are universal, but some may be unique to bi+ youth (e.g., not using substances to remain in control of one's behavior and not accidentally disclose one's sexual orientation). --- Decision Making Scholars have also applied theories of reasoned action and planned behavior to adolescent risk behavior, including substance use. These theories describe a path to adolescent risk behavior in which decision making is a deliberative process that involves consideration of behavioral options and anticipated outcomes (Gerrard et al., 2008). In such instances, the behavior and associated risks are considered, and the decision to engage in the behavior is reasoned and intentional. Still, little is known about this deliberative process with respect to how youth make decisions about substance use (e.g., what and how much to use, when and where to use). Understanding the various factors that youth consider when making decisions about substance use has the potential to inform the development of more effective prevention and harm reduction strategies by providing youth with the necessary skills to make thoughtful and deliberate decisions about their health and behavior. A small qualitative study of 13 adolescents who used alcohol and marijuana simultaneously found that decisions about substance use were influenced by various social and contextual factors (e.g., where they were, who was present) (Wolf et al., 2019). For example, participants tended to use alcohol at parties with large groups of peers, whereas they tended to use marijuana at small gatherings with close friends. In addition, they described using alcohol when they wanted to be more social, whereas they described using marijuana when they wanted to maintain greater control over their behavior (e.g., in the presence of adults). Participants were also more likely to use alcohol and marijuana simultaneously when their peers were doing so. This study highlights the various factors that youth consider when making decisions about substance use. While it is likely that some of these factors are relevant to the experiences of both heterosexual and bi+ youth (e.g., whether adults are present), bi+ youth may consider additional factors when making decisions about substance use (e.g., whether other sexual minority youth are present). However, this has yet to be explored. --- Consequences of Substance Use Substance use tends to increase from adolescence into young adulthood (Marshal et al., 2009), and many adults with substance use disorders started using substances as adolescents (United States Department of Health and Human Services, 2016). The consequences of substance use during adolescence are well documented (e.g., mental health and suicide risks, exposure to violence, sexual risk behavior; CDC, 2020b). Adolescence also marks a critical period for brain development and maturation, during which substance use can lead to persistent changes in brain function (Salmanzadeh et al., 2020). While there is substantial evidence of the consequences of substance use during adolescence, little is known about the specific consequences experienced by sexual minority youth and bi+ youth in particular. It is likely that heterosexual and bi+ youth experience many of the same consequences of substance use, but some consequences may be exacerbated for bi+ youth. For example, sexual minority youth are disproportionately affected by adverse mental health outcomes (e.g., depression, suicidality; Lucassen et al., 2017;Marshal et al., 2011), and these disparities are often greatest for bisexual youth (di Giacomo et al., 2018). As a result, bi+ youth may be particularly vulnerable to experiencing affective consequences of substance use, such as depressed mood. In addition, sexual minority youth are disproportionately exposed to violence (e.g., childhood sexual abuse, parental physical abuse) and, again, these disparities are often greatest for bisexual youth (Friedman et al., 2011). Therefore, bi+ youth may be at risk of experiencing harsher discipline by their parents if they are caught using substances. Finally, sexual minority youth report higher rates of sex while intoxicated than do heterosexual youth (Herrick et al., 2011), and may be more likely to experience unwanted sexual encounters as a consequence of substance use. --- The Current Study In sum, the goal of the current study was to understand substance use experiences among bi+ male youth including motivations for using and not using substances, how they make decisions related to substance use, what consequences they experienced, and connections between substance use and experiences related to one's sexual orientation. Consistent with motivational models of substance use (Cooper et al., 2016), we conceptualized motivations as why participants used substances. In contrast, consistent with theories of reasoned action and planned behavior (Gerrard et al., 2008), we conceptualized decision making as how participants thought about what they were going to use, how much they were going to use, and when and where they were going to use. Of note, our sample was inclusive of transgender male youth and, as such, we were also able to explore connections between substance use and experiences related to being transgender. This is particularly important given that bi+ identities are especially common among transgender individuals (Galupo et al., 2017;Grant et al., 2011;Morandini et al., 2017), and transgender youth are more likely to experience victimization and use substances than their cisgender peers (Johns et al., 2019). --- Method --- Participants The current analyses used data from a larger mixed methods project focused on sexual risk behavior and substance use among bi+ male youth. 1 A total of 58 participants completed a survey and an interview, but the analytic sample included 54 participants because three were not asked about substance use (there was not enough time) and one interview was not recorded (the recording device failed). Participants in the analytic sample ranged in age from 14 to 17 years (M = 16.12, SD = 0.97). Approximately two-thirds (63%) identified as bisexual, 30% as pansexual, and 7% as queer. They were racially/ethnically diverse (39% White, 33% Latinx, 19% Black, and 9% a different race). Approximately three-quarters (76%) of participants were cisgender male youth (i.e., they selected "male/man" as their gender identity and "male" as their sex assigned at birth) and 24% were transgender male youth (i.e., they selected "trans male/trans man" or "male/ man" as their gender identity and "female" as their sex assigned at birth). Of the 13 transgender participants, nine selected "trans male/trans man" as their gender identity and four selected "male/man." --- Procedure All data were collected in 2019, prior to the start of the COVID-19 pandemic. Participants were recruited online (e.g., Facebook, Instagram), in-person (e.g., at local community-based organizations serving LGBTQ + youth), and using a participant registry maintained by the Institute for Sexual and Gender Minority Health and Wellbeing at Northwestern University. The study was broadly described as being about sexual orientation and health. Potential participants were directed to an online eligibility survey. Eligibility criteria included: (1) 14-17 years old; (2) identified as male (regardless of sex assigned at birth); (3) identified as bisexual or another bi+ identity (e.g., pansexual, queer); (4) reported being HIV-negative or not knowing one's HIV-status; (5) lived in the U.S.; and (6) agreed to have their interview audio recorded. Those who were eligible were contacted using their preferred contact method (email, text message, or phone call) and provided with a link to a consent form. If they consented, they were automatically directed to the online survey. As part of the survey, participants were asked to report on their lifetime and past-year use of alcohol, marijuana, hallucinogens, inhalants, cocaine, ecstasy, methamphetamine, heroin, and poppers. For alcohol, participants were asked how many days they had at least one drink of alcohol in their lifetime and in the past year (0 days, 1 or 2 days, 3 to 9 days, 10 to 19 days, 20 to 39 days, 40 to 99 days, or 100 days or more). Participants who reported any past-year alcohol use were also asked to report the largest number of alcoholic drinks they had in a row (i.e., within a couple of hours) within the past month (1 or 2 drinks, 3 drinks, 4 drinks, 5 drinks, 6 or 7 drinks, 8 or 9 drinks, or 10 or more drinks); this was used to determine whether or not participants had engaged in binge drinking within the past month. 2 For all other substances, participants were asked how many times they had used each substance in their lifetime and in the past year (0 times, 1 or 2 times, 3 to 9 times, 10 to 19 times, 20 to 39 times, or 40 or more times). Responses were dichotomized for descriptive analyses. These questions came from the 2017 Youth Risk Behavior Survey; questions about lifetime use were adapted to assess past-year use, and questions about the use of poppers were added. After completing the survey, they were contacted to schedule their interview. Interviews were conducted remotely (by phone or video chat) or in-person depending on the participant's preference. The interviews were semi-structured, conducted by the primary investigator and two research assistants, and approximately 90 min in length. All of the interviewers were members of the LGBTQ + community. The primary investigator is a clinical psychologist with extensive experience conducting research on the experiences and health of LGBTQ + individuals across the lifespan, and the research assistants were graduate students in social work and medicine who were also involved in conducting research on LGBTQ + populations. As part of the interview, participants were asked a series of questions about their use of alcohol, marijuana, and other drugs. First, they were prompted with: "Tell me a little about your use of [alcohol, marijuana, other drugs]" (depending on which, if any, they reported ever using in the survey). As participants responded, interviewers probed for additional information by asking questions such as: "How often do you 1 All of our participants selected "male/man" or "trans male/trans man" in response to a question about gender identity. In pilot work with the target demographic, participants perceived the term "men" as referring to adults and they perceived the term "boys" as infantilizing. Colloquially, they preferred the term "guys", but they did not self-identify as such. Given these concerns, we decided on "male youth" as the most appropriate term to describe our participants. 2 The National Institute on Alcohol Abuse and Alcoholism defines binge drinking as a pattern of drinking that brings blood alcohol concentration (BAC) to 0.08 percent or higher. The typical cutoffs used to operationalize binge drinking are 4 + drinks (for female adults) and 5 + drinks (for male adults) in about 2 h (NIAAA, 2018). However, fewer drinks in the same timeframe result in the same BAC for youth (3 drinks for girls and 3-5 drinks for boys, depending on their age and size; Chung et al., 2018). Of note, it is unclear if the same cutoffs can be used with cisgender and transgender people. For these reasons, we report a range of 11-19% for past-month binge drinking in our sample (19% reported 3 + drinks in a row, 17% reported 4 + drinks in a row, and 11% reported 5 + drinks in a row). use it? Who do you use it with? What kinds of situations do you use it in?" Then, they were prompted with: "Tell me about the most recent time you used [alcohol, marijuana, other drugs]." Again, as they responded, interviewers probed for additional information by asking questions such as: "Who were you with? What were you doing? How did you decide to use it?" For any substances that they had never used, they were prompted with: "You said that you haven't used [alcohol, marijuana, other drugs]. Tell me about that." As they responded, interviewers asked follow-up questions about whether they had experienced any peer pressure and, if so, how they had resisted it. Finally, participants were asked two questions about the potential connection between sexual orientation and substance use: "How does being bisexual/pansexual/queer influence your use of alcohol and drugs?" and "Some research has found that bisexual guys are more likely to use alcohol and drugs than other guys, including gay guys. Do you have any ideas why that might be?" Of note, 10 participants did not report any lifetime substance use on the survey. With the exception of the first prompt described above ("Tell me a little about your use of [alcohol, marijuana, other drugs]," these participants were asked the same questions during the interview as the participants who did report lifetime substance use. The interviewers conducted member checks throughout the interviews by paraphrasing participants' responses and asking whether their own understandings of the responses were accurate. Participants who completed the survey and the interview received a $30 Amazon gift card as compensation. All procedures were approved by the Institutional Review Board at Northwestern University, and a waiver of parental permission was granted for the study. --- Data Analysis Interviews were transcribed, reviewed for accuracy, and analyzed using a reflexive thematic analysis approach (Braun & Clark, 2006, 2019, 2020). As such, our analysis reflects the data as well as the backgrounds, experiences, and knowledge of our research team. Our research team was knowledgeable about general and LGBTQ + specific theories of substance use (e.g., motivational models of substance use, minority stress theory) as well as research on the unique experiences of bi + individuals relative to other sexual minority individuals. Throughout the analytic process, we met to discuss how our own experiences with and assumptions about substance use could influence our interpretations of participants' responses to interview questions. Consistent with Braun and Clark's approach, the first and second authors read the relevant sections of all of the transcripts, noting key ideas in participants' responses. Notes were compared across transcripts and similar ideas were grouped together as codes. Codes were consolidated if several codes overlapped conceptually. By using this inductive approach, the codes were developed based on participants' responses rather than a pre-existing framework. Once the codebook was finalized, two authors (first and second) independently coded the transcripts in Dedoose. After they coded 25%, inter-rater reliability was tested. The pooled Cohen's kappa was 0.98, suggesting excellent inter-rater reliability (Hruschka et al., 2004). When disagreements were identified, they were discussed by the two coders and the last author until consensus was reached. The remaining 75% were divided between the coders. Frequencies and percentages for codes are reported in Table 1. For most percentages, the denominator was the total analytic sample size (N = 54). In exceptions, for "motivations for using" and "negative consequences" the denominator was the number of participants who reported any lifetime substance use (N = 44), and for "sources of stress" the denominator was the number of participants who reported using substances to cope with stress (N = 17). Percentages do not add up to 100% because each participant could receive codes for multiple sub-themes within a theme and not all participants contributed relevant data to every theme. Participant quotes were selected to represent themes. Quotes are presented verbatim with the exception of minor edits to facilitate readability. --- Results --- Frequencies and Contexts of Use In regard to any lifetime use, 74% of participants reported alcohol use, 46% marijuana use, and 20% other drug use (11% hallucinogens, 11% inhalants, 4% cocaine, 4% ecstasy, and 2% methamphetamine). No participants reported ever using heroin or poppers. Ten participants did not report any lifetime substance use on the survey. In regard to past-year use, 32% of participants reported alcohol use, 39% marijuana use, and 13% other drug use (9% hallucinogens, 6% inhalants, 4% ecstasy, 2% cocaine, and 2% methamphetamine). Five participants did not report any past-year substance use. In addition, 11-19% of participants reported past-month binge drinking. Participants described using substances with friends (59%), with family members (26%), alone (22%), and with romantic/ sexual partners (7%). --- Motivations for Use The 44 participants who reported any lifetime substance use described diverse motivations for using substances. The most commonly coded motivations were to cope with stress (39%), to experiment (32%), to have fun (23%), and to conform to social norms (7%). Other motivations (e.g., to relax, to feel more comfortable, to rebel) were described by 18% of participants, but only one or two participants described each one. Using to Cope. Nearly two-fifths of participants described using substances to cope with stress. Of the 17 participants who reported using substances to cope with stress, the most common sources of stress were mental health problems (59%), school (41%), and family (18%). Other sources of stress (e.g., work, romantic relationships) were described by 24% of participants, but only one or two participants described each one. Of note, no participants described using substances to cope with stress related to their sexual orientation and only two of the 13 transgender participants described using substances to cope with stress related to their gender (e.g., dysphoria, being misgendered). As an example of using substances to cope with mental health problems, one participant (16, cisgender, bisexual, White) said: "There are points where I just feel absolutely empty inside. And it's those points that I'll pretty much just do anything to end it. Not end my life but to end feeling that way. And weed basically like numbs my mind." Another participant (16, transgender, pansexual, White) similarly described using marijuana to cope with depression and anxiety, noting that it was more effective than prescription medication: "I do smoke weed pretty often...I kind of use it for medicinal purposes. I have depression and anxiety and when I smoke it, I feel a little better. And every time I've worked with a prescription medicine, it hasn't helped." Although mental health problems were the most commonly described source of stress, participants also described using substances to cope with external stressors (e.g., school, family). For example, one participant (17, transgender, pansexual, Latinx) described a recent incident where he drank to cope with academic stress: "I'm rather stressed out about my school right now because I'm currently going through the college application process and she [his friend] offered to bring alcohol and I jumped at the opportunity. I kind of drank beyond my own limits that day." Another participant (17, cisgender, bisexual, White) said that he liked drinking "...because it kind of numbs me and makes me forget about depression, stress, and stuff." Then, when asked what caused him stress, he talked about his parents, saying that "...they fight a lot...and it stresses me out..." For most percentages, the denominator was the total analytic sample size (N = 54). In exceptions, for "motivations for using" and "negative consequences" the denominator was the number of participants who reported any lifetime substance use (N = 44), and for "sources of stress" the denominator was the number of participants who reported using substances to cope with stress (N = 17). Percentages do not add up to 100% because each participant could receive codes for multiple sub-themes within a theme and not all participants contributed relevant data to every theme Using to Experiment. Approximately one-third of participants described their substance use as experimentation. When they did, they were typically describing only using a substance once or twice. For example, one participant (16, cisgender, pansexual, White) described drinking alcohol and smoking marijuana once: "It was just kind of to see what the hype is about. People like it and I wanted to see what people like about it." Another participant (16, transgender, pansexual, White) reflected on his experimentation with inhalants, saying: "That was just me being a dumb kid because they would always talk about it like, 'Oh, inhalants are very dangerous.' And I was like, 'Oh, are they? Well let me just do that.' I was a stupid kid...It's just like, 'This is something that I'll try'." Using to Have Fun. Nearly one-quarter of participants described using substances as a way to have fun. These descriptions tended to be brief and to the point. For example, one participant (16, cisgender, bisexual, White) said: "Usually I drink and smoke just to like have fun." Another participant (16, cisgender, bisexual, White) described drinking a few times a month and getting drunk at least one of those times, noting: "I've been told that I'm more fun when I'm drunk." Sometimes the desire to have fun led to using more than intended. For example, one participant (17, cisgender, bisexual, Latinx) described a situation where he had at least five drinks in a row, saying: "I was just having a lot of fun and I wasn't keeping track." Using to Conform to Social Norms. Finally, a few participants described using substances because their peers were doing it. For example, one participant (17, transgender, pansexual, White) described only drinking in social situations, saying: "If I'm hanging out with people and they start drinking, I'll start drinking with them just because it's not fun to be the only sober person there." Another participant (17, cisgender, bisexual, White) described using marijuana because his friends were doing it: "When I first starting using it, I did it because I thought it was cool and I was trying to fit in with my friends." Peer Pressure. Approximately one-third of all participants (35%) experienced pressure from their peers to drink or use drugs, but nearly all of them reported that they were able to resist it. For example, one participant (15, cisgender, bisexual, Latinx) described being pressured by his friends to smoke marijuana at a party: "They were like, 'Come on, I'm gonna give it to you free' and 'It's a good feeling,' but I just told them no. And then they were telling me how's it's going to make me feel, that it was going to make my life better, but I just texted my friend to come get me." Even participants who used substances described experiencing pressure to use in greater quantities. For example, one participant (16, cisgender, bisexual, White) said he experienced pressure to drink more than he wanted to, but he went on to say: "I haven't really caved in. I only drink what I'm comfortable with and if I think I've had too much, I'll stop...I don't need people telling me what I need to do to have fun." Although most participants who experienced peer pressure were able to resist it, some participants described giving in at times. For example, one participant (17, cisgender, bisexual, Latinx) said: "I did use marijuana, but that was only once because I was pressured into it, and I never did it again." The same proportion of participants (35%) had not experienced peer pressure. Some of them described not associating with people who used substances, while others described having friends who used substances but who didn't pressure them. For example, one participant (17, cisgender, bisexual, Black) said: "There are those at my school that do that kind of stuff. So I try not to talk to them as much or I avoid spending time with them because I don't want to be pressured into doing that and I don't know if I would do it if I was pressured into it." Another participant (16, transgender, pansexual, White) said: "People are very chill about it. Especially people that I hang out with. They're like 'You don't have to do anything you don't want to do'." --- Decision Making Nearly one-third of all participants (31%) described thinking about what substances they were going to use, how much they were going to use, and when and where they were going to use them prior to doing so. For example, some participants described using marijuana, but not other drugs, because they perceived marijuana as safer. One participant (16, transgender, pansexual, White) said he used marijuana, but not other drugs, because: "I prefer things that can't hurt me and I like to just stay more safe and other drugs really can just mess you up, like mess up your brain....I don't really think of weed as a drug." Another participant (17, cisgender, bisexual, White) expressed similar sentiments, but specifically said that he would only use marijuana in the form of edibles because he perceived them as safer than smoking: "I had been thinking about it, [that] it's something that would be fun to try...especially since it was an edible so I didn't have to worry about the effects of the smoke at all." When participants described rules for when and where they used substances, the rules generally had to do with only using substances in safe places and with people who they trusted. For example, one participant (17, transgender, bisexual, White) described his rules for drinking as follows: "It has to meet all three: comfortable place, comfortable people, comfortable mindset. Cause it always ensures that I'm not doing it because I'm depressed or anything." He went on to describe similar rules for when he smoked marijuana: "...it was only around people who I really, really trusted and felt comfortable with." Similarly, another participant (15, cisgender, bisexual, Latinx) described only drinking in certain situations: "Usually it's me and a couple of my friends at home. We're never out. We're never driving." Participants also described asking their friends for help with keeping their substance use within certain limits. For example, one participant (16, cisgender, bisexual, White) said: "I know when to cut myself off....When I start slurring words together or stuttering more than I usually do, I tell the most sober one in the room to stop letting me have alcohol." As demonstrated, even though this participant was thoughtful about not exceeding certain limits, he still drank to the point where his behavior was affected. --- Negative Consequences Of the 44 participants who reported any lifetime substance use, nearly one-third (32%) described a range of negative consequences related to their substance use (e.g., getting sick, getting in trouble with their parents, putting themselves in potentially dangerous situations), but only one or two participants described each one. In addition, a small subset of these participants (9%), all of whom were transgender, described their substance use as getting to a point where it became problematic. For example, one participant (15, transgender, bisexual, White) said: "I used to have a problem with alcohol. I used to drink a lot. It became a serious problem and I think it might've contributed to my depression..." He went on to say: "I would drink to the point where I couldn't walk up the stairs....It was a horrible experience. It wasn't fun because I started feeling like I needed it..." Similarly, another participant (17, transgender, bisexual, White) described a cycle of drinking and feeling depressed: "I just became a really depressed drunk and then every time I was drinking I would just get really depressed...and then I started drinking because I was depressed." One participant (17, transgender, pansexual, Latinx) went as far as to say: "I tend to refer to myself as an alcoholic waiting to happen, because I'm very, very susceptible to drinking whenever I feel upset. And if I had more access to it, I would most definitely probably be an alcoholic." --- Motivations for Not Using Substances Participants described diverse motivations for not using substances often or at all including being aware of and concerned about the consequences (46%), seeing them negatively affect others (35%), not liking them (31%), having limited access (17%), and personally experiencing negative consequences (15%). Being Aware of and Concerned About the Consequences. Nearly half of participants described not using substances because they were aware of and concerned about the physical, psychological, and legal consequences. For example, one participant (17, cisgender, bisexual, Black), said: "I've never used alcohol and drugs....I don't think you should ever really mess with [drugs] because anything that changes your state of mind and leaves you vulnerable to be taken advantage of is not good for you." Similarly, another participant (16, transgender, pansexual, White) said: "My future is very important to me and I don't want to compromise that by doing anything and getting in legal trouble or getting in mental trouble." Seeing Them Negatively Affect Others

Sexual minority youth are at increased risk of substance use compared to their heterosexual peers, and bisexual youth appear to be at greatest risk. However, little is known about their motivations for and against using substances, how they make decisions, and what consequences they experience. We used qualitative data from a study of 54 cisgender and transgender male youth (ages 14-17 years) who reported attractions to more than one gender or regardless of gender (i.e., bisexual, pansexual, or queer; collectively referred to as bi+) to explore these aspects of substance use. Participants completed a survey and an interview, and interviews were thematically analyzed. Qualitative analyses revealed that participants described diverse motivations for using substances (e.g., to cope with stress, to experiment, to have fun) and for not using them (e.g., concern about consequences, not having access). The most common sources of stress were mental health problems, school, and family. They did not describe sexual orientation-related stress as a motivation for their use, but they acknowledged that it could influence others' use. Participants also described thinking about when, where, and with whom they were going to use prior to doing so (e.g., only using in safe places and with people who they trusted). Finally, they described a range of consequences they experienced (e.g., getting sick, getting in trouble), and a subset of transgender participants described experiencing dependence symptoms. These findings suggest that substance use prevention and harm reduction interventions for bi+ male youth should address diverse motivations for use, including general stressors, which are often overlooked compared to minority-specific stressors. Further, interventions should approach youth as capable of making decisions. Findings also highlight the particular need to address substance use among transgender youth.

upset. And if I had more access to it, I would most definitely probably be an alcoholic." --- Motivations for Not Using Substances Participants described diverse motivations for not using substances often or at all including being aware of and concerned about the consequences (46%), seeing them negatively affect others (35%), not liking them (31%), having limited access (17%), and personally experiencing negative consequences (15%). Being Aware of and Concerned About the Consequences. Nearly half of participants described not using substances because they were aware of and concerned about the physical, psychological, and legal consequences. For example, one participant (17, cisgender, bisexual, Black), said: "I've never used alcohol and drugs....I don't think you should ever really mess with [drugs] because anything that changes your state of mind and leaves you vulnerable to be taken advantage of is not good for you." Similarly, another participant (16, transgender, pansexual, White) said: "My future is very important to me and I don't want to compromise that by doing anything and getting in legal trouble or getting in mental trouble." Seeing Them Negatively Affect Others. Approximately one-third of participants described not using substances because they had seen them negatively affect other people's lives. For example, one participant (16, cisgender, bisexual, White) said: "I live in an apartment facility and I have grown up surrounded by cigarettes, heroin needles, dirty condoms, marijuana, everything. And I don't want to do that. I see how it affects the people." They went on to say: "I've had people die up where I live because of drugs. I don't want to get myself involved in that. I'd rather stay sober and clean." Another participant (14, transgender, bisexual, White and Asian) described how his older sister's drug use affected him and his family: "I haven't really been inclined to do so [drink], but especially since my older sister...she was doing hard drugs and it left a really big impact on all of my family. And my younger sister and I, we've talked about it and we're not planning on drinking at all because we're so afraid that we're going to get addicted like our older sister." Not Liking Them. Nearly one-third of participants described not using substances often or at all simply because they didn't like them. For example, one participant (16, cisgender, pansexual, White) said: "I wish I liked drinking because you know, it's kind of the norm in society....It's just not something that I'm really into...I don't like the taste of it." Similarly, another participant (17, cisgender, queer, Latinx) said: "I spend most of the time with the Mexican side of my family. And they drink alcohol a lot. So I've always had access to alcohol, but I don't really like alcohol that much. I don't like the taste of it or anything about it. And so I never really drink alcohol." Other participants expressed similar sentiments about marijuana. For example, one participant (15, cisgender, bisexual, Latinx) said: "A lot of my friends say weed smells good, but to me it smells really bad. I heard the feeling is good, but my friends told me that if I ever did weed, I'll be the type to be very paranoid." Limited Access. Some participants described not having access to substances as a reason for not using them often or at all. For example, one participant (16, transgender, pansexual, White) said: "I drink sometimes, but it's a very rare occurrence. Usually I don't even have the opportunity to if I wanted to." Other participants were more specific in saying that they only used substances when they were with people who had access to them. For example, one participant (17, cisgender, pansexual, White) said: "I've only drank a few times...and it was with older family members, like older cousins, that had access to it." Similarly, another participant (15 cisgender, pansexual, American Indian/Alaska Native) described smoking marijuana once in a while with specific friends who would offer it to him. He went on to say: "It only happened once in a while because they weren't wealthy enough [to get it more often]." Experiencing Negative Consequences. Some participants stopped using substances or cut-back because of the negative consequences they experienced. For example, one participant (16, cisgender, bisexual, White) described getting in trouble with his parents for smoking marijuana and he said: "I'm cutting everything off since I don't want to get in trouble again." Another participant (16, transgender, pansexual, Latinx) who stopped using substances said: "I started drinking by myself and that was when I realized that there was danger involved... and that was when I stopped. It was still never very, very frequent, but it was something I started using as a coping mechanism." --- Connection to Sexual Orientation When first asked about the potential connection between their sexual orientation and their substance use, 52% of participants said that their substance use was not related to being bi+ in any way. However, half of them (and 46% of participants in total) went on to describe how being bi+ could be connected to substance use for other people through stigma (i.e., bi+ people experiencing stigma and using substances to cope with it). Other connections were described by 15% of participants (e.g., bi+ people being more open to new experiences or more rebellious), but only one or two participants described each one. As an example of how being bi+ could be related to substance use through stigma, one participant (17, cisgender, bisexual, Black), said: "Being bisexual is super stressful.... You're not accepted by straight [community] members. Sometimes you're not accepted by gay people. Sometimes you're in your own league. You're an outsider to all these people. So, if a bottle is the only person who accepts you, then the bottle is." Another participant (17, transgender, bisexual, White) described how being bisexual was related to substance use through internalized stigma for one of his friends: "I have a friend who is bisexual but he'll only say it if he is drunk cause that's the only time he ever feels comfortable enough to say it. And it's the only time he'll ever do things [sexually] with a guy, if he is drunk, because of that." Although most of these participants were not describing their own experiences, one participant (17, transgender, pansexual, Latinx) described struggling with his sexuality when he was younger and he went on to say: "When I was younger, it happened a couple of times where I just completely broke down because of something someone said to me and I just wanted to calm down, to numb the pain, and ended up taking the drink or two." --- Discussion The goal of the current study was to understand substance use experiences among bi+ male youth, including motivations for using and not using substances, decision making, negative consequences, and connections between substance use and experiences related to one's sexual orientation. While substance use was not required to participate in the study, rates of substance use were similar to national rates among sexual minority youth (CDC, 2020a(CDC,, 2020b;;Jones et al., 2020). For example, rates of past-month alcohol use, lifetime marijuana use, and lifetime illicit drug use were 32%, 46%, and 20% in our sample (compared to 34%, 50%, and 28% among sexual minority youth in the 2019 Youth Risk Behavior Survey). Our findings can be used to inform the development of substance use prevention and harm reduction programs to address the high rates of substance use among bi+ male youth. --- Motivations for Using and Not Using Substances Our participants described diverse motivations for using substances (e.g., to cope with stress, to experiment, to have fun), which paralleled the motivations that have been described in the broader literature (Bos et al., 2016;Cooper et al., 2016;Lo & Globetti, 2000;Palmqvist et al., 2003). The most commonly described motivation was to cope with stress, but sexual orientation-related stress was not described as a source of stress. Instead, the most common source of stress was mental health problems (e.g., depression, anxiety), followed by school (e.g., applying to college), and family (e.g., parents fighting with each other). It is possible that sexual orientation-related stress played a role in these other stressors, but that was not explicit in participants' narratives. Although our participants did not describe sexual orientation-related stress as a motivation for their own substance use, they did acknowledge that it could contribute to substance use for other bi+ male youth (e.g., if they did not feel accepted or if they were not comfortable with their sexual orientation). Therefore, it is possible that our participants experienced relatively low levels of sexual orientation-related stress and, for that reason, they did not perceive it as having an influence on their substance use. It is also possible that our participants lacked insight into the extent to which sexual orientationrelated stress influenced their substance use. Regardless, our findings highlight the importance of attending to other sources of stress in order to understand substance use among bi+ male youth. In addition, although peer groups can have a powerful influence on adolescents' engagement in potentially risky 1 3 behaviors, including substance use (Miller & Prentice, 2016;Neighbors et al., 2007), very few of our participants described using substances to conform to social norms. Approximately one-third of our participants did report that they had experienced pressure from their peers to drink or use drugs, but nearly all of them described being able to resist it. These findings are consistent with evidence that factors other than peer pressure (e.g., personal choice, curiosity) may have greater influences on youth substance use than do peer pressure and the desire to conform (McIntosh et al., 2003). Those who had not experienced peer pressure explained that they either did not associate with people who used substances or they had peers who used substances but who did not pressure them. It has been suggested that sexual minority youth may be at increased risk of substance use because they may be more likely to associate with peers who use substances (Dermody et al., 2016). While this may be true of some sexual minority youth, our findings suggest that others may be equipped to resist peer pressure. Our participants also described diverse motivations for not using substances, which were consistent with the limited prior research in this area (Hardcastle et al., 2019). Many of these were related to the consequences of substance use (e.g., being aware of and concerned about the consequences, having personally experienced consequences, having seen others experience consequences), suggesting that both personal experience and vicarious learning through others' experiences can function to keep youth from using substances. However, other motivations for not using substances were not related to the potential consequences (e.g., not liking the taste, not having access). In particular, lack of access suggests that some of our participants would have used substances, or used them more, if they had greater access to them. As such, it may be particularly important to direct substance use prevention and harm reduction interventions to these youth in an effort to prepare them for when they do have greater access. --- Decision Making Consistent with theories that describe a path to adolescent risk behavior in which decision making is a deliberative process (Gerrard et al., 2008), approximately one-third of our participants described thinking about what substances they were going to use, how much they were going to use, and when and where they were going to use them prior to doing so. For example, they described using marijuana, but not other drugs, because they perceived marijuana as safer, and they described only using substances in safe places and with people who they trusted. These findings are consistent with the results of a prior qualitative study, which also found that adolescents had rules for when, where, and with whom they used alcohol and marijuana (Wolf et al., 2019). These findings stand in contrast to the common framing of adolescent substance use as impulsive and thoughtless. However, it is important to acknowledge that even with rules, adolescents may still engage in unsafe behavior. For example, one of our participants explained that he asked his friends to stop him from drinking once he had reached a certain point, but the point at which he thought he should stop was once his behavior was already affected (e.g., once he started slurring his words). As such, these findings are encouraging in that they suggest that some bi+ male youth are approaching substance use in a thoughtful way, but they also suggest that youth may need support to ensure that they use substances as safely as possible. --- Negative Consequences Some youth described a range of consequences related to their substance use (e.g., getting sick, getting in trouble with their parents, putting themselves in potentially dangerous situations). In addition, a small subset of participants, all of whom were transgender, described their substance use as getting to a point where it became problematic (e.g., feeling dependent on alcohol, substance use having a negative influence on one's mental health). Some of our transgender participants also described using substances to cope with gender-related stress (e.g., dysphoria, being misgendered). These findings are consistent with evidence that transgender youth are more likely to experience victimization and use substances than their cisgender peers (Johns et al., 2019), and they highlight the critical need to identify and address substance use problems among transgender youth. --- Limitations The current findings should be considered in light of several limitations. First, while our sample was relatively large for qualitative analyses, the extent to which the current findings generalize to the broader population of bi+ male youth remain unclear. It will be important for future quantitative studies with larger samples to continue to examine bi+ male youth's experiences with substance use, including motivations for using and not using substances, decision making processes, negative consequences, and connections between substance use and experiences related to one's sexual orientation. Second, our findings were based on self-report and it is possible that some participants may not have felt comfortable disclosing the details of their substance use or they may not have accurately remembered all of their experiences using substances. It may be useful to collect information from other sources, such as parents or peers, to gain a more comprehensive understanding of substance use from the perspectives of all those involved. Third, some of our participants received a subset of interview questions due to insufficient time. As such, some of the proportions of participants who received certain codes were based on a subset of participants. Finally, two transgender participants described using substances to cope with stress related to their gender, but we did not specifically ask about this and it is possible that more transgender participants would have described similar experiences if they had been asked. --- Conclusion Despite limitations, the current findings provide important insights into bi+ male youth's experiences with substance use and can be used to inform substance use prevention and harm reduction interventions for this population. While a number of interventions have been developed to reduce substance use (typically in conjunction with sexual risk behavior) among sexual minority young adults (for a review, see Newcomb & Feinstein, 2019), we are only aware of one intervention that was specifically developed to address substance use among sexual and gender minority youth (Schwinn et al., 2015). Schwinn and colleagues developed a three-session, online drug abuse prevention program that focused on teaching skills related to minority stress management, decision making, and drug use refusal, and they tested it in a randomized trial with a no intervention control condition. At three-month follow-up, youth who completed the intervention reported lower stress, peer drug use, and past 30-day other drug use (i.e., inhalant, club drug, steroid, cocaine, methamphetamine, prescription drug, and heroin use), as well as higher coping, problem solving, and drug use refusal skills compared to youth who did not complete the intervention. Based on the current findings, it is important to not only focus on minority stress, but to also attend to general stressors that all youth experience (e.g., academic pressure, family conflict). In addition, while focusing on stress and coping is important, our findings highlight the range of motivations that bi+ male youth have for using substances. When substance use is motivated by other desires, different intervention strategies may be needed. For example, it has been suggested that those who drink to cope may benefit from stress reduction and coping skills training, whereas those who drink to enhance pleasure may benefit from alternative sources of pleasure and restructuring expectancies for alcohol's enhancing effects (Cooper et al., 1995). Finally, our findings also suggest that not all substance use among youth is impulsive. Instead, our participants described thinking about which substances they used, how much they used, and when, where, and with whom they used. As such, when it comes to helping bi+ male youth make decisions about substance use, it is important to approach them as being capable of making thoughtful decisions related to substance use. Funding This project was supported by a grant from the National Institute on Drug Abuse (K08DA045575; PI : Feinstein). Research reported in this publication was also supported by the National Institute on Minority Health and Health Disparities of the National Institutes of Health under Award Number T37MD014248. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. --- Data Availability De-identified data (including survey data and coding of interview data) are available from the senior author upon request. Survey and interview materials are also available from the senior author upon request. Code Availability Not applicable. --- Declarations --- Conflict of interest The authors do not have any conflicts of interest to disclose. Human or Animal Rights All study procedures were approved by the Institutional Review Board at Northwestern University prior to implementation. The study was performed in accordance with the ethical standards in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed Consent Online informed consent to participate was obtained from all participants.

Sexual minority youth are at increased risk of substance use compared to their heterosexual peers, and bisexual youth appear to be at greatest risk. However, little is known about their motivations for and against using substances, how they make decisions, and what consequences they experience. We used qualitative data from a study of 54 cisgender and transgender male youth (ages 14-17 years) who reported attractions to more than one gender or regardless of gender (i.e., bisexual, pansexual, or queer; collectively referred to as bi+) to explore these aspects of substance use. Participants completed a survey and an interview, and interviews were thematically analyzed. Qualitative analyses revealed that participants described diverse motivations for using substances (e.g., to cope with stress, to experiment, to have fun) and for not using them (e.g., concern about consequences, not having access). The most common sources of stress were mental health problems, school, and family. They did not describe sexual orientation-related stress as a motivation for their use, but they acknowledged that it could influence others' use. Participants also described thinking about when, where, and with whom they were going to use prior to doing so (e.g., only using in safe places and with people who they trusted). Finally, they described a range of consequences they experienced (e.g., getting sick, getting in trouble), and a subset of transgender participants described experiencing dependence symptoms. These findings suggest that substance use prevention and harm reduction interventions for bi+ male youth should address diverse motivations for use, including general stressors, which are often overlooked compared to minority-specific stressors. Further, interventions should approach youth as capable of making decisions. Findings also highlight the particular need to address substance use among transgender youth.

Background Workers' value of their health and safety imposed by contextual factors might be associated with the amount of occupational effort and resources used to protect workers. In the context where workers are more valued, more attention will be given and greater quantity and quality of safety protections will be offered for their safety and health. Alternatively, where workers are less valued, they might suffer from greater workplace hazards and extreme outcomes such as occupational fatalities due to lesser safety protections. A key motivation for not providing adequate safety education and training is to reduce organizational resources, saving time and money [1,2]. Even while adhering to federal safety and health standards (e.g., improving safety environments, providing safety equipment), organizations may divert organizational resources away from safety and health practices in order to focus on positive business outcomes (e.g., increasing productivity, meeting deadlines). In fact, many organizations emphasize organizational success over matters of safety [3] such that workers can be exposed to hazards they were previously safeguarded against. Sensemaking theory [4] contends that people rationalize and give meaning to their experiences based on pertinent contextual information. It helps explain how people develop a consensus on values and beliefs which can be used as plausible reasons for their behaviors in a given context. According to sensemaking theory, people develop shared perceptions on which behaviors and practices are acceptable or not acceptable throughout the ongoing process of retrospective assessment of social norms/standards and common practices. Haas and Yorio [5] reframe the sensemaking model in terms of risk assessment and state that sensemaking can be viewed as a risk management process which allows everyone in the workplace to identify hazards, communicate risks, and respond accordingly. Sensemaking begins with an observation of organizational cues and workplace behaviors by employees. If any disparity is noted, employees engage in the enactment of addressing this disparity by aligning behaviors to the organizational cues or devaluing, overlooking, and annulling the cues. Behaviors that are reinforced or not penalized would be retained and referenced to as the base of an organization's policies and procedures. This process applies to employers as well such that they can make sense of an adequate level of managerial standards in the promotion of occupational safety and health by observing societal norms and common practices of appraising workers' value. As long as their current practices do not incur any notable backlash from governing authority and organizational members, policies and procedures of their organization would be maintained. In turn, organizations tend to reinforce and solidify their cultures through a series of attraction, selection, and attrition processes [6,7]. The present study, on the continuum of the pilot study of Lee and Klos [8], posits the different standards and practices of workers' value among the 50 United States and the federal District of Columbia (D.C.) can lead employers to have a certain sense as to the value of their workers, which will be referred to as worker value throughout the remainder of this study. Specifically, we focus on minimum wage and workers' compensation benefits. Among the many indicators of worker value, minimum wage was chosen because it is a set value and oftentimes the basis of the calculation of labor cost [9]. Also, all 50 states of the U.S. and D.C. have jurisdictional guidelines for minimum wages. A number of socioeconomic factors are associated with minimum wage. Low minimum wage might be indicative of affordability of labor as well as easiness to find substitutes when incumbent workers become unavailable or lose their workability. Moreover, low minimum wage itself has been noted as potential occupational safety and health hazards because of its impact on suboptimal access to quality medical care [10][11][12]. Additionally, minimum wage can reflect a state's protection policies towards workers. Another indicator of worker value considered in the present study is workers' compensation benefits. Workers' compensation is a form of insurance providing financial resources to cover lost wages, medical costs, and ongoing care expenses to injured workers during employment. The employers are the policy holders of workers' compensation insurance. There is no federal oversight of workers' compensation programs, which are regulated by each state [13]. The American Public Health Association [13] found that many employers believe the incidence rate of injury has plateaued or declined, implying that hazardous work conditions are no longer a meaningful threat to workers. To avoid increased premiums in a seemingly safer work environment, companies can misclassify workers and underreport payroll in an effort to obtain a lower premium [14]. Despite the workers' compensation cuts and the consistently decreasing costs of workers' compensation, self-insured employers argue that high workers' compensation costs will not attract economic growth and in order to stay competitive, workers' compensation benefits costs must stay relatively low [13,14]. In sum, workers' compensation is necessary to protect workers, but more adequate amounts of workers' compensation benefits pose greater financial burden to employers. As a result, the amount of workers' compensation benefits can imply the worker value over competing business demands. Overall, people make decisions in consideration of contextual factors on acceptable and desirable behaviors [15]. Accordingly, local governments, organizations, and employers may interpret minimum wage and workers' compensation benefits as specific contextual factors regarding worker value and use this information to leverage human resources to achieve industrial and economic progress. If workers are inadequately valued, it does not violate common sense to treat workers as expendable commodities [16]. --- Hypotheses The present study aimed to explore how minimum wage and workers' compensation benefits in 2015 are associated with the average fatality rates across 50 United States and D.C. in 2015 through 2017. To this end, the following hypotheses were examined: Hypothesis 1. Minimum wage is significantly and negatively associated with average fatality rates. Hypothesis 2. Average workers' compensation benefits for major body part loss is significantly and negatively associated with average fatality rates. --- Methods The present study utilized archival data on five statelevel variables (n = 51) categorized into three groups including key-study, contextual, and control variables. Key-study variables were 1) minimum wage in USD units ($), using data from 2015; 2) average of workers' compensations for the loss of an arm, hand, leg, or foot in 2015. --- Minimum wage We used the 2015 minimum wage data from the U.S. Department of Labor [17]. The 2015 minimum wage data had a range of $2.00 (Oklahoma) to $10.50 (D.C.) with a mean of $7.67 (SD = $1.34). For states with different minimum wage standards for large and small employers, we used the minimum wage standards that had the smallest value. In fact, a sizeable portion of workers are hired through small businesses in 2017, such as in Minnesota (47.8%), Montana (65.2%), Nevada (42.0%), Ohio (46.0%), and Oklahoma (52.8%), while 47.5% of the private workforce in the U.S. was employed by small businesses [18]. --- Average of workers' compensations for the major body part loss The four body partsarm, hand, leg, and footwere chosen to be the focus for looking at average workers' compensation for a few key reasons. These four body parts are major body parts that are commonly injured during work and the amount of workers' compensation for the loss of other body parts is generally highly correlated with the workers' compensation for the loss of the four chosen body parts (e.g., correlations with the workers' compensation for the loss of eye ranged from.88 to.95). Also, data were available in all 50 states and D.C. for all four of these body parts (i.e., the amount of workers' compensation for the loss of eye info was not available from 10 out of 50 states). The average of the workers' compensation benefits for the loss of the four major body parts can be indicative of the overall amount of financial support from the employer for workers who are permanently disabled and lost workability due to their work. The data for all four body parts were obtained from a ProPublica article by Groeger, Grabell, and Cotts [19]. ProPublica calculated the maximum benefit injured workers can receive for the total loss or amputation of various parts by researching the law for all 50 states and D.C., following each state's provided formula. The maximum benefit was determined by taking a 100% loss of each body part for a worker who earned enough to qualify for the state's current maximum compensation rate. In cases where states assigned higher values for amputations, or if the injury occurred on a dominant hand, the highest value was used [20]. In this data, the average of workers' compensation benefits for a permanently injured hand, arm, leg, or foot (for damage to body part and future lost wages) were $40,205 in Alabama and $568,027 in Nevada (national mean = $157,944, SD = $97,767) and for each state the maximum benefit was used. This is illustrated for each of the four individual body parts in Fig. 1. --- Contextual variables Contextual variables were 1) education level per state defined by the % college degree earned in 2015 [21]; 2) GDP (gross domestic product) per capita in 2015 [22]. These variables were included because they can respectively serve as representatives of job type (i.e., more/less protected) [23] and regional economic development (see Fig. 2). A combination of both education and GDP per capita offers a useful snapshot of the primary industries in a particular state. For instance, Alaska has a low education level, but high GDP, which is reflected in its major industries being high risk with high return (e.g., natural resource development, fishing, and logging). Massachusetts has a high education level and GDP with its major industries being low risk with high return (e.g., healthcare, education, and finance). Mississippi has low education levels and GDP with its major industries being high risk with low return (e.g., agriculture and retail). Also, it is worth noting that minimum wage might be influenced by the standard of living in the state while it oftentimes is reflected by GDP per capita [24]. --- Control variable The control variable for the present study was the population for each state according to the national census data from 2010 [25]. This variable was included because it can serve as the surrogate for availability and cost of labor as suggested by the demand-supply model [26,27]. A linear regression analysis was conducted to examine the prospective relationship between these five variables and the average occupational fatality rates for the years 2015 through 2017 [28], operationalized as the number of workers killed at work per 100,000 workers. The dependent variable, fatality rates, was log-transformed.. The analysis was conducted with archival data using the statistical software R. --- Results Descriptive statistics and correlations of the study variables are presented in an additional file [see Additional file 1]. As summarized in Table 1 and Fig. 3, both workers' compensation benefits (B = -0.09, SE = 0.04, p = 0.042) and minimum wage (B = -0.11, SE = 0.05, p = 0.024) in 2015 were significantly and negatively associated with the average fatality rates in 2015 through 2017. The outcome variable of fatality rates was logtransformed, so the results from the linear regression analysis were exponentiated to allow for a more meaningful interpretation of the main variables of interest. A to -1.5% [= 1exp.(-0.15)]). These findings supported hypotheses 1 and 2, as seen in Fig. 4A andB. Regarding the contextual and control variables, education level was negatively associated with fatality rates (B = -0.33, SE = 0.06, p = 0.000) while GDP per capita was positively associated with fatality rates (B = 0.22, --- Discussion The present study showed that state-level quantitative indicators of worker value, namely minimum wage and workers' compensation benefits, were significantly and negatively associated with the average fatality rate from 2015 to 2017. Within a culture of capitalism, in which profit can be prioritized over worker safety and workers can be inadequately valued, injured and diseased workers are more likely to be considered expendable [29]. Additionally, while workers' compensation benefits are intended to protect workers, workers can still suffer from a loss of income due to decreased workability after incurring an occupational injury that warrants workers' compensation [30]. Monetary informational and other regulatory indicators of worker value can serve as contextual factors that provide reference points for linking beliefs and ideas to broader networks of meaning which can augment both employers' and employees' sensemaking processes about occupational safety and health [15,31]. These informational cues can be exacerbated considering that contracts are designed to efficiently defend against fraudulent claims by overinsuring small losses and underinsuring large injuries, thus restricting the monetary value that can be assigned to workers' health [32]. Due to sensemaking, employers' and employees' project themselves onto the situation, which is often characterized by low minimum wage and insufficient workers' compensation benefits. Subsequently, they observe the negative outcomes in their workplace such as improper safety leadership, training, and protections, and learn which organizational behaviors are acceptable [33]. Presumably, it is unrealistic to inflate minimum wage and workers' compensation benefits arbitrarily in a short period of time. However, we can think about alternative protections for workers through social and organizational systems. For example, we may want to try to enhance the awareness of employers and employees regarding the inadequacy of the extant minimum wage and workers' compensation benefits in the protection of workers. Raising awareness on workers' compensation benefits is important as it has been shown that those who are in more precarious employment situations are not only more likely to get injured and need access to workers' compensation benefits, they are less likely to be aware of workers' compensation and the assistance it provides [34]. Additionally, many past studies have shown that a large number of workers do not file claims even when they are aware of these benefits and qualify for them [35][36][37]. Raising awareness could help workers receive needed protections such as workers' compensation benefits. Additionally, more prevention efforts can be made throughout the United States. There can be more collaborative efforts between organizations and government agencies to provide safety and health training/programs as well as access to occupational safety services (e.g., safety monitoring and hazard assessment). It has been shown that overtime, safety initiatives increase safety performance and reduce accidents, which in turn helps lower insurance costs for organizations [38]. Haley-Lock and Shah [39], describe that while employers are already minimally incentivized by public policy to participate in supportive employment practices, research has found that those who chose to participate in additional high involvement human resource management strategies helped reduce expenses related to turnover, hiring, and training. Furthermore, policy makers can focus on addressing the gaps in current policies and loopholes in implementation of these policies to mitigate the workplace safety and health related disparity between lower and higher income workers [40]. Additionally, public policy can put in place programs to incentivize employers to be more involved in the safety and health of their workers and provide resources to help these employers train and educate their workers on occupational safety. In order to extend the present study conducted at the state-level, future research is needed to examine whether the minimum wage and workers' compensation benefits at the state-level are indeed interpreted as worker value at the individual-level across the samples of employees and employers. Also, to address the limitation of the study findings based on archival data, a more controlled experimental approach utilizing a series of likely scenarios of judgment and decision making in terms of occupational safety and health investment as well as the provision of various financial support including salaries and workers' compensation benefits can be adopted. Additionally, the present study utilized data on workers' compensation benefits that did not differentiate between the two types of payouts for workers' compensation benefits which are a lump sum or long-term payments over time. The present study utilized the maximum payment offered per injury, making the results a conservative estimate. Future research can investigate if there are differences in occupational safety and health outcomes based on these two different payout systems. Finally, the present study utilized three contextual variables, GDP, education level, and population, to control for the potential variation in common industry sectors and labor market situations across the 50 states and D.C. Alternative indicators more directly reflecting the characteristics of industries or workforce can be considered for future studies. --- Conclusions Minimum wage and workers' compensation benefits, which were chosen for the present study as the two state-level quantitative indicators of worker value, were significantly and negatively associated with fatality rates. The study speaks to the importance of contextual factors and their relationship with worker value, as they can be one of the many factors that affect outcomes of health and safety, culminating at a state-level. Further investigation into these contextual factors is needed to fully understand these relationships. --- Availability of data and materials The dataset generated and/or analyzed during the current study are attached to the submission under supplemental materials. The data used is also publicly available and accessible at the following sources: U.S. Department of Labor [17], ProPublica [19], Bureau of Economic Research [21], Bureau of Economic Analysis [22], U.S. Census Bureau [25], and U.S. Bureau of Labor Statistics [28], --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12889-021-11117-9. --- Additional file 1. Authors' contributions LK was the leader and a major contributor in writing of the manuscript with FG, SS, MA, & JL also being major contributors in writing the manuscript. JL analyzed and interpreted the data regarding the results of this study. All authors read and approved the final manuscript. --- Authors' information Not Applicable --- Declarations Ethics approval and consent to participate Not Applicable. --- Consent for publication Not Applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: The present study aims to observe how societal indicators of workers' values at the state-level are related to health and safety outcomes, particularly major injuries and fatalities in the U.S. Underscoring workforce flexibility and workability over workforce stability and safety might be indicative of the worth of workers which can be associated with occupational safety and health concerns. Methods: Linear regression analysis with a log-transformed dependent variable was adopted to examine how the state-level indicators of worker value in terms of 1) minimum wage, using data from 2015; 2) average of workers' compensations for the loss of an arm, hand, leg, or foot in 2015 were concurrently and prospectively associated with occupational fatality rates averaged across 2015, 2016 and 2017. Socioeconomic contextual variables such as education level, GDP per capita, and population at the state-level were controlled for. Results: The present study showed that state-level quantitative indicators of how workers are valued at work, namely minimum wage and workers' compensation benefits, were significantly and negatively associated with fatality rates in the following year. Conclusions: The present study illustrates the gap in how workers are valued across the U.S. The study speaks to the importance of contextual factors regarding worker value, as they can affect outcomes of health and safety culminating at a state-level.

Background In the last few decades, there has been consistent and mounting evidence for the pervasiveness and health consequences of intimate partner violence facing women worldwide. The World Health Organization's (WHO) definition of intimate partner violence (IPV) is a, "behaviour within an intimate partner relationship that causes physical, sexual or psychological harm, including physical acts of aggression, sexual coercion, psychological abuse or controlling behaviours" [1]. One-third of everpartnered women globally, and nearly 40% in the South and East Asian regions, reported experiencing physical and/or sexual violence from an intimate partner in their lifetime in 2013, according to the WHO [2]. Research in India and elsewhere has shown that women who sell sex, or female sex workers (FSWs), experience particularly high rates of violence from intimate partners [3][4][5][6]. In a study among a representative sample of 1750 FSWs in north Karnataka state, India, 29% of women reported experiencing physical violence from an intimate partner in the last 6 months alone [7]. Research to date has identified a host of important risk factors for partner violence using multivariate analyses. Efforts to extend this analysis to the context of sex work have largely focussed on exploring the impact of client violence on women's health and wellbeing [8][9][10][11]. A few studies have qualitatively explored violence and condom use in intimate relationships among female sex workers, but never with a community-based approach or among both women and their intimate partners [3,6,12,13]. A deeper exploration of how social influences interact to shape the experience of violence among women in sex work and their intimate partners is required if health interventions are to effectively address the underlying processes that heighten IPV and HIV/AIDS vulnerability. We conducted a qualitative community-based participatory research (CBPR) study to explore the experience and understandings of violence and condom use within intimate relationships among 38 purposively selected participants, including female sex workers and male intimate partners, in Bagalkot district of north Karnataka state, India. This study was part of a mixed method evaluation study of the intimate partner violence prevention program called Samvedana Plus being implemented in north Karnataka by the non-governmental organization, Karnataka Health Promotion Trust (KHPT), and their partnering community-based organization (CBO), Chaitanya AIDS Tadegattuva Mahila Sangha, in Bagalkot district. In the tradition of community-based participatory research, this study worked with women in sex work as co-learners together with other researchers. The results have been used to adapt the Samvedana Plus program over its implementation cycle. In this paper we present our findings, and discuss the implications for IPV and HIV prevention by drawing on a "structural vulnerability" theoretical framework. We aim to contribute to the nascent literature on how an array of social, cultural, economic and political issues interact to perpetuate intimate partner violence and HIV among women in sex work, at a time when population health research and program implementers are increasingly approaching the interwoven social roots of health issues through a complex systems lens [14,15]. --- Methods --- Study design In response to local interests and needs, the study drew on a community-based participatory research (CBPR), interpretive qualitative methodology [16]. A CBPR approach aims to combine the strengths of community and academic partners to research important issues and inform culturally-appropriate and sustainable interventions [13,17]. The action-oriented, collaborative approach involved in CBPR aligns well with interpretive forms of inquiry by seeking to understand social phenomena according to those who experience them. This is based on the assumption that "social reality" is necessarily constructed and context-specific, and must therefore be interpreted through meaning-making processes [16,17]. Hence our methods were collaborative in nature, developing questions able to explore experiences from women's and men's perspectives in relevant ways by working closely with community partners, and together undertaking analyses that allowed for multidimensional interpretations on how participants conceptualized IPV and low condom use, the root causes, and related ways to address them. The study was conducted through in-depth collaboration with a Research Committee of women selected by and from the sex worker-led community-based organization Chaitanya in Bagalkot. They were involved in the design of the study, oversight of data collection, the selection and training of two female community research investigators (CRIs) to conduct the interviews with female participants, data analysis, interpretation and knowledge translation. A trained male research investigator (RI) from the local region who had been hired by KHPT for previous studies conducted the interviews of male intimate partners (IPs). --- Strategies for enhancing rigour We aimed to contextualize general guidelines for rigour in terms of pursuing authenticity in both the methods and results of this study. Further details on our approach to pursuing authenticity in CBPR can be found in another article [18]. Briefly, the research team drew on their previous experience and prolonged engagement with each other to develop relevant and sensitive research questions, tools, methods, and modes of analysis that could enhance credibility, or "verisimilitude" between the data and the phenomena they aimed to represent. To ensure dependability of the results, the perspectives of the interviewers and research team, as well as all methodological choices and processes, were recorded by the first author at each stage in an audit trail [19]. We also relied on peer debriefing as we undertook meaning-making workshops with the local CBO and NGO teams after initial coding by the researchers to ensure confirmability or "reproducibility" of the results. Transferability, or the ability to link the results to other contexts, was pursued through our use of purposive sampling that aimed to maximize variation on characteristics reflective of the community socio-demographics, and through detailed description of the research investigators' observations of the interviews and surrounding context. We also triangulated results on reasons for violence from KHPT's quantitative behavioural survey among a representative sample of FSWs in the district. In these ways, we aimed to garner trustworthy interpretations on the meaning of the social phenomena under study [16,19]. The study received ethics approval from St. John's Medical College in Bangalore, India, the Research Ethics Board at London School of Hygiene and Tropical Medicine, and the Human Research Ethics Board at University of Manitoba. --- Study setting The study was conducted in the two sub-districts of Mudhol and Jamkhandi in Bagalkot district in north Karnataka state, India [20]. According to the 2011 Census, the population of Bagalkot was 1.9 million, with almost 70% living rurally. Compared to southern and western Karnataka, literacy and employment rates in 2011 were lower in the northern district of Bagalkot, at 79% and 54% respectively among men, and 58% and 32% respectively among women [20]. Mudhol and Jamkhandi have been long-term sites for KHPT's work on HIV prevention, including efforts to develop and strengthen peer-led community-based organizations among FSWs, such as Chaitanya Mahila Sangha. This was a response in part to curb the prevalence of HIV in the late 20th century, which was highest in the state and linked to high rates of sex work in the region. This emerged with the historical Devadasi tradition that began centuries ago, in which devadasis (or'servants of the god or goddess' in Sanskrit) were traditionally dedicated or "married" to a god or goddess in the temple, often the goddess Yellamma, to perform various religious duties and sometimes sexual services for priests or others [21]. While Devadasi women are not permitted to marry legally there are many who have intimate partners in various forms. These may include men who are long-term clients also providing nonmonetary or emotional support; men considered husbands, lovers or cohabiting or non-cohabiting nonpaying partners; and sometimes pimps with control over their work and finances [5,22,23]. Alongside changes in socio-religious practices, and particularly after it was officially banned by the Indian government in the 1980s, Devadasis have come to practise more commercialized sex work and now hold a lower socio-economic status than historically [24]. Almost all participants in the current study identified as Devadasis; however women who sell sex but who are not traditionally dedicated often adopt the label 'Devadasi' in this region due to some enduring cultural status. --- Participant selection Participants in this study were purposively selected from KHPT's intervention records, which contained demographic information on each woman and man enrolled in the Samvedana Plus program in 2012 who had reported chronic IPV. This was to maximize variation based on inclusion criteria representing the community from which it was sampled, and found to be related to IPV in past routine monitoring data and according to the Research Committee. These included: those reporting poor condom use in intimate relationships; those with or without alcohol use among either partner; having children or not; both below and above 25 years of age (accepted local cut-off for being younger or older as a sex worker); being married or not; working in sex work only or in other work; and women practising different typologies of sex work (home-based, brothel etc.). The final sample consisted of 38 interviews, including 10 couples (interviewed separately), 13 individual female sex workers, and 5 individual male intimate partners, all of whom provided full participation. --- Research tools The in-depth interview tools were developed in multiple stages. After the overarching research questions were agreed upon, workshops were held with the community Research Committee to develop theme areas of interest and questions for each. These included: participants' relationships and supports, problems, hopes and dreams, and expectations in the relationship, followed by experiences, acceptance and methods to address violence and low condom use. The tool was finalized in regional Kannada, the local language, based on the input of our community Research Committee, NGO and academic partners in a three-day participatory workshop led by the university researchers with the community research investigators. The CRIs then completed two pilot in-depth interviews and underwent further training with the second author to optimize the data collection process. The interview guides are available as Additional files 1 and 2 for female and male participants respectively. --- Data collection The female community research investigators went to the villages to meet the potential participants in person and informed them about the study. If they were interested, the CRIs requested their consent. If the potential participant had an intimate partner and she gave her consent to be interviewed, her consent was requested for the male research investigator to meet her intimate partner to seek informed consent for him to be interviewed. Data collection was completed between July and October 2014. The research investigators conducted semistructured, one-to-one, in-depth interviews in Kannada using audio recorders. These were conducted in the participants' homes between the interviewer and participant at a single time point for an hour to an hour and a half. Individual in-depth interviews were used for data collection because of the sensitive and exploratory nature of the study questions that aimed to better understand complex issues surrounding norms, beliefs and behaviours [25]. The CRIs also took down detailed field notes based on the training they received, to cover contextual details, nonverbal cues from the participants or circumstances during the interview that may be relevant to the analysis of the transcript. The recordings and field notes were brought back to KHPT's regional office for transcription and translation by a locally trained translator familiar with the interview guide and data collection processes. The interviews were reviewed for accuracy and completeness in English by the second author prior to analysis, who consulted with the research investigators for clarity and meaning where needed. Further interviews were not collected as the first round of coding revealed recurring patterns on each area of the topic guide, indicating that data saturation was sufficiently achieved. --- Data analysis All de-identified interviews were imported into NVivo 10.0 using pseudonyms (which are used in the Results), and coded by the first and second authors based on the interview questions and evaluation outcomes, adding new codes or sub-codes when necessary. To prevent selectivity based on outside biases or perspectives, the authors who completed coding met with the male and female research investigators to go over their field notes for each interview and to ascertain what they felt were key findings for each interview question across participants. The CRIs presented these key findings in Kannada to the Research Committee and intervention team for further triangulation to assess the important categories. Data was not shared back to the participants because it was felt that this could create issues for maintaining confidentiality around their participation in the study. However we ensured stakeholder input was a central part of the interpretation of results, by relying heavily on a collaborative meaning-making process adapted from participatory methods of card-sorting and flow diagram exercises [25]. For this, the main findings were written onto two sets of cards in Kannada, and two groups placed these on separate poster boards in order of proximity around the central themes of violence and condom use, which were then explained and discussed between the groups. The outputs and discussions of the collaborative meaning-making analyses were used to prioritize and interpret the relationships between codes, and thereby to answer the research questions. --- Results --- Intimate partner relationships, supports and expectations The length of intimate relationships varied from less than one to 30 years in length, with an average of 13 years. Half of the male intimate partners started as a client of the female partner. All but two women had children, and a quarter of these women had children with someone other than their current intimate partner. Only a few women said they saw him more than once a week; most said that it was at least once in a month or two. Table 1 presents the socio-demographic characteristics of the participants. Social circumstances surrounding intimate relationships were important in shaping the relationship dynamics themselves, and the occurrence of violence and condom use within them. All female respondents were responsible for caring and providing for their children, parents, siblings, or grandparentsa reality that sprung from their inability to marry, as one woman expressed: I have shouldered the whole responsibility of the family like a man... I am leading the family by shouldering all the responsibilities even more than a male member. (Sharmistha) Meanwhile, the support that male and female partners provided each other fell squarely within common gender role expectations. Women often said their partners were "taking responsibility of", "maintaining" or "running" her life by providing regular sums of money. Frequently, she also hoped he would pay for clothes or jewellery, her children's births or marriages, or for building her a house or shop. In return, men widely expected their partners to cook and clean for them, satisfy his sexual desires, maintain fidelity by abandoning sex work, stay at home, and generally be "good" with him. Not only were gender role expectations important in intimate relationships, but they were likened to a marriage. Women almost always expected his level of affection and support to be equivalent with his wife's if he were married: He treats me similar to how others look after their wives, and he gets me things that I desire, also he stays in our house for a day and the next day he will stay at their home [with his wife]... For many women, particularly those in long-term relationships, financial support reflected her partner's love and commitment, as Kaveri shared: He does everything from the bottom of his heart. It is proven in his actions. No one supports either financially or in any other way without having love. He doesn't come here just to satisfy his lust. He comes here thinking that he has family over here. For that he offers everything. The framing of the relationship as "like marriage" was also related to aspirations of social acceptance. This was explained clearly by Sharmistha, who observed that many Devadasis were shifting away from sex work in a sociopolitical context where it is criminalized and blamed for the spread of HIV. As a result many have started to rely more on financial support from intimate partners than clients. The importance of having a male partner to support her, as in a marriage, also strongly derived from women's hopes to improve their and their children's futures, as Shanta expressed: I have to give a good life to my children; I have to send them to a good family... People should praise me that, even though she is a Devadasi she gave a good education to her children. I have that desire that they should not face the words as 'her mother is a Devadasi and they too became a Devadasi'. They have to praise me that, she had a lover and he is good with her so he gave a good life to her and her children. Hence, the socio-economic and political circumstances that participants faced were important in shaping the cultural expectations that they and their families held for their lovers. --- Experiences of intimate partner violence Women frequently reported emotional or verbal violence and controlling behaviour, as well as some cases of sexual coercion, though these were not always interpreted as "violence". Respondents most commonly reported physical violence in terms of slapping or hitting, which was often mutual and considered unserious. Usually longer term partners said they had adjusted or matured, leading to less fights or violence. In only a few cases women said that physical violence occurred routinely, with reported incidents of a bloody head or pulled hair. Some men said there was no violence that occurred, even when the female partner reported it, while other men said there were occasional fights where he had slapped or hit her. --- Understandings of intimate partner violence and condom use Gender role expectations and "mistakes" Our analysis revealed that unmet expectations of their partners were most commonly the triggers of fights or beating. Conflicts often occurred when men could not fulfil women's expectation that their partner should be present and provide financial support or material goods, as Sangita responded: We quarrel while giving or taking a few things. I shout at him if he delays his arrival after fixing the time for going to the market or elsewhere. A major reason why violence was said to occur was when the female partner contravened the gender role expectations her male partner had for her, which were universally classified as "mistakes", as Ravi explained: Any man will get angry if women do'mistakes'. He will not get angry if she cooks and serves properly and she obeys him. One woman shared how conflict arose when her partner expected to have intercourse more than she wished: He wants to have sex every other day. I cannot do it frequently. I say to him to come, be with me, spend some time with me and go. Then he argues with me. Women also said violence was triggered if they disregarded men's expectations around modesty, as Sangita stated: They [partners] doubt women, they question, 'why does she get ready beautifully, why does she go out?' They force women to stay at home, they restrict their mobility, so [then] violence takes place. --- Trust, fidelity and sex work Among the most common of the unmet expectations said to cause violence were trust and fidelity between partners. In a few cases, a woman fought with her partner if he showed affection or talked to others, as Ajay related: If my phone remains busy when she was calling, she doubts me saying that I am in another relationship with someone. I become upset then. She blames me like this. I smack her then. She doesn't eat anything for couple of days, she keeps quiet. After a couple of days, we again become normal. Distrust from the man's side related most greatly to his disapproval of her engagement in sex work, which he expected to stop when he was supporting her, as Rani's experience demonstrates: He shouted at me saying that I was not at all trustworthy to him despite the fact that he had done so much for me and I had looked for other clients... It has been happening always. He is catching me redhanded every now and then. He scolds me, beats me, I cry. He makes up with me... How can his love towards me allow him to avoid me completely? A number of women indicated the distress and fear they experienced when their partners were monitoring them through friends or recording their phone calls. Almost all men said they believed that their partner had stopped sex work, but many women continued covertly for financial reasons. There was also noticeable stigma when men and women both discussed how sex work was viewed, using terms like "dark side", "vulgar work", or "whore". Hence, stigma around sex work converged with suspicions around infidelity to produce emotional turmoil and violence within intimate relationships. --- Condom use negotiation Suspicions of infidelity also related to women's requests to use condoms, which could lead to disagreements and violence. Though partners often started as clients, condoms were discontinued once they considered it an exclusive relationship or rejected outright, as one woman reported: He said, 'Hey, don't you know to whom it [condom] is given [for sex work]? Throw that away, why are you giving that to me? I am not that kind of person'. He was shouting like this. He was not using the condoms... Once he came to beat me when I asked him to use condoms; I think I tried couple of times. He never listened to me. --- Indu expressed how condoms symbolized HIV prevention, and by extension, non-exclusive or client sexual relationships: They [lovers] question us on whether [we think] they have any diseases like AIDS and why should we want them to use the condoms. They show their anger if we force them, and say that if we force them they would leave us... What can I do now, if I insist that my lover use condoms he threatens me like that. I can insist that my clients use condoms when I do sex work secretly, but if we force the lover he would leave me; how can I lead my life then? Thus lack of condom use was due to a combination of expectations for fidelity, their perceived association with sex work and HIV, as well as women's fears of violence or losing the man's presence and support. --- Acceptance of violence and its perpetuation Another major issue said to contribute to IPV was the acceptance of violence by women, their partners and wider community; however in some accounts non-acceptance of violence was also seen to cause violence. Overall, violence was seen to be acceptable among both men and women if the female partner had made a "mistake" while the partner was supporting her, as Nandhini stated: Because men will give everything for the household, so they have that right to beat us, but they should beat us only if we do the'mistake'. This appeared to be rooted in normative expectations of a marital relationship, which Ravi expressed: She should not have any contact with other men, because I have taken the responsibility of her life. So I will beat her if she commits'mistakes'. I will be like a husband with [lover's name] as how you [interviewer] are with your wife. Further, women and men stated how violence was acceptable because it was a sign of love, as one intimate partner stated: When we fight each other, I show more love towards her... Yes sir, there is a saying, 'Where there is love, there is a fight'. One woman shared how others also told her to accept violence if he was supporting her, and that she viewed it as a sign of love: Interviewer: Okay when you fight like this, do other women tell you that you should accept it? Respondent: Yes, they tell that I should accept though he beats me, hits me, as he looks after everything for me. They told me to adjust with him... I feel as if I was beaten by my husband, no one else. I: You believe that he is like your husband though he had beaten you and you don't mind at all? R: Yes, because when he becomes angry, he hits me, blames me, but I lead my life believing that he is like my husband and no one else. I think as if he hits me because he loves me. Conversely, a handful of women said that they did not see violence as correct but accepted it for pragmatic reasons such as fear that he would leave, or shame to be exposed in public, as Rani stated: When we accept the man whom we love we have to accept everything from him, he might send us out, we have to accept beating from him and we have to go to him when he calls us on the bed. For that sake they keep us. Further, a couple women shared that if women don't accept violence it will lead to more fights: Those who do not accept violence, they become a victim [of her partner] and those who accept violence they will be happy after compromise. (Parvati) Interviewer: What happens if you won't accept violence from your lover? Respondent: That will be the basic cause for a fight. (Sita) In the male partners' views, many felt violence was only justified if to correct her behaviour, as Ravi said: It is correct to beat when they commit the mistake. If we beat them without any reason or without them doing any wrong, my hands will be cut-off; I will get wounds on my hand. Ajay also discussed that despite its role to correct behaviour, violence was seen as undesirable if frequent and unjustified: If we come and beat the wives every day, it doesn't look good. We will be in the bad books of our neighbours too... Life shouldn't be like 'dosa [flat bread] at everyone's home has holes'. Though less common, a few women felt that they should never accept violence. A distinctive account was given by Kaveri, a long-time CBO peer educator, who said that violence was fundamentally wrong: [We should not accept it] because we treat them very well bearing in mind that they are like a husband. They should also respect us and treat us in a dignified manner. Do you think we have kept them to behave like that? We don't. That is the reason. In a similar way, Rani's lack of acceptance for violence was fuelled by her belief that she needed to remain independent as a Devadasi woman: We have been offering them everything; they take care of us by arranging everything for our needs. That is it. Why should we bend? Whatever it may be, they cannot take care of us as same as they do for their wives. Though anything goes wrong, we won't get the status of wife. Everyone including our lover treats us as 'whore' on one or the other day. In such conditions, why should we get scared about them? Now he has love and affection towards me, who knows, what would happen tomorrow? Thus, the women least likely to accept violence were those who either felt prepared for him to leave, or were confident enough that he would not. --- Methods to address intimate partner violence The participants' levels of acceptance aligned with their recommendations for dealing with violence. As Sita explained, if the mistake was from the woman's side then she must change her own behaviour. However, if she had not made a mistake then her family would come to intervene: If the mistake is from the woman's side it will stop [she will adjust and the dispute will end]; if the mistake is from his side, her family member's will not become quiet. One male intimate partner shared a revealing account of the challenges that Devadasis faced in dealing with violence as unofficial partners: If there is violence or a fight between a wife and a husband, we can go to them and counsel them not to do that. It is our duty as an elder, and it is correct to say so in such a relationship. However, no one says anything to lovers. Similarly, Indu told the interviewer that the neighbours did not lend support because it was considered an illegitimate relationship: When we quarrel like that, we don't let anyone of the neighbours come to know this because they start undervaluing us, saying that we are not wife and husband and we are quarrelling like this as we are in an extra-marital affair. In this context, Kaveri expressed that the power to deal with violence had to come through collective strength within the Devadasi community: We have all combined together. The life of our women is like that only, everyone sees us contemptibly. So we are strong enough as a group. Don't allow any violence to take place because it hurts everyone if any of us is attacked. So we don't get adjusted with everything. We don't like violence from anyone on us. A few women likewise received assistance from the CBO helpline, information on the Domestic Violence Act or advocacy from peer workers for preventing and dealing with violence. --- Discussion Our community-based qualitative study to explore violence and condom use in the intimate relationships of women in sex work sheds light on the complex linkages between vulnerability to IPV and HIV, and broader factors including societal gender norms, expectations and unequal power relations, as well as socio-economic constraints, stigma and illegality surrounding sex work. Many of these echo the factors that have been found to be quantitatively associated with IPV and HIV risk among women generally, though the strength and mechanism of effects surely vary by context [2,3,5,6,9,[26][27][28][29]. Although the inclusion of couples has been rare in past research on IPV and particularly among women in sex work, taking this approach shed new light on the norms and expectations related to condom use and violence held more broadly in society by virtue of being referenced, though often challenged, in intimate relationships. The exploration of the unique situation of Devadasi women, who are most often unable to marry and primary breadwinners in their families, likewise helped to reveal what was considered "normal" or "ideal" behaviours for men and women in relationships. Importantly in this study and similar to others, violence and lack of condom use stemmed greatly from the challenges that both intimate partners faced in reconciling her vocation in sex work with the desire to assume socially-accepted gender roles for exclusive, even marital, relationships [6,28,[30][31][32]. Participants' expectations closely echoed others' findings on gender roles and norms in marital relationships, including the role of men as providers and women as modest, care-taking partners that have been related to domestic violence elsewhere in India [33][34][35][36]. The ascription to marital relationships was also demonstrated when some male and female respondents in our study explained that violence was considered a sign of love, because it reflected that the man cared enough to discipline his partner when she made mistakes as one would in a marriage. Similarly in Kapadia-Kundu et al.'s study among young married women, their husbands and mothers-in-law in south India, women's "mistakes" around gender role expectations, particularly care-taking, modesty and sex, were seen as the primary causes that justified domestic violence against them [30]. Such resonance in the associations between gender roles and violence between study contexts contributes to Bottorff and colleague's call for improved understandings not only in terms of individuals' interactions, but how these speak to meso-and macro-level representations of gender relations in society, and their influences on health [37]. Increasingly researchers have aimed to move beyond individual-level frameworks and a focus on immediate "risk factors" towards understanding socio-structural influences on health. To do this, some have started to apply frameworks of "structural vulnerability" that are derived from theories of structural violence [38][39][40]. Drawing on the concept of "structural vulnerability" encourages a broader appreciation of how one's experience of violence is shaped not only at the individual or interpersonal level, but also through the influence of community and societal factors that involve relations of power [28,38]. In this study, socio-economic, cultural, political and legal issues that have been independently associated with violence in quantitative studies appeared to interact with one another and with individual participants' experiences to shape the degree of "structural vulnerability" to IPV and HIV that they faced [41,42]. First, male intimate partner's financial support was crucial for women who experienced stigma and being low caste as sex workers, within cultural and legal structures preventing them from marrying and criminalizing them, and the economic challenges of supporting their families as unmarried women. Yet women in sex work reported having more difficulty fulfilling the gender role expectations for women and thus were more likely to make "mistakes", particularly by engaging in sex work. These factors also made it more likely for them to comply with the intimate partner and stay with them, in spite of violence or inability to negotiate condoms. This was based on a reportedly widespread societal acceptance of violence if women made "mistakes", as well as the association of condoms with non-exclusive and client relationships. The latter may have been inadvertently fuelled by past HIV prevention messages that focussed solely on condom use with clients, as others have found [5,28,32]. Finally, female respondents felt that society especially blamed them for experiencing violence because the same legitimacy was not attributed to non-formal partnerships as to marital relationships. These results help explain why the categories reported as most contributing to IPV in KHPT's representative quantitative survey of FSWs were their "engagement in sex work", "financial constraint", followed by their "subordinate position in society" [43]. Our findings on the multi-levelled issues shaping "structural vulnerability" to IPV and HIV confirm that continued reliance on behaviour change models to help women avoid "mistakes" or reduce their conditional acceptance of violence are insufficient to address, and may even exacerbate, violence and HIV [38,40,44]. Here it is relevant to draw on the growing application of theories on the interrelationship of social structures and individual agency when trying to understand social influences on health [45,46]. Connell's relational theory provides a pertinent example in which gender as a social structure, which others have described as both rules and resources, can be understood as shaping as well as being shaped by individuals' practices (e.g. IPV) through their position in interpersonal power relations and social conditions [45,47,48]. Related conclusions have been made by Jewkes and colleagues from their research on violence and gender, who state that, "[m]asculinities are embodied and reproduced across the social ecology, and thus [violence] interventions must seek changes at multiple levels" [49]. Our findings suggest that future research should conceive of health issues like IPV not only in relation to multiple levels of factors, as outlined in important socioecological models for IPV [50], but also the interplay across levels, to better conceptualize how sociostructural factors influence and are influenced by individual and collective practices [51]. Effective structural interventions to address IPV vulnerability will likely be those that partner with communities to address pre-existing arrangements of power within the social context; analogous approaches have already been found to be effective for HIV prevention among FSWs [9,52,53]. Through a "structural vulnerability" lens, the involvement of male partners in conflict and anger management must be coupled with work on societal-level platforms to cause sufficient change in structural rules such as gender norms that promote acceptance of violence [54,55]. Further, such structural interventions would continue to work to augment women's access to important material and social resources through access to entitlements, advocacy against stigma and discrimination, economic assistance, and just enactment of laws [54,56]. --- Limitations We aimed to maximize representation in our purposive sample, but selection bias could have arisen if those who consented shared milder cases or were less afraid to discuss violence. However the original intervention database from which the sample was derived only included couples known to have reported IPV and low condom use previously. There was likely some social desir

Background: Research has increasingly documented the important role that violence by clients and the police play in exacerbating HIV vulnerability for women in sex work. However few studies have examined violence in the intimate relationships of women in sex work, or drawn on community partnerships to explore the social dynamics involved. A community-based participatory research study was undertaken by community and academic partners leading intimate partner violence (IPV) and HIV prevention programs in Bagalkot district, Karnataka state, India. The purpose was to explore the experience and understandings of intimate partner violence and HIV/AIDS among women in sex work and their intimate partners in Bagalkot that would inform both theory and practice. Methods: A community-based, interpretive qualitative methodology was used. Data was collected between July and October 2014 through in-depth interviews with 38 participants, including 10 couples, 13 individual female sex workers, and 5 individual male intimate partners. Purposive sampling was done to maximize variation on socio-demographic characteristics. Thematic content analysis was conducted through coding and categorization for each interview question in NVivo 10.0, followed by collaborative analysis to answer the research questions. Results: The results showed that an array of interrelated, multi-level factors underlay the widespread acceptance and perpetuation of violence and lack of condom use in participants' intimate relationships. These included individual expectations that justified violence and reflected societal gender norms, compounded by stigma, legal and economic constraints relating to sex work. The results demonstrate that structural vulnerability to IPV and HIV must be addressed not only on the individual and relationship levels to resolve relevant triggers of violence and lack of condom use, but also the societal-level to address gender norms and socio-economic constraints among women in sex work and their partners.

conceptualize how sociostructural factors influence and are influenced by individual and collective practices [51]. Effective structural interventions to address IPV vulnerability will likely be those that partner with communities to address pre-existing arrangements of power within the social context; analogous approaches have already been found to be effective for HIV prevention among FSWs [9,52,53]. Through a "structural vulnerability" lens, the involvement of male partners in conflict and anger management must be coupled with work on societal-level platforms to cause sufficient change in structural rules such as gender norms that promote acceptance of violence [54,55]. Further, such structural interventions would continue to work to augment women's access to important material and social resources through access to entitlements, advocacy against stigma and discrimination, economic assistance, and just enactment of laws [54,56]. --- Limitations We aimed to maximize representation in our purposive sample, but selection bias could have arisen if those who consented shared milder cases or were less afraid to discuss violence. However the original intervention database from which the sample was derived only included couples known to have reported IPV and low condom use previously. There was likely some social desirability bias or non-disclosure in the accounts, particularly around the extent of conflict and condom use. Community research investigators' perceptions on each respondent's level of comfort and openness in the interviews likely aided in ensuring trustworthiness in these areas. Though the interviewers had to develop their skills in a short period, it was clear that after the first few interviews they became much more in-depth with support from the research team. --- Conclusions Our community-based research study among female sex workers and their male intimate partners in Karnataka exposed the ways that gender role expectations between intimate partners, by virtue of reflecting those widely-held for marital relationships, can lead to IPV and low condom use. Such interpersonal issues were further compounded by the complex interaction of broader factors like stigma and socio-economic constraints, making it more difficult for FSWs to fulfil their partners' expectations, and thereby heightened their structural vulnerability to HIV and IPV. Our results support the potential for sensitively involving not only women but men and community members in violence research and programs, as their views and actions were not isolated, but referenced wider perspectives on gender relations and violence. Future research to understand how complex and multi-levelled issues shape structural vulnerability to health issues by joining forces between community and academic partners will help to better bridge knowledge and action, and thereby inform relevant and effective prevention programs in other contexts. --- Availability of data and materials Original interview transcripts and observation notes have been stored in locked file cabinets at the KHPT Regional Office. De-identified interview data and analyses are held on a secure electronic network. To ensure anonymity and confidentiality, interviews will not be made publicly available. --- Additional files Committee members of Chaitanya AIDS Tadegattuva Mahila Sangha provided inputs on study design and planning, data collection, analysis, interpretation and application of the results. SR was involved in the data analysis and application of results. RT, SM, PJ, PP, and SI were involved in the study design and implementation. SGB, MC, LH, and PB provided technical expertise on the project design, data collection, analysis, and application of results. All authors reviewed, provided input, and approved the manuscript. --- Ethics approval and consent to participate Ethical approval was received from the St. John's Medical College in Bangalore, India, the Research Ethics Board at London School of Hygiene and Tropical Medicine, and the Human Research Ethics Board at University of Manitoba (ethics file #: HS17598 (H2014:108)). Informed written consent was received from all participants to participate, using consent forms approved by these research ethics boards. --- Consent for publication During the informed consent process, all participants' permission was obtained to publish data anonymously from their interviews. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Research has increasingly documented the important role that violence by clients and the police play in exacerbating HIV vulnerability for women in sex work. However few studies have examined violence in the intimate relationships of women in sex work, or drawn on community partnerships to explore the social dynamics involved. A community-based participatory research study was undertaken by community and academic partners leading intimate partner violence (IPV) and HIV prevention programs in Bagalkot district, Karnataka state, India. The purpose was to explore the experience and understandings of intimate partner violence and HIV/AIDS among women in sex work and their intimate partners in Bagalkot that would inform both theory and practice. Methods: A community-based, interpretive qualitative methodology was used. Data was collected between July and October 2014 through in-depth interviews with 38 participants, including 10 couples, 13 individual female sex workers, and 5 individual male intimate partners. Purposive sampling was done to maximize variation on socio-demographic characteristics. Thematic content analysis was conducted through coding and categorization for each interview question in NVivo 10.0, followed by collaborative analysis to answer the research questions. Results: The results showed that an array of interrelated, multi-level factors underlay the widespread acceptance and perpetuation of violence and lack of condom use in participants' intimate relationships. These included individual expectations that justified violence and reflected societal gender norms, compounded by stigma, legal and economic constraints relating to sex work. The results demonstrate that structural vulnerability to IPV and HIV must be addressed not only on the individual and relationship levels to resolve relevant triggers of violence and lack of condom use, but also the societal-level to address gender norms and socio-economic constraints among women in sex work and their partners.

INTRODUCTION Exposure to particular circumstances and experiences across the lifecourse may have a bearing on premature mortality and disease. Knowing which exposures contribute most, and when, can help design and target preventive measures. Systematic reviews of several studies have shown that worse socioeconomic circumstances in earlylife are associated with higher risk of all cause and cardiovascular disease (CVD) mortality [1] and CVD risk. [2] Additional adjustment for adult socioeconomic circumstances and/or adult risk factors generally attenuated these relationships but some association of early-life socioeconomic circumstances remained. Adult socioeconomic factors have been related to mortality in several countries. [3,4] Whilst personal risk factors such as smoking, raised body mass index and raised blood pressure are clearly associated with all cause mortality and CVD, there have been fewer studies on the relationships between parental risk factors and outcomes in adult offspring. There is some evidence of transgenerational effects: parental height was associated with lower risk of offspring CHD, [5] Creactive protein was higher in non-hypertensive offspring of hypertensive parents, compared with non-hypertensive offspring of parents without hypertension; [6] higher parental body mass index was associated with less favourable levels of offspring CVD risk factors; [7] and non-obese offspring had higher C-reactive protein and higher renin if they had obese parents compared with non-obese offspring with non-obese parents. [8] Cardiovascular risk factors are known to track across generations and persist into adult life. [9] Many studies have shown parental lifespan to be related to mortality or survival, for example in Japan, [10] the USA, [11,12] China, [13] Sweden [14] and Iceland. [15] However, comparisons between parental lifespan and other risk factors across the lifecourse have not been made. In this paper, we aimed to find out which type of factors were the most important for determining mortality and CVD risk: mid-life biological and behavioural factors, mid-life socioeconomic factors, parental biological and behavioural factors, early-life socioeconomic factors or parental lifespan. We used a study based in Scotland with information at different stages of the lifecourse, and excellent information on both parents. --- METHODS The Midspan Family Study began in 1996 [16] and involved adult offspring of couples who were both part of the Renfrew & Paisley prospective cohort recruited in 1972-1976. [17] Renfrew & Paisley participants (7049 men and 8353 women) were residents of the two towns, aged 45-64 years at screening, and included 4064 known married couples. The offspring cohort consisted of 2338 participants (1040 men and 1298 women, aged 30-59) from 1477 families, a 73% individual and a 84% family response. [18] Participants in both studies completed a questionnaire and attended a screening examination. The questionnaire included questions on smoking habit, occupation and home address for both generations, and alcohol consumption, exercise, accommodation, car availability, childhood accommodation, car availability in childhood, education and number of siblings for the offspring generation only. Smoking was defined as never, current or former. Social class was derived from occupation [19,20] and used as a continuous variable from 1-6. Social class was defined by the Registrar General's Social Class Schema of I (Professional etc), II (Intermediate), IIIN (Skilled non manual), IIIM (Skilled manual), IV (Partly-skilled) and V (Unskilled). As father's social class was missing for 20 offspring, mother's social class was used for 15, and offspring-reported father's social class used for five offspring. Carstairs deprivation category was derived from the home address and defined as a continuous variable from 1 (least deprived) to 7 (most deprived). [21] Alcohol consumption was obtained from a detailed report of the previous week's drinking and translated into units per week. [22] High alcohol consumption was defined as >28 units per week for men and >21 units per week for women. No exercise was classified as being not very or not at all physically active during usual daily activities and being physically active outside work less than once a week or never. [16] Accommodation in adulthood and childhood was defined as owner-occupied or not, and overcrowding as number of residents greater than or equal to number of rooms. [23] Car availability was defined as one or more cars in the household versus none, and childhood car availability as parental use of a car when the participant was under 16 years. [23] Education was defined as the highest level completed (tertiary or school), years of education, number of Standard grades or O levels (qualifications at age 16) and number of Highers or A levels (qualifications at age 17 or 18). [24] At the screening examination for both generations, blood pressure, height, weight and forced expiratory volume in one second (FEV1) were recorded and non-fasting plasma cholesterol measured from a blood sample. [16,17] Body mass index was defined as weight (in kg) divided by height (in m) squared. Percent predicted FEV1 was defined as actual FEV1 as a percentage of expected FEV1, derived from regression equations based on healthy participants. [25,26] Additional variables, more recently identified as risk factors were measured only in the offspring cohort: highdensity lipoprotein (HDL) cholesterol, triglycerides, C-reactive protein, creatinine, glucose, leg length, waist and hip from which waist-hip ratio was derived. [27,28] Offspring and parents were followed-up for mortality and embarkation (leaving the UK) by flagging at the NHS Central Register which provided dates and causes of death to the end of 2011. Offspring were linked to the Scottish Morbidity Records (SMR) database from screening to the end of 2011. This is a computerised database of all hospital discharges in Scotland. International Classification of Diseases (ICD) version 9 codes 390-459 or ICD version 10 codes I00 -I99, G45 or R58 defined CVD deaths and hospital discharges. Two offspring who did not give permission to follow progress through medical records were excluded from CVD analyses. Father's and mother's lifespan were defined as age at death if deceased or age at end of 2011 if still alive, as in previous studies. [12,29] --- Statistical methods Non-normal variables (triglycerides and C-reactive protein) were log transformed. Variables were assigned to five groups: mid-life biological and behavioural factors, mid-life socioeconomic factors, parental biological and behavioural factors, early-life socioeconomic factors and parental lifespan. Cox proportional hazards models were used to analyse the contribution of the factors in explaining all cause mortality or CVD (defined as main diagnosis of a hospital discharge or CVD death) in offspring. Survival was from date of screening to death, embarkation or the end of 2011 for mortality analyses, and additionally to hospital discharge for the CVD analyses, whichever was first. Proportional hazards assumptions were verified by inspection of Schoenfeld residuals. As tests for interaction with sex were not significant (p=0.27 for all cause mortality and p=0.24 for CVD), models were run with both sexes combined. Null models adjusting for age and sex were run first. Next, each risk factor was added separately to the null model. As there were missing data for some variables, the null models were re-run excluding participants with that missing variable. The best variables were selected by inspecting statistical significance and the decrease in Akaike's Information Criterion (AIC). [30] AIC is a measure of goodness-of-fit of models, with better fit indicated by smaller AIC. It is defined by (-2 x maximised log likelihood) + (2 x number of parameters estimated). The best variables were added to the null model for each of the five groups separately and the decrease in AIC noted. To ensure comparability, these analyses were restricted to participants with no missing data for all the selected variables. Both father's and mother's lifespan were included in parental lifespan models. Hazard ratios were calculated for one standard deviation increase for most continuous variables. Analyses were carried out using Stata release 11, adjusting for clustering of offspring within families. Excluded from all analyses were 30 offspring who had been adopted or were step children, leaving 2308 participants in this study. --- RESULTS There were 2092 (90.6%) offspring whose fathers had died between the ages of 47 and 98, and 1770 (76.7%) offspring whose mothers had died between the ages of 50 and 98. Fathers were still alive for 216 (9.4%) offspring (and aged between 83 and 100) and mothers were still alive for 538 (23.3%) offspring (and aged between 81 and 99). There were 132 (5.7%) deaths in 2308 offspring in the follow-up period. For each variable, AIC for the null model (adjusted for age and sex), AIC for the model which included the variable, and the decrease in AIC between the two models are shown in table 1. Particularly large decreases in AIC were seen for C-reactive protein, FEV1, % predicted FEV1, smoking and car availability. Some variables did not improve the model fit (eg cholesterol, exercise). All the mid-life socioeconomic factors improved the model fit. Variables which were significant and improved the model fit in the individual variable models were selected for the next set of models using groups. The variable resulting in the largest decrease in AIC was chosen where there were similar variables which could be highly correlated, such as systolic and diastolic blood pressure. Analyses for all groups were conducted with complete data for all the selected variables (1997 participants with 111 deaths in the all cause mortality analysis; 1736 participants, 276 with CVD in the CVD analysis). Hazard ratios and 95% confidence intervals for all cause mortality for each group of variables (mid-life biological and behavioural factors, mid-life socioeconomic factors, parental biological and behavioural factors, early-life socioeconomic factors and parental lifespan) are shown in table 3 * For continuous variables, hazard ratios represent 1 standard deviation increase unless otherwise stated Table 5 summarises the decreases in AIC for each model compared to the AIC for null models, from tables 3 and 4. For both all cause mortality and CVD, the largest decrease in AIC was for mid-life biological and behavioural factors, meaning that this group was the most important for both these causes. In both cases, the decrease was substantially greater than with the other groups. For all cause mortality, the next most important was for mid-life socioeconomic factors, followed by parental biological and behavioural factors, early-life socioeconomic factors and parental lifespan. Apart from the most important group, results for CVD were different from those for all cause mortality, with the second most important group being early-life socioeconomic factors, followed by parental biological and behavioural factors, mid-life socioeconomic factors and parental lifespan. These last three groups had markedly smaller decreases in AIC (6.7, 3.7 and 0.4), compared with the first two (37.8 and 17.3). --- DISCUSSION In this well-defined cohort study of adult offspring with information at different stages of the lifecourse and on both parents, biological and behavioural factors in mid-life were the most important factors for risk of all cause mortality and CVD. Although this was not unexpected, the large size of the decrease in AIC compared to the other groups of factors was of interest. Mid-life socioeconomic factors were the next most important for all cause mortality, but early-life socioeconomic factors were the next most important for CVD. Previous studies have shown relationships between adult socioeconomic factors and all cause mortality and CVD, [3,4] and earlylife socioeconomic factors and all cause mortality and CVD. [1,2] In Finland a study of nearly 24 000 men and women found childhood adversity was associated with incident CVD (hospital admission or death) in adulthood, especially in women. [31] In the British Regional Heart Study of 5552 men aged 52-74, the effect of adverse childhood socioeconomic circumstances on fatal or non-fatal coronary heart disease (CHD) risk persisted in older age. [32] Parental biological and behavioural factors were the third most important group for both all cause mortality and CVD, performing better than early-life socioeconomic factors for all cause mortality, and better than mid-life socioeconomic factors for CVD. There have been some studies of intergenerational effects. In the 1958 British Birth Cohort, higher parental body mass index was associated with less favourable levels of offspring CVD risk factors, such as C-reactive protein. [7] In a previous analysis of this cohort, greater parental height was associated with lower risk of offspring CHD, more strongly in mothers than fathers, suggesting possible intra-uterine mechanisms. [5] In the current study, mother's height was included in the all cause mortality model but not in the CVD model. Mother's body mass index, mother's FEV1 and mother's diastolic blood pressure were selected for the CVD model, but not mother's height which had resulted in a very small decrease in AIC. In the current study, parental lifespan explained the smallest amount of all cause mortality and CVD, compared to the other groups of factors. Other studies have shown parental lifespan to be related to mortality or survival. A large study in Japan found inverse associations between mortality from all causes (and from CVD) by father's and mother's age at death. [10] A US study of adults found a survival benefit to offspring for each extra decade of parental survival. [12] A study from China found that familial transmission of longevity existed at very old ages. [13] A study of over 6000 men in Sweden found an inverse association between mortality and father's age at death and a weaker association with mother's age at death. [14] A study of the whole population of Iceland including ancestors suggested a familial component to longevity which could be genetic. [15] Parental lifespan, especially mother's lifespan, was positively associated with better cognitive functioning and inversely associated with self-reported chronic diseases in later life in a cohort of older men and women. [29] The usefulness of parental lifespan as a predictor of mortality depends on what other factors are available; in this study other factors have been shown to be better predictors. From a public health perspective, reduced parental lifespan cannot be altered, but could act as a spur to behaviour change, and to intervention where an early parental death was heritable. The majority of the mid-life biological and behavioural factors are modifiable at an individual level, suggesting action on these factors could help reduce mortality and CVD risk. Whilst early-life socioeconomic factors are not modifiable at an individual level, action can be taken at a societal level, for example in education and accommodation. Negative mid-life socioeconomic factors are also modifiable with action at a policy level rather than by individuals. It is not possible to change one's parental biological and behavioural factors. It is encouraging that the group of factors with the biggest apparent impact on mortality is probably the easiest to modify. --- Strengths There were several more recently identified risk factors available (for example triglycerides, Creactive protein and waist-hip ratio). Unlike other studies, this study did not depend on adult recall for parental risk factors, parental lifespan and some early socioeconomic markers (father's social class and father's deprivation category). Adult recall of father's social class has been shown to underestimate the real associations [33] and offspring recall of parental lifespan could be incorrect. [29] Its main strength is the availability of data for both parents including parental lifespan, in addition to lifecourse data on the participants. --- Limitations The Family study is not fully representative of the general population since its participants were offspring of parents who had both taken part in an earlier study. Since that study included men and women aged 45-64 years, they had to have survived to at least 45 years to take part. Family study participants were likely to be more advantaged and healthier than people who did not grow up with both parents. [34] The participants were offspring from a regional cohort in the west of Scotland, so these results may not be generalisable to other populations. They were healthier than participants of Scottish and English studies conducted around the same time. [16] The main analyses were complete case analyses but no differences were found between the group of participants with missing data and the group included in the analyses, except for sex in the CVD analysis, where 53.0% (95% confidence interval 48.9% -57.1%) of the group with missing data were men and 41.6% (39.3% -43.9%) of the group included in the analysis were men. Thus the results and conclusions were unlikely to have been affected by the exclusions. Since only 5.7% of participants have died, any associations may be different with longer-term followup. The associations found in this study cannot be considered causal. Biological and behavioural factors, such as smoking, are known to be socially patterned, whether by adult or early-life socioeconomic circumstances, [35][36][37] so our groups are not independent, and biological and behavioural factors may be on causal pathways influenced by socioeconomic or cultural factors. [38] Biological and behavioural factors were measured in mid-life but some, such as height and FEV1, are due to influences across the lifecourse. Risk factors, especially when measured longitudinally can explain part of the social gradient in mortality [39] and the current study was limited to one screening. Although AIC may not be able to detect non-linearities,[40] it is suitable for comparing models as in this study. --- Conclusions These analyses have shown that there are multiple influences on health across the lifespan. As midlife biological and behavioural factors best explained both all cause mortality and CVD, continued public health action to reduce these appears warranted. Targeting adverse socioeconomic factors in mid-life and early-life may contribute to reducing all cause mortality and CVD risk respectively. --- ACKNOWLEDGEMENT The Privacy Advisory Committee of ISD Scotland gave permission for use of the hospital discharge data. --- ETHICAL APPROVAL OF RESEARCH Ethical approval for the Family study was granted by the Argyll and Clyde Local Research Ethics Committee (Ref LREC 11/95). Participants gave full consent before taking part. --- DATA SHARING Permission for use of data is via the Midspan steering committee --- Licence for Publication The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive licence (or non exclusive for government employees) on a worldwide basis to the BMJ Publishing Group Ltd to permit this article (if accepted) to be published in JECH and any other BMJPGL products and sublicences such use and exploit all subsidiary rights, as set out in our licence (http://group.bmj.com/products/journals/instructions-for-authors/licence-forms). --- COMPETING INTERESTS --- None declared --- BOX What is already known on this subject? Socioeconomic, behavioural and biological risk factors have all been associated with mortality and cardiovascular disease (CVD) at different times of the lifecourse. Parental risk factors and parental lifespan may also have effects. --- What this study adds? For both all cause mortality and CVD, own biological and behavioural factors were the strongest factors, and parental lifespan the weakest. Of next importance were mid-life socioeconomic factors for all cause mortality and early life socioeconomic factors for CVD. This suggests continued public health action to reduce own biological and behavioural factors. Targeting adverse socioeconomic factors in mid-life and early-life may help reduce mortality and CVD risk respectively.

We aimed to identify which personal and parental factors best explained all cause mortality and cardiovascular disease (CVD).In 1996, data were collected on 2338 adult offspring of the participants in the 1972-1976 Renfrew and Paisley prospective cohort study. Recorded risk factors were assigned to five groups: mid-life biological and behavioural (BB), mid-life socioeconomic (SE), parental BB, early-life SE and parental lifespan. Participants were followed up for mortality and hospital admissions to the end of 2011. Cox proportional hazards models were used to analyse how well each group explained all cause mortality or CVD. Akaike's Information Criterion (AIC), a measure of goodness-of-fit, identified the most important groups.For all cause mortality (1997 participants with complete data, 111 deaths), decreases in AIC from the null model (adjusting for age and sex), to models including mid-life BB, mid-life SE, parental BB, early-life SE and parental lifespan were 55.8, 21.6, 10.3, 7.3 and 5.9 respectively. For the CVD models (1736 participants, 276 with CVD), decreases were 37.8, 3.7, 6.7, 17.3 and 0.4. Mid-life BB factors were the most important for both all cause mortality and CVD; mid-life SE factors were important for all cause mortality, and early-life SE factors were important for CVD. Parental lifespan was the weakest factor.As mid-life BB risk factors best explained all cause mortality and CVD, continued action to reduce these is warranted. Targeting adverse SE factors in mid-life and early-life may contribute to reducing all cause mortality and CVD risk respectively.

INTRODUCTION ''I used to think the top environmental problems were biodiversity loss, ecosystem collapse and climate change. I thought with 30 years of good science we could address those problems. But I was wrong. The top environmental problems are selfishness, greed and apathy... And to deal with these we need a spiritual and cultural transformation -and we scientists don't know how to do that.'' -James Gustave Speth Sustainability science has come a long way in the last 20 years. Since Kates et al. (2001) published their pioneering essay, sustainability science has burgeoned as an integrative and applied discipline. Bringing together economics, social science, ecology and technology studies (Komiyama and Takeuchi 2006), the quest began to solve the most pressing practical and ethical challenges facing the planet and to address them via appropriate policies. Indeed, sustainability has moved from a buzzword to a mainstay concept in nearly all areas of society. However, despite the prominence of sustainability as a concept, planetary trajectories remain deeply unsustainable (e.g. WWF 2016). Now that sustainability science is well established as a field of scholarship, it is timely to consider how it has progressed and where the field needs to go in the future. This article contends that despite substantial analytical advancement, sustainability scholarship has not catalysed the necessary change. The vast majority of sustainability science has focused on the external world of ecosystems, economic markets, social structures and governance dynamics. In doing so, a critical second dimension of reality has been neglected: the inner lives of individuals. We argue here that our inner worlds, such as our emotions, thoughts, identities and beliefs, lie at the root of sustainability challenges and are fundamental to the solutions to some of the world's greatest challenges. Yet, apart from a few scattered examples (e.g. Wolf 2012;Horlings and Padt 2013), the inner life has evaded explicit analysis within mainstream sustainability science because it cannot be understood via traditional scientific tools, approaches and terminologies. Some fields of knowledge have long recognised the importance of inner dimensions of human experience. Aristotle's concept of Phronesis (or 'practical wisdom') is an important concept in classical philosophy. Practical wisdom has an inner source. One acquires an intuitive kind of knowledge, borne of experience, that enables action in uncertain or unprecedented situations (Harding 2009). Another foundational philosophical theory is David Hume's theory of motivation (Hume 1975). Hume asserted that the motivation to perform some action is dependent on both an inner belief that the action is right, and the desire to perform it. However, only more recently have environmental and sustainability scholars started to attend to inner worlds. A topic that has received considerable attention is the notion of value shift as integral to combating the environmental crisis. Martin et al. (2016, p. 6105) suggested that ''we need fundamental shifts in values that ensure transition from a growth-centered society to one acknowledging biophysical limits and centered on human well-being and biodiversity conservation''. This is a call for change progressing from the inside out (see O'Brien 2013). Value shift also is a current topic of debate in conservation science (Manfredo et al. 2017;Ives and Fischer 2017). However, most of this discourse remains focused on interior change at the collective group (or societal) scale-that is, communities, and societies as a whole hold certain collective values which may or may not be conducive to sustainability. To date, scholars seem to have neglected the importance of individual inner lives, including their own. Yet, the inner lives of individuals have been (perhaps unsurprisingly) highlighted by those outside academic circles, especially in spiritual arenas. For example, Pope Francis in his Church Encyclical Laudato Si (On care for our common home) suggests ''the ecological crisis is also a summons to profound interior conversion... I am interested in how such a spirituality can motivate us to a more passionate concern for the protection of our world'' (Pope Francis 2015). Similarly, in ''Ethics for the New Millennium'', the Dalai Lama (1999) argued that greater attention to our inner worlds would both lead to greater individual happiness, as well as provide a sound foundation for a more ethical and sustainable global community. Against this background, our aims for this article are twofold: to highlight the neglect of our inner worlds in sustainability scholarship and practice, and to stimulate discussion of how engaging with our inner worlds may help effect change towards sustainability. We seek to speak as'mainstream' sustainability scientists to other colleagues in our field, hoping to encourage members of our own field to begin to engage more deeply with the notion of inner worlds. In due course, this will necessarily entail bridging gaps to existing work from other disciplines, such as extensive scholarship on individuals' inner worlds from branches of philosophy and psychology. Here, we do not try to complete this journey, but rather lay down arguments for why it will be worthwhile to start taking steps in that direction. To begin, we explore four realms of enquiry and how they have been emphasised in sustainability science over time. --- VIEWING SUSTAINABILITY SCIENCE THROUGH FOUR REALMS OF ENQUIRY Sustainability science has emerged as an integrative arena that brings together many disciplines with a focus on understanding the connections between human and natural systems so as to generate solutions for pressing planetary challenges. Sustainability science has been described as 'use-inspired basic research', highlighting its dual role of generating fundamental understandings of the world and providing practical solutions (Clark 2007). Yet, some domains of reality have been neglected in sustainability science. To understand this more fully, we distinguish between two dimensions of reality: an internally versus externally experienced dimension; and an individually versus collectively experienced dimension. Following Wilber (2000), we recognise that combining these two dimensions yields four domains of human experience, or four ways of generating knowledge about the world. These four dimensions can be labelled as follows: (1) 'it'knowledge of exterior and individual phenomena, (2) 'they'-knowledge of exterior and collective phenomena and their interactions, (3) 'we'-knowledge of internal and collective phenomena and their interactions, and (4) 'I'knowledge of internal and individual phenomena and experiences (Esbjo <unk>rn-Hargens 2010). We show below how sustainability science relates to each of these four dimensions, and argue that the fourth dimension-'I'-has been largely neglected to date. A summary of the four realms of enquiry is outlined in Table 1. --- It: Exterior individual The 'it' domain might be understood as empirical enquiry into the outside world. It focuses on understanding external phenomena, often in a quantitative way, and adopts an objectivist epistemology, which ensures the researcher is kept at a distance from the subject. Questions that are answered through this form of enquiry might relate to the chemical composition of a substance or its behaviour in different settings. This type of knowledge is sometimes connoted with the 'pure sciences', and has important contributions to make to sustainability. The 'it' quadrant is closely connected with 'environmental science', a precursive discipline to sustainability science. Topics of interest may include the amount of carbon stored in soil or the mineralogy of bedrock underlying a river basin. --- They: Exterior collective This dimension is closely related to systems thinking. Sustainability science was established as a field that seeks to ''understand the fundamental character of interactions between nature and society'' (Kates et al. 2001, p. 641). In this way, a systems perspective has been central to the development of the field, focusing on relationships among system elements. These include the biotic and abiotic elements of ecosystems and the influence of social structures such as institutions and policies on these elements. Questions in this domain may include 'what is the effect of the use of agricultural pesticides on river ecosystems?' or 'how do fishing quotas lead to recovery of fish populations?'. In this way, the 'exterior-collective' domain has been the primary focus of sustainability science to date. Major advances in sustainability science have been possible through employing systems thinking (Fischer et al. 2015). --- We: Interior collective The ''we'' dimension describes collectively experienced, internal phenomena, such as social values. In recent years, sustainability scholars have begun to emphasise the importance of intangible and internal dimensions of human experience. Miller et al. (2014) for example, argued for the need to move beyond simply the analysis of sustainability problems to also consider social values. They state that ''inquiries into values are largely absent from the mainstream sustainability science agenda. Yet, at its core, sustainability is a fundamentally ethical concept raising questions regarding the value of nature, responsibilities to future generations and social justice''(p. 241). This recognition of values has been framed in the context of collective groups, and has been tied closely with discourses of reflexive governance and participatory decision-making (Reed et al. 2010;Smith and Stirling 2017). The central argument has been that robust decisions for sustainability in a 'post-normal' world (Funtowicz and Ravetz 1994) require the careful integration of scientific knowledge with diverse and plural stakeholder values and perspectives (Colloff et al. 2017). The assessment of social values has therefore become a rapidly growing field of enquiry in sustainability and conservation (Ives and Kendal 2014;Kenter et al. 2015;Tadaki et al. 2017). Indeed, as Miller et al. (2014, p. 241) state ''As soon as values become a core part of the sustainability research agenda, then the need for participatory approaches follows, since decisions can no longer be based solely on technical or scientific criteria (the domain of expert knowledge) alone''. Questions relevant to this domain include 'what visions for sustainability do different stakeholders have?' and 'what sets of values are embedded in policy frameworks?'. Navigating a plurality of values, in turn, has major benefits for uncovering socially robust trajectories towards environmental sustainability (Kenter et al. 2015;Scholz and Steiner 2015). --- I: Interior individual Finally, the ''I'' dimension relates to the inner worlds of individual people. Unlike the previous three domains, the interior-individual domain has been almost entirely neglected in sustainability science. The inner landscape of both sustainability scholars and members of communities that researchers investigate has been largely overlooked or seen as inaccessible. And yet, we argue that there is a fundamental relationship between our inner lives and the kind of sustainable future that we aspire to create. Science typically removes the subject of research from the investigator, but there is a need for greater integration. We concur with Wamsler et al. (2017) In talking about individuals' inner worlds, we acknowledge that terminology is difficult and often ambiguous. We consider inner worlds to encapsulate entities of values, thoughts, emotions, identities, beliefs and worldviews, amongst others. As such, the term is broad and inclusive, so as to invite exchange of ideas and insights from across academic disciplines. We distinguish inner worlds from phenomena that exist in the 'it', 'they' and 'we' dimensions, which have been the primary focus of sustainability science to date. We recognise that the four domains we outline are a simplified abstraction for the purpose of aiding analysis: often it is in the connections between different domains that human experience of the world is understood. For example, many religious traditions engage interior dimensions via physical, embodied expressions of spirituality in community with other people. Indeed, Buber (1958) famously argued that human experience is summed up in interactions between individuals and objects (I-it relationships) and individuals and other people or the divine (I-thou relationships). Thus, while we discuss the four dimensions discretely, we consider it important to explore relations among these dimensions in the future. --- INNER WORLDS AS A REALM OF TRANSFORMATION Our inner worlds underpin much of how systems function, yet are commonly 'beneath the surface'. One useful image to communicate this is by drawing on the analogy of an iceberg (Fig. 1). According to systems thinking, the deepest and most influential levels of a system are the underlying'mental models': ''the filters through which we interpret our experiences, evaluate plans and choose among possible courses of action'' (Nguyen and Bosch 2013, p. 109). These are invisible but inform the questions we deem appropriate to ask, and underpin the structures, patterns and ultimately events that are observed and measured by scientific methods. The capacity for individuals to suspend assumptions, critique their mental models and potentially adopt new paradigms thus is one of the most powerful ways to dramatically influence sustainability outcomes (Meadows 1999). We suggest that the sustainability crisis is in large part an emergent property of the state of our inner worlds. If we consider only external solutions to 'out there' problems (such as biodiversity loss, climate change, resource exploitation), we will fail to identify some of the most powerful and effective solutions that begin 'in here'. It might be said that the scale of the sustainability crisis extends all the way from planetary systems to the heart and soul of every human being. In this way, we consider the inner life as both an underexplored means to change, and an end in itself. In short, since our inner lives underpin external change, we argue that change in the world must occur (in part) from the inside-out. Yet change must also occur from the outside-in: our inner lives must be shaped by the reality of the social and environmental injustices that are occurring in the world today. In this way, taking our inner lives seriously does not mean separating ourselves from external reality as a form of escapism. Rather, we argue for inner lives that reflect more closely the challenges of sustainability that are before us. --- The inner life as a means to sustainability outcomes There are signs of an opening up of scientific horizons in sustainability science that could accommodate such an appreciation of inner lives. For example, effective action for sustainability is increasingly understood to require not only systems knowledge (technical knowledge of how systems function) but also normative knowledge (how systems ought to be), and transformative knowledge (how to change systems to more desirable states) (ProClim-Forum for Climate and Global Change 1997; Abson et al. 2014). The call for transformative science is premised on a commitment to not only study processes of transformation but to activate them, which necessarily involves shifts in the mindsets of many individual stakeholders, including sustainability scientists themselves (Schneidewind et al. 2016). The strongest step in this direction thus far is in sustainability science education and teaching (Caniglia et al. 2016;Wiek et al. 2016). The Aristotlean concept of Phronesis (practical wisdom) has also been recognised as essential for sustainability transformations (Fazey et al. 2018). We support these recent efforts to expand thinking in sustainability science and suggest that a focus on 'inner worlds' could help to create coherence in this emerging area of thought. How can our inner lives influence sustainability? One vital area is through the motivational resources that exist in our inner lives. This includes deep awareness, building of empathy, and willingness to transcend paradigms. Awareness of our deepest motivations and experiences is perhaps the most fundamental (and grossly neglected) aspect of our inner worlds. Practices of individual reflection reveal awareness of society's values and goals, our own values and goals, and differences between the two. Reflection can also help build empathy and compassion towards others by seeing matters from others' points of view. Contemplation can even enable an expansion of empathy to include people from different cultures and locations, and non-human subjects (wildlife, ecosystems), which has been found to relate to pro-environmental behaviour (Berenguer 2003). This'shifting perspectives' is a fundamental skill in enabling personal paradigms and mental models to be transcended. It is the malleability of personal paradigms that is the most powerful tool for transformative change (Meadows 1999;O'Brien 2018). Inner life, with its values, goals and (often subconscious) desires, can be understood as the deepest driver of behaviour and behavioural change. Because sustainability ultimately requires behaviour shift (Schultz 2011), revealing, understanding (and potentially influencing) inner life is critical for developing strategies for change. Empathy cultivated via contemplation can be translated into action (Ericson et al. 2014). Paying attention to the inner life can 'tap into' something bigger than oneself. Such 'transcendent' motivation is common to all religious traditions, and has sustained action for profound social change throughout history. Nevertheless, while the inner life is a deep driver of behaviour, it is unlikely to be sufficient to generate the profound systemic change necessary for addressing global sustainability challenges in isolation. Any exploration of inner worlds within sustainability science must be done in conjunction with analysis of institutional structures, social context and politics (see O'Brien 2018). A healthy and compassionate inner life as a sustainability goal Not only are our inner lives fundamental to the pursuit of social and environmental well-being, we suggest that the state of our inner lives ought also to be regarded as something worthwhile in its own right. In relation to the image of the iceberg, sustainability is greater than simply the events that occur (such as the use of renewable energy, or the provision of adequate housing). It necessarily includes the systems and structures that enable sustainability to be realised. A society free from violence thus cannot be called'sustainable' if 'peace' is maintained through an oppressive dictatorship. In this way, sustainable actions and outcomes are not truly sustainable if motivated by greed or inner discord. At present, many sustainability strategies do not challenge the underlying values that contribute to it, but seek to work with these values (Manfredo et al. 2017). Tax incentives for 'green' products (e.g. electric vehicles) implicitly appeal to greed and materialism in order to shift behaviours. Similarly, sustainability scholars and activists can be driven by insecurity, fear or hubris just as much as other professionals. What if we extended to our own lives the aspiration of wellbeing and flourishing that we strive for in our sustainability work? Exploring inner lives, and working towards sustainability from the inside out, may reveal immaterial sources of lasting contentment and well-being, with positive flow-on effects for the world at large. --- HOW COULD INNER LIFE BE APPROACHED IN SUSTAINABILITY SCIENCE? Increasing recognition of the inner life in sustainability science is likely to be a long process. This article does not presume to provide a simple blueprint for how to address the neglect of the inner life. Yet, we offer below some starting points to a new pathway, which we hope will open conversation among sustainability scholars and practitioners. We consider that the concept of 'leverage points' for sustainability transformation (Abson et al. 2017;Fischer and Riechers 2019) is a useful framework by which this can be explored. According to Meadows (1999), complex systems possess different 'leverage points' whereby interventions can affect a certain amount of change. Shallow leverage points focus on existing system parameters. They are easily acted upon but unlikely to bring about transformative change. In contrast, deep leverage points tackle underlying worldviews, paradigms and values-they are more difficult to work with, but have much stronger transformative potential. We argue that a focus on the inner life has major potential to function as a domain for deep leverage for change. To operationalise this, we therefore call for (i) an expansion of the language used in framing sustainability, (ii) greater consideration of the inner life in sustainability research, and (iii) enhanced awareness and cultivation of the inner life in practice. --- Framing and language The language used to articulate sustainability concepts and problems often betrays highly normative perspectives on the framing of sustainability. Lakoff and Johnson (1980) demonstrate that the language we use gives us clues to deep and collectively-held conceptual frameworks (and thus to the paradigms that shape them). We suspect that language contributes to a cycle, either virtuous or vicious: language expresses paradigms, and reinforces them. A change of language, in turn, has potential to challenge deeply held beliefs, and potentially shift them. Indeed, language might be considered a 'deep leverage point', acting to influence system paradigms. For instance, the term ''sustainability science'' implies a rational approach to the pursuit of maintenance. In contrast, other terms might connect with a deeper desire and inspire us to seek and create the futures we want. Rabinow (2011, p. 217) refers to a ''flourishing'' existence, supported by a science of ''care''-ful ''practices, relationships and experiences''. Stengers (1997, p. 113) writes about (re)awakening a ''jouissance'' in science, which has potential to bridge the gap between the ''intensity'' of scientific discovery, and the ''sterilizing'' language often used to express it. Wahl (2016) also promotes the concept of ''regenerative cultures'' over sustainability. Even use of the term ''the environment'' has recently been challenged within public discourse (Monbiot 2017). Given the importance of language, we call for a greater exploration and expansion of terminology in sustainability that engages both the head and the heart. The term we introduced in this article-inner worlds-is deliberately broad and encompasses many dimensions of internal human phenomena; including, as we outlined above, emotions, thoughts, identities and beliefs. While traditional science typically strives for great conceptual precision, seeking to create sharp boundaries between related concepts (e.g. the distinctions between attitudes, beliefs and values in psychology; Rokeach 1968), such precision can at times constrain integrative enquiry and thereby obscure important insights. Scientific language has also not arisen to develop mindfulness and empathy. We offer the term ''inner worlds'' as a way of holding together multiple dimensions of ''human being'' that are otherwise neglected in sustainability science. Similarly to the term ''resilience'', the vagueness of the term ''inner worlds'' thus could be considered an asset, in accordance with Strunz's (2012) argument that a certain degree of conceptual vagueness fosters creativity and enables integration across different knowledge domains. The term ''inner worlds'' thus could help to bring together existing insights, and perhaps generate new ones, with tangible benefits for both sustainability research and practice. --- Research There are a number of potential research questions salient to how our inner worlds connect with sustainability. We explore a few here, recognising that this list is nowhere near exhaustive. The first set of questions refers to how inner lives of individuals relate to individual behaviours towards sustainability. One dimension of the inner life that is particularly pertinent is that of values. While much has been written in social psychology on the relationship between personal values and behaviours (Dietz et al. 2005;Steg and Vlek 2009), the focus in the context of sustainability has been on values as they exist in a certain population or in a collective sense. For example, there is a voluminous literature on the structure and persistence of human values across different cultures and socio-political contexts (e.g. Schwartz 1994;Inglehart et al. 1998). In contrast, there has been little exploration of personal values as preconditions for action in support of transformative change for sustainability (Ives and Fischer 2017). The importance of personal values in the context of organisational leadership is one area where the relationship of personal dimensions to higher level systemic change is directly relevant (e.g. Hemingway and Maclagan 2004). Of course, values are only one facet of the inner life and should not necessarily be separated from other dimensions of inner experience. There is also a need to explore how other conditions of people's 'inner' lives (such as emotional wellbeing, or capacity for reflection) can enable and motivate actions for sustainability. One area of promising research is the relationship between personal character strengths and virtues and sustainable behaviour (Corral-Verdugo et al. 2015). The second field of research is how inner worlds can be shaped and transformed to align more with sustainability outcomes. The capacity for personal values to be shaped and shifted intentionally is gathering greater interest (see Raymond and Kenter 2016), and there is a need to explore how such value shift might enable sustainability transformation (Ives and Fischer 2017). The fostering of 'virtues' is another growing field of study that relates deeply to sustainability. Traditional western virtues include humility, kindness, patience, diligence, temperance and charity. Individuals who have inner lives characterised by these qualities may, arguably, be positioned to pursue sustainability passionately and persistently. The need to emphasise virtues in education is increasingly recognised, as the inadequacy of knowledge and skills alone in contributing to a healthy and flourishing society is acknowledged (see Arthur et al. 2017). How such virtues might be cultivated within individuals and how they relate to change for sustainability is therefore an arena ripe for further research. A third arena for further research is how institutions and organisations that relate to the inner life might promote sustainability. This includes religious groups and communities, and their institutionalised practices such as mindfulness, meditation and contemplation. With 84% of the global population professing some kind of religious faith (Pew Research Centre 2017), religious institutions are ideally positioned to engage with the inner lives of individuals as they relate to sustainability and to promote inner change. There is therefore a need for research into how various spiritual and religious beliefs and practices might motivate or constrain action for sustainability (Hitzhusen and Tucker 2013). While research has shown somewhat complicated relationships between religiosity and pro-environmental behaviour (Gifford and Nilsson 2014), there is undoubtedly a need to engage spirituality with the sustainability crisis, and religious institutions are ideally situated to do this. As Orr (2002) noted ''The transition to sustainability will require learning how to recognize and resolve divergent problems, which is to say a higher level of spiritual awareness''. Finally, there is a need for research on how inner worlds relate to existing theories of social change. Many theories have been proposed, investigated and operationalised. These do not need to be superseded by a 'new' theory of change focused on inner worlds, but rather, understanding inner worlds and their relationship to other quadrants (outlined in Sect. 2) opens up a broader perspective from which new questions can emerge. We have already introduced the concept of 'leverage points' as a theory of change grounded in systems thinking (Abson et al. 2017). In this context we see inner worlds as sources of leverage as they can connect observation to realisation and action. They enable dynamics in the other quadrants to be seen and their significance felt, including dynamics of power (the interior-collective dimension 'we'), systems of injustice and unsustainability (the exterior-collective quadrant 'they'), and changes in the biophysical world (the exteriorcollective quadrant 'it'). Actively incorporating inner worlds into our analyses would mitigate against the risk of divorcing interior and individual catalysts for change from the larger set of contexts deserving of change. Similar conceptual and empirical research should be done to relate inner worlds to other theories of social-ecological change. Below are a few examples. First, psychologically-grounded causal theories of behaviour, such as Ajzen's Theory of Planned Behaviour (Ajzen 1991) and Stern's Value, Belief, Norm Theory (Stern and Dietz 1994) continue to dominate literature on behaviour-change policy. Consideration of inner worlds could inform how deeply held values are formed and shaped over time, and in response to human interaction and various contexts. Second, social practice theory emphasises the importance of routines and behaviours within distinct social contexts (Shove 2010). There is an opportunity to explore how such practices stem from and influence individuals' inner lives. Finally, social innovation theory explores the emergence of new social solutions to problems within various institutions (Moore and Westley 2011). Considering inner worlds could highlight the inner 'preconditions' for innovation and the meanings of these innovations as they emerge. --- Practice Of equal importance to undertaking research on the inner life and its relevance to sustainability is the fostering of healthy inner lives of sustainability professionals. In essence, there is a need to 'lower the water line' of the iceberg (Fig. 1)-to increasingly expose those invisible dimensions (such as mental models and emotions) that influence the external activities and events we pursue. Structural change in academic institutions may be necessary to combat the increasingly competitive, output-driven and performance-oriented cultures in many universities (Fischer et al. 2012a) to help promote inner health and well-being of faculty staff. Practically, this may entail providing opportunities (both places and times) for reflection and informal exchange with colleagues (Fischer et al. 2012b), promotion of training and development in inner virtues and inner transitions (both for faculty and students), and prioritising aesthetics and meaning in work. Such a shift is may be enabled and reinforced by modifying existing systems and processes. These could include criteria for academic honours and promotion incorporating elements of personal character strengths, or funding bodies looking beyond criteria related to academic output and external 'impact' to also reward sensitive, respectful modes of working and provide resources for cultivation of inner health and well-being. Innovative teaching programmes are likely to be an important part of a sector-wide shift towards appreciation of inner worlds, both within traditional institutions (e.g. teaching on sustainability and inner transformation at Lund University, or the role of inner worlds in environmental leadership at the University of Nottingham), as well as pioneering educational platforms (e.g. Ubiquity University's Wisdom School). Personal practices are also likely to be important in embracing inner worlds in sustainability. In the context of a ''post-truth'' society that is increasingly skeptical or dismissive of scientific evidence, there is a need for sustainability scholars and practitioners to take time to create space to build the inner resources that will sustain action over the long term. Practices of solitude and silence have long been held as vital to inner health and wellbeing amongst many religious traditions. Mindfulness techniques have been shown to reduce stress and promote mental health (Grossman et al. 2004), and the potential for these to contribute to sustainability has been recognised recently (Wamsler et al. 2017;Wamsler 2018). We are interested in how participation in these practices could help bring together the inner reality of our lives with the kind of world that sustainability scientists aspire to see. --- CONCLUSION The persistent degradation of the biosphere despite growing scientific knowledge suggests that there is a need for sustainability science to take a look at some of the deeper drivers of anthropogenic planetary change. We have argued that sustainability science has neglected an important dimension of human experience-the inner worlds of individuals. These have the potential to fundamentally shape human behaviour and possibly even the functioning of social systems. We call for greater recognition of the inner life in sustainability science and for a new agenda of research and practice that highlights the inner revolution that is needed. With a greater awareness and activation of inner resources for sustainability, we might just locate the transformative capacity to bring about the change necessary for a safe, just and sustainable future for humanity and the planet.

In the context of continuing ecosystem degradation and deepening socio-economic inequality, sustainability scientists must question the adequacy of current scholarship and practice. We argue that preoccupation with external phenomena and collective social structures has led to the neglect of people's 'inner worlds'-their emotions, thoughts, identities and beliefs. These lie at the heart of actions for sustainability, and have powerful transformative capacity for system change. The condition of people's inner worlds ought to also be considered a dimension of sustainability itself. Compassion, empathy and generosity, for example, are personal characteristics that mark individual expressions of sustainability. Sustainability science must take inner life more seriously by considering how language shapes and is shaped by paradigms about the world, prioritising enquiry into how spirituality, contemplation and sustainability transformation relate, and encouraging scholars and practitioners to intentionally cultivate their inner worlds to strengthen inner resources necessary for addressing sustainability challenges.

INTRODUCTION Colorectal cancer (CRC) is the third most common cause of cancer and the second most common cause of cancer death in the USA. Although CRC incidence and mortality rates have been declining, reflecting improvements in early detection and treatment, 1 ethnic/racial and socioeconomic disparities still persist. These disparities are due, at least in part, to disparities in CRC screening uptake. --- Conceptual Model This research is guided by the social contextual model put forth by Sorensen et al., 11 which illustrates the role of social contextual factors in influencing health behaviors across multiple levels of influence. Provider's recommendation for CRC screening has consistently been one of the strongest predictors of CRC screening behavior. [12][13][14][15][16][17][18][19] However, the association of screening status and the provider's understanding of patient's social context, independent of provider's recommendation for CRC screening, has not yet been examined. To address this gap, we examined the relative contribution of provider's understanding of patient's social context to current CRC screening adherence, independent of receiving a screening recommendation. We hypothesized that participants who had received a recommendation for CRC screening would be more likely to be current with CRC screening than those who had not received a recommendation. We further hypothesized that those who reported that their provider knew their social context well would be more likely to be current with CRC screening than those who reported that their provider did not know their social context well, controlling for provider's recommendation. --- METHODS --- Study Design and Sample This study used baseline data collected in 2004-2005 from Open Doors to Health, a CRC prevention cluster randomized control trial. Twelve urban public housing sites in Boston were the primary sampling units, and individuals within housing sites served as secondary sampling units. Participants provided informed consent and completed an interviewer-administered survey in either English or Spanish. The study protocol was approved by the Harvard School of Public Health Institutional Review Board. Study recruitment and sampling has been described in full detail elsewhere. 20,21 Eligibility criteria included (1) residence in a participating housing site, (2) at least 18 years of age, and (3) fluency in English or Spanish. Baseline surveys were obtained on 1,554 participants. Analyses included Black, White, and Hispanic participants aged 50 years and over who had completed the baseline survey, including questions pertaining to CRC screening and provider's understanding of the patient's social context (n=695). Due to cognitive or other types of disabilities, 12% of participants completed a shorter version of the survey which did not include these questions. Ninety-seven percent of the overall sample had health insurance that covered CRC screening. --- Measures Provider's recommendation for CRC was assessed by asking participants whether a health provider ever told them that they should be screened for colon or rectal cancer (yes/no). Participants reported on their regular provider's understanding of their social context, operationalized as how well they felt their regular provider knew (a) their responsibilities at work, home, or school; (b) their worries about health; and (c) them as a person and their values and beliefs. 11,22 Responses included not at all, a little, somewhat, and very well for each item. A point was given for each question for which the participants felt their provider knew their social context somewhat or very well, and a summary score (range, 0 to 3) was computed. A median split of two categories was then created (zero to one item=did not know patient's social context well; two to three items=knew patient's social context well) for these analyses, based on the distribution of responses. Sociodemographic variables including age, race, gender, income, education, immigrant status, English as a first language, and number of doctor visits in the last year were also assessed. The outcome variable was CRC screening adherence. 23 Patients were considered current if they reported having a fecal occult blood test (FOBT) within 1 year of the survey, flexible sigmoidoscopy within 5 years, and/or colonoscopy within 10 years, as per American Cancer Association Screening guidelines. 24 Data Analysis Based on the cluster design, data for all analyses were weighted up to the population size within each housing site (weighted sample size=1,029). Based on the bivariate associations and consideration of potential confounders, cluster randomized multivariable logistic regression models were estimated for the dependent variable, using SUDAAN 9.01 and SAS 9.1 statistical software. 25 --- RESULTS --- Sociodemographic Analyses Table 1 presents the demographics for the sample. Participants were predominately non-White, female, insured, and had low levels of education. --- Bivariate and Multivariable Analyses for Current CRC Screening Table 2 presents the bivariate analyses. Participants who had received a recommendation were significantly more likely to be current with their CRC screening compared to those who did not receive one (p=0.001). Similarly, participants who reported that their provider knew their social context well were significantly more likely to be current with CRC screening compared to those who reported that their provider did not know their social context well (p=0.004). Table 3 displays the odds ratios and 95% confidence intervals for the adjusted multivariable model of current CRC screening. After adjusting for potential confounding factors, the association between receiving a recommendation and being screened remained significant (p=0.001). The association between provider's understanding of patient's social context and current CRC screening also remained significant after adjusting for age, number of times visited the doctor per year, and provider's recommendation. --- DISCUSSION The aim of this article was to examine whether the provider's understanding of the patient's social context was associated with current CRC screening, independent of receiving a recommendation to be screened among a sample of low-income, multiethnic adults. Consistent with previous findings, [12][13][14][15][16][17][18][19] provider's recommendation emerged as a strong predictor of CRC screening. Additionally, as hypothesized, provider's understanding of patient's social context was significantly associated with CRC screening, independent of recommendation. Several limitations must be noted. Participants were asked about screening recommendations made by any health care provider, while provider's understanding of patient's social context was assessed specifically for one's regular provider. of the high level of health insurance coverage among this sample of public housing residents, access to health care was not generally a barrier and thus may have impacted on the study findings. However, this is an ideal situation in which to study the association between provider's understanding of patient's social context and health care utilization because the results are less likely to be influenced by issues of access. Our findings indicate that provider's understanding of their patients' social context may facilitate screening adherence. Although longitudinal data would be needed to assess any causal patterns, it is possible that providers who know their patients better are more effective at participating in the shared decision making that is an important part of addressing cancer screening. These providers may also be more effective at removing barriers to screening because they have a better understanding of their patients' social context. Thus, our findings are useful in emphasizing important aspects of patients' social context to which providers need to attend to in their clinical encounters in order to increase CRC screening. --- Although we considered restricting the recommendation question to the current regular provider, it is possible that another provider would make a CRC screening recommendation (e.g., a nurse at a blood pressure clinic), which would cue the patient to talk with their regular provider about screening. Currently, in most states, CRC screening requires a referral from one's primary care provider. Thus, it is likely that there was high convergence between the referent for the provider and screening questions. This study is also cross-sectional, and thus, causal interpretations cannot be made. The results can only be generalized to similar populations. Of note, likely because

The primary aim of this paper was to explore whether provider's understanding of patient's social context is associated with screening uptake, independent of provider's recommendation. Baseline data were collected in 2004-2005 from a cluster randomized control trial in 12 low-income housing sites. Participants included 695 lowincome, multiethnic adults aged 50 years and over who were primarily insured (97%). Provider's recommendation was significantly associated with current adherence to colorectal cancer (CRC) screening. Provider's understanding of patient's social context, as operationalized by how well participants felt that their provider knew (a) their responsibilities at work, home, or school; (b) their worries about health; and (c) them as a person and their values and beliefs, was also significantly associated with current adherence to screening, independent of provider's recommendation. Participants who reported that their provider knew them well on two or three items were significantly more likely to be current with CRC screening compared to those who reported their provider knew them well on only one or none of the items (odds ratio=1.56; 95% confidence interval=1.06, 2.29). Our findings indicate that provider's understanding of patient's social context, independent of provider's recommendation for CRC screening, contributed to adherence to CRC screening in this low-income, multiethnic population.

Introduction The healthcare system in China has undergone dramatic reform since the establishment of the People's Republic of China in 1949. The most recent round of healthcare reform started in the year 2009, aiming to provide universal coverage of basic healthcare by the end of 2020 and emphasizing government-led, people-centered healthcare [1]. Although China has made great achievements in increasing healthcare accessibility, improving financial protection, and reducing health inequity since 2009, satisfaction of the health system did not increase from 2010 to 2016 [2]. The increasing burden associated with an ageing population leads to gaps in fulfilling the needs of healthcare services [3]. According to the seventh national census in 2020, the population aged 60 and older accounted for 18.70% of the population in China [4]. China is moving toward a superannuated society, bringing about a heavy disease burden and increasing the need for care among older people with chronic diseases and disabilities [5]. As the proportion of the older population dramatically expands, difficulties in the accessibility and affordability of healthcare services exist in terms of high health expenses, low benefits of health insurance, and a shortage of healthcare services [6]. The difficulties in the accessibility and affordability of healthcare services have posed persistent challenges to the Chinese government ever since the 1990s [7]. In China, the discontent of the public towards the healthcare system is well-known as Kan Bing Nan, Kan Bing Gui ("<unk>, <unk>"), which means an "insurmountable access barrier to health care, insurmountably high health costs" [8]. The saying has been adopted in the government policy documents, and solving the difficulties in accessibility and affordability has been treated as the main goal of healthcare reform [9]. However, as the major policy outcomes of healthcare reform, little is known about the public perception of the accessibility and affordability of the healthcare system after years of reform, especially among vulnerable groups such as older adults. Therefore, this study aims to explore the following questions: First, how do older adults perceive the accessibility and affordability of the healthcare system in China? Second, what are the factors that are associated with their perceptions of accessibility and affordability? --- Literature Review 2.1. Healthcare System Reform in China The social problem of Kan Bing Nan, Kan Bing Gui caused widespread social discontent with healthcare and gained serious political attention at the end of the 1990s [7,8]. In response to the public discontent, after 2009, China's efforts to build up Universal Health Coverage (UHC) emphasized pursuing better equity, quality, and affordability [10]. To improve access and quality while lowering the cost of healthcare, the new round of healthcare reform in 2009 focused on three primary areas, which are health insurance, drug pricing, and public hospitals [9]. Many policies have been implemented to expand insurance coverage and improve the efficiency of healthcare delivery by establishing primary healthcare (PHC) facilities [3]. First, to solve the problem of affordability, the government endeavored to expand the coverage of healthcare insurance. Three insurance schemes, i.e., the New Rural Cooperative Medical Scheme (NRCMS) in rural areas and the Urban Employee Basic Medical Insurance (UEBMI) and Urban Resident Basic Medical Insurance (URBMI) in urban areas, together constituted China's basic social medical insurance system [11]. These schemes are divided into individual accounts and pooled funds. The medical expense incurred in a settlement period (generally 1 year) will be calculated by stages and paid cumulatively according to the hospital grade (community, primary, secondary, tertiary). Then, the pooled fund and individuals share the expenses following the prescribed proportion [10]. Second, to solve the problem of accessibility, the government also increased investment in primary healthcare facilities to promote equal access and universal healthcare coverage [12]. As a major target of health reform in 2009 and a core component of the government's "Healthy China 2030" strategy [1], primary care facilities are expected to alleviate the over-utilization of secondary and tertiary care by tackling frequently encountered medical conditions at community levels [13]. The reform has had significant achievements in increasing health insurance coverage, lowering out-of-pocket expenditure, and providing essential public health services to all citizens through primary healthcare facilities, which make the healthcare services be more affordable and accessible [14]. However, many challenges still exist in the affordability and accessibility of healthcare services in China. Generally, the current healthcare system seems not to favor vulnerable groups such as older adults. First, benefits and reimbursement rates vary across different health insurance schemes, which suggests disparities in benefits between rural and urban residents, different socio-economic groups, formal sector employees and others, and disparities due to age differences of insured individuals in each pool [8]. Rural and unemployed residents receive fewer benefits from healthcare insurance schemes because NRCMS for rural residents and URBMI for urban unemployed (including the elderly and children) have a lower actual reimbursement rate than that of UEBMI for urban employees (and retired employees). According to an existing study, the estimated actual reimbursement rate for urban employees exceeded 63% while the estimated actual reimbursement for urban residents and rural residents was about 37% in the year 2016 [10]. Second, people still prefer to seek medical services in hospitals rather than primary care institutions, which leads to overcrowding in larger hospitals and rapidly rising service costs [15]. The underutilization of primary care facilities is very common countrywide [16]. Older adults are susceptible to the impacts brought about by these disparities in health insurance benefits and tiered healthcare service utilization. Generally, the reimbursement rate of primary health institutions in many provinces is higher than that of tertiary institutions, and local government intentionally tries to encourage patients to utilize primary health institutions through such medical insurance policies [17]. However, elderly patients are more likely to have diseases requiring comprehensive checkups, treatments, and medicines provided by large hospitals, which may not be listed as reimbursable items in the healthcare insurance scheme they enrolled in and therefore receive lower actual reimbursement rates [18]. Due to declining physical and mental conditions, they have an increasing demand for inpatient and other services for disabilities/chronic diseases in tertiary hospitals, which make them prone to catastrophic health expenses [19]. Health care for the elderly is under pressure due to the disparities of different healthcare insurance schemes, insufficient resources available for elderly care, as well as the demand for longterm medication and outpatient coverage [20,21]. According to these existing studies, it seems that older adults are likely to encounter difficulties in the accessibility and affordability of healthcare services due to their vulnerability during the aging process. --- Public Perceptions of the Healthcare System The social problem of Kan Bing Nan, Kan Bing Gui is not only about actual difficulties in accessibility and affordability faced by the public, but it also represents public perceptions constructed by academic literature and policy proposals vacillating between the state or market debates [22]. Therefore, in addition to the objective evaluations of accessibility and affordability, it is also important to examine the perceived accessibility and affordability as a dimension of the public perceptions of the healthcare system. One of the prolific research areas exploring public perceptions of the healthcare system is public satisfaction research. Public satisfaction is a general subjective evaluation of the healthcare system for all citizens, including users and non-users of healthcare services [23]. Existing studies found that objective measures, such as access and expenditure, as well as the subjective expectation of these objective measures constitute the subdimensions of satisfaction with the healthcare system [24]. In other words, satisfaction with the healthcare system may depend more on specific cultural and political factors external to the actual performance of the system itself [25]. For example, a study on public opinion in 61 countries found that peoples' perceptions of accessibility and affordability were significant dimensions of overall public opinions towards the healthcare system [26]. Citizens' expectations and political debates may influence the public satisfaction with the overall healthcare system or with specific domains such as affordability and accessibility [27]. These cross-national studies revealed the influence of subjective factors on the formation of the public perceptions of the affordability and accessibility of the healthcare system. Some empirical studies in the Chinese context demonstrated the role of subjective factors, such as the beliefs of the healthcare system, in constructing the public perception of the healthcare system. In China, satisfaction with the healthcare system is strongly associated with the subjective belief in personal responsibility for meeting healthcare costs while negatively associated with perceptions of unequal access as well as unethical service providers [28]. Mass media in China have long been reporting negative coverage and dramatizing unethical misconducts of hospitals and health professionals, which create negative public opinions towards healthcare services [29][30][31]. Generally, both crossnational studies and existing empirical studies in the Chinese context revealed that public perceptions were socially and culturally constructed beyond the actual performances of the healthcare system. --- Patient Experience and Public Perception Patients are users of healthcare services, and thus patient experience is associated with the public perception of the healthcare system. For example, in a study of 21 European countries, those who had used healthcare services in the last 5 years were more satisfied than those who had not [32]. Patient experience can be conceptualized as both patients' experiences of care and as feedback from patients about those experiences [33]. There are many debates about the extent to which satisfaction with healthcare services is explained by patient experience. Some studies have concluded that much of the variation in the satisfaction with health services is explained by patient experience, while other studies posit that patient experience accounts for only a small fraction of the variation in health service satisfaction [34]. Patient experience can be measured by various items, such as waiting time, doctor-patient communication, and staff responsiveness [33]. Current studies on patient experience in China mainly explore the doctor-patient relationship and communication. In China, public hospitals are not fully sponsored by the government to pursue public interests or privately owned profit-maximizing entities. As a result, sometimes, the healthcare providers tend to pursue their own interests through overtreatment, overprescription, and getting red-envelope money (i.e., a very big tip intended to ensure good service) from patients [35]. These misconducts caused difficulties in the accessibility and affordability of healthcare services for patients while bringing about negative patient experiences. For instance, a prevalently and commonly observed phenomenon is called 'guanxi jiuyi' (medical guanxi), which involves patients leveraging their 'guanxi' (personal connections) to establish mutual relationships with physicians when seeking healthcare. Patients and their families believe this facilitates access to experienced medical specialists and reduces the financial and medical risks of overdiagnosis and overtreatment [36]. Such distorted doctor-patient relationships usually lead to a perception of unfairness and injustice in the medical care system [37]. However, current studies mainly focus on trust issues in doctor-patient relationships while seldom discussing other aspects of patient experience, such as patients' feedback on waiting time and service quality. Moreover, little is known about the relationship of patient experience at the individual level to overall public perception or a specific dimension of public perception, such as perceived accessibility and affordability at the macro level. --- Conceptual Framework In sum, our research aimed to explore the perceived accessibility and affordability of the healthcare system, which also refers to the social problem of Kan Bing Nan, Kan Bing Gui among older Chinese adults. Based on existing studies, we explored the objective and subjective factors in constructing the public perceptions of the accessibility and affordability of the healthcare system among both elder users and elder non-users of healthcare services, including healthcare expenditure variables, patient experience variables, financial protection variables, and social evaluation variables. Objective factors refer to the factual data, including out-of-pocket medical expenditure in the last year and healthcare insurance schemes enrolled in by the respondents. Subjective factors are self-reported and evaluative data, including the self-reported medical burden, patient experience, and social evaluation variables. As we have reviewed the existing literature above, we generalize some main factors that may influence the perceived accessibility and affordability from existing studies to construct our conceptual framework in Figure 1. We categorize the factors into four sets of variables, which include both subjective and objective factors and correspond to our literature review from Sections 2.1-2.3. --- Method --- Data and Sampling The data for this study were obtained from the Chinese Social Survey (CSS) open database. The CSS is a comprehensive cross-sectional survey project conducted by the Institute of Sociology at the Chinese Academy of Social Sciences and provides nationally representative data. The CSS survey is a biannual longitudinal survey to collect labor and employment data, family and social life information, social a itudes, and other aspects. A multi-stage stratified probability sampling method was used to select and interview a representative sample that reflected the profile of households across China. To ensure data quality and a high response rate, a research team from universities and scientific research centers was established, who provided 3-5 days of training to the data collectors and developed the quality control system. Wri en informed consent was obtained from all respondents, and the study objectives were also clearly introduced. The study utilized the unaltered dataset from the CSS 2021, encompassing 100,136 individuals residing in 584 villages/communities and 152 counties/districts across 64 provinces/cities/autonomous regions in China. Given the focus on older Chinese adults, the selection criteria were limited to respondents aged 60 years or older (born in or after 1961). Subsequently, the variables were carefully screened, matched, and processed, with the respondents' birth year recorded in the questionnaire, resulting in a final sample size of 2169 data points. --- Variables --- Dependent Measure In this study, the perceived affordability and accessibility of the healthcare system was the dependent variable. Respondents were asked to answer the question: "What do you think is the most pressing social problem in our country?" The dependent variable was constructed as a binary variable and was assigned a value of 1 if the older adults --- Method --- Data and Sampling The data for this study were obtained from the Chinese Social Survey (CSS) open database. The CSS is a comprehensive cross-sectional survey project conducted by the Institute of Sociology at the Chinese Academy of Social Sciences and provides nationally representative data. The CSS survey is a biannual longitudinal survey to collect labor and employment data, family and social life information, social attitudes, and other aspects. A multi-stage stratified probability sampling method was used to select and interview a representative sample that reflected the profile of households across China. To ensure data quality and a high response rate, a research team from universities and scientific research centers was established, who provided 3-5 days of training to the data collectors and developed the quality control system. Written informed consent was obtained from all respondents, and the study objectives were also clearly introduced. The study utilized the unaltered dataset from the CSS 2021, encompassing 100,136 individuals residing in 584 villages/communities and 152 counties/districts across 64 provinces/cities/autonomous regions in China. Given the focus on older Chinese adults, the selection criteria were limited to respondents aged 60 years or older (born in or after 1961). Subsequently, the variables were carefully screened, matched, and processed, with the respondents' birth year recorded in the questionnaire, resulting in a final sample size of 2169 data points. --- Variables 3.2.1. Dependent Measure In this study, the perceived affordability and accessibility of the healthcare system was the dependent variable. Respondents were asked to answer the question: "What do you think is the most pressing social problem in our country?" The dependent variable was constructed as a binary variable and was assigned a value of 1 if the older adults selected Kan Bing Nan, Kan Bing Gui as China's most pressing social problem and a value of 0 if not. --- Independent Measures This study identified four sets of independent variables. The first set is healthcare expenditure variables, including medical expenditures and self-reported medical burdens. Medical expenditure is the objective indicator of household out-of-pocket medical expen-diture. The medical expenditure variable was developed by asking respondents to fill in the total amount their family spent on healthcare in the last year (2020) after insurance reimbursement and therefore was constructed as a continuous variable. The self-reported medical burden is the subjective indicator of unaffordable payment. The self-reported medical burden variable was developed by asking respondents to report whether in the last twelve months they or their family has had difficulties in paying for healthcare services and feel the payment is unaffordable. This was constructed as a binary variable with two categories (No = 0, Yes = 1). The second set is patient experience variables, including difficulties in distance, waiting time, cost, and quality, which were developed by asking respondents to report to what extent they encountered difficulties the last time when seeking healthcare services in terms of distance, waiting time, cost, and quality and which were constructed as a 4-point Likert scale variable ranging from 1 ("very serious") to 4 ("did not encounter this difficulty"). The third set is financial protection variables, including insurance coverage and insurance type. The insurance coverage variable was developed by asking respondents to report whether they had enrolled in any medical insurance schemes provided by the government; this was constructed as a binary variable with two categories: 1 (Yes) and 0 (No). For the insurance type variables, the older adults were asked to report whether they had enrolled in the following types of insurance schemes: Urban Employee Basic Medical Insurance (UEBMI) (No = 0, Yes = 1), Urban Resident Basic Medical Insurance (URBMI) (No = 0, Yes = 1), Government Insurance Scheme (insurance scheme for civil servants, GIS) (No = 0, Yes = 1), New Rural Cooperative Medical Scheme (NRCMS) (No = 0, Yes = 1), and critical illness insurance for urban and rural residents (CICURR) (No = 0, Yes = 1). The fourth set comprises social evaluation variables, including satisfaction with healthcare insurance, satisfaction with local healthcare service delivery, and the perceived fairness of public healthcare services. The satisfaction with healthcare insurance variable was developed by asking respondents to report to what extent they were satisfied with the healthcare insurance provided by the government, which was constructed as a 10-point Likert scale variable ranging from 1 ("very unsatisfied") to 10 ("very satisfied"). The satisfaction with local healthcare service delivery variable was developed by asking respondents to report to what extent they were satisfied with the local government's performance in healthcare service delivery, which was constructed as a 4-point Likert scale variable ranging from 1 ("very unsatisfied") to 4 ("very satisfied"). The perceived fairness of healthcare services variable was developed by asking respondents to report to what extent they perceived the fairness of the public healthcare services, which was constructed as a 4-point Likert scale variable ranging from 1 ("very unfair") to 4 ("very fair"). --- Statistical Analysis First, a descriptive analysis was employed to gain an overall profile of the measurement variables and the older adults. Subsequently, the point-biserial correlation was utilized to investigate the connections between social evaluation and the affordability and accessibility perceived by older adults. Furthermore, considering that the dependent variable, perceived affordability and accessibility of the healthcare system, was presented as a binary variable, binary logistic regression analysis was conducted to examine how the four sets of intended independent variables impacted older adults' perceived affordability and accessibility of the healthcare system. All data analyses were carried out using IBM SPSS 25. --- Findings --- Descriptive Characteristics of the Sample The descriptive statistics of the variables of interest in this study are presented in Table 1. Finally, this study obtained a sample comprising 1042 male and 1127 female older adults, with an average age of 64.92 years (SD = 2.745). The descriptive results showed that 54.5% of the older adults perceived Kan Bing Nan, Kan Bing Gui. We further con-ducted a descriptive analysis of China's fourteen pressing social problems. The fourteen pressing social problems included unemployment, Kan Bing Nan, Kan Bing Gui, old-age security, education expenses, inequality between rich and poor, rising prices, expensive housing prices, public security, decreasing social trust, corruption, environmental pollution, the safety of food and drugs, injustice in land expropriation, and unequal treatment of rural-urban immigrant workers. It should be noted that Kan Bing Nan, Kan Bing Gui has become the most pressing issue among the older adults in China, as more than half of the older adults selected this as the most pressing societal issue in China (54.5%), followed by rising prices (25.9%) and inequality between the rich and poor (25.5%), calling for our attention to improve older adults' affordability and accessibility to healthcare services. --- Logistic Regression Model of Perceived Affordability and Accessibility Point-Biserial correlation analyses were conducted to investigate the association between the social evaluation variables of the older adults and their perception of affordability and accessibility. Given that perceived medical expenditure violated the assumption of normality, it was not included in the correlation analysis. The results showed that the older adults' perception of affordability and accessibility was significantly and negatively related to all of the social evaluation variables, including perceived difficulties in distance (r = -0.119, p <unk> 0.001), perceived difficulties in waiting time (r = -0.150, p <unk> 0.001), perceived difficulties in cost (r = -0.301, p <unk> 0.001), perceived difficulties in quality (r = -0.136, p <unk> 0.001), satisfaction with healthcare insurance (r = -0.180, p <unk> 0.001), satisfaction with local healthcare service delivery (r = -0.146, p <unk> 0.001), and perceived fairness of healthcare services (r = -0.201, p <unk> 0.001). The results indicated that the more older adults perceived Kan Bing Nan, Kan Bing Gui, the more likely they perceived difficulties in terms of distance, waiting time, cost, and quality in seeking healthcare services, the less likely they were satisfied with healthcare insurance and local healthcare service delivery, and the less likely they perceived fairness of public healthcare services. To further examine the factors influencing the older adults' perceived affordability and accessibility of the healthcare system, binary logistic regression analysis was utilized to examine the effects of the intended independent variables on perceived affordability and accessibility while controlling for demographic characteristics. In the base model (Model 1), demographic variables, including gender, age, education level, total annual income, and residency, were identified as common confounding factors and were examined. To investigate the impact of the four sets of independent variables, namely, healthcare expenditure variables (Model 2), patient experience variables (Model 3), social evaluation variables (Model 4), and financial protection variables (Model 5), separate models were constructed. The results of the Hosmer and Lemeshow chi-square test indicated that all models, i.e., Model 1 (<unk> 2 (8) = 12.42, p = 0.13), Model 2 (<unk> 2 (8) = 7.90, p = 0.44), Model 3 (<unk> 2 (8) = 4.75, p = 0.78), Model 4 (<unk> 2 (8) = 16.26, p = 0.04), and Model 5 (<unk> 2 (8) = 2.48, p = 0.96), did not demonstrate statistical significance, indicating that all models had a good fit. Moreover, the results of the likelihood ratio chi-square test showed that Model 1 (<unk> 2 (6) = 18.02, p <unk> 0.001), Model 2 (<unk> 2 (8) = 88.62, p <unk> 0.001), Model 3 (<unk> 2 (12) = 157.11, p <unk> 0.001), Model 4 (<unk> 2 (17) = 117.74, p <unk> 0.001), and Model 5 (<unk> 2 (20) = 129.01, p <unk> 0.001) had a significant improvement in fit over the intercept-only null model. The descriptive results of the binary logistic regression models were also summarized and are presented in Table 2. The demographic variables were included as common confounding variables in Model 1; the results showed that all of the demographic characteristics did not significantly affect the older adults' perception of affordability and accessibility except for the education level in Model 1 and Model 2. Moreover, the Nagelkerke R 2 of Model 1 was only 0.01, and the explanatory power increased from 0.01 to 0.22 in Model 5, indicating that the inclusion of the four sets of intended independent variables better explained the dependent variable. For Model 2, only healthcare expenditure variables were included. The results showed that medical expenditure (B <unk> 0.01, OR = 1.00, p > 0.05) did not significantly affect the older adults' perception of affordability and accessibility, whereas the older adults' self-reported medical burden significantly influenced the dependent variable. More specifically, with the use of no medical burden as the reference group (B = 0.82, OR = 2.28, p <unk> 0.001), for every one unit increased in the older adults' self-reported medical burden, the odds ratio of increasing the older adults' perception of affordability and accessibility by one additional level increased by 128%. For Model 3, the healthcare experience variable was also included in the model. In addition to the medical expenditure variable, the older adults' encountered difficulties in seeking healthcare services in terms of distance (B = -0.01, OR = 0.99, p > 0.05), waiting time (B = -0.05, OR = 0.95, p > 0.05), and quality (B = -0.02, OR = 1.02, p > 0.05) did not significantly affect the dependent variable. Nevertheless, the self-reported medical burden (B = 0.48, OR = 1.61, p <unk> 0.001) was a significant and positive predictor of the dependent variable. In addition, the cost of healthcare services (B = -0.52, OR = 0.60, p <unk> 0.001) was a negative and significant predictor of the dependent variable. When the cost of healthcare services as a very serious difficulty was used as the reference group, for every one unit increase in the older adults' difficulty in cost, the odds ratio of increasing the older adults' perception of affordability and accessibility decreased by 42%. In other words, if the older adults did not encounter any difficulty in the cost of healthcare services, they were not likely to perceive Kan Bing Nan, Kan Bing Gui as the most pressing social problem in China. For Model 4, we included the financial protection variables in terms of the enrollment of the five insurance types, and the results revealed no significant effects of the financial protection variables on the dependent variable. Moreover, whereas the Nagelkerke R 2 increased from 0.06 to 0.15 from Model 2 to Model 3, the R 2 had only a very limited improvement from Model 3 (0.15) to Model 4 (0.17). Finally, the social evaluation variables were included in Model 5. Similarly, the results also supported the reported medical burden as a significant and positive predictor (B = 0.24, OR = 1.27, p <unk> 0.05) and encountered difficulty in the cost of healthcare services as a significant and negative predictor (B = 0.60, OR = 0.55, p <unk> 0.001), whereas none of the financial protection variables had a significant effect on the dependent variable. As for the social evaluation variables, satisfaction with healthcare insurance (B = -0.04, OR = 0.96, p > 0.05) and satisfaction with local healthcare service delivery (B = -0.11, OR = 0.89, p > 0.05) did not have a significant association with the dependent variable. In contrast, the older adults' perceived fairness of public healthcare services was strongly associated with the older adults' perceived affordability and accessibility (B = -0.37, OR = 0.69, p <unk> 0.01). More specifically, for every one unit increase in older adults' perceived fairness of public healthcare services, there was a 31% decrease in the odds ratio of increasing their perception of affordability and accessibility, indicating that the more the older adults perceived healthcare services as fair, the less likely they perceived Kan Bing Nan, Kan Bing Gui. --- Discussion Our research finds that 54.5% of the older adults considered Kan Bing Nan, Kan Bing Gui as the most pressing social problem in China, which is far more than those who selected other social problems listed in the CSS 2021 questionnaire. This implies that due to the declining health conditions in the later stage of life, older adults tend to be more concerned about the social problems related to healthcare services rather than other social issues related to earlier stage of life, such as housing, employment, and education. It should be noted that our research on perceived accessibility and affordability includes both users' and non-users' perceptions of the healthcare system. Some respondents had not used healthcare services in the past year and incurred no payments. The variable of medical expenditure is more likely to be influenced by the extreme value of the payment amount. Thus, we adopted medical expenditures as objective indicators and the selfreported medical burden as a subjective indicator of healthcare expenditures. Our research found that only the self-reported medical burden significantly impacted the older adults' perceived affordability and accessibility of the healthcare system, while actual household medical expenditures in the last year did not. The results indicate that those older adults who perceived themselves as having generated unaffordable expenditure last year were likely to perceive inaccessibility and unaffordability in the healthcare system. Such results also implied that the perceived affordability and accessibility of the healthcare system were closely related to the users who utilized the healthcare services. However, it is far from certain to conclude that objective medical expenditure in the past year did not matter to perceived accessibility and affordability. There are many objective measures of patients' financial risks and medical burdens. For example, the World Health Organization defined catastrophic health expenditure as an out-of-pocket health payment for health care equaling or exceeding 40% of a household's capacity to pay [19]. The complexity lies in the social context that China has very high individual savings rates, and the Chinese elderly tend to accumulate savings throughout their whole life cycle to deal with the reduced income and increasing medical expenditures in their later life [38]. Some existing studies that adopted objective measures such as healthcare expenditure/disposable income per capita to measure the economic burden of healthcare also found that it was not the major driving factor for healthcare satisfaction [39]. We argue that it is possible because the precautionary savings of a family reduce medical expenses and alleviate the dissatisfaction with affordability. An existing study using the health-cost-to income ratio as the objective measure and perceived financial difficulty as the subjective measure of the financial burden of lung cancer patients found that the proportion of patients with financial burdens was different when measured using objective or subjective means, which indicated that it could be helpful to capture the actual financial burden by examining both dimensions [40]. We suggest that it is important to pay attention to the subjective evaluation of affordability among older Chinese adults. If the respondents perceived unaffordable medical burdens, it could be an alarming signal that the family's savings have run out for healthcare services, and catastrophic healthcare expenditure may occur. In our study, a fairly large proportion of the older adults reported that they perceived medical burdens (41.4%), which were significantly related to the perceived affordability and accessibility of the healthcare system. Our research results further demonstrated the relationship between patient experience and perceived accessibility and affordability. According to a study on patients' assessment of healthcare services in the Chinese context that categorized 34 indicators into strong, medium, weak, and undetermined indicators, the researchers found that due to the shortage of medical resources in China, patients paid little attention to indicators, such as waiting time, the crowdedness of the hospital, and the convenience of transportation. Rather, they were more concerned about the convenience of paying, the transparency of the fees, and the rationality of the medical fee [41]. Our results also found that, compared to the cost of healthcare services, other indicators of patient experience, such as distance, waiting time, and quality, did not have a significant relationship with perceived accessibility and affordability. Thus, although the saying Kan Bing Nan, Kan Bing Gui, which means "insurmountable access barrier to health care, insurmountably high health costs [8]" and is usually mentioned in parallel in the same sentences in government documents, it is the financial barrier that makes the older adults feel difficulty in gaining access to and paying for healthcare services. In other words, the difficulty in utilizing healthcare services lies in the difficulty in affordability. Our research results also resonated with existing studies on the public perception of the healthcare system in China based on the CSS database, which showed that the perceived affordability of the last medical visit was strongly associated with a patient's perception of the health system, while accessibility measures (e.g., waiting time and distance) were less likely to influence the public perceptions of health care [42]. Our results show that relieving patients' financial burdens is still one of the key challenges for the successful implementation of healthcare reform in China. Our results revealed no significant effects of the financial protection variables on the older adults' perceived accessibility and affordability of the healthcare system. According to existing studies, segmentation by urban-rural and employment status relates to different benefits of healthcare insurance packages in China [43]. Because of the fragmentation in the healthcare insurance system, the reimbursement levels and the benefit packages have disparities, which drove the government to launch pilots for the integration reforms of healthcare insurance in several locations in 2007 and deepened the reform in 2016 [44]. Fragmentation in social health insurance schemes is an important factor for inequitable access to health care and financial protection for people covered by different health insurance schemes [45]. Some existing studies showed that enrollment in healthcare insurance signif-icantly impacted the public perception of the healthcare system [28]. However, the results of our study resonated with other studies that did not find a significant relationship between enrollment in healthcare insurance schemes and the public perception of the healthcare system. For example, some studies found that although the population preferred a lower out-of-pocket expense, the ratio of out-of-pocket expenses had no significant impact on public satisfaction with the healthcare system [23]. Medical insurance enrollment was not a significant factor in evaluating the accessibility of medical resources [46]. The possible explanation could be that due to the vulnerability of older adults, the financial protection of these healthcare insurance schemes for the older adults was still insufficient. Existing research showed that, due to the effect of cream-skimming, the biggest gap lies in hospital access, and the payment of China's healthcare service indicates disparities between senior cadres and the general public [47]. Our results indicated that older adults experienced comparable difficulties in accessibility and affordability irrespective of the type of healthcare insurance scheme in which they enrolled. It is also possible for the cream-skimming effect that satisfaction with healthcare insurance was not significantly associated with perceived accessibility and affordability. Interestingly, satisfaction with local healthcare service delivery was not significantly associated with the older adults' perceived accessibility and affordability of the healthcare system, while the older adults' perceived fairness of public healthcare services was strongly associated. In China, the distribution of healthcare resources has regional disparities between urban and rural and developed and less developed provinces. Since the government expenditure on healthcare at sub-national levels accounts for <unk>90% of the total government expenditure on healthcare, inequity in public expenditure exists at subnational levels [48]. Due to health finance decentralization and uneven economic development in China, local governments in less developed regions cannot fulfill their responsible investments in the healthcare sector [49]. However, our results implied that compared to satisfaction with healthcare services within immediate reach at the local level, the older adults placed more weight on the perceived fairness of the overall healthcare services. Our results are in line with the existing studies, indicating that perceived social equality is a promoting factor in the accessibility evaluation of medical resource allocation [46]. Different from equality or equity, which refers to the gap in distribution, fairness in Chinese usually includes the meaning of distributive justice, which implies that one gets what one deserves based on a perspective of self-interest [50]. Fairness in Chinese is inclined toward egalitarianism, which is a feature of privilege deprivation rather than redistribution [51]. Egalitarianism has been ingrained in the traditional Chinese cultural context. In China, the ancient saying Bu Huan Gua Er Huan Bu Jun ("<unk>") means "no worry about scarcity but unevenness" [52]. Although perceived fairness in healthcare increased from 2006 to 2009 [42], many problems with allocating and distributing healthcare resources still exist. For example, profit-orientated practices widely exist in public institutions, and guanxi is more often used to access hospital-based services, which could be a potential factor contributing to the inequality of health resource allocation [53]. In addition to significant inequality in the geographic distribution of health resources, the rich are more likely to use well-resourced hospitals for outpatient care, while poorer people are more likely to use poorly resourced primary care institutions for inpatient care [54]. Due to the limited effect of health insurance on reducing the inequality in healthcare utilization, the pro-rich horizontal inequities in both the probability and frequency of health services among middle-aged and older adults exist [55]. Such uneven distribution and allocation of healthcare resources may generate the perception of unfairness and injustice. --- Conclusions There are many research gaps in examining the current public perception of the healthcare system in China, especially accessibility and affordability, which has long been considered the major goal and outcome of healthcare reform by policymakers. To fill the gap, our research investigated the perceived accessibility and affordability of the health-care system among older adults and revealed the key determinants of these perceptions. Based on 2169 sample data points from the CSS 2021 survey, the results of descriptive analyses revealed that 54.5% of the older adults considered

The difficulties in accessibility and affordability of healthcare services have posed persistent challenges to the Chinese government ever since the 1990s. This research aimed to explore the perceived accessibility and affordability of the healthcare system, which is also referred to as the social problem of Kan Bing Nan, Kan Bing Gui among older Chinese adults. The research, based on 2169 sample data points from the Chinese Social Survey (CSS) 2021 database, explored the objective and subjective factors in constructing the public perceptions of accessibility and affordability of the healthcare system among both older adult users and older adult non-users of healthcare services, including healthcare expenditure variables, patient experience variables, financial protection variables, and social evaluation variables. The study utilized binary logistic regression analysis to investigate how four different sets of independent variables impact the perceived affordability and accessibility of the healthcare system by older adults. The research found that Kan Bing Nan, Kan Bing Gui was considered the most pressing issue among the older adults in China. Self-reported medical burdens, the cost of healthcare services, and the perceived fairness of public healthcare services were significantly associated with older adults' perceived affordability and accessibility of the healthcare system.

anxi is more often used to access hospital-based services, which could be a potential factor contributing to the inequality of health resource allocation [53]. In addition to significant inequality in the geographic distribution of health resources, the rich are more likely to use well-resourced hospitals for outpatient care, while poorer people are more likely to use poorly resourced primary care institutions for inpatient care [54]. Due to the limited effect of health insurance on reducing the inequality in healthcare utilization, the pro-rich horizontal inequities in both the probability and frequency of health services among middle-aged and older adults exist [55]. Such uneven distribution and allocation of healthcare resources may generate the perception of unfairness and injustice. --- Conclusions There are many research gaps in examining the current public perception of the healthcare system in China, especially accessibility and affordability, which has long been considered the major goal and outcome of healthcare reform by policymakers. To fill the gap, our research investigated the perceived accessibility and affordability of the health-care system among older adults and revealed the key determinants of these perceptions. Based on 2169 sample data points from the CSS 2021 survey, the results of descriptive analyses revealed that 54.5% of the older adults considered Kan Bing Nan, Kan Bing Gui as the most pressing social problem in China; the results of binary logistic regression analyses revealed that self-reported medical burdens, perceived difficulties in cost, and perceived fairness of healthcare services influenced the older adults' perception of affordability and accessibility. The results reflected that although healthcare reform in China has achieved great progress in the past years, for vulnerable groups such as older adults, perceived difficulties in affordability continue to be the main barrier for them to access healthcare services. Our research showed that enrollment in different healthcare insurance schemes did not significantly impact perceived affordability and accessibility. To provide better financial protection, expanding the coverage of healthcare insurance, raising the reimbursement rate of healthcare services, and reducing the out-of-pocket burden are reasonable measures to solve the affordability problem. However, some studies found that the current healthcare insurance schemes contributed to increasing pro-rich inequity in health service utilization [55]. The healthcare insurance schemes did not significantly reduce the outof-pocket payments, and the schemes with more benefits even tended to induce higher health costs [56,57]. Thus, we argue that in addition to increasing coverage and the reimbursement rate, effective measures should be taken to reduce the pro-rich inequity and cream-skimming effects to ensure the most vulnerable and disadvantaged older adults get better financial protection. Increasing the affordability of healthcare services among vulnerable groups may be a main challenge for healthcare reform in China. Our research also illustrated that it is necessary to consider the subjective evaluation of healthcare expenditures such as the self-reported medical burden for a better understanding of the actual financial burden among the older adults in China. Due to the high individual saving rates in China, the function of precautionary savings in buffering the medical expenditure burden should be addressed. Subjective evaluation, such as selfreported medical burden, could be a signal of running out of lifetime savings for healthcare services. Thus, we further call for more nuanced studies on the mediating role of household precautionary savings in the relationship between medical expenditure and household consumption to provide more effective financial protection. Our research also indicates that subjective beliefs may play a significant role in constructing the public perception of the healthcare system. We found that the perceived fairness of public healthcare services was strongly associated with perceived accessibility and affordability among the older adults. Therefore, we argue that to alleviate the perceived unfairness, it is important for policymakers to balance the allocation of healthcare resources and establish a unified healthcare insurance scheme reducing disparities in different funding pools. It is also important to eliminate the privilege and misconduct in access to healthcare services such as medical guanxi. Our research has some limitations. First, in the CSS 2021 database, a lack of details about the actual reimbursement rate of healthcare service utilization in the last year makes it difficult to evaluate the effectiveness of financial protection of each healthcare insurance scheme. In China, the actual reimbursement rate of healthcare services not only depends on enrollment in different schemes but also depends on various factors such as inpatient or outpatient service utilization, medicine prescribed, treatment received, level of hospital, and local or non-local service spots. The actual reimbursement rate may cause the disparities of benefits and financial protection among different medical insurance schemes. Therefore, it is necessary for more databases to collect these data for future research. Especially, since 2016, the New Rural Cooperative Medical Scheme (NRCMS) and Urban Resident Basic Medical Insurance (URBMI) began to merge into the new Urban and Rural Resident Basic Medical Insurance scheme [58]. It is important to explore the change of the actual reimbursement rate under the new scheme to evaluate the performance of a unified medical insurance system. Second, the CSS 2021 database collected data in 2020 and 2021. The outbreak of the COVID-19 pandemic may have restricted healthcare service utilization among older adults and thus reduced healthcare expenditures in the past year. Further, the public perception of the healthcare system is more likely to change during the pandemic. However, due to a lack of specific details about healthcare-seeking behavior and public perception during the pandemic, we did not study the impacts of COVID-19 on the perceived accessibility and affordability of the healthcare system. Third, this study used the open-source CSS 2021 dataset, which mostly measures the identified variables with observed items. A future study could use mixed methods to develop a more comprehensive measurement scale to assess the older adults' experience and perceptions of healthcare systems. For example, the respondents were asked to answer the question, "What do you think is the most pressing social problem in our country?" for 14 social problems; the value of the answer was "Yes" or "No", which we think may simplify the different extent to which respondents are concerned about the accessibility and affordability of the healthcare system. Thus, we call for more nuanced measurements for perceived accessibility and affordability in the future. Moreover, although this study has revealed a representative understanding of the factors that influence perceived accessibility and affordability among older adults, this study was not able to reveal an in-depth understanding of the different health beliefs and health-seeking behaviors at the micro-level and the mechanisms behind such differences. Future research is encouraged to conduct a qualitative study (such as a case study) to unpack the complex phenomenon. Finally, since causal analysis is beyond the scope of our research, we did not incorporate the analysis on the mediating effects of self-rated SES on the relationship between perceived fairness and perceived accessibility and affordability. --- Data Availability Statement: This study was authorized to use a publicly archived dataset from the Chinese Social Survey 2021 (link: http://csqr.cass.cn/index.jsp (accessed on 16 March 2023)). --- Conflicts of Interest: The authors declare no conflict of interest.

The difficulties in accessibility and affordability of healthcare services have posed persistent challenges to the Chinese government ever since the 1990s. This research aimed to explore the perceived accessibility and affordability of the healthcare system, which is also referred to as the social problem of Kan Bing Nan, Kan Bing Gui among older Chinese adults. The research, based on 2169 sample data points from the Chinese Social Survey (CSS) 2021 database, explored the objective and subjective factors in constructing the public perceptions of accessibility and affordability of the healthcare system among both older adult users and older adult non-users of healthcare services, including healthcare expenditure variables, patient experience variables, financial protection variables, and social evaluation variables. The study utilized binary logistic regression analysis to investigate how four different sets of independent variables impact the perceived affordability and accessibility of the healthcare system by older adults. The research found that Kan Bing Nan, Kan Bing Gui was considered the most pressing issue among the older adults in China. Self-reported medical burdens, the cost of healthcare services, and the perceived fairness of public healthcare services were significantly associated with older adults' perceived affordability and accessibility of the healthcare system.

Background The importance of subjective measures of oral health is well-recognized in dental research [1][2][3]. Theoretical models have provided the framework that links clinical conditions with patient perceptions and impacts on their oral health-related quality of life (OHRQoL) [4,5]. Evidence shows that individuals' perceptions of their dental condition is closely related to OHRQoL, [6] and may confer greater impacts than the actual clinical conditions [1]. The United States (US) Surgeon General's report on Oral Health in America underscores and emphasizes the importance of OHRQoL, and its improvement on a population-level is defined as a goal [7]. For these reasons, subjective oral health (SOH) instruments have been used to capture the multi-dimensional concept of OHRQoL [8,9] and are used to quantify patient outcome experiences, monitor oral health status on national level, and identify dental public health goals [10,11]. During this past decade the critical role of health literacy in medicine and public health has gained considerable attention [12,13]. The multi-level consequences of low health literacy have been reviewed extensively and include negative health behaviors, reduced utilization of preventive health services, and poorer adherence to therapeutic protocols [14,15]. Data from the most recent National Adult Literacy Survey (2003) indicate that an alarming proportion of US adults are functionally illiterate [16], and there exists evidence connecting low literacy with poorer health-related quality of life [17]. Health literacy is now considered an underlying cause of health disparities and has become a national health priority [18,19]. Although much is known about health literacy in the medical context, little is known about oral health literacy (OHL) and its relationship to clinical conditions, patients' subjective assessments, and OHL's perceived impacts on daily life in the community. A working group of the National Institutes of Dental and Craniofacial Research (NIDCR) defined OHL as "the degree to which individuals have the capacity to obtain, process, and understand basic oral health information and services needed to make appropriate health decisions" [20]. Horowitz and Kleinman recently proposed that "oral health literacy is the new imperative for better oral health" as health literacy is now considered a determinant of health [21]. An accumulating body of evidence links low OHL with worse oral health outcomes such as oral health status [22,23], dental neglect [24] as well as sporadic dental attendance [25]. In a investigation among a group of Indigenous Australians, Parker and Jamieson [26] found that although low OHL was not associated with selfreported oral health status, it was associated with increased prevalence of OHIP-14 impacts (proportion of items reported fairly/very often). Noteworthy, in a recent study among child-caregiver dyads in the US, caregivers' OHL modified the association between children's oral health status and child OHRQoL impacts, with low-literacy caregivers reporting less impacts [27]. Previous pilot studies have explored the patterns of association between OHL and measures of OHRQoL using the Test of Functional Health Literacy in Dentistry (TOFHLiD) [28] and the Rapid Estimate of Adult Literacy in Dentistry (REALD-99) [29]. Interestingly, as in the Parker and Jamieson study, Richman and colleagues reported that while OHL was not associated with dental health status, higher OHL scores were significantly associated with less perceived OHIP-14 impacts, indicating better OHRQoL [29]. In the validation study of the short form of the REALD (REALD-30) among patients in a medical clinic setting, Lee et al [24] reported an inverse association of REALD-30 with OHIP-14 scores; however, the authors noted that because the data were collected on a convenience sample of health care-seeking subjects, future work is warranted on a larger, more diverse sample, as recommended by the NIDCR proposed research agenda [20]. To this end, the aims of the present study were to investigate the association between OHL and OHRQoL using REALD-30 in a large and more diverse and noncare seeking sample of subjects, and to explore any differences in this association between racial groups. --- Methods --- Study population and recruitment This investigation relied upon interview data from the Carolina Oral Health Literacy (COHL) Project [30], a study exploring OHL in a low-income population of caregivers in the Women, Infants, and Children's Supplemental Nutrition Program (WIC) in North Carolina (NC). Non-random WIC sites in 7 counties in NC were selected using certain criteria including geographic region, rural/urban makeup, population demographics, active WIC clinics and established working relationships. Study staff members were deployed in the selected WIC clinics and approached consecutive individuals to ask if they would answer eight questions from the study eligibility screening instrument. Eligibility criteria included being: a) the primary caregiver of a healthy (ASA I or II) and Medicaid-eligible infant/child 60 months old or younger, or expecting a newborn within the next 8 months, b) 18 years or older and c) Englishspeaking. Caregivers that met these criteria and agreed to participate were accompanied to a private area for a 30-minute in-person interview with one of the two trained study interviewers. Purposeful quota sampling [31] was employed to ensure that minority groups would be well-represented in the study sample. In this approach, individuals in pre-determined minority groups (African Americans and American Indians in the COHL study) are targeted preferentially and recruited into the study until adequate representation in the final sample is achieved. From 1,658 subjects that were screened and determined eligible 1,405 (85%) participated and provided data in the domains of socio-demographic information, dental health and behaviors, OHRQoL, selfefficacy, and OHL. For the current analysis we excluded men (n = 49 or 3.5% of total), Asians (n = 12, or 0.9%), those who did not have English as their primary language at home (n = 79 or 5.6%), and those who had not yet reached age 18 (n = 2 or 0.1%). Therefore, our analytic sample included White, African American (AA) or American Indian (AI) female caregivers, whose primary language was English (N = 1,278). --- Variable Measurements Additional demographic characteristics included age and education. Age was measured in years and coded as a quintile-categorical indicator variable. Education was coded as a four-level categorical variable where 1: did not finish high school, 2: high school or General Education Diploma (GED), 3: some technical education or some college, 4: college or higher education. Dental attendance was self-reported as the time since the last dental visit and coded as a four-level categorical variable where 1: <unk> 1 year, 2: 12-23 months, 3: 2-5 years, 4: > 5 years or never. OHRQoL impacts were assessed with the use of the short form of the Oral Health Impact Profile (OHIP-14) index [32]. Consistent with previous investigations [11], three OHIP-14 estimates were derived from subjects' responses: Severity (cumulative OHIP-14 score), prevalence (proportion of subjects reporting fairly/very often one or more items) and extent (number of items reported fairly/very often) of impacts were calculated as measures of OHRQoL. In terms of interpretation, the authors acknowledge Locker's critique that the OHIP may not fully satisfy the criteria for 'quality of life' measures [33], to be consistent with previous publications, however, have adopted the widely used term of OHR-QoL in this manuscript. OHL was measured with the previously validated word recognition test (REALD-30) [23]. The REALD-30 includes 30 words of dental context (e.g. fluoride, plaque, caries, halitosis, temporomandibular, etc.) arranged in order of increasing difficulty. The criteria used to determine word difficulty were based on word length, number of syllables, and difficult sound combinations, as well as results from 10 pre-test interviews that had been conducted prior to the REALD-30 validation study [23]. The study participant is asked to read each word out loud with one point given for each word that is pronounced correctly, resulting in a 0-30 cumulative score where 0: lowest and 30: highest literacy. Although the REALD-30 is a word recognition test and may be capturing only some aspects of literacy skills, it has been shown to be highly correlated with functional health literacy [28] and to possess good psychometric properties [23]. Norms or thresholds for what constitutes "low OHL" have not been established, however in previous investigations [27,34] a threshold of <unk> 13 on the 30point REALD-30 scale was used to define a "low OHL" group. --- Analytical Strategy We used bivariate tabular methods to display the distribution of the three OHRQoL estimates (severity, prevalence and extent) by strata of socio-demographic variables. We calculated Spearman's correlation coefficients (rho) and 95% confidence limits (CL; obtained with bootstrapping, N = 1,000 repetitions) to quantify the associations between OHL scores and prevalence, severity, and extent. Although the inverse association between OHL and OHRQoL has been shown in previous investigations [23,26], no information has been reported regarding the shape and gradient characteristics of this relationship. For this reason, we used polynomial smoothing functions (LPSF) and corresponding 95% CL to illustrate the relationship between the OHL scores and OHIP-14 estimates. LPSF are non-parametric and data-adaptive functions [35,36] that are flexible in displaying an association without prior assumptions about its shape, gradient, or monotonicity, while minimizing biases from misspecification that could be introduced by traditional modeling applications. Further, to examine the association between "low" OHL and OHRQoL we used the <unk> 13 REALD-30 score threshold, representing the lowest quartile of the distribution, to define the "low OHL" stratum. We obtained crude and adjusted differences and ratios of OHIP-14 impacts using Poisson models. Because severity is the OHIP-14 estimate that arguably carries the most information (no items or scoring schemes are arbitrarily collapsed) and the entire range of the instrument scale (0-56) [11], we chose this measure for subsequent analytical iterations. To further quantify the association between OHL and severity, we used Zero-Inflated Negative Binomial modeling (ZINB). This analytical approach was used because of the distribution characteristics of severity, which followed a negative binomial type distribution with "excess zeros" (Figure 1). The ZINB explicitly specifies two models that are fit simultaneously, one that models the "probability of zero" and one that models the count outcome, using a negative binomial distribution. These models have gained popularity in analyses of count outcomes with high proportion of zeros, but their selection and applicability can be data-specific [37,38]. For this reason and to determine the best fit, we considered other analytical approaches including the negative binomial (NB) and the zero inflated Poisson (ZIP) model. The appropriateness of ZINB versus the NB or the ZIP model was tested and confirmed with diagnostic model-fit statistics, using a Vuong test (ZINB favored over NB, P <unk> 0.05) and a likelihood ratio test (ZINB favored over ZIP, P <unk> 0.05) [39]. The exponentiated coefficient of the negative binomial component of the model corresponds to a Prevalence Rate Ratio, which in this analysis we interpret as ratio of reported impacts (problems), or "Problem Rate Ratio" (PRR) as in a previous study [40]. To facilitate interpretation, we report model coefficients that correspond to one standard deviation change in OHL, which in our study was 5.3 units on the 30 unit REALD-30 scale. In other words, the PRR correspond to the change in reported cumulative OHIP-14 impacts that is associated with one standard deviation change in REALD-30 (expressed as ratio). Inclusion of confounders in the Poisson and the ZINB models was determined by likelihood ratio tests, comparing nested (reduced) models with the referent (full) model using a criterion of P <unk> 0.1. Interpretation of the model coefficients was based on effect estimation rather than hypothesis testing [41]. We employed three (race-specific) multivariate models to explore the possible heterogeneity of the association between OHL and OHRQoL between racial groups. Consistent with our aims, we considered race as an a priori modifier of the examined association and therefore, these three models were identical to the "main effects" model but were restricted to strata of Whites, AAs and AIs. To determine whether race modified the association between literacy and quality of life, we compared these model-obtained race-specific estimates of the association between OHL and severity. The rationale for conducting comparisons of stratum-specific estimates as opposed to testing the hypothesis in the context of statistical interaction is based on the fact that the former approach does not assume covariate effecthomogeneity across racial groups. This could be a source of non-negligible bias when quantifying a weak main effect (e.g. OHL) in the presence of strong confounders (e.g. education), unless all potential interaction terms are included. To that end, we first conducted a global Wald X 2 test of homogeneity or "a common PRR across racial groups" using a conservative criterion of P <unk> 0.2. We further examined post hoc differences in estimates between racial groups by calculating three pairwise homogeneity Z-scores (Z homog ) using the formula: Z homog = |b x -b y |/(se x 2 +se y 2 ) 1/2, where b x/y/z and se x/y/z are the ZINB model-obtained race-specific coefficients and standard errors respectively [42]. Two-tailed P-values corresponding to the Z-scores were obtained using the normal distribution function of the Stata 12.0 (StataCorp LP, College Station, TX) statistical program. A P <unk> 0.05 criterion was used for the pairwise tests. --- Results The demographic characteristics of our final analytic sample (N = 1,280) with corresponding OHIP-14 prevalence, extent, and severity scores are presented in Table 1. Participants' mean age in years was 26.6 (median = 25). Sixty percent had a high school education or less. Seventy-five percent reported a dental visit within the last two years. The OHL score was distributed normally [30] with a mean (SD) REALD-30 of 15.8 (5.3), with 25% of participants (N = 316) scoring less than 13, classified as "low OHL". Pronounced OHL gradients were noted relative to education as follows: less than high school-13.0 (4.8), high school or GED-15.0 (4.9), some technical or college-18.0 (4.7) and college degree or higher-20.1 (4.8). Differences by race were also evident: whites-17.4 (4.9), AA-15.3 (5.1), AI-13.7 (5.3). The mean OHIP-14 severity and extent scores were 10.6 (95% CI = 10.0, 11.2) and 1.35 (95% CI = 1.21, 1.50), respectively. Thirtyseven percent reported at least one oral health impact fairly or very often (prevalence), while AIs had the highest severity score. A strong gradient was found with decreasing age and OHIP-14 scores. Some age and racial differences were noted, with older subjects and AIs reporting more impacts. OHL showed weak correlations with all three OHIP-14 estimates: prevalence rho= -0.14 (95% CI = -0.20, -0.08), extent rho = -0.14 (95% CI = -0.19, -0.09), and severity rho = -0.10 (95% CI = -0.16, -0.05). These bivariate associations are illustrated in Figures 2a,b, and 2c with local polynomial smoothing functions and 95% confidence intervals. In these illustrations the inverse, non-linear association between OHL and the OHRQoL estimates was evident. Although the negative gradient was more apparent for prevalence, the inverse relationship of all three OHRQoL measures with OHL was more "profound" at the lower end of the OHL range. This was confirmed by the contrast of the "low" versus the "high OHL" group (Table 2), where the former group had consistently worse OHRQoL estimates. "Low OHL" was associated with significant absolute and relative increases in all OHRQoL dimensions, with relative prevalence estimates ranging from +11% for severity to +34% for extent. Multivariate analysis adjusting for age, race, and education revealed that the weak inverse association between OHL and severity across the entire sample persisted: PRR = 0.91 (95% CL = 0.86, 0.98). Table 2 presents estimates obtained from the stratified (racespecific) multivariate models, where: Whites-PRR = 1.01 (95% CL = 0.91, 1.11), AA-PRR = 0.86 (95% CL = 0.77, 0.96) and AI-PRR = 0.92 (95% CL = 0.80, 1.05). By comparing these estimates ensemble we rejected the assumption of homogeneity (Wald X 2 = 4.6; degrees of freedom = 2; P <unk> 0.2). Subsequent pairwise comparisons of the race-specific estimates confirmed that the measures of association among AAs and Whites departed from homogeneity (Z homog = 2.06; P <unk> 0.05). In fact, no association between OHL and OHIP-14 severity was found among Whites whereas weak associations were found among AAs and AIs. --- Discussion This investigation provides the first report of the association between OHL and OHRQoL (as measured by OHIP-14) in a multi-racial community-based sample. This study was restricted to a non-probability sample of low-income female caregivers participating in the WIC program in NC; however, we believe that this homogeneity is advantageous because strong income-gradients have been identified in oral health impacts on the population level [43,44]. Moreover, recruitment of subjects from a non-dental clinical environment reduces the potential for selection bias and, within the limitations of the sampling procedures and target population, increases the generalizability of our findings. It is noteworthy but not surprising that the OHL levels in this study were considerably lower than those reported for dental patients seeking care in private practice [REALD-30 (SD): 23.9 (1.3)] [22] or a dental school setting [20.7 (5.5)] [45], and comparable to those found among a community-based sample of indigenous Australians [15.0 (7.8)] [26]. It has been acknowledged that minority individuals and those towards the lowest end of the literacy distribution may be underrepresented in oral health research [46] and this can be even more exacerbated in literacy investigations. Interestingly, the most profound negative gradients between OHL and OHRQoL measures were observed at the lower end of the OHL spectrum, with subjects scoring <unk> 13 on the 30-point REALD-30 scale reporting significantly more OHRQoL impacts versus those with higher literacy. This finding is consistent with conceptual frameworks that consider skills such as conceptual knowledge and OHL as pre-requisites of appropriate decision-making [47]. It is likely that OHL exerts strong influences on oral health-related outcomes when below a certain threshold, but it may be a less impactful determinant at higher levels. The high representation of AAs and AIs that were enrolled in COHL offered us an opportunity to examine for any underlying heterogeneity in the association of OHL with SOH between racial groups. We found a weak negative association between OHL and OHIP-14 severity for AAs and AIs, but not Whites. While AAs have been shown to report worse OHIP scores in the US [10] and patterns of OHRQoL changes have been shown to differ by race [48,49], this finding warrants further investigation; race may be a proxy of unmeasured mediating factors between OHL, oral health status, and perceived impacts [50]. The fact that the dimensionality of OHR-QoL [8] may differ between diverse populations or ethnic groups may amplify this phenomenon; therefore, we acknowledge the limitation of our analytical sample that was restricted to low-income WIC-participating female caregivers. Replication of our main as well as race-specific findings should be undertaken on a population-based representative sample. Lawrence et al [51] recently demonstrated that OHIP-14 scores show good correlation with clinical oral health Table 2 Oral health-related quality of life (OHRQoL) differences [mean difference and prevalence ratios (PR) with corresponding 95% confidence limits (CL)] between participants with "low" (<unk> 13 REALD-30; referent category) and "high" (<unk> 13 REALD-30) oral health literacy in the Carolina Oral Health Literacy study (N = 1,278) 1: Mean differences and ratios of OHIP-14 impacts were calculated using the "high literacy" category as referent. 2: Adjusted differences and ratios were obtained using a Poisson model controlling for race, age, education level and dental attendance. status, independent of gender and socioeconomic inequalities in oral health. Among our community-based caregivers, the prevalence of oral health impacts (36.5%) was higher compared to nationally representative samples from other studies including the US (15.3%) [10], Australia (dentate subjects-18.2%), United Kingdom (dentate subjects-15.9%) [11] and New Zealand (23.4%) [51]. However, the extent and severity estimates reported here are lower compared to these samples. One possible interpretation of this finding is that our study group was limited to young, low-income, poorly educated, WIC participants with relatively low education. The young mean age (26.6 years) may explain the low severity and extent estimates while the low-income and low-education level status may explain the high prevalence of at least one impact reported as fairly/very often. Considering the high prevalence of impacts revealed in the study population, the significance of lower OHL is demonstrative. Using our "main effects" model coefficients, we estimate that a one standard deviation increase in OHL (5.3 REALD-30 units) corresponds to a 9% decrease in OHIP-14 severity [PRR (95% CL) = 0.91 (0.86, 0.98)], whereas (using race-specific estimates from Table 3) this decrease is more pronounced (14%) among AA [PRR (95% CL) = 0.86 (0.77, 0.96)]. On the other hand, this finding provides a foundation to consider interventions to enhance OHL, or rather improve the readability of written materials and accessibility to dental services to an appropriate literacy level [30]. It remains uncertain whether improvement in OHL is feasible and if so, whether this would lead to better oral health status and subjective oral health. Although education and income arguably remain the strongest correlates of oral health and disease, and literacy is one of numerous other distal determinants, OHL may be part of causal mechanisms that lead to worse oral health [21]. Accumulating evidence linking poor OHL with adverse oral health outcomes among caregivers [24] and their young children [27,34] supports the introduction and implementation of rapid OHL screening tools [52] in clinical practice, dental research and public health surveillance. Moreover, we suggest that more studies exploring the association between OHL and OHRQoL be undertaken in multi-racial community based samples to confirm or reject this study's finding of effect measure modification by race. --- Conclusions We found a high prevalence of perceived oral health impacts in this sample of low-income female WIC participants. Although the inverse association between OHL and OHRQoL across the entire sample was weak, subjects in the "low" OHL group reported significantly more OHR-QoL impacts versus those with higher literacy. Within the limitations of our study among low-income female caregivers, our findings indicate that the association between OHL and OHRQoL appears to be modified by race. --- Authors' contributions KD conducted the data analysis and prepared the first draft of the manuscript. JL conceived the study, overviewed the data analysis, contributed to the interpretation of results and assisted in preparation of the first draft of the manuscript. ADB participated in data collection, and critically revised the manuscript. WFV contributed to the interpretation of results and critically revised the manuscript. All authors read and approved the final manuscript. --- Competing interests The authors declare that they have no competing interests.

Background: To investigate the association between oral health literacy (OHL) and oral health-related quality of life (OHRQoL) and explore the racial differences therein among a low-income community-based group of female WIC participants. Methods: Participants (N = 1,405) enrolled in the Carolina Oral Health Literacy (COHL) study completed the short form of the Oral Health Impact Profile Index (OHIP-14, a measure of OHRQoL) and REALD-30 (a word recognition literacy test). Socio-demographic and self-reported dental attendance data were collected via structured interviews. Severity (cumulative OHIP-14 score) and extent of impact (number of items reported fairly/very often) scores were calculated as measures of OHRQoL. OHL was assessed by the cumulative REALD-30 score. The association of OHL with OHRQoL was examined using descriptive and visual methods, and was quantified using Spearman's rho and zero-inflated negative binomial modeling. Results: The study group included a substantial number of African Americans (AA = 41%) and American Indians (AI = 20%). The sample majority had a high school education or less and a mean age of 26.6 years. One-third of the participants reported at least one oral health impact. The OHIP-14 mean severity and extent scores were 10.6 [95% confidence limits (CL) = 10.0, 11.2] and 1.35 (95% CL = 1.21, 1.50), respectively. OHL scores were distributed normally with mean (standard deviation, SD) REALD-30 of 15.8 (5.3). OHL was weakly associated with OHRQoL: prevalence rho = -0.14 (95% CL = -0.20, -0.08); extent rho = -0.14 (95% CL = -0.19, -0.09); severity rho = -0.10 (95% CL = -0.16, -0.05). "Low" OHL (defined as < 13 REALD-30 score) was associated with worse OHRQoL, with increases in the prevalence of OHIP-14 impacts ranging from 11% for severity to 34% for extent. The inverse association of OHL with OHIP-14 impacts persisted in multivariate analysis: Problem Rate Ratio (PRR) = 0.91 (95% CL = 0.86, 0.98) for one SD change in OHL. Stratification by race revealed effect-measure modification: Whites-PRR = 1.01 (95% CL = 0.91, 1.11); AA-PRR = 0.86 (95% CL = 0.77, 0.96). Conclusions: Although the inverse association between OHL and OHRQoL across the entire sample was weak, subjects in the "low" OHL group reported significantly more OHRQoL impacts versus those with higher literacy. Our findings indicate that the association between OHL and OHRQoL may be modified by race.

percent from 2017 to 2020, reaching an estimated 171 billion US dollars in 2020 1. In the first half of 2020, there were more than 10 million live streams and more than 400,000 active live streamers. During the live streaming, the streamers show how products are developed and used according to viewers' needs, clarify any areas of their confusion, and interact with them in real-time to stimulate immediate purchases. To keep viewers watching longer, live streamers make their shows as engaging and immersive as possible, and express right emotions in order to foster viewers' active attitudes and behaviors in communications and transactions. During the interactions, displaying organizational desired emotional expressions and suppressing others are very important for live streamers who interact with customers to promote their product. Such psychological regulation process of employees during service delivery is characterized as emotional labor (Ashforth & Humphrey, 1993;Grandey, 2003;Hochschild, 1983). It can be achieved either while live streamers alter inner true feelings to match the required emotions of the work (deep acting), or while they fake the required emotional expressions (surface acting) (Grandey, 2000(Grandey,, 2003)). Several meta-analyses have shown that deep acting is more likely to be linked with positive outcomes on employees, customers and organizations, but surface acting has a negative impact on the above outcomes (Chen et al., 2021;Fang et al., 2019;Xu et al., 2020). Therefore, how to address live streamers' emotional labor issues and hereby improve corporate performance will be important in the theoretical and practical domain. The recent studies have adopted the conservation of resources (COR) theory to explore the individual and organizational antecedents of emotional labor (Wen et al., 2019;Xu et al., 2020). According to the COR theory, individuals often tend to conserve and gain available resources and try to avoid threats that may result in exhaustion of resources (Hobfoll, 1989). And resources are further divided into external resource that individuals hope to gain (e.g., social support) and internal resources (e.g., objects, time, knowledge, motivation, personal characteristics or inner energy) (Hobfoll, 2001). While they lose internal resources or fail to gain external resources outside, they would experience stress or emotional exhaustion which prompts them to take action to conserve resources by changing their emotional labors. Social support is among the most important antecedents that have been investigated in recent years (Lam & Chen, 2012). Usually, the organizations are trying to provide various supports that enable the frontline employees to cope with stress (Hwa, 2012;Lam & Chen, 2012;Moon et al., 2013). Social support from organization, coworker, supervisor, family and etc. is regarded as a kind of external resource, which can supplement an individual's inner energies by producing positive affective perceptions (Lam & Chen, 2012;Moon et al., 2013;Wen et al., 2019). However, the findings in existing studies has indicated that the relationship between social support and emotional labor strategies is complex and inconsistent (Xu et al., 2020). And more importantly, those studies provide little evidence on how and why social support can impact emotional labor strategies. Therefore, there is a need to understand how social support relates to different emotional labor strategies (e.g., surface acting and deep acting), including mediating and moderation effects that explain this relationship. To address this research gap, our paper uses the COR theory to explain the effects of social support on emotional labor strategies (e.g., surface acting and deep acting), and investigates potential underlying mechanisms and boundary conditions. From this perspective, we argue that social support, as an important external resources, may impact live streamers' emotional labor strategies directly by providing tangible or psychological assistance. In addition, social support is predicted to have an indirect effect on emotional labor strategies through the mediating influence of employees' self-concept in the form of professional identity. Specifically, we propose that social support shapes how live streamers perceive the meaning of the job itself, which is identity of profession. As a selfconcept for the career, professional identity can help employees to better recognize emotional rules of organization, thereby regulate their emotional labor (Guan et al., 2016). Moreover, this perception of professional identity may also boost a person's self-esteem, enhance a person's motivation to regulate affective reactions in terms of deep acting instead of surface acting, especially if that person can achieve a positive professional identity (Shahzad et al., 2021;Zhang et al., 2019). Finally, our paper further extends existing research by proposing emotional intelligence as a moderator of the professional identity -emotional labor relationship and of the indirect relationship of social support on emotional labor through professional identity. Our study aims to make several contributions to extant literature. First, our paper contributes to the emotional labor literature by providing empirical evidence in e-commerce live streaming industry that social support not only enhances deep acting of streamers but also reduces their surface acting during the interaction with viewers, highlighting the importance of social support in e-commerce live streaming industry. Second, we respond to the call of Xu et.al (2020) by examining the mechanism of how and why social support can impact emotional labor strategies (Xu et al., 2020). Specifically, by introducing professional identity as a mediator, we could better understand the process of how social support influence emotional labor strategies via shaping one's perceptions regarding a professional role. Third, our paper examines emotional intelligence as a moderator to the mediation effects of professional identity discussed above. Previous studies suggest that emotional intelligence may buffer the negative effects of surface acting by supplying resources employees need and is important to their emotional labor (Wong & Law, 2002;Xu et al., 2020). Our study furthers the existing findings by clarifying how emotional intelligence interacts with professional identity to influence the employees' recognition of emotional rules and commitment to their profession, thereby reduce surface acting and increase deep acting. Finally, our study also provides practical guidance for online marketing practitioners on improving live streamers' emotional labor. --- RESEARCH CoNTEXTS ANd LITERATURE REVIEw --- E-Commerce Live Streaming in China In the early 2000s, live streaming of video games and singing/dancing first became popular in the live streaming platforms such as Huya, Douyu and Panda TV. The streamers acted as information sources and content contributors, delivering the live video content to viewers over the internet. Viewers participated in on-line community by writing online reviews and buying virtual gifts to the streamers, which appear as little animations on the screen in real time. Then, the streamers respond to the gifts in real time, engaging with the viewers who pay generous sums and thanking them directly. In April of 2016, the e-commerce giant Taobao established an in-app platform called "Taobao Livestream", which was the first to link up live streaming with an e-commerce store to allow viewers to watch and shop at the same time. Thereafter, the idea of live streaming goes beyond entertainment to e-commerce which is one of the world's fastest growing industries (Ahluwalia, & Merhi, 2020;Hou et al., 2021;Mou et al., 2020;Yuan et al., 2021;Zhao et al., 2020). A number of originally e-commerce platforms (e.g.,Taobao, Pinduoduo and JingDong), social networking sites (e.g.,Weibo, Wechat and Xiaohongshu), and entertainment live streaming apps (e.g., Yizhibo, and Kuaishou) have now also incorporated e-commerce functionalities. According to a report released by China National Internet Information Office in September 2020, it is expected that by the end of 2021, the transaction volume of e-commerce live streaming is expected to exceed 2.5 trillion RMB yuan, accounting for about 20% of the total amount of Internet e-commerce. In the year of 2020, 6939 e-commerce live streaming companies emerged across the country, a 360.8% year-on-year growth, and the number of e-commerce live streamers rocketed from 2.7 million to 12.3 million <unk>. These live streaming companies are distributed in Taobao Live, TikTok, Kuaishou, Xiaohongshu, Pinduoduo and other 200 live streaming platforms, acting as a powerful marketing tool at an extremely low cost. E-commerce live streaming entails the real-time broadcast of video content to viewers, which provides viewers with a FaceTime-like interface with the streamers. With this feature, streamers are able to demonstrate different perspectives of the products and respond to the pressing questions of potential customers, making promotional activities more convincing and effective. In a fashion live streaming, the seller showcases its products to viewers by a live model (most of time, the streamer herself) dressing up in the merchandise besides of presenting the product information in the traditional form of texts and images on the platform. The viewers can have a live chat with the streamers, ask them to show the products according to their needs. Upon the request, the streamers will demonstrate the apparels in a variety of poses or any kind of angles to suggest how best to mix and match the clothes while also suggest new additions. Meanwhile, the streamers literally interact with viewers through a chat function or reaction buttons every second, which make viewers to thumb up, generate electronic word-of-mouth (e-WOM), etc. Although streamers can attract traffic to watch the live streaming via Social Network Sites (SNS) and other advertisings, the views do not actually have a strong sense of participation. To retain viewers and achieve sales performance, the streamers need to answer viewers' queries and react to their requirements while broadcasting. They need to create a fun, unique and interactive emotional space, which makes the show as engaging as possible and motivates potential customers to watch longer and place order eventually. Moreover, the streamers can further enhance emotional connections with viewers by setting up fan community. Such intimate relationship between streamers and viewers can ensure a more stable sales. --- Emotional Labor: Surface Acting and deep Acting This process of emotional operations to attract and retain reviews could be regarded as emotional labor of e-commerce live streamers. In most cases, streamers must consciously portray themselves in a certain way to meet the emotional needs of a wider viewers and the rules of live streaming platforms (Yang, 2021). Therefore, streamers have to design a new emotion carefully by suppressing emotions and behaviors that are inconsistent with platform rules and catering to those values that the viewers have recognized, even if these emotions conflict with their real feelings. Efforts to conceal their true emotions, to feel expected emotions or to behave as if they are feeling these emotions involves a pattern of effort to do the job, which is called "emotional labor" (Grandey, 2000;Hochschild, 1983). Briefly, it is defined as the efforts spent to display emotions expected by the organization. According to Hochschild (1983), emotional labor can be further differentiated with regards to different acting strategies, known as surface acting and deep acting (Hochschild, 1983). In surface acting, live streamers spend their efforts to pretend to feel or fake the expected emotions during the live interaction with customers although they do not feel that way (Grandey, 2000). Usually, they will adjust their emotions by changing outward appearance such as voice tone, body gesture, facial expression, and motions. However, the emotion actually perceived by live streamers does not conform to the rules of the organization's emotions (Hochschild, 1983). In deep acting, they would try to change their emotions which do not meet the performance of organizational requirements by utilizing their preparation, imagination or past experience to express the circumstance fitted feelings (Lee et al., 2019). And they actually experiencing the organizational expected emotions because they believe it is the case (Grandey, 2003). Existing studies demonstrate that an individual's external resources or inner energies can help change his emotional labor behaviors in works, such as the adoption of deep acting or surface acting (Kim et al., 2012;Lam & Chen, 2012;Yin et al., 2013). For example, Grandey (2000) has demonstrated that social support is one of the predominant variables affecting the way emotional labor is expressed (Grandey, 2000). Yin et al. (2013) find that emotional intelligence, as an emotional ability or resource for service employee, is positively related with deep acting strategy (Yin et al., 2013). Kim et al.(2012) find that emotional intelligence is positive associated with deep acting of hotel frontline employees, but negatively related with surface acting (Kim et al., 2012). As a deep psychological processing, deep acting thus requires both internal and external resource to generate a series of positive emotions and support an individual's emotive effort. More importantly, although both emotion acting strategies are internally false, they represent different intentions or motivations of an individual for work (Grandey, 2000(Grandey,, 2003)). Deep acting highlights regulating the inner feelings to meet requirements of the work (Grandey, 2000;Hülsheger and Schewe, 2011), which can be regarded as intrinsic motivation of an individual. This good motivation may not be present in employees who do not accept their work roles (Ashforth & Humphrey, 1993). It reflects that employees have emotional attachment to the organization's goals and values as well as a willingness to help the organization achieve its objectives (Meyer et al., 2002). This intrinsic motivation can be derived from the meaning of the work itself, such as an identification with the profession. However, unlike the intrinsic motivation of deep acting, surface acting emphasizes shaping outward appearance to meet the requirements of the organization rather than altering the inner true feelings (Grandey, 2000). On the one hand, the persons using surface acting are merely concerned with the benefits of conforming to the display rules, such as keeping the job and improve their feedback on performance. On the other hand, there is no need for affective commitment and self-identity to the organization and their profession. Therefore, while the employees experience more incongruences between emotional rules of organization (what they follow rules to do) and self-identity of their profession (what they commit to do), they will be more likely to fake their emotions using surface acting strategy during the interaction with customers. --- THEoRETICAL FoUNdATIoN ANd HyPoTHESES --- Social Support and Emotional Labor Social support refers to the extent to which an individual's desire recognition, understanding, respect, and support from family members, friends, coworkers and organizations (Wills, 1991). That means the individual may obtain all kinds of tangible or psychological support in the surrounding environment. Specifically, it may consist of communicating constructive feedback and guidance while providing tangible assistance and expertise to enable employees to complete the job, expressing empathy, and caring about the needs of employees (Zhang et al., 2018). According to COR theory, individuals give priority to protect internal resources and seek the support of external resources and the amount of employees' resources may impact the choice of emotional labor strategies (Lam & Chen, 2012). Social support can both enhance one's available resources and can replace or provide other resources that have been lacking (Hobfoll, 1989). For example, the empirical study of Nam and Kabutey (2021) finds that social media use at work may help employees conserve them of personal resources (Nam & Kabutey, 2021). Therefore, as an external resource, it can timely compensate streamers for the internal resources consumed in emotional labor during the live streaming. Those high in social support may access to a variety of resources which may help to buffer them against stressful events in emotional labor (Lam & Chen, 2012), thus tend to engage in deep acting. On contrast, those who lack resources would choose to minimize resource consumption and conserve energy through surface acting. The relationship between social support and emotional labor has also been uncovered in several studies. For example, Lam and Chen (2012) found that social support prompts service providers to engage in less surface acting and more deep acting during service interactions in hospitality industry (Lam & Chen, 2012). Therefore, we hypothesis that higher levels of social support lead streamers to perform deep acting, while lower levels of social support lead them to fake their emotions through surface acting. H1a: Social support has a negative effect on surface acting. H1b: Social support has a positive effect on deep acting. --- Professional Identity and Emotional Labor Professional identity is defined as the relatively stable and enduring constellation of attributes, beliefs, values, motives and experiences that individuals use to define themselves in their professional role (Dobrow & Higgins, 2005;Schein, 1978;Zheng, Bai & Cross, 2021). A streamer may gradually understand and recognize the professional meaning and his role or social-based work identities in the course of career. It will shape an individual's attitude, affect and perception towards his professional role and work purpose, then the subsequent professional behavior (Johnson et al., 2012;Siebert & Siebert, 2005). We propose that professional identity may help live streamers to better recognize emotional rules of live streaming platform or live streaming company, and lessen their attentions towards utilitarian gains from fake emotions. During the interactions with customers, emotional rules of the live streaming company or streaming profession generally require streamers to show positive emotions and behavior to customers. And the live streaming companies always provide a series of incentives to encourage the streamers who follow the requirements of streaming profession. But streamers with low professional identity may less identify with the emotional rules of organization. In order to avoid this incongruence, they have to hide or disguise their own emotions. On the contrary, streamers with high professional identity are more likely to comply with the emotional rules in natural and spontaneous way, without hiding and faking (Ashforth & Humphrey, 1993). Therefore, they need not worry to regulate their internal feelings or display the required emotions, and are less likely to act superficially in live streaming. Ashforth and Humphrey (1993) also find that one of the reasons why an individual adopts the surface acting is that the low level of professional identity does not fit with the requirements of the profession (Ashforth & Humphrey, 1993). Furthermore, professional identity may help individuals to define the work's purpose and assign meaning to what they do in the work (Haslam, 2012). Once a positive professional identity has been formed, individuals are more committed to values and practices of their profession (Haslam, 2012;Hirschy et al., 2015;Wang et al., 2020). And because of the internalized congruence between one's personal and professional values, those high in professional identity will be intrinsically motivated to invest personal resources (e.g., time, efforts, emotional and psychological energies) (Hirschy et al., 2015), and foster meaningful deep acting activities with customers by showing warmth, empathy and consideration. Accordingly, we hypothesis that: H2a: Professional identity is negatively related to surface acting. H2b: Professional identity is positively related to deep acting. --- The Mediating Effect of Professional Identity Recent research shows that social support is effective to reduce emotional exhaustion (Nam & Kabutey, 2021), because it helps to reinforce the positive aspects of the self, in essence self-concept when stressful times have led one to lose sight of them (Swann & Predmore, 1985). As a kind of self-concept, professional identity reflects one's perceptions regarding a professional role (Dobrow & Higgins, 2005). It will be related to the variety of support individuals receive from others in their careers (Dobrow & Higgins, 2005). First, an employee with good social support means more respect and understanding from others, which may help to establish the positive links between self and profession (Strauser et al., 2008). For instance, Wei et.al (2021) find that prompting recognition, understanding, respect, and support for nursing students is crucial for their professional identity (Wei et al., 2021). Second, employees learn who they are as a professional by seeing themselves in the eyes of others (Roberts et al., 2005). The mutual trust, interdependence, and reciprocity that characterize relationships in the surrounding environment offer a great intermediary for reflecting and shaping their professional identity (Swann, 1987). Third, social support may provide access to different kinds of resources and information, which may reinforce an individual's professional identity exploration (Dobrow & Higgins, 2005). Identification with the profession will trigger individuals to have an internal sense of obligation to cooperate to work in return (Prosser, 2006). Through this process, streamers may consciously abide by the emotional rules of the platform or organization, and when they realize that their current emotions do not meet the emotional requirements, they may manage and adjust in a positive way rather than by surface acting to hide the real emotions. In addition, profession identity can influence individuals' perceptions of a job environment and can thereby have a positive influence on the commitment to work, including enhancing a sense of belonging, self-image building, and professional satisfaction (Piil et al., 2012). It is a key way that streamers assign meaning to themselves, and it shapes their attitudes, affect and emotional behavior to meet requirements of the work. Combining all of the above evidence, we argue that social support reinforces streamers' positive aspects of the self through enhanced understanding and recognition of professional significance and value towards live streaming, in turn fosters a positive professional identity and thus influences their emotional labor strategies. Therefore, we propose the following hypothesis: H3a: Professional identity mediates the relationship between social support and surface acting. H3b: Professional identity mediates the relationship between social support and deep acting. --- The Moderated Effect of Emotional Intelligence As we discussed above, streamers with low professional identity may not recognize the emotional rules of live streaming profession or lack a sense of self-identity towards their profession. But in order to meet the requirements, they will take a hidden or disguised way to express their emotions by surface acting. Emotional intelligence, as an ability of individuals to observe, recognize, manage and regulate their own emotions as well as those of others to solve problems (Abraham, 1999;Wong & Law, 2002), plays a crucial role in the process of emotional labor (Goleman, 1995). Live streamers with high emotional intelligence tend to enjoy a high degree of interaction with their coworkers and supervisors, which helps them to better understand and recognize the emotional rules of live streamer so as to reduce their faking emotions. In addition, individuals with high emotional intelligence are better equipped to manage and control their emotions to meet the requirements of organization than those with low emotional intelligence (Xu et al., 2020). Therefore, emotional intelligence should help to enhance the negative effects of professional identity on surface acting. Besides, streamers with high professional identity are more likely to be intrinsically motivated to invest personal resources and thereby foster meaningful deep acting activities. As a kind of internal (emotional) ability, emotional intelligence can also promote employees' faith, abilities to manage emotions and accumulation of other related individual resources (Wen et al., 2019). Research has shown that such emotional abilities can help individuals to adopt deep acting to replace surface acting in their work. For instance, Yin et al.(2013) found that teachers' emotional intelligence has a significant impact on deep acting, but it is not a significant predictor of surface acting (Yin et al., 2013). The meta-analysis of Xu et al.(2020) also found a positive effect of emotional intelligence on deep acting and the negative effect of emotional intelligence on surface acting (Xu et al., 2020). Therefore, both professional identity and emotional intelligence can supplement the emotional resources consumed by emotional labor of streamers in the work. When individuals work in positions required with emotional labor, they would gradually consume their emotional, psychological and physical energies derived from their professional identity. If there is no alternative way to recovery their personal resources, the commitment to values and practices of their profession may be the important way to be motivated and avoid resource exhaustion. However, if an individual has other inner resources or emotional abilities like emotional intelligence, the role of alleviating resource exhaustion resulted from professional identity will be substituted, which in turn reduce the positive effect of professional identity on deep acting. In other words, emotional intelligence can function as an important inner resources to emotional labor and counteract on the effect of professional identity, thereby negatively moderate the positive relationship between professional identity and deep acting. H4a: Emotional intelligence moderates the negative relationship between professional identity and surface acting, in that the higher the streamers' emotional intelligence, the stronger the negative relationship between professional identity and surface acting. H4b: Emotional intelligence moderates the positive relationship between professional identity and deep acting, in that the higher the streamers' emotional intelligence, the weaker the positive relationship between professional identity and deep acting. --- The Moderated Mediation Effect of Emotional Intelligence The above hypotheses of 3a, 3b, 4a and 4b reveal that professional identity mediates the relationship between social support and surface acting/deep acting, and emotional intelligence moderates the impact of professional identity on surface acting/deep acting. As the moderation effect of emotional intelligence is integrated it into the indirect path of social support and emotional labor, it further reveals the moderated mediation effects of emotional intelligence in our proposed model. Specifically, emotional intelligence moderates the indirect effects of social support on surface acting/deep acting through professional identity, by changing the relationship between professional identity and surface acting/deep acting. Therefore, we propose the following hypothesis: H5a: The indirect effect of social support on surface acting through professional identity is moderated by emotional intelligence, such that the indirect effect is stronger at high levels of emotional intelligence. H5b: The indirect effect of social support on deep acting through professional identity is moderated by emotional intelligence, such that the indirect effect is weaker at high levels of emotional intelligence. --- RESEARCH METHodS --- Sample and Procedures Located in most prosperous coastal regions, Zhejiang, Jiangsu, Guangdong, Liaoning province have developed a large number of living streaming platforms and enterprises. The local municipal governments make great effort in promoting the development of living streaming industries in the regional economy. Therefore, we collected data from 32 e-commerce live streaming companies located in Zhejiang, Jiangsu, Guangdong and Liaoning provinces. These e-commerce live streaming companies had on-line stores on Taobao Live, Douyin and other e-commerce live streaming platforms in order to reach as many potential customers as possible. With the assistance of surveyed company, we randomly selected 370 live streamers from 64 workgroups in 33 live streaming enterprises and distributed the questionnaire through social media platforms (such as Wechat) and email. All participants were informed with a declaration of confidentiality in advance, promising that the survey would be used only for academic research. By the end of the survey period, 355 questionnaires were recovered and 24 questionnaires were excluded due to excessive missing data. Our final sample comprised 331 live streamers with a response rate of 93.23%, the size of which ensured a good reliability of the study (Bakker et al., 2012). Due to the nature of live streaming industry, the study's sample predominantly comprised females (66.77%) and was characterized by young people aged below 31 years (78.85%). The mean tenure in live streaming industry was 2.34 years (SD =0.91) and the education level of participants were high school or below (33.53%) and technical school diploma (46.83%), bachelor and above(19.64%). --- Variable Measurement The measurements of this research include social support, professional identity, emotional intelligence, and emotional labor. We design our questionnaire using existing scales which are translated from English into Chinese after a back-translation process to avoid cultural bias. Further, as the respondents are e-commerce live streamers, we followed the way of prior studies to modify some wording to ensure the applicability of e-commerce live streaming context (Sila, 2019;Sohaib et al., 2019;Zhang et al., 2018).The questionnaire items regarding social support are measured on a seven-point Likert scale ranging from "1" (strongly disagree) to "7" (strongly agree), while the remaining variables are measured by five-point Likert scales ranging from "1" (strongly disagree) to "5" (strongly agree). 1. Social support (SS): It is measured using 12 items from Perceived Social Support Scale (PSSS) adapted by Zimet et.al (1988) (Zimet et al., 1988). Sample items include: "There is a special person who is around when I am in need", "There is a special person with whom I can share my joys and sorrows", "My family really tries to help me". (<unk>=0.94). The Cronbach alpha coefficient of scale is 0.94. 2. Professional identity (PI): It measures the level of identification of an individual with his profession. We adapt Mael et al.'s (1992) Professional Identity Scale (PIS) (Lui et al., 2001) and substitute the word "profession" for the word "live streamer profession" in scale items. This scale contains 6 items. Two example items are: "When someone criticizes this live streamer profession, it feels like a personal insult", and "When I talk about this live streamer profession, I usually say 'we' rather than 'they'" (<unk>=0.89). The Cronbach alpha coefficient of new scale is 0.89. --- Emotional intelligence (EI): To measure emotional intelligence of live streamers, we use Law et al. (2004) Emotional Intelligence Measurement Questionnaire (WLEIS) (Law et al., 2004). A total of 16 items are included, some example of which are: "I have good understanding of the emotions of people around me", and "I am a good observer of others' emotions" (<unk>=0.94). The Cronbach alpha coefficient is 0.94. 4. Surface acting (SA) and deep acting (DA): Surface acting (SA) and deep acting (DA) are measured using the Affective Labor Scale (ALS) from Grandey (2003) (Grandey, 2003), 5-item scale for surface acting (SA) and 6-item scale for deep acting (DA). On a 5-point scale ranging from very slightly or not at all to very frequently, respondents are required to indicate the extent to which the items describe them at work. A sample item for surface acting is "Just pretend to have the emotions I need to display for my job" (<unk>=0.90). A sample item for deep acting is "Work hard to feel the emotions that I need to show to others" (<unk>=0.90). 5. Control variables: Since previous studies suggest that the employees' demographic background may affect emotional labor (Zhang et al., 2018), our paper also control for gender (1= male; 0= female), age (years), industry tenure (number of years in live streaming industry) and education (3 = bachelor and above, 2 = technical school diploma, 1 = high school or below). --- RESEARCH RESULTS --- Confirmatory Factor Analysis and Common Method Bias This study uses Mplus 7.0 conduct CFA on five variables: social support (SS), professional identity (PI), emotional intelligence (EI), surface acting (SA), and deep acting (DA). As shown in Table 1, the five-factor structure of Model 1 fits the data best in fit indexes among Models 1-7, with <unk> 2 /df=1.14, RMSEA=0.02, CFI=0.98, TLI=0.99, and SRMR=0.04. These results suggest the scale used in this study has a good distinguishing validity and the five constructs capture distinctiveness as expected. Since the data for both independent and dependent variables comes from the same respondent, it may raise concerns of common method variance in our paper. We adopt two statistical techniques to verify whether there is possible common method variance problem. First of all, this study use the Harman's single factor test to verify whether the common method bias is high. The single-factor model (Model 7 in Table 1) shows an unsatisfactory fitting index (<unk> 2 /df=6.55, RMSEA=0.13, CFI=0.41, TLI=0.38, SRMR=0.19), indicating that the common method variance in this study is not serious. Furthermore, the unmeasurable latent factor model is also used to further test the common method bias. It is shown that the fitting index of unmeasurable latent factor model (Model 2) is <unk> 2 /df=1.14, RMSEA=0.02, CFI=0.99, TLI=0.99, SRMR=0.04, achieving no significant improvement compared with the five-factor model (Model 1). So the common method bias in this study is not serious. --- descriptive Statistical Analysis Table 2 shows the mean, standard deviation and correlations between the control variables, social support, professional identity, emotional intelligence, surface acting and deep acting. The results show social support is positively related to professional identity (r=0.48, p <unk>0.01) and deep acting(r=.051, p <unk>0.01), and negatively related to surface acting(r= -0.46, p <unk>0.01). Professional identity is found to be positively related to emotion intelligence (r=0.16, p <unk>0.01) and deep acting (r=0.53, p <unk>0.01), and negatively related to surface acting(r=-0.44, p <unk>0.01). --- The Main Effect of Social Support and Mediating Effect of Professional Identity The main effect of social support (SS) and mediating role of professional identity (PI) is tested by path analysis and bootstrap methods using the process v3.5. The results are reported in Table 3 andshow: 1. The path coefficient of social support<unk>surface acting (<unk>= -0.19, p<unk>0.001) indicates that social support has significant negative effects on surface acting. The path coefficient of social support<unk>deep acting (<unk>=0.20, p<unk>0.001) indicates that social support has significant positive effects on deep acting. Therefore, Hypotheses 1a and 1b are supported. 2. The path coefficient of professional identity<unk>surface acting (<unk>= -0.29, p <unk>0.001) indicates that professional identity significantly negatively affects the surface acting. The path coefficient of professional identity<unk>deep acting (<unk>=0.39, p <unk>0.001) indicates that professional identity has significant positive effects on deep acting. Overall, these results support Hypotheses 2a and 2b. 3. Moreover, the indirect effect of social support on surface acting via professional identity is significant (<unk>= -0.08, p <unk>0.001), with 95% confidence interval (CI)= [-0.11,-0.05], and the indirect effect of social support on deep acting via professional identity is significant (<unk>=0.11, p <unk>0.001), with 95% confidence interval (CI)= [0.07,0.15]. Since all confidence intervals exclude zero, assuming that Hypothesis 3a and 3b are supported. --- The Moderated Effects of Emotional Intelligence (EI) The interaction term of professional identity and emotional intelligence on surface acting is significant (<unk>= -0.40, p<unk>0.001), indicating that emotional intelligence significantly moderates the relationship between professional identity and surface acting. The interaction term of professional identity and emotional intelligence on deep acting is also significant (<unk>= -0.34, p <unk>0.001), indicating that emotional intelligence significantly moderates the relationship between professional identity and deep acting. This paper further examines the moderation effect of emotional intelligence using simple slope plot in Figure 2 and Figure 3. As emotional intelligence improves, the negative impact of professional identity on surface acting is enhanced (<unk> low. = -0.135, t = -1.960, p <unk>0.05; <unk> high. = -0.744, t = -10.770, p <unk>0.001),

Drawing on the conservation of resources (COR) theory, the authors investigate whether and how social support impacts the emotional labor of live streamers through professional identity. They also explore the boundary conditions by focusing on the moderation effect of emotional intelligence. Based on a sample of 331 live streamers in the content industry, the results show that social support weakens (enhances) live streamers' surface acting (deep acting) by enhancing their professional identity. Emotional intelligence significantly moderates the professional identity-emotional labor relationship. In addition, they find that emotional intelligence strengthens the negative indirect effect of social support on surface acting through professional identity but weakens the positive indirect effect of social support and deep acting through professional identity. They also discuss theoretical contribution in emotional labor literature and practical implications for live commerce.

social support on deep acting via professional identity is significant (<unk>=0.11, p <unk>0.001), with 95% confidence interval (CI)= [0.07,0.15]. Since all confidence intervals exclude zero, assuming that Hypothesis 3a and 3b are supported. --- The Moderated Effects of Emotional Intelligence (EI) The interaction term of professional identity and emotional intelligence on surface acting is significant (<unk>= -0.40, p<unk>0.001), indicating that emotional intelligence significantly moderates the relationship between professional identity and surface acting. The interaction term of professional identity and emotional intelligence on deep acting is also significant (<unk>= -0.34, p <unk>0.001), indicating that emotional intelligence significantly moderates the relationship between professional identity and deep acting. This paper further examines the moderation effect of emotional intelligence using simple slope plot in Figure 2 and Figure 3. As emotional intelligence improves, the negative impact of professional identity on surface acting is enhanced (<unk> low. = -0.135, t = -1.960, p <unk>0.05; <unk> high. = -0.744, t = -10.770, p <unk>0.001), and Figure 2 shows that the linear slope at a high emotional intelligence level is greater than that at a low emotional intelligence level. Therefore, it confirms that emotional intelligence positively moderates the negative relationship between professional identity and surface acting. In Figure 3, it is found that as emotional intelligence improves, the positive impact of professional identity on deep acting is diminished (<unk> low. =0.858, t =12.966, p <unk>0.001; <unk> high. =0.328, t =4.961, p <unk>0.001), and the slope at a low emotional intelligence level is greater than that at a high emotional intelligence level. Therefore, emotional intelligence negatively moderates the positive relationship between professional identity and deep acting. Hypothesis 4a and 4b are supported. --- The Moderated Mediation Effects of Emotional Intelligence (EI) To gain a deep insight of how the above indirect effects differ by the levels of emotional intelligence (EI), we adopt the bootstrapping procedures to quantify the indirect effects at lower (mean-1SD) and higher (mean+1SD) levels of emotional intelligence (EI) by using process v3.5. As shown in Table 4, --- dISCUSSIoN ANd CoNCLUSIoN Drawing on the COR theory, this study examines the mechanism of how social support has an impact on the emotional labor using a questionnaire survey in e-commerce live streaming industry. The main findings of study conclude as follows: First, social support and professional identity has a negative effect on surface acting, and simultaneously, a positive effect on deep acting. Second, on the mediating effect of professional identity, we find that social support weakens (enhances) live streamers' surface acting (deep acting) by stimulating their professional identity. Third, emotional intelligence can enhance the negative indirect effect of social support on surface acting through professional identity. Our findings also indicate that emotional intelligence can weaken the positive indirect effect of social support on deep acting through professional identity. Compared with previous studies, our study not only emphasizes the direct effect of social support of live streamers on their emotional labor, but also examine the mediate effect of professional identity and moderating effect of emotional intelligence on the above relationship. These findings have important theoretical contribution to the existing emotional labor literature and have important managerial implication for e-commerce live streaming industry. Our empirical findings confirm that social support will have a significant effect on both surface acting and deep acting but in opposite directions. There is a consensus in emotional labor literature that social support is positively related to deep acting (Lam & Chen, 2012), however several studies indicate that employees engage in surface acting for complicated reasons and the relationship between social support and surface acting is also inconsistent and even conflicted (Xu et al., 2020). For example, Xu et.al (2020) argue that employees who get more social support may engage in surface acting because the supportive environment may make them feel free from punishment even their faked smile is detected by customers (Xu et al., 2020). Using e-commerce live streaming as a new context, our paper finds social support may reduce surface acting of live streamers, which is opposite with the arguments of Xu et.al (2020). One of the possible reasons is that supportive environment may also enhance the identification of streamers' profession and recognize the meaning of work. Therefore, future studies may examine these conflicting arguments using more fined empirical research design. Prior studies emphasize that internal energies (e.g., emotional intelligence, organizational commitment or motivation) can be used to cope with stress from emotional labor (Lam & Chen, 2012;Xu et al., 2020;Yin et al., 2013). Conforming to the COR theory, our study considers this point further by proposing that employees may increase deep acting or reduce surface acting when they understand and recognize the professional meaning and work-related identities. Specifically, gaining a high faith towards professional identity can help to better recognize emotional rules of live streamers and commit to their work so as to reduce live streamers' faking emotions. More importantly, we find the mediating effect of professional identity on the relationship between social support and emotional labor. It finds that social support can influence streamers' perception towards professional identity of live streaming and can thereby exert an effect on emotional labor by increasing deep acting and reducing surface acting. This finding may indicate that the influence of professional identity occurs before that of social support and professional identity is a more invisible but deeper factor affecting emotional labor strategies of live streamers. Furthermore, our paper highlights the moderation effect of emotional intelligence on the relationship between social support and emotional labor. Previous studies indicate that emotional intelligence, as an important personal resources, can help individuals handle negative emotions and facilitate employees' emotional labor (Chen et al., 2019;Grandey & Melloy, 2017). For example, several studies find that emotionally intelligent individuals normally tend to perform deep acting and reduce surface acting (Kim et al., 2012;Wen et al., 2019;Yin et al., 2013). However, there is limited research that has investigated emotional intelligence from an interaction perspective where emotional intelligence acts an individual characteristics that enable live streamers' social support that regulates the emotional strategies to have a stronger or weaker effect. It could be one of the possible reasons for inconsistent findings in the relationship between social support and emotional labor strategies because of neglecting some important contextual variables such as emotional intelligence (Xu et al., 2020). This paper finds emotional intelligence can enhance the mediating effect of professional identity on the relationship between social support and surface acting but may reduce the mediating effect of professional identity on the relationship between social support and deep acting. Our findings suggest even if employees have not yet established a high level of professional identity, higher emotional intelligence helps them to reduce surface acting by identifying, understanding their own and audiences' emotions, and adjusting to the required emotional rules. However, although both emotional intelligence and professional identity may foster meaningful deep acting activities (Ashforth & Humphrey, 1993;Wen et al., 2019), but these two kind of internal resources of individuals may has a substitution effect with each other on providing sufficient emotional abilities and resources to engage in deep acting. Our study has a number of limitations that are necessary to be considered in future studies. First, because this study selects e-commerce live streamers in China as our samples, the majority of respondents are female aged below 30, which limits the generalizability of our results. Since emotional labor exists in various service industries, whether the findings of this paper is applicable to other service industries and other regions or not needs to be further investigated in future studies. Second, we only use live streamers' self-reported and cross-sectional data, which may be subject to common method bias. In future studies, it may be necessary to use longitudinal studies, experimental designs or collect data in time lagged way. Third, the influence of social support on emotional labor may also be mediated by other factors. In this paper, professional identity is the key exploratory variable. Future research may explore other potential mediators of the influence of social support on live streamers' surface acting and deep acting. Finally, we only investigate emotional intelligence as potential moderator, and more research is needed to explore the moderating effect of other factors such as emotion regulation and self-efficacy (Deng et al., 2017). --- FUNdING AGENCy Open Access Funding for this article has been covered by the authors of this manuscript.

Drawing on the conservation of resources (COR) theory, the authors investigate whether and how social support impacts the emotional labor of live streamers through professional identity. They also explore the boundary conditions by focusing on the moderation effect of emotional intelligence. Based on a sample of 331 live streamers in the content industry, the results show that social support weakens (enhances) live streamers' surface acting (deep acting) by enhancing their professional identity. Emotional intelligence significantly moderates the professional identity-emotional labor relationship. In addition, they find that emotional intelligence strengthens the negative indirect effect of social support on surface acting through professional identity but weakens the positive indirect effect of social support and deep acting through professional identity. They also discuss theoretical contribution in emotional labor literature and practical implications for live commerce.

INTRODUCTION Among the factors associated with obesity, significant attention has been given to birth weight. High birth weight (HBW), especially greater than or equal to 4 kg, 1 has been correlated with overweight and/or obesity over the long term, in epidemiological studies. [2][3][4][5] Particularly in developed countries, HBW is the main risk factor for obesity among children and adolescents whose mothers were affected by gestational diabetes mellitus, because of the elevated number of adipose cells acquired by the infant. 6 However, other factors may overlap the association between birth weight and overweight/obesity, such as the mother's body mass index (BMI), 7 both the mother's and the father's BMI, 8 family income 3,8 and the type of school attended, with regard to developing countries. 9 These factors may reduce the importance of the birth weight variable in multivariate analysis. 3 Systematic reviews have shown that correlations between birth weight and overweight/obesity among children not only have presented contradictory results, but also have differed regarding the methods used. Thus, these reviews have demonstrated that research on this subject needs to be conducted with the proper methodological rigor, in order to identify factors that might reduce the effect of birth weight on overweight or obesity among children and adolescents. 10,11 Because of the contradictions in these studies, it is important to clarify the association between birth weight and overweight/obesity, so as to determine the stage of life at which this relationship first appears. Longitudinal studies have revealed these associations in adulthood. 12,13 A study conducted in the city of Florianópolis in the year 2002 by de Assis et al. 14, among children aged 7-9 years, found that the prevalence of overweight (including obesity) was 22.1%, in accordance with the criteria of Cole et al. 15 Data from this city's live births registry, obtained in 2005, showed that the prevalence of low birth weight (LBW) among live births in Florianópolis was 8.3%. 16 These proportions are similar to those found in developed countries, where an increase in HBW has been observed with a simultaneous decrease in LBW. 17,18 Florianópolis was classified as the fourth highest city in Brazil in terms of human development in 2000 and the third city in 2013. 19,20 Thus, the importance of the present study lies in the fact that in Florianópolis, elevated HBW prevalence that could be correlated with high rates of overweight and obesity would be expected. --- OBJECTIVE The purpose of this study was to investigate whether overweight and obese children and adolescents aged 7 to 14 years living in Florianópolis, Santa Catarina, southern Brazil, were born with low or high birth weight, taking into consideration the possible interactions between socioeconomic factors and other biological variables. --- METHOD Design, setting and ethics This study was conducted on a retrospective cohort from the year 2007. An assessment was made on a probabilistic sample of schoolchildren aged 7 to 14 years who were enrolled in public and private elementary schools in Florianópolis, Santa Catarina, southern Brazil. The schoolchildren in this investigation were included after obtaining consent from their parents or legal guardians, who signed a free and informed consent statement. This study had previously been approved by the Ethics Committee for Human Research of the Federal University of Santa Catarina (Universidade Federal de Santa Catarina, UFSC) through project number 028/06. --- Sample The sampling was divided into two stages. Firstly, the 221 schools in Florianópolis were listed according to their geographical location in the city (north, south, east, center or mainland) and their kind of institution (public or private). The number of schools selected in each of the four geographical areas was defined taking into account: the proportion of schools in each area compared to the totality of schools in the city and the proportion of private and public institutions in each geographical area, totalizing 17 schools (11 public and 6 private). The selection of the 17 schools was made randomly, by simple draw, but taking into account the previous list (stratified selection by geographical area and kind of school). Subsequently, students in each school were selected taking into consideration the ratios of schoolchildren registered in the 2004 school census in Florianópolis (53,595 individuals) in the following categories: geographical location of the student's home, kind of school, gender and age group. Sample size was calculated taking the prevalence of overweight (including obesity) among children aged 7 to 9.9 years to be 10%, 21 and 17% for adolescents aged 10 to 14 years, 22 with 95% confidence levels and a two-tailed sampling error of 2%. The design effect was estimated to be 1.3 and the power was taken to be 80%. This calculation resulted in a requirement for a sample size of 1,100 children aged 7 to 9.9 years and 700 adolescents aged 10 to 14.99 years. In addition, all the children who had participated in a previous study conducted in 2002, 14 and who in 2007 were adolescents enrolled in these randomly selected schools were also included. It was possible to locate 30% of all the students who had participated in the previous study in these randomly selected schools. Considering the error margins for losses in tests, the total sample was estimated to be 1,200 children and 1,900 adolescents (800 new adolescents plus 1,100 adolescents from the previous study). Data on 2,863 students were collected. Those younger than seven years of age (n = 18) and those who were 15 years or over (n = 16) were excluded, as were those for whom no valid data for weight (n = 2) or birth weight (n = 131) was available. Thus, the final sample investigated comprised 2,696 students (857 children and 1,839 adolescents), i.e. 94% of the total. --- Data collection Biological data (gender, age, birth weight, gestational age, weight, height, and subscapular and triceps skinfolds of the schoolchildren; and age, weight and height of their parents) and socioeconomic data (kind of school, i.e. public or private, family income level and parental educational level) were collected. Information relating to birth weight, gestational age, age, parents' weight and height and socioeconomic data were collected by means of a self-administered questionnaire that was sent to the parents and legal guardians of the students. Anthropometric data on the students were collected by previously trained anthropometry technicians, 23 following a protocol based on the recommendations of Lohman. 24,25 Weight was assessed using a Marte electronic scale, model PP 180 (Marte Cient<unk>fica, Santa Rita do Sapuca<unk> -MG, Brazil). with a capacity for 180 kg and precision of 100 grams. Height was measured using an AlturExata stadiometer (AlturExata, Belo Horizonte -MG, Brazil) with 1.0 millimeter precision. Skinfolds were measured using a Cescorf caliper (Cescorf Equipamentos Antropométricos, Porto Alegre -RS, Brazil) with 0.1 millimeter precision. The children and adolescents were measured without shoes and wearing light clothes. A pilot study was conducted among subjects who were not included in the sample and, following the recommendations from the World Health Organization, 26 the intra-rater technical measurement error (TME) for skinfold (SF) measurements was calculated. The intra-rater TME showed a reliability coefficient (R) greater than 0.95, which showed that all of the anthropometry technicians made skinfold measurements properly, thus resulting in low variability in the data. 23 The data were entered into EpiData 3.2 and were fully checked by the duly trained data entry team, and automatic consistency and amplitude checks were made. --- Statistical variables and analysis The outcome variables were overweight and obesity. Overweight was defined as BMI <unk> 85 th percentile, according to age and gender, as proposed by Must et al. 27 and as recommended by the Brazilian Ministry of Health until the year 2008. 28 Obesity was defined as BMI <unk> 85 th percentile, according to age and gender, as proposed by Must et al., 27 along with triceps and subscapular skinfolds in millimeters (mm) <unk> 90 th percentile as proposed by Johnson et al. 29 The criteria of Must et al. 27 were chosen because they are widely used in the literature and have been recommended by the World Health Organization. 24 These were also recommended by the Brazilian Ministry of Health 28 for evaluating nutritional status among children and adolescents until 2008, when this study was conducted. Birth weight and birth weight according to gestational age were the independent variables (or exposure) and were classified as follows: a) birth weight was classified as a single piece of data, as LBW (<unk> 3,000 g), appropriate birth weight (ABW; between 3,000 and 3,999 g) and HBW (<unk> 4,000 g); 24,30 and b) birth weight was correlated with gestational age in order to classify schoolchildren as small for gestational age (SGA), i.e. below the 10 th percentile; appropriate for gestational age (AGA), i.e. between the 10 th and the 90 th percentiles; and large for gestational age (LGA), i.e. above the 90 th percentile. 30 The World Health Organization classifies children as presenting insufficient birth weight if they are born weighing between 1,500 and 2,999 g and as presenting low birth weight if they are born weighing 2,500 g or less. 24 In our study, these categories were unified because the number of responses was low and this could have hidden an association between the outcome and the independent variables. The BMI of the students and their parents was obtained by dividing the weight measurement (in kg) by the square of their height (in meters). Non-elderly parents were classified as overweight if their BMI was between 25 and 30 kg/m 2 or obese if their BMI was <unk> 30 kg/m 2, as proposed by the World Health Organization, 24 and elderly parents were classified as overweight if their BMI was <unk> 27 kg/m 2, as proposed by the American Dietetic Association. 32 Both of these sets of criteria are recommended by the Brazilian Ministry of Health. 28 An analysis on the consistency of the database was made using the Stata version 9.0 statistical package. A descriptive analysis was conducted to show the prevalence ratios of overweight and obesity for each independent variable, and the prevalence of LBW, HBW, SGA and LGA. In an inferential analysis, models divided by age group and gender were used to calculate associations among children aged 7 to 9.9 years and, separately, among adolescents 10 to 14.9 years, because of the possible effect of sexual maturation among the adolescents and the differences between genders regarding the prevalences of overweight and obesity. In order to compare the prevalences of overweight and obesity for the different categories within each variable studied, an analysis was performed using Pearson's modified contingency coefficient, based on chi-square statistics. Poisson univariate analysis was used to investigate association between birth weight (independent variable) and overweight and obesity (outcomes). Multivariate analysis was performed to ascertain the extent to which the exposure variables influenced the outcomes. Additionally, 95% confidence interval (95% CI) and P-values were estimated. The Poisson analysis model was used because for high-prevalence outcomes in cross-sectional studies (more than 10%), odds ratio estimates are said to either overestimate or underestimate associations with outcomes, in comparison with prevalence ratios. 33,34 All the analysis took into consideration the effect of the sampling design, through the svy command in the Stata software, which is used to analyze data from complex samples. Associations among the variables for which the P-value was <unk> 0.05 were considered to be statistically significant. --- RESULTS Data on 2,863 children (aged 7-9 years) and adolescents (aged 10-14 years) were collected. The proportion of the data that comprised refusals or exclusions was 5.8%. Students less than 7 years of age (n = 18) and more than 14.9 years of age (n = 16) were excluded because these ages were outside of the study range. Invalid weight data (less than 10 kg, n = 2) and birth weight (lower than 800 grams, n = 131, and higher than 6 kg) were also excluded. Occurrences of no response or discrepant values were considered to be invalid data. Thus, 2,696 students (857 children and 1,839 adolescents) were studied. In relation to the initial number calculated (3,100), this study presented a data loss rate of 13%. Table 1 describes the data on the population studied. The LBW prevalences among the children and adolescents were respectively 8.2% and 7.8%. HBW was found in 7.4% and 7.9% of the children and adolescents, respectively. The prevalences of children and adolescents who were born SGA were 14.4% and 13.0%, respectively. Children and adolescents who were born LGA accounted for respectively 10.2% and 12.9%. The prevalence of overweight among the children was 31.5% and of obesity, 10.9%. Among the adolescents, the prevalence of overweight was 21.0% and of obesity, 6.0%. Table 2 shows the prevalences of overweight and obesity among the children according to gender and the variables investigated. In this table, it can seen that the prevalence of overweight among children was significantly higher in the following cases: boys born LGA (P = 0.02); children of both genders whose mothers were obese (P = 0.008 for boys and 0.001 for girls); and boys whose fathers were obese (P = 0.02). The prevalence of obesity was higher among girls whose mothers were obese (P <unk> 0.001); and among boys whose fathers were obese (P = 0.05). The other variables investigated did not demonstrate any significant association with overweight or obesity in children (Pearson chi-square test). Table 3 shows the prevalences of overweight and obesity among the adolescents (10-14 years of age). The prevalence of overweight was higher among the following: boys with HBW (P = 0.01) and LGA (0.01); boys in the top monthly per capita income quartile (P = 0.01); boys whose mothers (P = 0.003) and fathers (P <unk> 0.001) were obese; girls who attended public schools (P = 0.001); and girls whose mothers (P <unk> 0.001) or fathers (P = 0.002) were obese. The prevalence of obesity was higher among the following: adolescents with HBW (P = 0.002); LGA (P = 0.001); and those whose mothers (P <unk> 0.001) or fathers (P = 0.03) were obese (Pearson chi-square test). analysis (Table 2), it ceased to be correlated after controlling for confounding. The father's and mother's BMI were probably the variables that were most strongly associated with the outcomes. Table 5 shows the prevalence ratios for obesity among adolescents and their adjusted associations with birth weight. A significant association was found between birth weight and overweight among male adolescents, for those with HBW (PR = 1.14; 95% CI = 1.02-1.27; P = 0.03). However, it needs to be noted that this association observed between HBW and overweight among male adolescents was weak, because the prevalence of overweight among the male adolescents born with high birth weight was only 1.14 times greater than the prevalence of overweight found among adolescents born with appropriate weight. Hence, birth weight does not seem to be the principal factor that determines overweight in adolescence. Birth weight and birth weight according to gestational age were not significantly associated with These data show that a higher number of variables were associated with overweight and obesity among adolescents (12) than among children (6). Among children, the biological variables (birth weight for boys and parental BMI for both genders) were those that established a positive, proportional and significant association with the prevalences of overweight and obesity. Among adolescents, socioeconomic variables (kind of school and family income) were also associated in addition to the biological variables. The main variable of interest (birth weight) seemed to be associated only among males. Table 4 shows the prevalence ratios in adjusted analysis on the outcomes of overweight and obesity among children and their associations with the exposure variables. No association between birth weight and overweight or obesity was found among the children in multivariate analysis. Thus, although birth weight according to gestational age was correlated with these outcomes in univariate * P = 0.01; <unk> P = 0.001, <unk> P = 0.001, <unk> P = 0.001, || P = 0.01, <unk> P = 0.001, ** P <unk> 0.001, <unk> <unk> P = 0.003, <unk> <unk> P <unk> 0.001, <unk> <unk> P <unk> 0.001, |||| P = 0.002, <unk> <unk> P <unk> 0.001, *** P = 0.003, <unk> <unk> <unk> P = 0.03. --- DISCUSSION Investigating the influence of birth weight on overweight and obesity among children and adolescents is important, according to public health studies. This study in Florianópolis was justified by the city's high Human Development Index, which reached the fourth position in Brazil in 2000 and this index is continually increasing. Currently, Florianópolis is the third city in the country in terms of human development. 19,20 Moreover, the prevalence of newborns who were LGA (10.2% among the children and 12.9% among the adolescents), was similar, in 2000, to that found in developed countries. 17,18 The LBW prevalence found in this study (around 8%) is similar to that recorded in the Ministry of Health's Live Births Registry for the city of Florianópolis, which was 8.3% in 2005, 16 and the same has been found in developing countries such as Argentina and Mexico. 35 However, it must be pointed out that intrauterine growth restriction, which is an indication of poor nutrition during pregnancy, reached 27.4% in this study. The prevalences of overweight and obesity found in this survey (31.5% and 21% of the children and adolescents, respectively, were overweight and 10.9% and 6% were obese) were similar to those found in other studies on schoolchildren that used the same diagnostic criteria. For example, an assessment made by Dutra et al. 5 36,37 These findings may indicate that overweight and obese infants also tend to have overweight and obesity during childhood and adolescence. Regarding the influence of birth weight on the prevalence of overweight in the sample investigated, a significant association was found in the univariate Poisson regression analysis for male adolescents who were born LGA. No association between birth weight and overweight was found among children of both genders, or among female adolescents. In an analysis adjusted for the parental socioeconomic and biological variables, the most significant association found was between HBW and overweight among male adolescents. Additionally, associations with obesity were found in univariate analysis for male adolescents who were born LGA and for those with HBW. However, after inclusion of parental socioeconomic variables and parental BMI in the multivariate analysis, the associations with obesity ceased to be significant. Gilman et al. 4 conducted a cross-sectional study on a cohort 13 years and found that birth weight between 2,890 and 3,110 g and birth weight <unk> 3,110 g were associated with overweight according to the cutoff points set by Cole et al. 15 (OR = 2.85; 95% CI = 1.49-5.47; and OR = 7.03; 95% CI = 3.53-13.99, respectively), in a multivariate analysis. Monteiro et al. 3 obtained similar results to those of the present study, in a cross-sectional investigation embedded in a cohort of children born in the city of Pelotas in 1982. Using the same diagnostic criteria for overweight and obesity as used in this study, they also found through a univariate analysis that there was an association between birth weight according to gestational age and obesity (i.e. being born LGA) among adolescents aged 14 to 16 years. However, the association did not remain significant after adjusting for family income and maternal BMI. The present study showed that being born with high birth weight is a factor associated with overweight among male adolescents. However, this association seemed to be weakened by the variables of obesity in the mother and obesity in the father, in the multivariate analysis. In childhood, this association between birth weight and overweight/obesity was not observed. This seems biologically plausible, since there are reports in the literature showing a strong association between these variables in adulthood. 39 So, even though it was not so strong, the relationship between overweight in male adolescents and being born with high birth weight indicates that health services should promote nutritional monitoring among adolescents with a focus on lifestyle, in order to reduce the chances of overweight and obesity in adulthood. In addition, it is recommended that epidemiologists and other researchers should investigate whether adolescents with elevated birth weight were born from mothers who presented gestational obesity or gestational diabetes, because the origins of overweight may be found in intrauterine development. In relation to birth weight and parental weight and height, it should be stressed that this information was reported by the children's legal guardians through a self-administered questionnaire. This method was chosen in order to make it easier to collect data and administer the data collection instrument. Ara<unk>jo et al. conducted a validation study on the degree of agreement between reported birth weight information and birth weight measurements made immediately after birth, among eleven-year-old adolescents who were part of a cohort in Pelotas (Rio Grande do Sul) and found a high level of agreement regarding information on low birth weight (kappa = 0.73), with disagreements of the order of -20.0 g (standard deviation = 288.3). 40 However, when stratifying birth weight information, the authors found that reported information tended to be overestimated in the case of LBW children and underestimated in the case of HBW children. They pointed out that the linear relationship between birth weight and BMI was more consistent when birth weight was measured immediately after birth and not so precise when birth weight was reported. Therefore, validation for birth weight measurements in a subsample is highly recommended, so as to compare the reported variable with data gathered by hospital registries or on health record cards. In spite of these limitations, the present study had external validity, was probabilistic and had a complex sampling plan. Additionally, the researchers took care to train the anthropometry technicians, which resulted in reliable data. In addition to appropriate data gathering, care was taken in compiling the research data, which were fully entered and checked by a trained team, using software that enables data entry control. Therefore, the results from this study seem not to have been influenced by selection or measurement bias. --- CONCLUSION No significant association was found between high birth weight/being born large for gestational age and obesity after adjusting for the control variables, either among children or among adolescents. The same results were found for low birth weight and being born small for gestational age. Even though the association was not very strong, a relationship between high birth weight and overweight among male adolescents was observed. between the body mass index of schoolchildren aged between 7 and 14 years and that of their parents in the city of Florianópolis, in the State of Santa Catarina, Brazil]. Rev Bras Sa<unk>de Matern Infant.

CONTEXT AND OBJECTIVE: Being born heavier than 4 kg is associated with current overweight and obesity over the long term. The objective here was to ascertain whether birth weight was related to overweight or obese status, among 7 to 14-year-old schoolchildren, taking into consideration the possible interactions between socioeconomic factors and other biological variables. DESIGN AND SETTING: Retrospective cohort study on a probabilistic sample of 2,696 children and adolescents living in Florianópolis, Santa Catarina, Brazil.The following data were collected: anthropometric (student's weight, height and age; and parents' weight and height), socioeconomic (family income, number of people in house and parental schooling level), birth weight and gestational age. Overweight and obesity were classified using percentiles of body mass index and triceps and subscapular skinfolds. The outcome variables were overweight and obesity and the main explanatory variables were birth weight and birth weight according to gestational age. The control variables were the parents' nutritional status, their schooling level and the per capita family income. Poisson multivariate regressions were carried out. RESULTS: Higher prevalence of high birth weight was observed among overweight male adolescents (PR = 1.14; 95% CI = 1.02-1.27; P = 0.03), but this was not observed among obese male adolescents. Low birth weight and being born small for gestational age were also not associated with the outcomes. Among overweight and obese children, birth weight was not significantly different from that of normal-weight children. CONCLUSION: No significant association between birth weight and obesity was observed. However, there was a weak but significant association between high birth weight and overweight, among male adolescents. RESUMO CONTEXTO E OBJETIVO: Nascer com peso > 4 kg se associa a sobrepeso e/ou obesidade em longo prazo. O objetivo foi verificar se o peso ao nascer foi um fator relacionado ao estado de sobrepeso ou obesidade atuais, em escolares de 7-14 anos, levando em consideração interações possíveis com fatores socioeconômicos e outras variáveis biológicas. TIPO DE ESTUDO E LOCAL: Estudo coorte retrospectivo numa amostra probabilística de 2.696 crianças e adolescentes residentes em Florianópolis, Santa Catarina, Brasil. MÉTODOS: Coletaram-se dados antropométricos (peso, estatura e idade dos escolares, peso e estatura dos pais), socioeconômicos (renda familiar, número de pessoas na casa, escolaridade dos pais) e de peso ao nascer e idade gestacional. Sobrepeso e obesidade foram definidos utilizando-se os percentis de índice de massa corporal e as dobras cutâneas tricipital e subescapular. Sobrepeso e obesidade foram desfechos, e peso ao nascer e peso ao nascer/por idade gestacional foram variáveis explanatórias principais. As variáveis de controle foram: estado nutricional dos pais, seu nível de escolaridade e renda familiar per capita. Foram realizadas regressões de Poisson. RESULTADOS: Nos adolescentes do sexo masculino com sobrepeso, houve maior prevalência de elevado peso ao nascer (RP = 1,14; 95% IC = 1,02-1,27; P = 0,03). Naqueles com obesidade, não foi observado o mesmo. Baixo peso ao nascer e nascer pequeno para a idade gestacional também não se associaram aos desfechos. Nas crianças com sobrepeso ou obesidade, o peso ao nascer não diferiu daquele entre as eutróficas. CONCLUSÃO: Não se observou associação significativa entre peso ao nascer e obesidade. Porém, houve associação significativa, ainda que fraca, entre elevado peso ao nascer e sobrepeso, nos adolescentes do sexo masculino.

Background International health collaborations have been steadily increasing since the 1990s [1], often bringing together stakeholders from high-income countries (HICs) situated in the global North, where most funding sources are located, with stakeholders from lower-and middle-income countries (LMICs) located primarily in the global South. Whilst these collaborations may tackle a range of research questions, they often include questions about conditions that primarily affect people living in the global South [2,3]. Driving this growth in collaborations are expectations that such collaborations will play a significant role in mitigating global health disparities [4][5][6][7][8][9]. Another driving factor is a "desire to be socially responsible" [10]. International collaboration in health research may have the effect of increasing clinical and research capacity in global South contexts and afford scientists in resource-poorer countries an opportunity to participate in or lead innovative scientific research and to publish [7,11,12]. Notwithstanding these potential benefits and driving factors, a range of critiques has been levelled against international health collaborations. Such collaborative arrangements have been accused of being exploitative of Southern researchers and communities, with some researchers labelling such practices as neo-colonial [13][14][15][16]. And while many Northern researchers may not set out to reproduce inequalities based on a colonial past, often collaborations have been seen as paternalistic, creating what Okeke has called, "the little brother effect" [17]. To address concerns related to international collaborations re-entrenching unequal relations, several bodies of literature have provided guidelines, imperatives and suggestions for equity in international collaboration. These include substantial conceptual accounts about what equity is, why it is imperative, and how it can be achieved [18,19]. They draw on rich literature from political philosophy that explores the concept of equity and social justice and identify several components, including avoiding unfair power relations, recognition, inclusion in decision-making, and rights to self-development and adequate levels of well-being. Social justice means reducing unequal power relations such as subordination, exploitation, exclusion and violence [20][21][22] and includes recognition and affirmation of group difference [21,[23][24][25]. Three core aspects of misrecognition are: 1) cultural domination, 2) de-valuing and stereotyping social groups, and 3) rendering their knowledge and perspectives invisible [21,[26][27][28]. Accordingly, recognition encompasses affirmation of group difference, rendering the invisible visible and demonstrating respect. Social justice also means ensuring individuals and social groups, including those considered disadvantaged or marginalized, are included in making decisions that have a significant impact on their well-being [21,29,30]. Efforts should also be made to ensure that dominant power hierarchies are not reinforced, and those considered disadvantaged or marginalized are not included as tokens. Scholars often equate fairness with consensual and deliberative decision-making [21,[31][32][33]. Social justice further calls for ensuring self-development (understood as developing and exercising one's capacities) and human flourishing (understood as achieving an adequate level of well-being for all) [21,29,[34][35][36]. Philosophers argue for giving some priority to bringing disadvantaged or marginalized individuals and groups/communities up to an adequate level of well-being [34,37,38]. Ethics researchers have applied these and other concepts from the philosophy literature on health and social justice to explore what equity means for international research [39][40][41][42]. This conceptual work has proposed that international research collaborations should generate new knowledge to improve the health and well-being of LMIC populations, particularly those considered disadvantaged or marginalized; foster their and LMIC researchers' meaningful participation in decision-making about its conduct; and build research capacity in LMICs [39,40,43]. However, debates continue as to whether international research collaborations even have a responsibility to contribute to justice beyond the micro-level (i.e. achieving a fair balance of burdens and benefits during individual projects). Some scholars have argued that international research collaborations should not be expected to contribute to reducing global health disparities and building research capacity, whereas others strongly disagree [43][44][45]. Additionally, several guidelines and frameworks have been developed that seek to promote more equitable collaborations. These include the "Responsible Conduct in the Global Research Enterprise" [46], the Montreal Statement on Research Integrity in Cross-Boundary Research Collaborations [47], the guidelines developed by the Commission for Research Partnership with Developing Countries [48] and the COHRED Fairness Index for international collaborative partnerships [49]. Moreover, there have been several qualitative studies reporting on researcher's experiences within international collaborations [4,50]. While these papers provide necessary empirical evidence for what constitutes both inequity and equity, there has been no systematic attempt to synthesize empirical studies in this domain. The absence of such a synthesized understanding challenges researchers' ability to proactively engage in and understand how equity in research collaboration should be promoted. To address this gap, we conducted a scoping review, mapping and synthesizing research from qualitative studies investigating dimensions of equity in international health collaborations. --- Aim To develop, through synthesizing evidence from published articles, key areas considered critical to fostering equity in collaborations identified by stakeholders involved in international collaborations. --- Method Scoping reviews are generally used to map emerging evidence relating to broad topics [51][52][53] and could focus on: synthesizing the available evidence in a field; defining and clarifying concepts and ideas; identifying where more information is required or identifying a question for a systematic review [51]. For this review, we were interested in synthesizing evidence relating to equity and developing greater clarity on various dimensions considered important for international collaborations to be equitable. --- Identifying relevant data --- Study selection The third step of the framework involves developing and applying criteria to consistently and transparently select relevant studies. In this review, we applied the following criteria: studies which reported on a researcher's or scientist's perspectives or opinions on what they perceived made an international scientific collaboration (in) equitable, or studies which reported on empirical qualitative evidence about what made international scientific collaboration (in)equitable. Studies which reported on tools, frameworks, guidelines or regulations that could be implemented to ensure equity were excluded. Studies reporting on any normative or philosophical accounts of what constituted equity in international research collaborations were also excluded (Table 1). Two reviewers (MCF and NSM) applied the selection criteria, first at the title and abstract level and subsequently at the level of full-text articles. Each reviewer independently reviewed articles at the title/abstract and full-text stages. At the title/abstract review stage, articles, where both reviewers agreed, were included in the fulltext search. Where there was a discrepancy, a third reviewer (JDV) adjudicated, and decided if papers should be included for full-text review. For the full-text search, articles where both reviewers (MCF and NSM) agreed, full-text articles were included. A third reviewer (JDV) applied the selection criteria to all full-text manuscripts where the two reviewers (MCF AND NSM) disagreed or were unsure. --- Inclusion criteria For empirical studies to be included in this review, they had to meet the following criteria: qualitative studies study participants were researchers or those involved in research investigated experiences, ideas, concepts, values related to equity within international collaborations. For opinion or editorial pieces, the articles had to have a focus on equity within international research collaborations, particularly describing experiences, ideas, concepts and values relating to equity. Articles which did not meet these criteria were excluded from the review. --- Mapping the data Once studies were selected, the fourth step of the scoping review process is to map or "chart" the data [52]. While there are no commonly agreed methods for this step, what is important is that the method systematically organizes the data into relevant themes [52,53] not dissimilar to how conventional thematic analysis would operate in qualitative studies [54]. We approached this by firstly, extracting publication details such as author institutional affiliation and dates of publication. We then extracted data related to what studies reported as important for creating equitable international scientific collaborations. For this stage of the review, articles were read through several times and a thematic coding scheme was developed inductively by one reviewer (MCF) in discussion with another author (JDV). After themes were derived and papers coded using broad themes, sub-themes were developed. During each of stage of thematic development, themes and sub-themes were presented to two of members of the research team (NSM and JDV) who had also read through included papers and could provide feedback on the accuracy of themes. --- Results --- Search results We retrieved a total of 7611 results after duplicates were removed. After screening all papers at the title and abstract level, 49 papers were selected for fulltext screening. A total of seven papers met the inclusion criteria, with an additional four papers identified for inclusion after searching the bibliographies of the papers that matched the inclusion criteria. In addition to hand-searching, we also, through Google Scholar and the citation tracking function, checked for papers that had cited the papers included in our study but found none. A total of 11 articles were included in the final analysis, with 10 of these being qualitative studies, and one an editorial piece. (see Figs. 1 and2). --- Overall findings Most of the studies included in this review involved a range of participants in the research process, namely: bio-medical and social science researchers; university administrative staff and managers; community members; and government officials. Of the 10 empirical studies, three studies included participants from a single country in the global South. The remaining studies (n = 7) included participants based in the global North and South, with some participants being based in both contexts (see Tables 1 and 2 for a summary of demographics of included studies). All papers were published between 2003 and 2019. At the level of authorship, six of the 11 first authors were from LMICs, with six out of 10 senior 3 for a summary of authorship demographics. After conducting a thematic analysis for the 11 included papers, 10 themes relating to equity in international scientific collaborations were identified: funding; capacity building; authorship; sample ownership and export; research agreement; meeting health priorities in Southern-based contexts; trust; acknowledging inequality; recognition of all stakeholders; and communication. --- Funding Funding was a central determinant of equity and was reported on in 6 of the 11 articles included. Participants described that since research funding primarily originates from the global North, power is therefore located with the Northern partners [2,55]. This arrangement has important implications for equity. For example, two studies reported that their participants, based in the global South, often had little influence on how and where money was spent [2,56]. Another study reported that participants observed equitable collaborations when researchers based in LMICs can apply directly for funding and subsequently approach researchers based in HICs to collaborate [57]. Relating to dependency, participants in one study highlighted that researchers in the global South often struggle to identify calls for funding, and have insufficient capacity (particularly with regard to human resources) to develop funding proposals for international funding [58]. Another study noted that African researchers typically do not have access to sustainable funding which limits their ability to independently raise funds and pursue their own research agendas [3]. Participants lamented their own dependency on Northern funders since there was a serious lack of funding for research from local governments in the global South [2]. --- Capacity building A second key theme that emerged from our analysis was capacity building, which was identified in 10 out of 11 articles. For most studies, capacity building was defined as ensuring that the partner located in the global South received relevant training. Relevant training was considered training related to the current research project and [56]. In a different study, capacity building meant increasing infrastructural support [3]. Specifically, the articles emphasized that researchers based in the South should not simply be viewed as'sample collectors'. Two studies were critical of traditional notions of capacity building, asserting for instance that the objective of capacity building activities should not simply be to upskill researchers but to ensure that over time, Southern researchers come to rely less on the technical assistance of Northern partners to conduct research [3,59,62]. Importantly, two studies cautioned that capacity building was not only a one-way process where researchers based in LMICs are educated by researchers based in HICs, but rather that capacity building should be a reciprocal process where both partners learn from each other since both partners are contributing different skills and resources to the project [55,63]. Conceptualizing capacity building as a unidirectional activity risks obscuring the skills and knowledge of Southern researchers, while unwittingly propping up Northern researchers as intellectually or academically superior [63]. --- Authorship in scientific collaborations The third emerging theme we identified related to authorship. Five studies highlighted the importance of fair authorship practices [3,55,60,61,63]. The primary concern was that researchers based in LMICs are duly recognizedat a minimum, that they should be listed as co-authors, but ideally should be supported to take on first-or senior author roles. Such a change would involve empowering scientists from the South to lead analyses, initiate and manage the writing and publication processes. --- Sample ownership and export Another theme related to sample ownership, export and secondary uses of the samples [2,60,64].. Participants in two of the 11 studies articulated concerns that when samples were exported from LMICs for analysis in laboratories in HICs, there is little control over the use of those resources and the knowledge generated from those samples [3,57]. Of particular concern was that once samples were only stored abroad, it would be difficult to ensure that the researchers involved in the initial collection of these resources would continue to be involved or recognized in subsequent work or publications emanating out of the research [3,64]. And this was considered to undermine equity in international research collaborations. --- Research agreements The fifth theme we identified was the importance of clear research agreements to promote issues of equity. Four of the studies noted that for collaborations to be equitable, details of the collaboration, such as how funding will be spent, division of labour, decision-making arrangements and authorship policies, should be clearly articulated at the outset [3,57,58,63]. Studies also underscored the critical role of researchers based in LMICs in negotiating fair terms for such agreements [56,57,59]. --- Local priorities An important criticism of international collaborations is that research priorities in the global South are not taken seriously by Northern collaborators. Three studies reported that international collaborations should consider or focus on local health priorities as key objectives of studies as opposed to researchers based in HICs deciding what those objectives should be [3,55,61]. Studies also described the importance of local communities benefitting from international collaborations. While the study did not articulate what these benefits could be, the article described how difficult it was for Southern-based researchers to spend money on research, but not being able to financially assist their research participants who were often struggling to survive [56]. --- Trust Trust was another key relational aspect of developing equitable partnerships. Most notable was the relationship between trust and funding. In one study, the Northern and Southern participants reported that funders tend not to trust Southern researchers to manage funds and preferentially trust Northern partners to handle fund management [55]. Another paper reported that as a result of funding generally being located in the North, Northern partners often dictate how money is spent, without having to account to Southern partners [2]. Such issues of mistrust could be bypassed (to some extent) when Southern staff were managed by Southern researchers and managers who had the cultural sensitivity necessary to create a cooperative working environment [63]. However, trust often takes a long time to build between collaborators, and having open and honest discussions between collaborators is essential for doing so [56]. Additionally, there were two factors which signalled a collaborator could be trusted. First, it was the reputation "and the absence of undesirable qualities" and second, this was also extended beyond the individual to include the trust funders placed in institutions where researchers located [56]. Yet trust at the interpersonal level is insufficient to mitigate the asymmetrical power relations between collaborators that are rooted in uneven international funding arrangements. --- Acknowledging inequality Studies also reflected on the importance of acknowledging that most collaborations between researchers based in LMICs and HICs are inherently unequal, and it would be disingenuous to pretend that the inequality did not exist. For one study, this acknowledgement was about researchers from both the global North and South to acknowledge the different capacities each team brings to the collaboration and the limitations of their contributions [56]. Other studies reflected on the importance of acknowledging more material and structural inequalities that are inherent in international collaborations, such as differences in training, exposure to technology, skill sets, and funding and other resources, which is as a result of past and contemporary geopolitical arrangements [1,2,55,58]. --- Recognition of the contribution of all collaborators Recognizing the role of various stakeholders contributing to the success of the project was also seen as critical for equity. Five of the included studies reported that their participants felt that recognizing skills, abilities, and local expertise was an essential element of reducing inequality. In practice, this also entailed acknowledging actors who would usually not qualify for co-authorship [2,3,55,60,63]. From some studies, this recognition also entailed broadening the group of people who get recognized and should include stakeholders such as those collecting samples and data [56]. In line with views on capacity building, one study emphasized the importance of recognizing the limitations of researchers based in HICs as well [63]. Related to this was the importance of valuing equality. The studies described that what often stifles more equitable relationships is when some roles are de-valued, or other roles receive relatively greater recognition or prominence [56,63]. Such arrangements "undermine the opportunity for change when Northern personnel, as 'capacity providers' are unable to admit to need, and Southern researchers, as'receivers', are not acknowledged for existing capacity" [63]. --- Communication Communication emerged as another component for creating equitable relationships. Specifically, three studies observed that collaborations should be set up in ways that allow for open and honest communication between researchers at the outset of the collaboration [1,2,61]. For example, one study noted that open communication at the outset allowed all stakeholders to feel heard and be taken seriously [56]. Another study noted that transparent communication at the start of the collaboration was critical to clarify roles and expectations, and would help prevent conflict at a later stage of the collaboration [2]. --- Discussion This scoping review synthetized empirical evidence about the experiences and understanding of equity in international research collaborations. Our review identified 10 dimensions of international scientific collaborations that were considered to be important in promoting equity. At the level of structural aspects of the collaboration, issues relating to where funding is held and control over funding, authorship arrangements, two-way capacity building, sample ownership and export were described as important in fostering equity. Also essential were fair research agreements negotiated at the outset of the collaboration, with LMIC researchers being able to ensure that study benefits are aligned to local health priorities. The review also highlighted more relational aspects of research collaborations that underpin equity such as explicitly acknowledging and discussing the impact of existing inequality, and ensuring recognition of the work of all stakeholders in the research endeavourincluding the contributions for instance of the people organizing sample collection efforts. Moreover, trust was critical to fostering equity, alongside the need for researchers to be able to communicate openly and transparently. Structural and relational dimensions were interrelated in the sense that the articles revealed that to achieve equity with regards to a relational aspect, also requires attention to be given to a structural dimension. For example, in our reporting, we noted that while trustbuilding is a key relational dimension, it was often influenced or determined by funding arrangements, a structural element. Beyond the structural and relational, we also draw attention to the fact that the identified domains map onto five components of social justice identified in the political philosophy literature. Funding location and control over funding issues are consistent with unfair power relations of subordination, where a privileged few get to determine the rules and make decisions that apply to many others [20]. Subordination encompasses an unfair division of labour in the workplace between those who plan and those who execute [21]. This has been a feature of HIC-LMIC international research collaborations [3,63]. Acknowledging inequities, recognition, and authorship domains are consistent with social justice as recognition, which entails affirming group differences and valuing others, especially those who have been marginalized by social institutions and norms [21,23,24]. In the international research context, LMIC researchers have been marginalized by funding institutions and collaboration norms. Sample ownership issues were connected to unfair power relations (i.e., subordination or control by others) and recognition. Capacity building and trust speak to rights of self-development and achieving adequate well-being, specifically relational aspects of well-being: affiliation and commune. Commune refers to relations of harmony with others [65]. Where research collaborations are characterized by relations of discord (i.e. ill will, us versus them), they can negatively affect members' well-being. Finally, open communication and research agreements that are characterized by fair terms and transparency are consistent with norms of fair decision-making and deliberation. Inherent in existing accounts of deliberative democracy are norms such as inclusion, reciprocity, reasonableness, and publicity [29,31]. We thus propose that these five components of social justicenamely avoiding unequal power relations, group recognition and affirmation, inclusive decisionmaking, promoting everyone's well-being and ensuring self-development -(at least) are relevant for equity in international collaborations. Future conceptual work should explore what upholding particular components of social justice means for the relevant domains of equity in international research collaborations identified in this paper, i.e. what reducing unfair power relations calls for in terms of funding. How each domain should be defined can draw on rich work from political philosophy. In comparing our findings to the (limited) ethics literature that has applied theories of health and social justice to develop guidance on equity in international research collaborations, we note that relational matters such as recognition, communication, trust, and acknowledging inequities are largely not discussed. This paper's findings, however, indicate that relational aspects of equity in international research collaborations are important, and we suggest they, therefore, require further consideration and exploration. Matters of building research capacity, inclusive decision-making throughout the research process, and ensuring collaborations are responsive to the health care and system needs of LMIC populations are discussed. However, the implications of inclusive decision-making for developing research agreements have not been specifically focused upon [39,41,66]. Fair funding arrangements was a key theme which emerged as critical in promoting more equitable collaborations. A key criticism of funding arrangements was that funding originates in the North, and Northern partners often have control over how and where money is spent, which severely restricts Southern researchers' power in collaborations. Although there appeared to be a perception that funders tend to prefer Northern partners to oversee funding activities, it is not clear if this is indeed the case. For example, funding from national government agencies is more likely to require that the funds be held at an institution in that country, irrespective of where the research is carried out. This may differ in cases where the primary focus is research in LMICs. The dynamics of funding have recently started to be explored, though the issue of control identified in this paper does not always feature. For example, Pierson and Millum explore what individual health research funders are ethically obligated to do to help reduce global health disparities [67]. However, they are more focused on how funders should allocate their resources to various illnesses than to how resources should be allocated amongst researchers to reduce relations of subordination. Pratt and Hyder, in contrast, argue that equity means funders should incentivize LMIC ownership of international research agendas by restricting lead applicant eligibility to LMIC institutions [68]. An example of where that was the case is the H3Africa Consortium [69]. Moreover, with regards to structural elements, whilst the articles we reviewed, described the importance of capacity building and fair funding arrangements, these domains were considered almost exclusively between researchers or research groups. What is missing from almost all the papers included in this review are institutional forms of equity. For example, forms of capacity building described in the articles were training in research methods and new techniques, formal graduate training programmes and developing grant writing skills so that Southern-based researchers can attract their own funding. While such forms of capacity building are critical, they are only focussed on capacity building directly related to the research and only consider an individual or group. Only one study in this review reported on the importance of building more structural forms capacity, for instance, strengthening ethics review capacity [3]. The articles included in this review did not consider broader frameworks for building institutional or structural capacity [70,71], or the importance of also building social science expertise [72]. Similarly, while the articles we included did report on the importance of funding to support research in lowerand middle-income countries, they narrowly focused on direct funding for research. Indirect costs of research funding are critical for institutional capacity building, such administrative and financial offices which provide critical support for research to flourish [73]. Importantly, the under-funding of institutional capacity, especially at the administrative level, has the potential to erode capacity of Southern researchers and institutions [73]. Finally, one of the critiques of international research collaborations is that researchers in LMICs often do not lead publications. Bibliometric studies on authorship in global health research reveal that a significant proportion of articles reporting on international health research conducted in LMICs had an author affiliated to the LMIC where the research was carried out [74,75]. Overall, less than 15% of publications on global health research did not have a co-author from the LMIC of interest [75]. The problem seems to lie more with the authorship order, whereby authors from LMICs are less likely to be first or last authors [76,77]. This scoping review highlights a desire for authorship parity in international health research collaborations. --- Implications While the 10 dimensions we identified provide overall guidance about what necessary dimensions of equitable collaborations, the findings of this review also underline the highly complex nature of achieving equity in collaborations. This complexity relates, firstly, to the differing perspectives from researchers and other stakeholders about which domains are required to establish equitable relationships. Secondly, there is also a variety of perspectives about what each domain requires to achieve equity in collaboration. For example, capacity building often means different things to researchers depending on their context. While many themes are generally consistent across many studies, such variance presents important challenges to ensuring equity. What this scoping review perhaps suggests is that finding standard ethical arrangements across time and space is difficult, and these aspects should be worked out in different contexts and collaborations. While such variance is important, especially considering multiple contextual factors, having no baseline shared, and standardized understanding of equity may also potentially undermine equity within international collaborations [78]. While these dimensions cannot be standardized across all collaborations, our findings demonstrate these are the key areas which must be carefully considered for international collaborations to develop more equitable practices. In addition, what also needs to be considered is that achieving equity often requires both structural and relational obligations to be fulfilled beyond individual researchers and/or institutions. For example, as it relates to matters of authorship or building trust, these could be worked out between researchers and institutions. However, funding practices are often determined by funders, their policies and the constituencies to whom they are accountable, well beyond the control of individual researchers and research institutions. --- Limitations One important limitation of this study was the small number of papers included in the review. This could suggest that our search strategy did not yield adequate results. We developed a comprehensive search strategy that included as many relevant key terms as possible, searching in as many relevant databases as we had access to. The initial search yielded a large number of publications, yet surprisingly few papers were included. To expand our search, we hand-searched the bibliographies of all included studies for any papers that we may have missed in our original search, and we found an additional four papers which met our inclusion criteria. In addition to hand-searches, we also, through Google Scholar and the citation tracking function, checked for papers that had cited the papers included in our study but found none. We think that the relatively low number of papers on this topic suggests that there are simply very few empirical studies which focus on equity within international collaborationsa surprising finding especially considering the growth in collaborative practices in global health. The second limitation of this study is that we did not include normative accounts or other literature focussing on providing framework or guidelines related to equity in international collaboration and this is an equally important analysis that needs to be conducted going forward. The third limitation of this study is that our search was limited to studies in English, and may therefore have missed studies published in other languages. --- Conclusion International collaborations have increasingly become a defining feature of global health research. While international collaborations offer important benefits, they have also been described as re-inscribing unequal power relations in favour of researchers and institutions situated in the global North. While there have been several attempts in the literature to address equity within international collaborations, to date, there had not been any attempts to summarise this evidence. Our scoping review, mapping evidence from qualitative studies and opinion and editorial pieces, has articulated 10 dimensions of equity. We argue that these 10 dimensions form the key areas which must be considered when developing equitable international collaborations. --- Appendix --- Search strategy --- PubMed: Research AND "Empirical studies" OR "qualitative research" OR "qualitative methods" OR opinion OR perspective. EBSCO Host: Researcher OR investigator OR scientist. AND Collaboration OR collaborators OR cooperation OR "international collaboration" OR "decision making" OR "international scientific collaboration" OR "international research collaboration". AND Ethics OR ethical OR equity OR equitable OR equality OR fair OR fairness OR values OR justice OR "social justice". AND "Empirical studies" OR "qualitative research" OR "qualitative methods" OR opinion OR perspective. --- Authors' contributions MCF: Conceptualization, analysis, methodology, writing, reviewing editing. NSM: Conceptualization, analysis, reviewing, editing. NABN: Conceptualization, reviewing, editing. BP: Conceptualization, analysis, methodology, reviewing, editing. JVD: Conceptualization, analysis, methodology, reviewing, editing. The authors read and approved the final manuscript. --- Funding The research presented in this article was supported by a research grant from the Wellcome Trust (Title: Equity and the Human Cell Atlas; WT 219600/ Z/19/Z). --- Availability of data and materials --- Not applicable Ethics approval and consent to participate Not applicable. --- Consent for publication Not applicable. --- Competing interests None. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Whilst global health research often involves international collaborations, achieving or promoting equity within collaborations remains a key challenge, despite established conceptual approaches and the development of frameworks and guidelines to promote equity. There have also been several empirical studies documenting researchers' experiences of inequity and views on what is required to advance equity in global health collaborations. While these empirical studies provide critical insights, there has been no attempt to systematically synthetize what constitutes equity and how it can be achieved. To address this gap, we conducted a scoping review of qualitative studies, opinion and editorial pieces about what equity is and how it can be promoted in international collaborations. Methods: We conducted a scoping review to explore domains of equity in international health collaborations. This review included qualitative studies and opinion pieces or editorial pieces on equity in international health collaborations. We mapped the data and identified common themes using a thematic analysis approach. Results: This initial search retrieved a total of 7611 papers after removing duplicates. A total of 11 papers were included in this review, 10 empirical studies and 1 editorial piece. We conducted our search between October -November 2019. We identified 10 key domains which are important for promoting equity in international collaborations: funding; capacity building; authorship; sample ownership and export; trust; research agreement; acknowledging inequality; recognition and communication. Discussion: Our findings suggest that for international collaborations to be considered more equitable, it must at least consider the 10 domains we highlighted. The 10 domains map onto five key aspects of social justice theory, namely avoiding unequal power relations like subordination, group recognition and affirmation, promoting the well-being of all, inclusion in decision-making and ensuring self-development.

Introduction Social network users are constantly bombarded with digital content. While the sheer amount of information users have access to was unthinkable just a couple of decades ago, the way in which people process that information has also evolved drastically. Social networks have become a battlefield for information warfare, with different entities attempting to disseminate content to achieve strategic goals, push agendas, or fight ideological battles (Rowett 2018;Denning 1999). As part of this tactic, governments often employ "armies" of actors, operating from believable accounts and posting content that aims to manipulate opinion or sow public discord by actively participating in online discussions. Previous work has studied the involvement of state-sponsored accounts in divisive events, e.g., the Black Lives Matter movement (Stewart, Arif, and Starbird 2018) or the 2016 US elec-Copyright c 2020, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. tions (Badawy, Ferrara, and Lerman 2018;Boyd et al. 2018), highlighting how these entities can be impactful both on the information ecosystem and in the real world. In today's information-saturated society, the effective use of images when sharing online content can have a strong influence in whether content will catch people's attention and go viral (Berger and Milkman 2012;Jenders, Kasneci, and Naumann 2013;Khosla, Das Sarma, and Hamid 2014). Users often feel overwhelmed with how much content they are exposed to (Koroleva, Krasnova, and Günther 2010), and pay attention to each piece of information for short amounts of time, with repercussion to their attention span (Wrzus et al. 2013). In fact, previous research showed that 60% of social network users re-share articles on social media without reading them, basing their decision on limited cues such as the title of the article or the thumbnail image associated with it (Gabielkov et al. 2016). Therefore, as part of the efforts aimed to actively push agendas, state-sponsored accounts do not only use textual content, but also take advantage of the expressive power of images and pictures, e.g., using politically and ideologically charged memes (Rowett 2018). In Figure 1, we report some (self-explanatory) examples of images pushed by state-sponsored accounts on Twitter, showcasing their unequivocally political nature and how they can be used to push agendas. Nonetheless, the role of images in information diffusion on the Web has attracted limited attention from the research community, which has thus far mainly focused on textual content (Badawy, Ferrara, and Lerman 2018). In this paper, we begin filling this gap by studying the use of images by state-sponsored accounts, aka Russian trolls (Gadde and Roth 2018). In particular, we focus on the following research questions: 1. What content is disseminated via images by statesponsored accounts? 2. Can we identify the target audience of Russian statesponsored accounts by studying the images they share? 3. How influential are these accounts in making images go viral on the Web? How does this influence results compare to previous characterizations that look into the spread of news by these accounts? Aiming to address these questions, we use an image- processing pipeline, expanding that presented by (Zannettou et al. 2018), to study images shared by state-sponsored trolls on Twitter. More precisely, we implement a custom annotation module that uses Google's Cloud Vision API to annotate images in the absence of high-quality ground truth data, or for images that are not bounded to a specific domain (e.g., memes). We then run the new pipeline on a dataset of 1.8M images from the 9M tweets released by Twitter in October 2018 as part of their effort to curb state-sponsored propaganda (Gadde and Roth 2018). These tweets were posted by 3.6K accounts identified as being controlled by the Russian Internet Research Agency (IRA). Finally, we quantify the influence that state-sponsored trolls had on other mainstream and alternative Web communities: namely, Twitter, Reddit, Gab, and 4chan's Politically Incorrect board (/pol/). To do this, we use Hawkes Processes (Linderman and Adams 2014;2015), which allow us to model the spread of the images across multiple Web communities and assess the root cause of the image appearances. Main Findings. Along with a first-of-its-kind characterization of how images are used by state-sponsored actors, our work yields a number of interesting findings: 1. The sharing of images by the trolls coincides with realworld events. For instance, we find a peak in activity that is clearly in close temporal proximity with the Unite the Right rally in Charlottesville (Spencer 2017), likely suggesting their use to sow discord during dividing events. 2. Our analysis provides evidence of their general themes and targets. For instance, we find that Russian trolls were mainly posting about Russia, Ukraine, and the USA. 3. By studying the co-occurrence of these images across the Web, we show that the same images appeared in many popular social networks, as well as mainstream and alternative news outlets. Moreover, we highlight interesting differences in popular websites for each of the detected entities: for instance, troll-produced images related to US matters were mostly co-appearing on mainstream English-posting news sites. 4. Our influence estimation results highlight that the Russian state-sponsored trolls, despite their relatively small size, are particularly influential and efficient in pushing images related to politics to other Web communities. In particular, we find that Russian state-sponsored trolls were more influential in spreading political imagery when compared to other images. Finally, by comparing these results to previous analysis focused on news (Zannettou et al. 2019b), we find that trolls were slightly more influential in spreading news via URLs than images. --- Related Work Trolls and politics. Previous work has focused on understanding the behavior, role, and impact of state-sponsored accounts on the US political scene. (Boyd et al. 2018) perform linguistic analysis on posts by Russian state-sponsored accounts over the course of the 2016 US election; they find that right-and left-leaning communities are targeted differently to maximize hostility across the political spectrum in the USA. (Stewart, Arif, and Starbird 2018) investigate the behavior of state-sponsored accounts around the Black-LivesMatter movement, finding that they infiltrated both right-and left-leaning political communities to participate in both sides of the discussions. (Jensen 2018) finds that, during the 2016 US election, Russian trolls were mainly interested in defining the identity of political individuals rather than particular information claims. Trolls in social networks. (Dutt, Deb, and Ferrara 2018) analyze the advertisements purchased by Russian accounts on Facebook. By performing clustering and semantic analysis, they identify their targeted campaigns over time, concluding that their main goal is to sway division on the community, and also that the most effective campaigns share similar characteristics. (Zannettou et al. 2019a) compare a set of Russian troll accounts against a random set of Twitter users, showing that Russian troll accounts exhibit different behaviors in the use of the Twitter platform when compared to random users. In follow up work, (Zannettou et al. 2019b) analyze the activities of Russian and Iranian trolls on Twitter and Reddit, finding substantial differences between them (e.g., Russian trolls were pro-Trump, Iranian ones anti-Trump), that their behavior and targets vary greatly over time, and that Russian trolls discuss different topics across Web communities (e.g., they discuss about cryptocurrencies on Reddit but not on Twitter). Also, (Spangher et al. 2018) examine the exploitation of various Web platforms (e.g., social networks and search engines), showing that statesponsored accounts use them to advance their propaganda by promoting content and their own controlled domains. Finally, (Broniatowski et al. 2018) focus on the vaccine debate and study Twitter discussions by Russian trolls, bots, and regular users. They find that the trolls amplified both sides of the debate, while at the same time their messages were more political and divisive in nature when compared to messages from bots and regular users. Detection & Classification. (Badawy, Lerman, and Ferrara 2019) use machine learning to detect Twitter users that are likely to share content that originates from Russian statesponsored accounts, while (Im et al. 2019) detect Russian trolls using machine learning techniques, finding that these accounts are still very active on the Web. Also, (Kim et al. 2019) classify Russian state-sponsored trolls into various roles: left-or right-leaning or accounts that pose as news outlets. By applying their technique on 3M tweets posted by Russian trolls on Twitter, they find that despite the fact that trolls had multiple roles, they worked together, while for trolls that pose as news outlets, they find that they had multiple agendas. For instance, some were posting about violent news to create an atmosphere of fear, while others focused on posting highly biased political news. Remarks. Overall, unlike previous work, we focus on content shared via images by state-sponsored accounts. Indeed, to the best of our knowledge, ours is the first study performing a large-scale image analysis on a ground truth dataset of images shared by Russian trolls on Twitter. Previous research (Gabielkov et al. 2016) has showed that social network users usually decide what to share and consume content based on visual cues; thus, as state-sponsored accounts tend to post disinformation (Mejias and Vokuev 2017), studying the images they share provides an important tool to understand and counter disinformation. --- Methodology We now present our dataset and our methodology for analyzing images posted by state-sponsored trolls on Twitter. Dataset. We use a ground truth dataset of tweets posted by Russian trolls released by Twitter in October 2018 (Gadde and Roth 2018). The dataset includes over 9M tweets posted by 3.6K Russian state-sponsored accounts, and their associated metadata and media (1.8M images). Note that the methodology employed by Twitter for detecting/labeling these state-sponsored accounts is not publicly available. That said, to the best of our knowledge, this is the most up-to-date and the largest ground truth dataset of statesponsored accounts and their activities on Twitter. Ethics. We only work with publicly available data, which was anonymized by Twitter, and follow standard ethical guidelines (Rivers and Lewis 2014)-e.g., we do not try to de-anonymize users based on their tweets. Image analysis pipeline. To analyze the images posted by these state-sponsored accounts, we build on the image processing pipeline presented by (Zannettou et al. 2018). This relies on Perceptual Hashing, or pHash (Monga and Evans 2006), and clustering techniques (Ester et al. 1996) to group similar images according to their visual peculiarities, yielding clusters of visually similar images. Then, clusters are annotated based on the similarity between a ground truth dataset and each cluster's medoid (i.e., the representative image in the cluster). For this process, (Zannettou et al. 2018) --- Russian trolls Web entity detection using medoid image for each cluster Annotated clusters and images useful insights from open-domain images (i.e., not bounded to a specific domain like Internet memes). Figure 2 shows the extended pipeline. We perform the "Web Detection" task using Cloud Vision API, which provides us with two very useful pieces of information for each image: 1) a set of entities, and their associated confidence scores, that best describe the image (e.g., an image showing Donald Trump yields an entity called "Donald Trump"); and 2) a set of URLs on the Web that the same image appeared. To extract this information, the API leverages Google's image search functionality to find URLs to identical and similar images. Furthermore, by extracting data from the text of these URLs, the API provides a set of entities that are related to the image. These two pieces of information are crucial for our analysis as they allow us to understand the context of the images and their appearance across the Web. Running the pipeline. First, we extract a pHash for each image using the ImageHash library.2 This reveals that there is a substantial percentage of images that are either visually identical or extremely similar as they have the same pHashes (43% of the images). Next, we cluster the images by calculating all the pairwise comparisons of all the pHashes. This results in 78,624 clusters containing 753,634 images. Then, for each cluster, we extract the medoid, which is the image that has the minimum average Hamming distance between all the images in the cluster. Then, using each medoid, we perform "Web Detection" using the Cloud Vision API, which provides us with a set of entities and URLs, which we assign for each image in the cluster. This is doable since the average number of unique images per cluster is 1.8 with a median of 1 unique image per cluster (see Figure 3(a)). Pipeline Evaluation. To evaluate the performance of our pipeline, we manually annotate a random sample of 500 clusters. Specifically, the first author of this paper manually checked the 500 random clusters and the corresponding Cloud Vision entity with the highest confidence score to assess whether the entity is "appropriate" with respect to the images in the cluster. We find that the Cloud Vision APIbased annotation provides an appropriate entity in 83.7% of the clusters in the random sample. Thus, we argue this is a reasonable performance for the purposes of our study. --- Image Analysis We now present the results of our analysis. First, we perform a general characterization of the images posted by statesponsored accounts on Twitter and then an analysis of the content of the images. Also, we study the occurrence of the images across the Web. --- General Characterization We begin by looking at the prevalence of images in tweets by state-sponsored trolls. In Figure 3(b), we plot the CDF of the number of images posted per confirmed state-sponsored account that had at least one tweet (4.5% of the identified trolls never tweeted). We find that only a small percentage of these accounts do not share images (9.7% of the Russian troll accounts). Also, some accounts shared an extremely large number of images, 8% of the Russian trolls posted over 1K images. Furthermore, we find an average of 502.2 images per account with a median number of images of 37. Then, in Figure 4, we report the CDF of the number of images per tweet; we find that 19% of tweets posted by Russian trolls include at least one image. One explanation for this relatively large fraction is that Twitter automatically generates a preview/thumbnail image when you post a URL. Indeed, by inspecting the URLs in the tweets, we find that out of the 19% of the tweets that contained images, 11.8% of them Figure 5: Temporal overview of: a) all tweets and tweets with images as a percentage of all tweets; and b) all tweets with images as a percentage of all tweets that contained at least one image. contained automatically generated one, while the rest (7.2%) include images that are explicitly posted (i.e., not generated based on a posted URL). That said, we include all images in our dataset and analysis, as generated images too provide insight into the content posted by the state-sponsored accounts, especially considering their proclivity to post "fake news" (Mejias and Vokuev 2017) and the role images might play in catching people's attention. Temporal Analysis. Next, we look into how the tweets from Russian trolls are shared over time with a particular focus on the tweets that contain images. Figure 5(a) reports the percentage of tweets shared each week normalized by the number of all tweets, while Figure 5(b) the percentages normalized by the number of tweets that contained at least one image. The former shows that, in the early stages of their operations (before 2016), Russian trolls were posting tweets mostly without images, whereas, after 2016 it seems that they started posting more tweets containing images. This indicates that they started using more images in their tweets after 2016, likely because they started targeting specific foreign countries (e.g., the US (Mueller 2019)), suggesting the Russian trolls might believe the use of images can be better for pushing specific narratives. Figure 5(b) reveals an overall increase in the use of images after October 2016 with a peak of activity in use of images during the week leading to the Charlottesville rally in August 2017 (Spencer 2017), which led to the death of one counter protester (Caron 2017) and was a significant turning point in the use of online hate speech and anti-Semitism in fringe Web communities (Zannettou et al. 2020). This peak likely indicates that the use of images is an effective tactic used by Russian trolls to sow discord on social networks with respect to events related to politics, the alt-right, and white supremacists. --- Entity Analysis We now explore the content of images with a special focus on the entities they contain, which allows us to better understand what "messages" images were used to convey. To do so, we use the image processing pipeline presented in (Zannettou et al. 2018) to create clusters of visually similar images but leverage Google's Cloud Vision API to annotate each cluster (as discussed in the Methodology section). Then, for each image, we assign the entity with the highest confidence score as returned by the Cloud Vision API. We also associate the tweet metadata to each image (i.e., which image appears in which tweet). The final annotated dataset allows us to study the popularity of entities in images posted by state-sponsored accounts on Twitter. Popular Entities. We first look at the popularity of entities for the trolls: Table 1 reports the top 20 entities that appear in our image dataset both in terms of the number of clusters, as well as the number of images within the clusters. We observe that the two most popular entities for Russian trolls are referring to Russia itself (i.e., "Russia" and "Vladimir Putin" entities). Also, trolls are mainly focused on events related to Russia, Ukraine, USA, and Syria (their top entities correspond to these countries). Moreover, several images include screenshots of news articles (see entity "Web page") as well as logos of news sites (see entity "Logo"), hence indicating that these accounts were sharing news articles via images. This is because the state-sponsored accounts shared URLs of news articles, which do not include images, hence Twitter automatically adds the logo of the news site to the tweet. Finally, we find a non-negligible percentage of images and clusters that show memes, highlighting that memes are exploited by such accounts to disseminate their ideology and probably weaponized information via memes. Graph Visualization. To get a better picture of the spectrum of entities and the interplay between them, we also build a graph, reported in Figure 6, where nodes correspond to clusters of images and each edge to the similarity of the entities between the clusters. For each cluster, we use the set of entities from the Google Cloud Vision API and calculate the Jaccard similarity between each cluster. Jaccard similarity is useful here, because it exposes meta relationships between clusters. While images that appears within the same cluster are visually similar, there are likely to be other clusters that represent the same subjects, but from a different visual perspective. Then, we create an edge between clusters (weighted by their Jaccard similarity) with similarities below a pre-defined threshold. We set this threshold to 0.4, i.e., we discard all edges between clusters that have a Jaccard similarity less than 0.4, because we want to 1) capture the main connections between the clusters and 2) increase the readability of the graph. We then perform the following operations: 1) we run a community detection algorithm using the Louvain method (Blondel et al. 2008) and paint each community with a different color; 2) we lay out the graph with the Force Atlas2 layout (Jacomy et al. 2014), which takes into account weights of edges (i.e., clusters with higher similarity will be positioned closer in the graph); 3) for readability purposes, we show the top 30% of nodes according to their degree in the graph; and 4) we manually annotate the graph with representative images for each community, allowing us to understand the content within each community. In a nutshell, this graph allows us to understand the main communities of entities pushed by the state-sponsored accounts and how they are connected. Main Communities. From Figure 6, we observe a large community (sapphire) corresponding to clusters related to Vladimir Putin and Russia. This community is tightly connected with communities related to Donald Trump/Hillary Clinton/USA (green), Ukraine/Petro Poroshenko (light blue), and Sergey Lavrov (gray). Also, we observe that other big communities include logos from news outlets (pink) that are connected with communities including screenshots of articles (brown), images of documents (light green), and various other screenshots (emerald). Other communities worth noting are those including comics and illustrations (yellow) as well as images of products and advertisements (orange). Overall, these findings highlight that state-sponsored troll accounts shared many images with a wide variety of themes, ranging from memes to news via screenshots. --- Images Occurrence across the Web Our next set of measurements analyze the co-occurrence of the images posted by Russian state-sponsored accounts across the greater Web. Recall that the Cloud Vision API also provides details about the appearance of an image across the Web. This is useful when studying the behavior of state-sponsored accounts, as it either denotes that they posted the images on other domains too, or they obtained the image from a different domain, or that other users on the Web posted them on other domains too. Thus, studying the domains that shared the same images as state-sponsored Table 2: Top 20 domains that shared the same images as the trolls. We report the top domains both in terms of number of clusters and number of images within the clusters. #"! '! "! (!!!'" % "!"! "!'#"'#!! #& % "'% $ $'#!!!&! #"!&!"! "!!& #!" "!!& "!& #"!!"!!! % % % % % & & & #!!'!! accounts allows us to understand their behavior and potential impact on the greater Web. For instance, this information can be used to detect domains that are exclusively controlled by state-sponsored actors to spread disinformation. In Table 2, we report the top domains, both in terms of number of clusters and number images within the clusters, that shared the same images as the state-sponsored accounts. Unsurprisingly, the most popular domains are actually mainstream social networking sites (e.g., Pinterest, Twitter, YouTube, and Facebook). Also, among the popular domains we find popular Russian news outlets like ria.ru and riafan.ru, as well as Russian-owned social networking sites like livejournal.com and pikabu.ru. This highlights the efforts by Russian trolls to sway public opinion about public matters related to Russia. We further find both mainstream and alternative news outlets like theguardian.com and sputniknews.com, respectively (we use the list provided by (Zannettou et al. 2017) to distinguish mainstream and alternative news outlets). This provides evidence that the efforts of Russian trolls had an impact on, or were inspired by, content shared on a wide variety of important sites in the information ecosystem on the Web. Next, we aim to provide a holistic view of the domains while considering the interplay between the entities of the images and the domains that they also shared them. To do this, we create a graph where nodes are either entities or domains that were returned from the Cloud Vision API. An edge exists between a domain node and an entity node if an image appearing on the domain contained the given entity. Then, we perform the operations (1) and (2) as described in the entities analysis section (i.e., community detection and layout algorithm). We do this for the images posted by the trolls and present the resulting graph in Fig- ure 7. This graph allows us to understand which domains shared images pertaining to various semantic entities. We find popular Web communities like Twitter, Pinterest, Facebook and YouTube in the middle of the graph, constituting a separate community (light blue), i.e., they are used for sharing images across all entities. Entities mainly related to Russia are shared via Russian state-sponsored outlets like sputniknews.com (see orange community). Entities that are related to the USA and political persons like Donald Trump, Barack Obama, and Hillary Clinton are part of a separate community (pink) with popular news outlets like washingtonpost.com and nytimes.com. Finally, for matters Figure 7: Visualization of the interplay between entities and domains that shared matching images as the ones shared by the trolls. related to Ukraine (green community) most of the images co-appeared on popular Russian-owned social networks like livejournal.com and pikabu.ru. Overall, our findings indicate that the same images often appear on both their feeds and specific domains. Thus, statesponsored trolls might be trying to make their accounts look more credible and push their agenda by targeting unwitting users on popular Web communities like Twitter. --- Cross-Platform Influence Our analysis above studies the occurrence of images shared by Russian state-sponsored accounts on other domains, but does not encapsulate the interplay between multiple communities. In reality, the Web consists of a large number of communities that are not exclusively independent of each other, as communities naturally influence each other. For instance, a Twitter user might share an image on Twitter that she previously saw on Reddit: in this case, we see that the Reddit community has "influenced" the Twitter community with respect to the sharing of that particular image. In this section, we model and measure the interplay and the influence across Web communities with respect to the dissemination of images that were also shared by Russian state-sponsored accounts on Twitter. In other words, we aim to understand how influential the trolls were in spreading images to other communities. To do so, we rely on Hawkes Processes (Linderman and Adams 2014;2015), which allow us to estimate the probabilities that an appearance of an image on one community is due to that image previously occurring on the same or on another Web community. --- Hawkes Processes Hawkes Processes are self-exciting temporal point processes (Hawkes 1971) that describe how events occur on a set of processes. In our setting, events are the posting of an image, and processes are Web communities. Generally, a Hawkes model consists of K processes; each process has a rate of events that dictates the frequency of the creation of events in the specific process. The occurrence of an event on one process, causes impulses to the rest of the processes, temporarily increasing the rate of events in the other processes. The impulses comprise two useful pieces of information: the intensity of the increase in the rate, and how it is distributed and decays over time. By fitting a Hawkes model using Bayesian inference to data that describes the appearances of events in the processes, we obtain the parameter values for the impulses. This lets us quantify the overall rate of events in each process, as well as how much previous events contribute to the rate, at a given point in time. Naturally, we cannot possibly know what exactly triggered the creation of an event on a process, however, we can use Hawkes Processes to calculate the probability that the cause of an event is another process in the model, as also done by previous work (Zannettou et al. 2017;2018). Note that the background rate of the Hawkes Processes allow us to capture and model the interplay of external sources (i.e., platforms that we do not use in our analysis), hence the resulting probabilities encapsulate the influence of the greater Web via the background rates. --- Datasets Cross-Platform Dataset. We use a publicly available dataset consisting of 160M pHashes and image URLs for all the images posted on Twitter (using 1% Streaming API), Reddit, 4chan's /pol/, and Gab, between July 2016 and July 2017. 3 Then, we select the images that have the same pHashes with the ones shared by Russian state-sponsored accounts on Twitter. For each one of these images, we find all their occurrences on Reddit, /pol/, Gab, and Twitter. Next, we omit images that appear less than five times across all Web communities we study, ultimately obtaining a set of 90K pHashes. Finally, we annotate each pHash using the Web entities obtained from the Cloud Vision API. Since our dataset focuses primarily on the year before and after the 2016 US elections, we concentrate our analysis around this major event. Specifically, we want to assess whether Russian state-sponsored accounts were more effective in pushing images related to the Democratic Party or Republican Party. To do so, we select all the pHashes that have a Cloud Vision Web entity corresponding to "Democratic Party," "Hillary Clinton," and "Barack Obama" for the Democratic Party, and "Republican Party" and "Donald Trump" for the Republican Party. Using these entities, we find 9.9K images related to the Republican Party and 6K images related to the Democratic Party. Examples. To provide an intuition on what some of the politically charged images look like, we provide some ex- amples in Figure 8 and Figure 9 for the Democratic and the Republican party, respectively. These illustrate how Russian trolls use images to spread disinformation: for instance, Figure 8(b) shows Senator Robert Byrd meeting with Hillary Clinton and, in another occasion, wearing a Ku Klux Klan robe. The image with the robe is known to be fake as reported later by Snopes (Snopes 2016). We can also observe how state-sponsored accounts rely on edited/photoshopped images to make specific personalities look bad: e.g., Figure 8(a) is an edited image aimed at reinforcing the idea that Hillary Clinton was involved in the Pizzagate conspiracy theory (her face was edited and a baby was added in the picture). Finally, we find several memes that are meant to be funny; however they have a strong political nature and can effectively disseminate ideology. For instance, Figure 8 Events. Table 3 summarizes the number of events for our dataset. Note that we elect to decouple The Donald subreddit from the rest of Reddit mainly because of its strong political nature and support towards Donald Trump (Flores-Saviaga, Keegan, and Savage 2018). By looking at the raw numbers of events per category, we note that in general Russian state-sponsored accounts shared more content related to the Republican Party when compared to the Democratic Party. The same applies for all the other communities we study: in general we find 1.59 times more events for the Republican Party than the Democratic Party (385K vs 242K events). This indicates that content related to the Republican Party was more popular in all Web communities during this time period and that Russian state-sponsored accounts pushed more content related to the Republican Party, likely in favor of Donald Trump as previous research show ( Zannettou et al. 2019b). --- Results We create a Hawkes model for each pHash. Each model consists of six processes, one for each of Reddit, The Donald subreddit, Gab, Russian state-sponsored accounts on Twitter, and other Twitter users. Then, we fit a Hawkes model using Gibbs sampling as described in (Linderman and Adams 2014) for each of the 90K pHashes. Metrics. After fitting the models and obtaining all the parameters for the models, following the methodology presented in (Zannettou et al. 2018), we calculate the influence and efficiency that each community had to each other. The former denotes the percentage of events (i.e., image appearances) on a specific community that appear because of previous events on another community, while the latter is a normalized influence metric that denotes how efficient a community is in spreading images to the other communities irrespectively to the number of events that are created within the community. In other words, efficiency describes how influential the posting of a single event to a particular community is, with respect to how it spreads to the other communities. Overall Influence & Efficiency. Figure 10 reports the influence estimation results for all the events (i.e., all the images that were shared by Russian state-sponsored accounts and have at least five occurrences across all the Web communities we study). When looking at the raw influence results (Figure 10(a)), we observe that Russian state-sponsored accounts had the most influence towards Gab (2.3%), followed by The Donald subreddit (1.8%), and the rest of Reddit (1.6%), while they had the least influence to 4chan's /pol/ (0.2%). By comparing the influence of regular Twitter users, Figure 10: Influence estimation for all images shared by Russian state-sponsored accounts on Twitter: a) Raw influence between source and destination Web communities; and b) Normalized influence (efficiency) of each Web community as the results are normalized by the numbers of events created on the source community. The numbers in the cell can be interpreted as the expected percentage of events created on the destination community because of previously occurring events on the source community. with respect to the dissemination of memes, to the influence of the state-sponsored actors (see Figure 11 in extended version of (Zannettou et al. 2018)4 ), we observe similar patterns. That is, regular Twitter users were more influential on Gab (8%), followed by The Donald (3.6%), and the rest of Reddit (2.8%), while they had the least influence on /pol/ (0.7%). This comparison indicates that Russian trolls influenced other platforms similarly to regular Twitter users with the difference that the intensity of their influence is substantially lower (between 3.5x-1.5x times lower), mainly due to the fact that Russian trolls consist of a few thousands accounts. Furthermore, when comparing the results for Twitter against previous characterizations of Russian trolls on news URLs (see Figure 14 (a) in (Zannettou et al. 2019b)), we find that actually Russian trolls were more influential in spreading news URLs compared to images (1.29% for news URLs and 0.8% for images). When looking at the efficiency of Russian state-sponsored accounts (last row in Figure 10(b)), we find that they were most efficient in pushing the images on Twitter (6.5%) likely because it is the same social network. Also, they were particularly efficient in pushing images towards the rest of Reddit (2.9%), while again we find that they were not very effective towards 4chan's /pol/ (0.4%). Furthermore, we report the overall external efficiency of each community towards all the other communities (right-most column in Figure 10(b)). We find that the most efficient platform in the ones that we study is The Donald subreddit (68.4%), followed by the rest of Reddit (19

State-sponsored organizations are increasingly linked to efforts aimed to exploit social media for information warfare and manipulating public opinion. Typically, their activities rely on a number of social network accounts they control, aka trolls, that post and interact with other users disguised as "regular" users. These accounts often use images and memes, along with textual content, in order to increase the engagement and the credibility of their posts. In this paper, we present the first study of images shared by state-sponsored accounts by analyzing a ground truth dataset of 1.8M images posted to Twitter by accounts controlled by the Russian Internet Research Agency. First, we analyze the content of the images as well as their posting activity. Then, using Hawkes Processes, we quantify their influence on popular Web communities like Twitter, Reddit, 4chan's Politically Incorrect board (/pol/), and Gab, with respect to the dissemination of images. We find that the extensive image posting activity of Russian trolls coincides with real-world events (e.g., the Unite the Right rally in Charlottesville), and shed light on their targets as well as the content disseminated via images. Finally, we show that the trolls were more effective in disseminating politics-related imagery than other images.

1.5x times lower), mainly due to the fact that Russian trolls consist of a few thousands accounts. Furthermore, when comparing the results for Twitter against previous characterizations of Russian trolls on news URLs (see Figure 14 (a) in (Zannettou et al. 2019b)), we find that actually Russian trolls were more influential in spreading news URLs compared to images (1.29% for news URLs and 0.8% for images). When looking at the efficiency of Russian state-sponsored accounts (last row in Figure 10(b)), we find that they were most efficient in pushing the images on Twitter (6.5%) likely because it is the same social network. Also, they were particularly efficient in pushing images towards the rest of Reddit (2.9%), while again we find that they were not very effective towards 4chan's /pol/ (0.4%). Furthermore, we report the overall external efficiency of each community towards all the other communities (right-most column in Figure 10(b)). We find that the most efficient platform in the ones that we study is The Donald subreddit (68.4%), followed by the rest of Reddit (19.5%) and the Russian statesponsored accounts on Twitter (11.3%). Again, by looking at previous results based on news (see Figure 15 (a) in (Zannettou et al. 2019b)), we observe that Russian trolls were more efficient in spreading news URLs compared to images (16.95% external influence for news, while for images we find 11.3%). Politics-related Images. Next, we investigate how our influence estimation results change when considering only the politics-related images, and in particular the differences between the images pertaining to the Republican and Democratic Parties. Figure 11 reports our influence and efficiency estimation results for the images related to the Republican Party (R) and Democratic Party (D). NB: To assess the statistical significance of these results, we perform a two-sample Kolmogorov-Smirnov test to the influence distributions of the two samples and annotate the figures with an * for cases where p <unk> 0.01. We make the following observations. First, Russian state-sponsored accounts were most influential in pushing both Democratic and Republican Party-related images to Gab, The Donald subreddit, and the rest of the Reddit, while again were the least influential in spreading these images in 4chan's /pol/ (see last row in Figure 11(a)). Second, when comparing the results for both parties, we observe that on Twitter they have more or less the same influence for both Republicans and Democratic parties (1.3% vs 1.2%), on Gab they were more influential in spreading Democratic Party images when compared to Republican party (4.0% vs 3.1%). For The Donald and the rest of Reddit we observe the opposite: they were more influential in spreading Republican Party related images when compared to the Democratic Party (see last row in Figure 11(a)). Third, by looking at the efficiency results (Figure 11(b)), we find that again that Russian state-sponsored accounts were most efficient in spreading political images to big mainstream communities like Twitter and Reddit (see last row in Figure 11(b)). Fourth, by looking at the overall external influence of the communities (right-most column in Figure 11(b)), we observe that again The Donald subreddit had the bigger efficiency (over 60% for both parties), followed by the Russian state-sponsored accounts on Twitter and the rest of Reddit. Finally, by comparing the efficiency of state-sponsored trolls on all images vs the political-related images (cf. sian state-sponsored trolls were over 2 times more efficient in spreading political-related imagery when compared to all the images in our dataset (11.3% vs 23.6% and 25.7%). Most Influential Images. Since our influence estimation experiments are done with the granularity of specific pHashes, we can also assess which images the Russian statesponsored accounts were more influential in spreading. To do so, we sort the influence results for the Democratic and Republican Parties according to the external influence that Russian state-sponsored accounts had to all the other Web communities, and report the top three images with the most influence. Figure 12 and Figure 13 show the three most influential images shared by Russian state-sponsored accounts for the Democratic and Republicans party, respectively. Evidently, Russian state-sponsored accounts were particularly influential in spreading images "against" the Democratic Party: for instance, Figure 12(a) is an image that trolls Nancy Pelosi, currently serving as speaker of the US House of Representatives, while Figure 12(b) shares a political message against Hillary Clinton's chances during the 2016 US elections. On the other hand, the most influential images related to the Republican Party (Figure 13) are neutral and likely aim to disseminate pro-Trump messages and imagery. --- Discussion & Conclusion This paper presented a large-scale quantitative analysis of 1.8M images shared by Russian state-sponsored accounts ("Russian trolls") on Twitter. Our work is motivated, among other things, by the fact that social network users tend to put little effort into verifying information and they are often driven by visual cues, e.g., images, for re-sharing content (Gabielkov et al. 2016). Therefore, as state-sponsored accounts tend to post disinformation (Mejias and Vokuev 2017), analyzing the images they share represents a crucial step toward understanding and countering the spread of false information on the Web, and its impact on society. By extending the image processing pipeline presented in (Zannettou et al. 2018), we clustered the images and annotated them using Google's Cloud Vision API. Our analysis shed light on the content and targets of these images, finding that Russian trolls had multiple targets: mainly the USA, Ukraine, and Russia. Furthermore, we found an overall increase in image use after 2016 with a peak in activity during divisive real-world events like the Charlottesville rally. Finally, by leveraging Hawkes Processes, we quantified the influence that Russian state-sponsored accounts had with respect to the dissemination of images on the Web, finding that these accounts were particularly influential in spreading politics-related images. Also, by comparing our results to previous analysis made on news URLs, we find that these actors were more influential and efficient in spreading news via URLs when compared to images. Our findings demonstrate that state-sponsored accounts pursued a political agenda, aimed at influencing users on Web communities w.r.t. specific world events and individuals (e.g., politicians). Some of our findings confirm previous analysis performed on the text of the tweets posted by these accounts (Zannettou et al. 2019b), highlighting how statesponsored actors post images that are conceptually similar to their text, possibly in an attempt to make their content look more credible. Our influence estimation also demonstrated that Russian state-sponsored accounts were particularly influential and efficient in spreading political images to a handful of Web communities. Also considering the relatively small number of Russian state-sponsored accounts that were actually identified by Twitter, our analysis suggests that these actors need to be taken very seriously in order to tackle online manipulation and spread of disinformation. Naturally, our study is not without limitations. First, our pipeline relies on a closed-system (i.e., Cloud Vision API) with a relatively unknown methodology for extracting entities. However, our small-scale manual evaluation showed that the API provides an acceptable performance for our needs. Second, we study the images in isolation, without considering other features of the tweets like shared text, thus we may lose important knowledge that exists in the text like sentiment, entities that are referenced, toxicity, etc. Finally, our study relies on a dataset that is independently identified and released by Twitter, and the methodology for identifying these accounts is unknown and it is unclear on whether there are false positives within the dataset. Implications of our work. Overall, our study has several implications related to the exploitation of social media by Russian state-sponsored actors, who share weaponized information on divisive matters with the ultimate goal of sowing discord and influencing online discussions. As such, their activities should be considered as having broader impact than "just" political campaigns, rather, as direct attacks against individuals and communities, since they can lead to erratic real-world behavior outside the scope of any particular election-e.g., disease epidemics as parents are not vaccinating kids because of disinformation (Broniatowski et al. 2018;Mejias and Vokuev 2017). We also argue that the public should be adequately informed about the existence and the strategies of these actors, particularly their use of weaponized information beyond just "fake news," as a necessary step toward educating users in how to process and digest information on the Web. Our analysis also complements, to some extent, the Mueller Report (Mueller 2019). Although it represents the first comprehensive investigation of large-scale statesponsored "information warfare," much of the Report currently remains redacted. Even if it is eventually released in its entirety, it is unlikely to contain a quantitative understanding of how these state-sponsored actors behaved and what kind of influence they had. Furthermore, state-sponsored attacks are reportedly still on going (Barnes and Goldman 2019). While still awaiting scientific study, new campaigns, including for instance the Qanon conspiracy theory, have been launched by Russian trolls, and at least partially supported by the use of images initially appearing on imageboards like 4chan and 8chan (Collins and Murphy 2019). Overall, our work can be beneficial to policy makers, law enforcement, and military personnel, as well as political and social scientists, historians, and psychologists who will be studying the events surrounding the 2016 US Presidential Elections for years to come. Our scientific study of how state sponsored actors used images in their attacks can serve to inform this type of interdisciplinary work by providing, at minimum, a data-backed dissection of the most notable and effective information warfare campaign to date. Finally, the research community can re-use the tools and techniques presented in this paper to study image sharing by various teams or communities on the Web, e.g., state-sponsored accounts from other countries, bots, or any coordinated campaign. In fact, Twitter recently released new datasets for state-sponsored trolls that originate from Venezuela and Bangladesh (Roth 2019); our techniques can be immediately be applied on this data. Future Work. As part of future work, we plan to study the use of news articles and social network posts from statesponsored accounts with a particular focus on detecting possibly doctored images. Finally, we aim to build on top of our work to detect domains that are controlled by statesponsored actors and aim to push specific (disinformation) narratives on the Web. --- CASE project (GA No. 691025). Also, this work was partially supported by a Content Policy Research on Social Media Platforms award from Facebook Research.

State-sponsored organizations are increasingly linked to efforts aimed to exploit social media for information warfare and manipulating public opinion. Typically, their activities rely on a number of social network accounts they control, aka trolls, that post and interact with other users disguised as "regular" users. These accounts often use images and memes, along with textual content, in order to increase the engagement and the credibility of their posts. In this paper, we present the first study of images shared by state-sponsored accounts by analyzing a ground truth dataset of 1.8M images posted to Twitter by accounts controlled by the Russian Internet Research Agency. First, we analyze the content of the images as well as their posting activity. Then, using Hawkes Processes, we quantify their influence on popular Web communities like Twitter, Reddit, 4chan's Politically Incorrect board (/pol/), and Gab, with respect to the dissemination of images. We find that the extensive image posting activity of Russian trolls coincides with real-world events (e.g., the Unite the Right rally in Charlottesville), and shed light on their targets as well as the content disseminated via images. Finally, we show that the trolls were more effective in disseminating politics-related imagery than other images.

While the presence of social ties across the life course is fundamental, their benefits are contingent upon relationship quality: high-quality, supportive relationships foster health and well-being, whereas poor-quality or strained relationships can be more harmful long-term than relationship loss (Hawkins & Booth, 2005). Throughout the life course, individuals actively manage their social relationships to maintain health and well-being, expanding and contracting their social networks over time (Carstensen, Isaacowitz, & Charles, 1999). Older adults' efforts at socioemotional selectivity face challenges, however. Some reductions in network size are not freely chosen; loss of relationships to death, incapacity, or relocation are common (Rook, 2009). Moreover, although older adults strive to preserve emotionally meaningful relationships, not all meaningful ties are positive ones (Isaacowitz, Smith, & Carstensen, 2003). Both social isolation and negative relationships contribute to increased depressive symptoms among older adults, and depression is strongly associated with morbidity, mortality, and healthcare costs, in addition to being undesirable itself (e.g., Bock et al., 2017). The model of selection, optimization, and compensation (SOC) asserts that older adults who lose or otherwise The Gerontologist cite as : Gerontologist, 2018, Vol. 58, No. 6, 1109-1113doi:10.1093/geront/gnx151 Advance Access publication September 18, 2017 lack a close, high-quality relationship-particularly a marriage-will attempt to compensate by increasing the closeness or centrality of other social relationships, including with family and close friends (Carstensen et al., 1999). This perspective anticipates that perceived support from family and friends will play a larger role in the depressive symptoms of older adults who are widowed, never married, divorced/separated, or in lower-quality marriages than for older adults in high-quality marriages. [Perceived strain in family and friend relationships may be less likely to vary in importance across marital statuses, insofar as SOC is an intentional process by which people optimize the positive and discount the negative (Mather & Carstensen, 2005).] Marriages pose barriers to successful compensation, however. For instance, marriage can be "greedy," with friend and family ties weakening when adults select into marriage (Sarkisian & Gerstel, 2016). Further, it may be difficult for widowed or divorced older adults to rekindle relationships they had let lapse, particularly as social partners trim their own networks (Rook, 2009). Older adults may also lack the social connections necessary to initiate new relationships (Rook, 2009). Lastly, not all social ties are interchangeable; thus, the ability to compensate for losing or lacking a high-quality marriage by focusing on support in other network ties may be limited (Zettel & Rook, 2004). In short, later life attempts at compensation face numerous obstacles, while longer-term compensation behaviors may achieve greater success. Therefore, we expect family and friend support to most strongly affect the depressive symptoms of never married older adults, since their reliance on these relationships is part of a continuous, lifelong process of socioemotional selectivity rather than a late life compensation process wherein they make up for a lost or unsatisfactory relationship (Shapiro & Keyes, 2008). This study assesses compensation, by examining how relationships with family and friends influence depressive symptoms for older adults across multiple marital statuses. --- Method --- Participants The National Social Life, Health, and Aging Project began in 2005-2006 with a nationally representative sample of 3,005 community dwelling U.S. adults born between 1920 and 1947. The response rate was 75.5%. In 2010-2011, original participants, Wave 1 non-respondents, and a randomly-selected sample of spouses and coresident romantic partners were invited to participate in a second wave of self-administered questionnaires and in-person interviews. Overall, 87.8% of original participants, 23.1% of Wave 1 non-respondents, and 85.8% of spouses and partners participated in Wave 2. The completed second wave included interviews with 3,377 participants. Due to the unavailability of marital quality for 6 individuals, our analytic sample included 3,371 participants. --- Dependent Measure Depressive symptoms were measured using the 11-item Center for Epidemiological Studies-Depression (CES-D) scale (Payne, Hedberg, Kozloski, Dale, & McClintock, 2014). Responses ranged from 1 (rarely or none of the time) to 4 (most of the time). Items were averaged (<unk> = 0.79). Due to significant skew we took the natural log of scores, then standardized the measure for ease of interpretation. --- Type and Quality of Ties Marital status was self-reported as married/partnered, widowed, never married, or divorced/separated. Married/ partnered participants were asked about the extent to which they could (a) open up to and (b) rely upon their partner, and the extent to which their partner (c) criticized or (d) made demands on them (1 = never/hardly ever/rarely, 3 = often). We categorized the 61% of participants who reported that they could often open up to and rely upon their partner, and that their partner never, hardly ever, or rarely criticized or made demands upon them, as being in high-quality marriages. The remaining participants in lower-quality marriages were those who rated their marriage as less than the most desirable score on one or more of the four measures. Thus these "lower-quality" marriages may represent "less than perfect" rather than "poor-quality" relationships per se. Perceived family support was measured using two items concerning how often participants could open up to and rely on their family members, excluding their spouse/partner (1 = never/hardly ever/rarely, 3 = often). The items correlated at 0.44 (p <unk>.001) and were averaged. Perceived family strain was measured using three items concerning how often participants' family members, excluding their spouse/partner, got on their nerves, made too many demands, and criticized them (1 = never, 3 = some of the time/often). The items were averaged (<unk> = 0.57). Perceived friend support and perceived friend strain were measured using the same questions as family support and strain, asked about participants' friends. For friend support, items correlated at 0.56 (p <unk>.001) and were averaged. For friend strain, <unk> = 0.55 and items were averaged. --- Control Measures Network size measures included number of living children, including stepchildren; number of friends; and number of close relatives, excluding spouse/partner (friends and relatives: none, one, 2-3, 4-9, 10-20, and more than 20). Demographic measures included age (years); gender (male, female); race/ethnicity (White non-Hispanic, Black non-Hispanic, Hispanic, Other race/ethnicity); education (less than high school, high school only, some college or associate's degree, bachelor's degree or more); and income (split at $50,000 annually). --- Analytic Strategy The majority of cases-95%-were complete. Missing data diagnostics revealed no predictable patterns of missingness. We used multiple imputation by chained equations to address missing data, including all predictors and the outcome in our imputation equations. Analysis of variance tests (ANOVAs) examined mean differences in the number and quality of non-marital network ties by marital status and quality. Three ordinary least squares (OLS) regression models analyzed depressive symptoms. All three models adjusted for non-independence due to having both partners of some married couples in the sample. The first model tested the main effects of marital status and quality, adjusted for all demographic controls. The second model added the measures of non-marital network size and perceived support and strain. In a model not shown here, we estimated interaction effects between each of the marital status and quality categories and the measures of family and friend support. One interaction was statistically significant, and the third model shown adds that interaction term to the second model. We also tested interaction effects between marital status and quality and the measures of family and friend strain (not shown). As anticipated, none of these interactions was statistically significant. --- Results Descriptive statistics and ANOVA results for network size and quality by marital status are reported in Table 1. Table 2 displays the results of our regression models. Model 1 shows that older adults who were in lower-quality marriages (B = 0.29, p <unk>.001), widowed (B = 0.29, p <unk>.001), never married (B = 0.30, p <unk>.05), or divorced/separated (B = 0.22, p <unk>.01) all experienced significantly greater depressive symptoms than those in high-quality marriages. After accounting for non-marital network size and quality (Model 2) marital status differences in depressive symptoms were slightly attenuated, but only the coefficient for never married became non-significant. Perceived family support (B = -0.12, p <unk>.001) was significantly associated with fewer depressive symptoms, although friend support was not. Additionally, both perceived family strain (B = 0.23, p <unk>.001) and perceived friend strain (B = 0.13, p <unk>.01) were significantly associated with greater depressive symptoms. Model 3 incorporates the one significant interaction found, for perceived family support within the never married group (B = -0.46, p <unk>.01). The negative coefficient indicates that the link between perceived family support and reduced depressive symptoms was stronger for the never married than for those in high-quality marriages. The main effect of perceived family support was also significant (B = -0.11, p <unk>.001), revealing that perceived family support remains significantly-if less strongly-associated with experiencing fewer depressive symptoms for the other marital status groups, as well. This interaction is illustrated in Figure 1. --- Discussion In keeping with prior research, this study found that individuals in high-quality marriages experienced fewer depressive symptoms than widowed, never married, and divorced/ separated individuals, as well as those in lower-quality marriages (e.g., Mirowsky & Ross, 1992). Additionally, perceived family support, perceived family strain, and perceived friend strain were associated with older adults' depressive symptoms. The only difference in compensation by marital status was for the never married, among whom perceived family support had a significantly stronger effect. Whatever compensation previously married older adults and those in lower-quality marriages engage in does not appear to eliminate gaps in depressive symptoms according to marital status. That is, adults of all marital statuses are affected by the support and strain they receive from friends and family, and adults in high-quality marriages additionally benefit from marriage. This is consistent with research indicating that older adults find it difficult to make new friends and strengthen old ties (Rook, 2009;Zettel & Rook, 2004). Meanwhile, never married older adults have higher levels of depressive symptoms than those in high-quality marriages, and do not report greater perceived support from their friends or family (e.g., Mirowsky & Ross, 1992;Shapiro & Keyes, 2008). However, their depressive symptoms are more responsive to family support. This heightened influence may be due to family relationships being continuously central in the lives of never married adults, uninterrupted by marriage (Sarkisian & Gerstel, 2016). Yet never married adults were not more sensitive than their peers to family strain, because socioemotional selectivity does not simply heighten individuals' vulnerability to relationship cues of all types: People show a marked preference for positive information (Mather & Carstensen, 2005). There were also no differences by marital status in the effects of support from friends. It is possible that people of all marital statuses effectively employ socioemotional selection on their friend networks, cutting ties with poorquality or straining friendships and retaining supportive ones, since friendship is the most voluntary of personal ties (Blieszner & Roberto, 2004). This study retains a number of limitations. First, the heightened importance of family support to never married persons could be due to compositional differences in who comprises "family," as never married older adults had fewer adult children than ever-married persons did. Second, the data lacked direct measures of compensatory or supportive behaviors. Third, correlation and alpha coefficients for the support and strain measures were only moderate. Overall, our results suggest that perceived support from family and friends remains influential for depressive symptoms in later life across all marital status groups, and especially for never married older adults. Those in high-quality marriages, then, experience fewer depressive symptoms not only due to the quality of their marriages, but also thanks to their network of supportive family and friend relationships. --- Funding NSHAP is supported by the National Institute on Aging (grant numbers R01 AG021487, R01 AG033903, R37 AG030481), and by NORC, which was responsible for the data collection. The opinions expressed herein are those of the authors.

In later life, adults' social networks grow smaller through a combination of intentional selection and involuntary loss. This study examined whether older adults who lack a high-quality marriage compensate for this using support from other ties. We analyzed how relationships with family and friends are associated with depressive symptoms across multiple marital statuses. Design and Methods: Data from 3,371 older adults who participated in the most recent wave of the National Social Life, Health, and Aging Project (NSHAP) were analyzed using analysis of variance (ANOVA) and ordinary least squares (OLS) regression. Results: Individuals in high-quality marriages experienced fewer depressive symptoms than the widowed, never married, divorced/separated, and those in lower-quality marriages. Older adults' perceived family support, family strain, and friend strain were all significantly associated with depressive symptoms. The only difference in these effects according to marital status was for perceived family support, which was strongest for the never married. Implications: The never married may depend more on family and friends than the previously or unhappily married. Any compensation efforts among the latter failed to reduce depressive symptoms relative to happily married others. Older adults in high-quality marriages benefit from their marital relationship, and also benefit from supportive family and friend ties.

Introduction According to the last census, there are only 933 women for every 1000 men in India [1]. Gender bias favoring males is largely responsible for this [2]. Neglect of girls and women resulting in early death [3,4,5], female infanticide [6,7] and more recently, antenatal sex determination and female feticide [8], all contribute to it. Several reports suggest that sex selective abortion became more common in the 1990s [9,10] after ultrasound machines became available widely in the 1980s [11,12]. While the Nobel laureate Amratya Sen believes that the pattern of gender inequality shifted from'mortality inequality' to what he calls 'natality inequality' due to female feticide after the facility for antenatal sex determination became available [11,13], others suggest that parents are not substituting pre-natal for post-natal discrimination against girls, but combining the two strategies [14]. The relative contribution of these modes of discrimination, to the unbalanced sex ratio in India, is still unresolved [15]. It is important to resolve this issue so as to target remedies appropriately, before they threaten the stability and security of society [16]. The sex ratio in newborns as registered in the birth register (Registration of Births and Deaths Act 1969), is an indicator of the magnitude of the problem of female feticide as it does not include deaths due to neglect of girl children. However infanticides, in the first few days, are often reported as still-births [15] or not reported at all, within the incomplete birth registration system [7]. Data on sex ratio at birth in hospital records are therefore crucial, to estimate the influence of female feticide on the sex ratio, and which is not affected by other factors like infanticide and neglect of girl children. A study of hospital birth records over 110 years published by us previously has showed that there was an excess of boys born, if the previous child was a girl. If the first child was girl the sex ratio was 716 girls to 1000 boys and if the first child was boy it was 1140 girls to 1000 boys [17]. This finding apparently runs counter to the normal tendency of biological heterogeneity which results in families having a predilection to have children of the same sex. Biggar et al in Denmark found the probability of having another boy increases to 51.5, 51.6, 52.4 and 54.2 percent for families with previous one, two, three, or four prior boys, respectively [18]. The findings in second children in our previous study points to antenatal interventions and suggest that sex selective abortions are practiced quite commonly to overcompensate for the slight penchant in families to have babies of the same sex. However, we also found this excess of boys in families with a previous girl in a cohort of babies born in 1970s when ultrasound machines were uncommon and the only method of sex determination was amniocentesis and which was not available widely [17]. The methods used by parents to overcome the tendency to have second children of the same sex as the first are not fully understood. The sex of second children born to parents with a previous girl (or boy) has not previously been studied prospectively in India, to the best of our knowledge. This study was performed to look at the sex ratio in second children depending on the sex of the first child. By studying this prospectively and interviewing mothers it was hoped to gain insights into the practices for sex selection that are in vogue. A priori, it was known that because antenatal sex determination and sex selective abortions are proscribed by law, parents would not admit to these practices, but it was felt they would be more forthcoming about other 'legal' methods like use of traditional medicines, to promote birth of babies of the desired sex. The study hypothesis was that families with previous girls are more likely to use traditional medicines for sex selection and that mother taking these medicines are more likely to have a boy child. --- Methods This prospective study was done from 19 November 2008 to 18 November 2009 in a large hospital known for its maternal and child care services in Delhi. The hospital is century old charitable hospital in the heart of old Delhi catering to the poor and middle class people of the city. All mothers of live born babies delivering in hospital were eligible for inclusion. Informed consent for participation in the study was obtained from the mothers prior to inclusion. The primary focus of interest was mothers delivering their second and third babies. The sex of the child at birth, the sex of previous children was recorded. Mothers were interviewed after they had recovered from the strain of the delivery process -usually 12 hours after delivery. The lady researcher (SM) built up a rapport with the mothers and enquired from her if they knew of any methods or drugs used to get babies of a particular sex. They were also asked if they had utilized any of these methods. The data were recorded on an Excel spreadsheet. --- Sample Size Calculation The previous data [17] from 2005 in our hospital showed the sex ratio was 629 girls to 1000 boys if the first child was a girl. This gives an odd ratio of 0.49 for second child being female, if the first is female. To detect this odds ratio, in a case control design (case is a girl been born as a second child, control being boy born as a second child if the first is a girl child) with 5% alpha error and 90% power, we needed to study 182 cases and 182 controls. With 80% power this would be 136 cases and 136 controls. We assumed for simplicity, that the prevalence of exposurefirst child being female is 50%. As the remaining 50% first children are presumed to be males, we would need 182+182 cases and 182+182 controls, making a total of 728 babies born as second children, to look at significance in the two sexes separately. About 800 babies are born as second children in our hospital in a year, and therefore we planned to study this over a 1 year period. --- Statistical Methods Sex ratio was analyzed separately for primigravidas, and in multigravidas according to sex of their previous babies. We looked at the sex of the newborn against the methods they admitted to using for having babies of any particular sex, to see if any method influenced the sex of the baby. 95% confidence intervals (CI) for the sex ratios were calculated. Differences in proportions of the sex ratios were estimated. To look for proportions and their CI, and the difference in proportions with confidence intervals, we used the software 'Statistics with Confidence' (www.som.soton.ac.uk). Approval of the study protocol was obtained separately from the Hospital Research Committee and the Hospital Research Ethic Committee. --- Results A total of 3795 mothers delivered in the hospital that year. 48 had multiple pregnancies and were not included in the study. 2773 mother who gave birth to singleton babies, participated in the study. In the remaining, data could not be recorded on the account of early discharge from the hospital, before the researcher could interview the mother. A preliminary analysis was done looking at the sex ratio in the group that were not studied and this was no different from the sex ratio in the children studied (difference in proportion 0.009; 95% CI 20.015 to 0.033), suggesting that the drop-out of 974 babies did not bias the study. Further analysis was done on the sample of 2773 mothers who agreed to participate in the study and who signed the consent form. The results are tabulated in Table 1. The sex ratio in the study sample as a whole was 806 girls to 1000 boys. In primigravida mothers 866 girls were born to every 1000 boys. The sex ratio was 850:1000 in mothers with one previous child. However, there were only 255 girls to 1000 boys among mothers delivering their third child. This was significantly different from the overall sex ratio (difference in proportion 0.243; 95% CI 0.175 to 0.229). When looking at the sex ratio in the second babies, taking into account the sex of the first baby, we found that for every 1000 boys there were only 720 girls if the first was a girl and this rose to 1017 girls if the first was a boy. The difference was statistically significant (difference in proportion 20.086; 95% CI 20.148 to 20.022). There were 184 mothers with two previous children, 106 mother had two previous girls, 21 had two previous boys and 57 had each one girl and one boy. If the two previous children were girls, the sex ratio in the present pregnancy was 178 girls to 1000 boys. Those with two previous boys had a sex ratio of 615 girls for 1000 boys. Among the 1685 primiparous mothers, only 9 (0.5%) said that they had taken traditional medicine to help them get the baby of a desired sex. However among 978 mothers with 1 previous child, 58 (5.9%) of the mothers had taken these medicines and 54 out of 58 were from 510 mothers with a previous girl (10.6%) and 4 were from 486 mothers with a previous boy (0.8%). Among the 54 with a previous girl child, who had taken medication, there were 26 girls and 28 boys making the sex ratio 928 girls to 1000 boys. Among 184 mothers with previous two children, 106 had two previous girls and 42 of them had taken medication (39.6%); 21 had 2 previous boys and of them 1 had taken medication (4.8%); 57 had one girl and one boy previously, and of them 3 had taken medication (5.3%). --- Discussion Some researchers have suggested that the problem of sex selection and the status of women can be expected to be self correcting: as men begin to dramatically outnumber women, women's relative rarity will increase their value; their social status will rise and female offspring will become more desirable [19]. However, studies of societies with high sex ratio and a high proportion of males fail to support this prediction and there is evidence that such societies are disproportionately violent societies [19]. When there is a shortage of women in the marriage market the women can'marry up' inevitably leaving the least desirable men with no marriage prospects [20]. It is a consistent finding across cultures that an overwhelming percentage of violent crime is perpetrated by young unmarried low status males [16]. The need for avoiding this situation is self evident. We found that the overall sex ratio for deliveries at our hospital was 806 girls to 1000 boys. This is even lower than the sex ratio 865:1000 we reported from our hospital in the year 2005 [17]. We found that the sex ratio in the second babies, if the first was a girl, was even lower at 720. The sex ratio was 1017 girls to 1000 boys if the first was a boy. The previous retrospective study showed a similar trend where sex ratio was 716 (CI = 672 to 762) if the first child was girl and 1140 (CI = 1072 to 1212) if the first was a boy. The prospectively collected data in this study validated the finding of previous retrospective study and suggest that parents tend to manipulate sex of their offspring. 184 mothers had had two previous children and of these 106 had two previous girls and 21 had two previous two boys. The remaining 57 had had one boy and one girl. Sex ratio for newborns in families with two previous girls was as low as 178 girls to 1000 boys and this empathically underlines the inference of human interference. There were only 21 mothers with 2 previous boys who went on to have a third child compared to 106 who had 2 previous girls. The fact that there were more mothers with two previous girls than there were mothers with two previous boys suggests a tendency among mothers with girls to have more children in the hope of having a boy, while mothers with boy children tend to stop having more babies. In the natural course of events where sex ratio is not manipulated by human intervention, if there is a preference for males, the overall sex ratio will favor girls [18]. This is because of biological heterogeneity which results in families tending to have children of same sex. This phenomenon is not evidenced in India which suggests there is more direct manipulation of the sex ratio in India. Sex ratio in mothers with 2 previous boys was 615 compared to 1017 in those with 1 previous boy. The small sample size of mothers with 2 previous boys can be the reason for an artifactually low sex ratio here. Our findings are similar to the findings of Jha et al who looked at sex in second children in a household survey. They found the adjusted sex ratio for the second birth when the preceding child was a girl was 759 per 1000 males. By contrast, adjusted sex ratio for second births if the previous child was a boy was 1102:1000 [21]. One of the objectives of our prospective study was to enquire into the methods parents may be using to get babies of the desired sex. We were aware that parents are unlikely to incriminate themselves by telling the investigator about antenatal sex determination. However the use of other methods have not been proscribed by law, so we felt it was reasonable to enquire about them in this study. A study by Bandyopadhyay and Singh found that up to 46% mothers use sex selection drugs. They tested 7 samples of such medicines and found 3 contained testosterone one contained progesterone and one a natural steroid [22]. Our study found that more parents who have girls tend to take traditional medicine in the next pregnancy. Overall, some 0.5% mothers took such medication and this percentage increased to 10% if the first child was a girl. 40% of mothers with two girls took such medication. We found mothers with a previous girl child are more likely to take indigenous medication for sex selection, than mothers with previous a previous boy. Mothers with previous girls were also more likely to have a boy in the next pregnancy. On the face of it may appear that these traditional medicines help mothers to have more boys. However the sex ratio of newborns of mothers taking traditional medicines was 928 girls to 1000 boys. This was much higher compared to the overall ratio of 720 girls to 1000 boys in mothers with one previous girl child. It was also higher than the overall sex ratio of 806:1000. This suggests that the subset of mother who took these medicines perhaps relied on them and it prevented them from resorting to techniques like antenatal sex determination and sex selective abortions. Our study has one notable weakness. It relates to sex ratio of children born in a hospital. According to the National Family Health Survey 3 (2005Survey 3 ( -2006)), 60% of all deliveries in India take place at home and outside of the medical institutions. Our data cannot therefore be said to be representative of India. It may be argued that if the sex of the child is known antenatally, there is a greater chance that male fetuses will be brought to the hospital for delivery and this could alter the ratio. This may be seen as an antenatal extension of the practice wherein boys are presented earlier in their illness and more frequently to the hospital [4]. However the data from second children delivered at this hospital shows that the majority of children are girls if the previous child was a boy, and this militates against the suggestion that boy fetuses are selectively bought to hospital for delivery. In summary our study suggests that in spite of Pre-natal Diagnostic Techniques (Regulation and Prevention of Misuse) Act 1994, the sex ratio at birth continues to fall. Evidence from the second children clearly suggests this is the result of human interventions. The use of traditional medicines in mother with previous girl children does not influence the sex ratio and mother who use traditional medicine perhaps do not employ other means to manipulate the sex ratio. The exact method used by some parents to influence the sex of their children is yet not clear. The impulse is to blame it all on ultrasound machines but our previous work on sex ratio in second babies has shown that the skewed sex ratio in second children was common even in the 1970s before ultrasound machines were freely available. More research is needed to elucidate this.

Objective: Sex-ratio at birth in families with previous girls is worse than those with a boy. Our aim was to prospectively study in a large maternal and child unit sex-ratio against previous birth sex and use of traditional medicines for sex selection. Main Outcome Measures: Sex-ratio among mothers in families with a previous girl and in those with a previous boy, prevalence of indigenous medicine use and sex-ratio in those using medicines for sex selection. Results: Overall there were 806 girls to 1000 boys. The sex-ratio was 720:1000 if there was one previous girl and 178:1000 if there were two previous girls. In second children of families with a previous boy 1017 girls were born per 1000 boys. Sexratio in those with one previous girl, who were taking traditional medicines for sex selection, was 928:1000.Evidence from the second children clearly shows the sex-ratio is being manipulated by human interventions. More mothers with previous girls tend to use traditional medicines for sex selection, in their subsequent pregnancies. Those taking such medication do not seem to be helped according to expectations. They seem to rely on this method and so are less likely use more definitive methods like sex selective abortions. This is the first such prospective investigation of sex ratio in second children looked at against the sex of previous children. More studies are needed to confirm the findings.

Background Across the world, schizophrenia is a debilitating, persistent psychiatric disorder that adversely affects individuals with the disorder as well as their family members who provide support and care for people living with schizophrenia (PLS) [1][2][3]. Such family caregivers often serve as an extension of the mental health system [4,5]. It is now recognized that globally, family caregiving is likely to grow for PLS especially in lower-and middle-income countries with under-resourced mental health and social service systems [6]. In China, the number of PLS has increased significantly from 3.09 million in 1990 to 7.16 million in 2010 [7]; over 90% live with their family and receive family care due to insufficient community resources [8]. Numerous studies across the world have shown various negative impacts of caregiving on caregivers, among which caregiver burden and psychological distress have been the most extensively studied and widely reported across cultures [5,9]. Two types of burden have been identified: objective (e.g., manual tasks, household duties, etc.) and subjective (e.g., caregiver's perception of burden, distress, stigma, etc.), and each may be experienced as physical, mental, financial, or social demands [5,9]. Subjective burden has been found to be a major determinant of caregiver psychological distress, but objective burden, which usually depends on specific caregiving activities, appears not to have a direct influence on distress [10]. Psychologic distress has been a very common phenomenon among caregivers. Both qualitative and quantitative studies have shown depression and anxiety as the two most frequently reported psychological distress among caregivers [11][12][13][14][15][16][17]. Face-toface qualitative interviews conducted with a diverse sample of U.S. caregivers of PLS showed that the most frequently reported impact of caring for a person with schizophrenia was feeling "emotional", including feeling overwhelming, sad, frustrated, embarrassed, angry, stressed, anxious and worried; experiences that have also been described by caregivers with analogies like "rollercoaster" "white-water rafting" and "feeling underwater" [15]. Other quantitative studies consistently reported a high level of psychological distress for caregivers, with the prevalence of depression and anxiety at or above 80% [16,17]. Although most of literature has described caregiving as having negative effects, it is increasingly recognized that caregiving may have positive influences on both the caregiver and family [18]. Caring for a family member with schizophrenia may bring about a positive transformation in a caregiver's life and can promote positive inner feelings, such as improved self-admiration, self-affirmation, self-confidence, self-satisfaction and personal growth; referred to as caregiving rewarding feelings [10,19]. In addition, caregiving may also impact family functional dynamics and relationships in a positive way, especially in Asian countries that attach great importance to familism and collectivism consistent with Confucian cultural values [20][21][22][23][24]. Research has identified various aspects of positive family functional dynamics in caregiving families, including family cohesion, family connectedness, family resilience and family hardiness, which may be generally referred to as family functioning [23,24]. In wellfunctioning families, all family members will actively work together to manage concurrent demands and utilize available resources in adapting challenges and adversities to achieve a degree of balance [23]. Despite the voluminous amount of literature on the experiences and impacts of caregivers of PLS, there are several limitations in the previous studies. First, most research has not distinguished between objective and subjective burden. Second, limited research has been directed at examining specific caregiving activities as well as the positive impact of caregiving. Third, most caregiving research in mainland China has been conducted in urban areas using convenience sampling from hospitals, whereas research in rural communities is underrepresented even though rural caregiver experiences may vary from urban experiences due to differences in economy, education, family structure, and culture. And more importantly, most studies have not differentiated experiences and impacts among types of kinship caregivers, such as spouses and parents. Identifying impacts of caregiving and the differences among family caregiver types may identify vulnerabilities that inform targeted interventions, but to date, evidence for such differences is mixed or limited [25][26][27][28]. Conflicting evidence exists on caregiver burden between spouses and parents. For instance, the World Health Organization World Mental Health Surveys [6] have shown higher caregiver burden for spouses than parents across high-and upper-middle-income countries, yet higher caregiver burden for parents than spouses from lower-or middleincome countries [29]. One study on Chinese caregivers of PLS found that parent caregivers reported higher burden than spouse caregivers [30], while another study of African caregivers of PLS showed no significant difference in caregiver burden between spouse and parent caregivers [31]. These results indicate that caregiver burden may be culture-dependent and vary by country. Comparison of psychological distress between spouse and parent caregivers showed inconsistent results. For instance, Chang et al. [32] compared both depression and anxiety between caregivers of individuals with serious mental illness (with most diagnosed with schizophrenia) in Taiwan but found no statistically significant differences between the two groups in either depression or anxiety. Similar results were reported in Stanley et al.'s [13] study on caregivers of PLS in India where no significant differences were observed based on parent/spouse status in depression and anxiety. However, another similar study conducted in Singapore found much lower scores in the psychological domain in spousal caregivers as opposed to parental caregivers [14]. Since research on the positive impact of caregiving such as rewarding caregiving feelings and family functioning is limited, even less is known about the relative positive impacts between parent and spouse caregivers. In view of the wide range of caregiving impacts and inconclusive evidence on relative differences among kinship caregiver types, it is critical to examine a range of caregiving impacts between spouse and parent caregivers in a representative rural community of mainland China. The current study addresses this limitation by comparing caregiving activities, negative caregiving impacts including objective and subjective burden, and caregiver psychological distress such as depression and anxiety, positive caregiving impacts including caregiving rewarding feelings, and family functioning between rural spouse and parent caregivers. In the current study, spouses were defined as a wife or husband in marriage, as well as nonmarital partners who have accepted a social role similar to that of a spouse, but do not have rights and duties reserved by law to a spouse; parents were defined as a biological or adoptive father or mother, with caregiving responsibilities. Based on conflicting evidence on comparisons between parent and spouse caregivers, we did not specify a hypothesis as to specific differences of each group or which group would be better or worse off, but rather conducted exploratory analyses to examine such differences. --- Methods --- Participants and procedure This cross-sectional study was conducted in Ningxiang County of Hunan province from November 2015 to January 2016. Our sample was primary caregivers of PLS that were registered in the "Central Government Support for the Local Management and Treatment of Severe Mental Illnesses project", also named as "686 program", which is China's largest demonstration project for mental health services that seeks to integrate hospital and community services for serious mental illness [33,34]. The 686 Program provides multiple comprehensive mental health services including: patient registration and initial assessment, free medication and regular follow-up in the community, management of community emergencies, and free emergency hospitalization [33,34]. A one-stage cluster-sampling method was used to recruit primary caregivers of PLS from the 686 Program. First, three Towns and 1 Xiang (an administrative unit similar to a town but with lower socio-economic development) were randomly selected from Ningxiang County, and then followed by whole sampling of all communities within each Town and all villages within each Xiang, leading to a total sampling frame of 55 representative communities/villages. This sampling method was used to obtain a sample as representative as possible and minimize sampling bias. A local primary care physician within each community/village assisted in identifying eligible primary caregivers based on a list of PLS registered in the 686 Program. Eligibility criteria for primary family caregivers included: 1) caring for a family member with schizophrenia who is registered in the 686 Program; 2) the family member being cared for meets criteria of the Chinese Classification of Mental Disorders-3 (CCMD-3) or the International Classification of Diseases-10 (ICD-10) for schizophrenia; 3) the primary caregiver is living with the PLS at least for the past two years and assumes major responsibilities for caregiving; 4) the primary caregiver is older than 16 years of age, since by Chinese civil law, people at the ages of 16 and older who are engaged in labor (here referred to as "caregiving") and have a source of economic income are considered to have full capacity for civil conduct; and 5) the primary caregiver speaks Chinese, is literate, and not seriously disabled, thus enabling the person to understand and communicate in faceto-face interviews with the caregivers. Exclusion criteria included: 1) caring for family member with diagnosis other than schizophrenia such as depression and epilepsy; 2) primary family caregivers who do not speak Chinese or having serious physical or mental illness that makes them unable to communicate effectively in a face-to-face interview. A total of 352 primary family caregivers of PLS were identified in the final sample. The community/village primary care physician visited each study household with a member of the research team to obtain informed consent from the caregivers, who then completed a series of questionnaires (see measures below) in face-to-face interviews, which were all conducted in Chinese. All caregivers were reimbursed with small gifts equivalent to RMB <unk> 10 (approximately USD$1.5) in return for their participation. Details of the subject recruitment process have been described elsewhere [35,36]. Of the 352 caregivers approached, 327 participated in the interview (response rate: 93%). Among the 25 caregivers that refused the interview, reasons for refusal included no interest in the study (n = 11), concern about stigma (n = 9), too painful to share caregiving experiences (n = 3) and other reasons not disclosed (n = 2). No significant differences in socio-demographic background were observed between those who did or did not agree to participate based on age, gender, marital status, employment and education (available as Additional file 1). In the current study, only parents (n = 151) and spouses (n = 113) were included in the analyses, resulting in a final sample of 264. This sample size of 264 detected an effect size of 0.4 between parent and spouse caregivers by twotailed t-test, assuming <unk> = 0.05, <unk> = 0.10. --- Measures --- Socio-demographic information Demographic information of both the PLS and the primary caregiver was collected by asking the primary caregivers on a questionnaire designed for this study, which included gender, age, marital status, occupation, education, family financial status, and kinship between PLS and the primary caregiver. For the primary caregivers, we also asked if there were any co-caregivers, dependents, physical illnesses, and how care had been provided for the PLS. This information was collected through face-to-face interviews between the research team and primary caregivers. --- PLS functioning PLS functioning was assessed using the Global Assessment of Functioning scale (GAF) [37]. The GAF is an axis of the DSM-IV and assessed the person's overall social, occupational and psychological functioning in the past month from 1 (lowest) to 100 (highest), with examples provided for each ten-point interval. The GAF was first translated in Chinese by Zhang in 1984 [38] and proved to be a reliable and valid measure of psychiatric functioning [37,38]. For the current study, PLS GAF was assessed by the research team based on a combination of caregivers' report, direct assessment of the PLS, and a review of their clinical records, whenever available. --- Caregiving activities Specific caregiving activities were assessed by four "Yes-No" questions asking if the respondent was involved with the following four aspects of caregiving activities for the PLS: daily activities (e.g., eating, drinking, getting dressed), medication management (e.g., monitoring medication, buying medicine), hospital visits (e.g., registration, hospitalization), and financial help (e.g., spending and giving money). Responses of "no" were scored as 0, and "yes" as 1, and then followed by asking how often the respondent was involved with each caregiving activity, ranging from "occasionally" (scored as 1) to "always" (scored as 4). Detailed information about the questions and optional answers can be found in the Additional file 2. For the current study, questions on caregiving activities were administered by the research team to primary caregivers. --- Objective burden Objective burden was assessed using the Family Burden Interview Schedule (FBIS) [39] classified into six categories: financial burden, disruption of routine family activities, family leisure, family interactions, and effect on physical and mental health of others. The scale consists of 24 items rated on a 3-point Likert scale from 0 (no burden) to 2 (serious burden). The total score ranges from 0 to 48 with higher scores showing higher family burden. The FBIS was first translated into Chinese by Chien et al. [40] in 2004 and proved to be a reliable and valid measure of objective family burden. For the present study, FBIS was administered by the research team to primary caregivers and the Chinese version of FBIS showed acceptable internal consistency with a Cronbach's <unk> of 0.86 for the total scale and ranged from 0.63 to 0.86 for the subscales. --- Subjective burden Subjective burden was assessed using the Zarit Burden Interview (ZBI) [41]. The ZBI consists of 22 items scored on a 5-point Likert scale from 0 (never) to 4 (nearly always), except for the final item on global burden, rated from 0 (not at all) to 4 (extremely). The total score ranges from 0 to 88 with higher scores indicating higher perceived burden [42,43]. The ZBI was first translated into Chinese by Lu et al. [44] in 2009 and proved to be a reliable and valid measure of subjective caregiver burden. For the present study, ZBI was administered by the research team to primary caregivers and the Chinese version of ZBI showed acceptable internal consistency with a Cronbach's <unk> coefficient of 0.89. --- Caregiver depression Caregiver depression was measured using the 9-item Patient Health Questionnaire (PHQ-9) [45], which consists of 9 items scored in 4-point Likert scale from 0 (not at all) to 3 (nearly every day). The total score ranges from 0 to 27, with higher scores implying more depressive symptoms, and a cut-off point of 10 differentiating depression and non-depression [46,47]. The PHQ-9 was first translated into Chinese by Yeung et al. [48] in 2008 and proved to be a reliable and valid measure of depression. For the present study, PHQ-9 was administered by the research team to primary caregivers and the Chinese version of the PHQ-9 demonstrated good internal consistency with a Cronbach's <unk> coefficient of 0.89. --- Caregiver anxiety Caregiver anxiety was measured by the 7-item Generalized Anxiety Disorder Scale (GAD-7) [49] which detects the primary caregiver's anxiety symptoms during the past two weeks. Each item is rated on a 4-point Likert scale from 0 (not at all) to 3 (nearly every day). The total score ranges from 0 to 21, with a cut-off point of 10 differentiating anxiety and non-anxiety [50]. The GAD-7 was first translated into Chinese by He et al. [51] in 2010 and proved to be a reliable and valid measure of anxiety. For the present study, GAD-7 was administered by the research team to primary caregivers and the Chinese version of the GAD-7 demonstrated good internal consistency in the current study with a Cronbach's <unk> coefficient of 0.91. --- Caregiving rewarding feelings Positive feelings about caregiving were assessed using the caregiving rewarding feelings (CRF) scale which was designed for this study. CRF was initially developed based on qualitative interviews on a convenience sample of 30 pairs of PLS and their primary caregivers, then revised based on pre-testing and Delphi's method, and validated in a larger sample. The detailed development and validation process of the CRF has been described elsewhere [52]. The CRF was developed in Chinese and consists of 12 items asking about a range of possible rewarding feelings that caregivers may have during caregiving. Each "yes" response was scored 1 and "no" response 0, with total scores ranging from 0 to 12; higher scores indicate more positive feelings in caregiving. For the present study, CRF was administered by the research team to primary caregivers and the Chinese version of CRF showed acceptable reliability with a Cronbach's alpha of 0.77. --- Caregiver perceived family functioning Caregiver perceived family functioning was assessed using the Family Adaptation, Partnership, Growth, Affection and Resolve Index scale (APGAR) [53]. It consists of 5 items scored in 3-point Likert scale from 0 (hardly ever) to 2 (almost always). The total score ranges from 0 to 10 with higher scores indicating higher satisfaction with family functioning. The APGAR was first translated into Chinese by Chang et al. [54] in 1993 and proved to be a reliable and valid measure of family functioning. For the present study, APGAR was administered by the research team to primary caregivers and the Chinese version of APGAR showed acceptable internal consistency with a Cronbach's <unk> coefficient of 0.91. --- Statistical analysis Data analyses examined missing values, influential values and outliers, skewness, and kurtosis. Frequencies and percentages were displayed for categorical variables, and means with standard deviations or medians with interquartile ranges (IQR) displayed for continuous variables. We compared socio-demographic characteristics, caregiving activities, and six types of caregiving impacts (objective family burden, subjective caregiver burden, depression, anxiety, caregiving rewarding feelings, and family functioning) between spouse and parent caregivers using two-group tests. Depending on variable type and data distribution, we conducted various two-group tests including: independent two sample t-tests for normally distributed continuous data, Mann-Whitney U test for non-normally distributed continuous data, and Pearson's chi-square tests for categorical variables. In order to examine the predictive effect of kinship (spouse vs parent) on the six types of caregiving impacts, we further conducted six separate multivariate linear regressions, with kinship (spouse vs parent) as independent variables, and the six types of caregiving impacts as dependent variables, while controlling for all potential confounders (PLS functioning, gender, and education, as well as caregiver age, gender, marriage, employment, education, family income, whether have co-caregivers, whether there were additional dependents, whether caregivers have physical illness, and length of caring) to avoid analysis bias (Table 4). Data were analyzed using STATA software version 15.0, with p values smaller than 0.05 in two-tailed tests considered as statistically significant. --- Ethical considerations This study was approved by the Ethics Review Committee of the Xiangya School of Public Health of Central South University in China. The interviewed participants were informed verbally and in writing of the study's purpose, their right to refuse to participate, and the voluntary nature of their participation. All participants provided written informed consent before the interviews. --- Results --- Group comparison of sample characteristics Table 1 describes characteristics of the PLS. The median age was 44 for the PLS. Slightly less than half of PLS were female, married, with a primary education, and most were unemployed. PLS functioning as measured by the GAF showed a median score of 42, indicating serious impairment of functioning. We further compared the socio-demographic characteristics of PLS cared for by spouses and parents. Compared to PLS cared for by a spouse, PLS cared for by a parent had lower functioning (median: 35 vs 55), were younger (median: 38 vs 53), were more likely to be male (70% vs 29%), were less likely to be married (76% vs 100%) and had higher educational attainment (66% vs 42%). Table 2 describes characteristics of the primary caregivers. The median age was 62 for the primary caregivers. Slightly more than half of caregivers were female, employed, without co-caregivers and with additional dependents. Also, most caregivers were married, of a primary education 1 and with some physical illness and had been caring for the PLS for many years (median = 15 years). We also compared socio-demographic characteristics between spouse and parent caregivers. Compared to spouse caregivers, parent caregivers themselves were older (median: 64 vs 57) and had fewer years of caregiving (median: 14 vs 19); were more likely to be female (70% vs 32%), had co-caregivers (52% vs 28%) and had a physical illness (78% vs 67%); they were also less likely to be married (67% vs 100%), employed (44% vs 61%) and with middle and high school education (24% vs 42%). --- Group comparison of caregiving activities Table 3 shows the type and frequency of care provided by the primary caregivers to the PLS. Generally, most caregivers were actively involved with all four types of caregiving tasks, with participation rate ranging from 61% for financial help to 73% for medication management. For the frequency of each caregiving activity, a median frequency of 4 (always) was reported for all caregiving types except for daily activities (median frequency = 3). This difference may be because PLS in this sample had few physical disabilities that required assistance in daily activities as opposed to assistance with medication management, hospital visits, and finances. We also found no statistically significant difference between spouse and parent caregivers in their involvement in or the frequency of these four types of caregiving activities, indicating that parents and spouses were providing similar types of care for their loved one with schizophrenia. --- Group comparisons of caregiving impact Table 4 shows six types of caregiving impacts that includes both negative and positive impacts. Negative impacts include both objective family burden as measured by FBIS (median = 26) and subjective caregiver burden as measured by ZBI (mean = 44), as well as psychological distress including depression and anxiety, with a median of 9 for both. Positive impacts include caregiving rewarding feelings as measured by CRF (median = 7) and family functioning as measured by APGAR (median = 6). A further comparison of the six types of caregiving impact between spouse and parent caregivers showed statistically significant differences in all except for caregiving 81, P = 0.026), they also reported higher depression (12 vs 6.5, P <unk> 0.001), anxiety (11 vs 7, P <unk> 0.001) and lower family functioning (6 vs 8, P = 0.002). --- Multiple multivariate linear regressions of kinship on caregiving impacts. Table 5 shows the results of six separate multivariate linear regressions of kinship (spouse vs parent) on the six types of caregiving impacts (family burden, caregiver burden, depression, anxiety, caregiving rewarding feelings, and family functioning), while controlling for potential confounders (PLS functioning, gender and education, and caregiver age, gender, marriage, employment, education, family income, whether have co-caregivers, whether there are additional dependents, whether caregivers have physical illness, and length of caring). Although no statistically significant differences in objective family burden (FBIS) were found between spouse and parent caregivers, parent caregivers reported much higher subjective caregiver burden, with a coefficient as high as 7.94 (95% CI: 2.08, 13.80). Also, multivariate analyses showed that parent caregiving independently and significantly predicted higher depression (b = 3.88, 95% CI: 1.35, 6.41) and anxiety (b = 2.53, 95% CI: 0.22, 4.84). For positive impacts of caregiving, although no statistically significant difference in caregiving rewarding feelings were found between spouse and parent caregivers, parent caregivers reported significantly lower family functioning with a coefficient of -1.71 (95% CI: -2.85, -0.57). --- Post-hoc power analysis We did a post-hoc power analysis based on one primary outcome of subjective caregiver burden measured by Zarit Burden Interview (ZBI). Assuming <unk> = 0.05, the sample size of 113 for spouse caregivers (mean: 40.81) and 151 parent caregivers (mean: 46.18) were able to detect a significant difference with 95% power. --- Discussion This study compared the caregiving experiences of spouse and parent caregivers of PLS in China. The results show that both types of caregiversparents and spouses --report engaging in similar caregiving activities and comparable levels of objective burden. However, parent caregivers report significantly higher subjective burden, which is reflected in significantly higher depression and anxiety scores, and in reports of lower family functioning. Despite these differences, both groups of caregivers report comparable rewarding feelings about caregiving. As far as we know, no studies on caregiving of PLS have examined the specific caregiving activities of primary caregivers, a gap addressed by the current study through its examination of four specific types of caregiving activities: daily activities, medication management, hospital visits, and financial help. Our findings showed that parent and spouse caregivers are equally engaged in these activities, which is suggested by comparable reports of objective burdens across kinship caregivers. Our results are consistent with the extensive literature showing caregiving to be a highly-demanding experience that requires family caregivers to devote considerable time, energy and resources to provide round-the-clock care to a PLS, regardless of the specific activity [4,5,8,9]. However, parent caregivers reported significantly higher subjective burden than spouse caregivers, even though objective burden and caregiver activities were both comparable. This is an important finding because it suggests that objective burden and what caregivers actually do may not fully reflect the subjective burden of what they actually feel. The consistent relationship between subjective burden and depression and anxiety scores also suggests a potential role for subjective burden in mediating the relationship between caregiving and health [4,55]. Future longitudinal research should examine the relationship between subjective burden and health as well as the role of social supports in this relationship. Another intriguing issue raised by the present study is the intersection of family caregiving with the dominant belief in collectivism that is deeply rooted in Chinese culture. Family members in China attach great importance to specific roles and proper relationships among family members, thus maintaining family cohesion, harmony, and equilibrium [5,56]. As a result, different family members may perform similar caregiving tasks and experience similar objective burden since they are living in the same household and caring for the same person, yet their personal perception of burden may vary greatly due to the different role expectation they experience. For instance, parents, especially mothers, are more likely to be blamed for a child's sickness since they are expected to give birth to healthy children and nurture them through their parenting role [57]. A child with schizophrenia may contribute to a marriage crisis and family conflict [18,57,58]. The mother and father may blame each other for the child's condition, disagree about treatment plans, and argue about caregiving responsibilities. As a result, parents may perceive higher burden when caring for PLS due to the sense of guilt, blame, and anger that results from their situation [18,57,58]. In contrast, spouse caregivers may normalize caregiving as part of their marital role as reflected in marriage vows, such as "in sickness and in health" or "until death do us part" [59][60][61]. Spouses may thus experience less conflict between their caregiver roles and other family and work-related roles they have, and thus experience less subjective burden [62,63]. Differences in subjective burden between spouse and parent caregivers may also be related to differences in perceived obligation, duty, and responsibility for supporting an individual with mental health problems. For example, spouse caregivers may feel less of a keen sense of duty or obligation to care for a PLS than parents who may feel they have no choice but to continue caregiving. Future research should examine directly the role of culture and family caregiving for PLS. Related to the above issue is that parents reported being worse-off than spouse caregivers in both negative and positive caregiving impacts. Parents reported more perceived burden, more depression and anxiety symptoms, and lower perceived family functioning. These findings are consistent with the unique family-oriented culture common in Asian countries that emphasizes family cohesion and harmony, with special focus on familism and filial piety [20,56,64]. Chinese society is especially recognized for its strong sense of filial obligation, defined as: "the concept, desire, and behaviors of carrying on family line, being good, and obedient to, as well as taking care of one's parents". One key ideology of filial piety in China is the concept that "Bu Xiao You San, Wu Hou Wei Da" (There are three types of impieties, among which having no descendent is the worst) and "to have no posterity is unfilial" [65]. In a Chinese society that is based on close-knit family relationships, children are viewed as the only continuation of the family bloodline, thus representing the future prosperity of the entire family [66][67][68]. Children are expected to take care of their parents when they are old and frail [66][67][68], but a child with schizophrenia may mean that the family bloodline may be broken and that the parents have no one to depend on when they grow old. The situation is accentuated with the "one child per family" policy that persisted for over three decades before it was recently abolished [69]. --- Limitations This study has several limitations. First, its crosssectional design limits being able to examine differences among spouse and parent caregivers over time. Further longitudinal research is needed to look at such caregiving experiences and impacts, as family caregivers' views of their responsibilities for a mentally ill relative do evolve over time [70]. Second, many caregivers had been caring for their family members with schizophrenia for a long time, and thus the experiences of this sample may be different from others caring for a PLS after a first episode. Future research may seek to examine this issue. Third, although our total sampling frame included various kinships including spouse, parents, siblings, children and others, we only were able to examine spouse and parent caregivers due to an insufficient sample size to examine other kinship types. Future research should consider examining differences across more diverse caregiver types. Fourth, participants were approached through the 686 Program and thus excluding caregivers of PLS not registered into the system. It is likely that caregivers of PLS outside the 686 Program may have different caregiving impacts due to lack or lower quality of needed services. Future research may seek to include the caregivers of PLS outside the 686 Program to get a more representative picture of caregiving experiences and impacts for all PLS in China. --- Conclusions This study found that parent caregivers of PLS reported significantly higher subjective burden, depression, anxiety, and lower family functioning than spouse caregivers, even though both types of caregivers reported comparable caregiving activities, objective burden, and caregiving rewarding feelings. These results have several implications. First, caregiver intervention programs that target individual caregivers may need to address the specific vulnerabilities of parent caregivers of PLS, such as increased subjective burden, depression, and anxiety, and reduced family functioning. Second, family caregiver intervention programs should take into account kinship family dynamics when supporting family caregivers of PLS. One promising approach may be to offer psychoeducation to family members of the PLS as part of ongoing treatment and rehabilitation [71][72][73]. And third, this study suggests that it may be particularly useful to incorporate cultural values and beliefs in understanding the family processes at work for family caregivers of PLS. Although our study was conducted in mainland China, our findings are relevant to other countries, especially those that share Confucian beliefs common to Asian culture. --- Availability of data and materials The datasets used and/or analyses are available from the corresponding author with a reasonable request. --- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. 1186/s12888-020-02633-w. Additional file 1. Comparison of socio-demographics between respondents and non-respondents. Showing analysis results of comparison between the two groups. Authors' contributions YY, TL, SYX and JKT contributed to the conception and design of the study, YY, TL, YL, DQ, and SJX contributed to the research conduction and data collection, YY, TL and YL contributed to data analyses and interpretation. YY drafted the article while TL, YL, DQ, SJX, SYX and JKT critically appraised it and revised it. All authors approved the final version of manuscript for submission and publication and agreed to be accountable for all aspects of the work. --- Consent for publication Not Applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Conflicting evidence exists on whether parent or spouse caregivers experience better outcomes when caring for family members with schizophrenia. The current study aims to examine relative caregiving experiences and impacts of spouse and parent caregivers for people living with schizophrenia (PLS) in China. Methods: A cross-sectional study was conducted in a sample of 264 community-dwelling primary family caregivers of PLS. Face-to-face interviews were conducted to collect information on family caregiving activities; negative caregiving impacts including objective and subjective burden, and caregiver psychological distress such as depression and anxiety; positive caregiving impacts including caregiving rewarding feelings, and family functioning for spouse and parent caregivers. Results: Both types of caregivers report engaging in similar caregiving activities and report comparable levels of objective burden. However, parent caregivers report significantly higher subjective burden than spouse caregivers (b = 7.94, 95%CI:2.08, 13.80, P < 0.01), which is also reflected in significantly higher depression (b = 3.88, 95%CI:1.35, 6.41, P < 0.01) and anxiety (b = 2.53, 95%CI: 0.22, 4.84, P < 0.05), and lower family functioning (b = -1.71, 95%CI: -2.73, -0.49, P < 0.01). Despite these differences, both groups of caregivers report comparable rewarding feelings about caregiving. Conclusions: Our findings have implications for family caregivers globally, but especially for countries that adhere to Confucian cultural values and provide guidance for future family intervention programs. Such programs may do well to incorporate cultural values and beliefs in understanding caregiving and kinship family dynamics so as to support family caregivers, and in particular, the specific vulnerabilities of parent caregivers.

'conviviality', referring to the processes of 'cohabitation and interaction that have made multiculture an ordinary feature of social life in Britain<unk>s urban areas' (Gilroy 2004:xi). The paper describes how in a super-diverse context, conviviality, also understood in the sense of living together peacefully (Overing & Passes 2000), is characterised by a fine balance between building positive relations across difference and keeping a distance. While cultural diversity is seen as commonplace, there are differences in the ways people deal with cultural differences in public versus semi-public sites. I here use the differentiation between the public, parochial and private realm (Hunter 1985) to discuss these differences. While the public realm is the world in streets, parks, public transport or commercial spaces where one meets strangers, the parochial realm is characterized by more communal relations among neighbours, with colleagues in the workplace, or acquaintances through associations or schools (Hunter 1985;Lofland 1989). Importantly, the boundaries between these realms are fluid. For example, a corner-shop or a market where traders and costumers meet on a regular basis can take on the characteristics of the parochial realm because the social relations developed in these places can become habitual and frequent. The differentiation between the public, parochial and private realm is particularly useful when thinking about the degree to which interactions between people of different backgrounds are meaningful and contribute to intercultural understanding. In this article, I show how in the public realm, commonplace diversity is characterised by social interactions which are shaped by 'civility towards diversity' (Lofland 1989: 464), a concept which I discuss in further detail later in the paper and with illustrations from my fieldwork. Civility towards diversity in the public realm means that people treat everybody universally the same without acknowledging differences. In contrast, in the parochial realm, people's different backgrounds are being acknowledged and sometimes talked about. I show how especially in parochial space, civility towards diversity can be used as a strategy to both engage with difference and avoid conflict by ignoring difference. While using the concept of civility discussed by social scientists such as Sennett (2005) and Lofland (1989), the paper also draws on current discussions about the role of encounters regarding the enhancement of intercultural understanding and interaction. Although not conceptualized as 'parochial realm', there has been much discussion about the role of'semi-public' spaces such as schools or associations in regard to social relations and negotiations of difference, and the effectiveness of social contact across categorical boundaries in reducing possible conflict and tensions (e.g. Amin 2002;Blokland 2003b;Sandercock 2003;Sanjek 1998;Valentine 2008;Wood & Landry 2007). The demographic nature of a super-diverse context brings with it the emergence of numerous such 'zones of encounter' (Wood & Landry 2007). In these spaces, deeper and more enduring interactions between people who engage in shared activities and common goals can take place. Amin (2002) conceptualizes such spaces as'micropublics' where differences across ethnic, religious, class and other boundaries can be bridged and stereotypes broken. Such micropublics are crucial in shaping people's perceptions about each other. Wise (2007;2010) describes places in which people of different cultural backgrounds meet as 'transversal places' where intercultural encounters and relationships are formed. With examples of a bingo hall and an elderly people's club, Wise (2007: 7) shows how 'the simple fact of regular togetherness... can facilitate fleeting relations and sometimes friendships across difference'. As shown by Noble (2009: 52), this enables the creation of 'a set of relatively stable relations and ways of intercultural being which emerge out of sustained practices of accommodation and negotiation'. However, some writers caution against generalisations about the positive effects of regular encounters on intercultural understandings. For example, Valentine (2008:332) shows how 'positive encounters with individuals from minority groups do not necessarily change people's opinions about groups as a whole for the better'. She criticises discourses about conviviality and everyday multiculturalism as celebratory by demonstrating the co-existence of daily courtesies in public space and the continuity of privately held prejudiced views. Others similarly show how stereotypes and racism can co-exist with daily interactions in multicultural neighbourhoods (Wise 2005, Watson 2006, Swanton 2009, Noble 2011, Wilson 2013 [forthcoming]). In this paper, I show how conviviality in Hackney is characterized by both avoidance of deeper contact and engagement, and that civility towards diversity is a strategy to negotiate both positive relations and possible tensions. I begin the paper with a short history of Hackney's diversification and a description of how commonplace diversity evolved as a result of this diversification. The following section discusses patterns of conviviality in the public realm. Referring to concepts of civility, I describe how social interactions in public space are characterised by both intercultural competences and a pragmatic approach towards difference where everybody is treated the same, independent of their backgrounds, because so many people come from elsewhere. The following section examines how civility towards diversity works in parochial spaces where social relations are characterised by more regular contact. I show how, while diversity is acknowledged and sometimes talked about, actual engagement with difference remains limited and people rarely explore cultural differences more deeply. On the one hand, contact in parochial space can lead to more mutual understanding and acceptance of difference, sometimes leading to relationships of support, but at the same time, civility towards diversity is used to avoid possible tensions. The conclusion summarizes the differences of patterns of conviviality in the public and parochial realm and discusses how civility towards diversity and a certain indifference to cultural difference might be a mode of dealing with diversity. The conclusion also argues that issues surrounding inequality and poverty are far more relevant in the area than those relating to cultural diversity which, over the years, has become so commonplace. The paper draws on 18 months of ethnographic fieldwork in the London Borough of Hackney during the period of 2008-2010. The fieldwork included participant observation, for example in a weekly knitting group of elderly women, a youth club on an estate, a parents' group in a primary school, and an IT class for over 50s. All these groups were ethnically and socially mixed. Fieldwork also included participant observation in public spaces such as shops, parks and markets, as well as 28 in-depth interviews with local residents and key people such as councillors, teachers and social workers, and three focus groups. The people interviewed were of various ethnic and social backgrounds, including people of different age groups and legal statuses. At the time of the research, I had been living in Hackney for about four years already. As a local resident and mother, I was and continue to be participating in playgroups and other children-related activities, which has enabled me to deepen my understandings of informal social relations in public and associational space and to have numerous informal conversations with parents of various backgrounds. --- Hackney's history of diversification and the emergence of commonplace diversity If there is a general characteristic to describe Hackney, it is the continuity of population change over the past half century. With its population of 247,182, Hackney figures among the 10% most deprived areas in the UK, but it is currently seeing the arrival of an increasing number of middle-class professionals. It is also one of the most ethnically diverse boroughs in Britain, with only 36.2% of the population being white British. Jewish people have been settling in Hackney since the second half of the 17th century, and since the 1950s, sizeable groups of immigrants from West Africa, the Caribbean and South Asia have arrived. Turkish, Kurdish and Turkish Cypriot people started arriving in the area in the 1970s, both as labour migrants and political refugees (Arakelian 2007). Vietnamese refugees arrived in the late 1970s (Sims 2007). Among the biggest minority groups are Africans (11.4%), people of Caribbean background (7.8%), South Asians (6.4%), Turkish-speaking people (5.5%), Chinese (1.4%) and 'other Asian' (2.7%, many of whom come from Vietnam). 6.4% of the population identify as'mixed'. 35.5 % of Hackney's total population are foreign-born, and they come from 58 different countries, ranging from Zimbabwe, Cyprus, Somalia, Iraq, Albania to Denmark, Germany, etc. 1 Recently, there has been an increase in people from Eastern Europe, especially Poland (City and Hackney 2008), and Hackney has one of the largest refugee and asylum seeker populations in London, estimated to be between 16,000 and 20,000 (Schreiber 2006). The positive attitudes towards diversity are not only reflected in a general acceptance of diversity, but also in diversity not being seen as something particularly remarkable. For example, during my fieldwork in local associations, I noticed that newcomers are not usually asked about their origins, even if they look different or speak with an accent. When I asked whether I could do part of my fieldwork at a computer club for elderly people, the teacher of the club welcomed me there, but also told me that although his students came from many different places, diversity is not an issue in any of their conversations. They rarely ask each other where they come from and are not really interested in the other students' origins because everybody comes from elsewhere and it is therefore not a particularly special topic to talk about. In other words, diversity is so normal among the students in this computer club that it has become somewhat banal. This normalcy of diversity is what I conceptualize as 'commonplace diversity' (Wessendorf 2010). In his conceptualisation of'side-by-side citizenship ', Van Leeuwen (2011) This confirms Nava's argument that the familiarity between groups has'shifted the axis of belonging in much of contemporary London' (Nava 2007: 14). In his research in North London (including Hackney), Devadason (2010) has similarly shown that skin-colour no longer marks insider or outsider status. I have found that this also applies to dress-code and, to some extent, language, with African dresses or Indian saris as well as foreign accents not being perceived as worthy of mention. How is commonplace diversity reflected in the public realm when it comes to social interactions with people of different backgrounds? --- Conviviality in the public realm How do people who live in a super-diverse context deal with the fact that a large number of people whom they meet in public space differ in their cultural background? Do they attempt to adapt their behaviour according to the other's perceived background, or do they treat everybody universally the same? The skills necessary to communicate with people of different backgrounds have also been described as 'civility towards diversity'. In her discussions on patterns of behaviour and social life in the public realm, Lofland (1989) defines 'civility towards diversity' as one of the main 'interactional principles'. This principle...... specifies that in face-to-face exchanges, confronted with what may be personally offensive visible variations in physical abilities, beauty, skin colour and hair texture, dress style, demeanour, income, sexual preferences, and so forth, the urbanite will act in a civil manner, that is, will act 'decently' vis-à-vis diversity (Lofland 1989: 464-5). Importantly, Lofland states that this civility towards diversity does not necessarily imply a specific appreciation of diversity, but it means treating people universally the same, and it can emerge from indifference to diversity rather than from a specific appreciation of it. 2 Buonfino and Mulgan (2009: 16) take the definition of civility a step further and describe it as a 'learned grammar of sociability'. They compare these grammars of sociability with language. Although we are born with the disposition to speak a language, we still have to learn how to speak, read and write. Similarly, civility is based on existing dispositions, but it also has to be learned and cultivated. These grammars of sociability are important skills needed to get along in such a context, as most everyday interactions and conversations in public space take place with people of different backgrounds. These patterns of conviviality among people who differ are in line with Sennet's definition of civility as more than just good manners, but 'the capacity of people who differ to live together' (Sennet 2005: 1). In fact, this capacity is something that people living in Hackney consciously or unconsciously share. Civility towards diversity is a feature of public-space interactions which I observed on a daily basis during my fieldwork, as exemplified by the following vignette: I'm at a supermarket looking for a hair dryer. As I stand in front of the electronic household equipment, I observe an elderly Turkish woman asking a young white British shop assistant for advice. I hear him say: 'Do you understand?' She says, 'No, no English, only Turkish.' She calls someone on her mobile phone, indicating to the shop assistant to wait. Once she has spoken to the other person on the phone, she hands the phone to him. The person on the phone now seems to be doing the translation, and the phone is being handed back and forth between the shop assistant and the Turkish woman. It seems completely normal for the assistant to deal with a customer via an interpreter over the phone. He is very friendly all through the interaction and he seems in no way surprised about the translation service over the phone (Fieldwork diary, August 2008). Goffman describes the nature of such interactions with the concept of 'facework', referring to necessary mutual respect and recognition in social interactions, no matter across what kinds of perceived group differences (Goffman 1972). What differentiates a superdiverse context from other contexts with less categorical groups is the amount of information available about 'the other', information which could facilitate knowledge about what to expect from the other in a specific social interaction (Goffman 1971). In a super-diverse context, the'sign-vehicles' (Goffman 1971) available for understanding this information are much more complicated than in other contexts. Despite the presence of large minority groups in Hackney such as Turks and West Indians, many of whom share similar socio-economic backgrounds, migration histories and legal statuses, there exists a large number of people who are much more difficult to label. For example, the Muslim woman with a headscarf and Moroccan dress whom I met at a primary school turns out to be a native Italian who had come to London as a student, married a Moroccan and converted to Islam. Similarly, a South Asian-looking mother turns out to have been born in Zambia of Indian parents, and came to the UK via South Africa as a child. A black British Muslim nursery school teacher has her origins in Uganda. Her family is Christian and she is the only one in the family who converted to Islam. I have met countless people who surprised me with their unusual migration histories and backgrounds, and the more people I met, the more difficult I have found easy labelling and categorisation. In fact, one of my informants, a British woman in her 30s who came to Hackney from Northern England some eight years ago, said that when you meet a new person in Hackney, you cannot take anything as 'a given'. While she has a very positive attitude towards diversity, she also sometimes finds it tiring 'always having to consider where the person may be coming from when you meet them'. These difficulties of categorising strangers seem to lead to a certain pragmatism, where, in order to get around, get help to get on a bus, carry a buggy up the stairs, etc., you cannot afford not to be civil towards people who are different. This civility is usually expressed by way of treating people universally the same, as described by Lofland (1989). One of my elderly British informants told me that 'you cannot treat people differently according to their backgrounds because almost everybody comes from elsewhere'. 3 However, civility towards diversity can also ensure boundaries. People can be civil because they want to avoid further contact. This can apply to people of one's own groupwhether defined in terms of ethnicity, gender, religion, sexuality, etc. -and members of other groups. In the following section, I show how civility towards diversity plays an important role in parochial space where contacts with people who are different are more regular, and how this civility is used to both bridge differences and avoid tensions. --- Social relations and civility towards diversity in the parochial realm In parochial realm spaces such as associations, schools or among neighbours, cultural diversity is just as commonplace as in public space, but it is more concretely acknowledged and it is sometimes talked about, although rarely as an issue of contestation. As summarized in the introduction to this article, research has shown that social relations in the parochial realm can contribute to the reduction of stereotypes and the bridging of differences (Amin 2002;Wood & Landry 2007). At the same time, however, such stereotypes can persist in parallel to positive relations (Noble 2011;Valentine 2008;Wise 2005). The weekly coffee morning for parents at a local primary school presents an example where contact across difference can lead to the reduction of prejudice, but where civility towards diversity is also used to limit engagement across difference. The coffee morning is attended by mothers from a wide range of backgrounds: There is a white British woman who grew up in the area. Similarly, a British Pakistani and a British Guyanese mother have both grown up just around the corner. There is also an Italian woman who came to London in her 20s, married a Moroccan and converted to Islam. Then there is a white British woman who came to London from Northern England some eight years ago. Sometimes, two Nigerian women come to the coffee morning, one is Yoruba, the other from the area of Benin. And a Turkish mother is one of the most regular participants of the coffee morning. She is known among the mothers for her gardening and cooking skills. Although not all of these women attend the coffee morning every week, in a focus group discussion4, they emphasized how thanks to the children's centre in which the coffee morning takes place (and which is situated on the school ground), they get more opportunities to meet people and socialize. The women also emphasized that having children generally facilitated social contacts with other parents, be it at the school gates or the coffee mornings. Although during the focus group, they agreed that they appreciated interacting with 'people from different walks of life', the issues they talked about during coffee mornings were more often those which they shared, rather than their differences. Such shared themes are for example gardening or the education of their children. While conversations about gardening are characterized by the exchange of ideas and knowledge, when it comes to cooking, cultural differences form part of the conversations. Such differences also come up when they speak about religious traditions. The British Pakistani woman, for example, was fasting during Ramadan, an issue the other mothers listened to, but did not discuss any further. Similarly, the fact that the wearing of a head-scarf prevented the Turkish mother from getting a job at a fast-food chain formed the subject of a conversation. The other mothers agreed about the unfairness of it, but did not discuss the issue any further. Thus, while shared themes such as the children, being out of work and the changing built environment dominate the bulk of the conversations, cultural differences came up every once in a while. They were acknowledged as a matter of fact, but they were neither met with much curiosity nor with surprise or estrangement. Often, the conversations about these differences did not go very far, with few questions asked. Cultural and religious differences were thereby treated as personal characteristics of individuals, which the mothers shared with friends and family, but which, most of the time, they left at home. According to one of my informants, you do not ask too many questions about other people's backgrounds because 'you don't want to pry on people's lives'. Thus, also in parochial space, people make an effort to be civil towards diversity, expressed by a somewhat limited engagement with difference. As I discuss further below, this limited engagement can also be grounded in attempts to avoid tension and conflict. But even if differences are rarely talked about or engaged with, the existence of such spaces as the coffee morning facilitates contact across difference when needed. For example, the teacher who leads the coffee morning organised a cooking class for Turkish speakers and parents of Caribbean background in order to break down prejudice which some of the Turkish mothers held against Caribbean people because of gang violence in their area dominated by black male youngsters. The Turkish-speaking women had a much more differentiated view of the Caribbean mothers after the class than before.5 While the coffee morning is a more structured place of social interaction where contact goes beyond casual greetings, regular contact at the school gates or when dropping off children at nursery can similarly lead to more friendly and sustained relations across cultural differences (see also Jayaweera & Choudhury 2008, Wilson 2013 [forthcoming]). Sometimes, such relations can develop into important structures of mutual support. For example, a Turkish mother who was going through difficulties with her husband once asked Harriet, one of my British informants, whether she could look after her children for a night. Harriet, whom she knew from her children's school, was the only person whom she trusted enough to be able to provide support in a situation in which she did not want to call on her tight-knit family and kinship networks. Even if these two women are not close friends, the informal relations they have formed through the regular encounters at school provided this Turkish woman with an extra social resource for situations of emergency. Also, even if informal relations between parents do not go beyond these specific places of schools and often fade away once the children grow up, they contribute to a sense of being part of a community and being able to communicate with people who are different. This confirms the so-called 'contact hypothesis' (Allport 1958) developed in social psychology, according to which positive contact 'under cooperative interactive conditions' helps to change stereotypes and negative attitudes towards 'outgroup members' (Brewer & Gaertner 2001: 455). However, the emphasis, here, is on positive, and not all encounters in parochial space are necessarily of a positive nature. As mentioned earlier, civility towards diversity can also ensure boundaries. This was exemplified by a Congolese informant when describing relations with his neighbours. Christian, who has lived in the UK for 20 years, explains how you 'need to be diplomatic' when dealing with people of other backgrounds, both regarding good relations and in regard to ensuring boundaries. When I ask him about his neighbours, he tells me the following: That's another issue. Neighbours; you know we are different. We live in a place where left: British; right: Caribbean; on top: Asian; bottom: Turkish. Look at that scenario! So you need to be diplomatic you know, especially as a parent, you need to be diplomatic. (...) you need to understand the temper of the neighbour. I know my neighbour, every time he's angry, I just say hello to him, that's enough. Close the door finish (...). You need to be careful (...) to be open, but sometimes closed. Thus, although Christian says that he has good relations with his neighbours and generally with people of other backgrounds, he also describes how you have to be consciously civil towards people who are different. He says that politeness is one of the most important things to get on in this country. His description of dealing with difference by 'being open, but sometimes closed' aptly summarizes the fine balance between engaging with diversity, and keeping positive relations by way of avoiding contact. Thus, in the parochial realm where more regular encounters take place, dealing with difference is characterized by both avoidance and engagement. The more regular and sustained contact in the parochial realm enables people (at least those who choose to participate in interactions in such spaces) to ask each other questions when differences do come up in their conversations. Importantly, however, the bulk of conversations focus on commonalities rather than differences, with people attempting to create a shared understanding of the complexities that come with living in a super-diverse and continuously changing urban context. This focus on shared themes could also be interpreted as defining factor of conviviality. Furthermore, focussing on shared themes helps to avoid addressing --- Conclusion: Convivial relations and polite distance What are the differences in patterns of conviviality in the public and the parochial realm? While both realms are characterized by commonplace diversity, in the parochial realm cultural differences are acknowledged, while in the public realm, people treat everybody universally the same and somewhat ignore other people's differences. Lofland (1989) has also described this as indifference towards diversity. This indifference is exemplified by rather limited knowledge or curiosity about other residents' life-worlds and cultures. This also applies to the parochial realm where difference is rarely discussed, as exemplified by the parents' coffee morning. My informants' own interpretation of this was that asking people about their background might imply treating them like outsiders. Even if competent in communicating and interacting with people of various categorical groups, Hackney residents of all kinds of backgrounds often have very little idea about other residents' cultural backgrounds. They are perfectly comfortable with muddling through the neighbourhood in day-to-day life and somehow communicating with various types of people, but many of them know little about other people's ways of life. Neal et al. (2013:318) describe these'mundane competencies for living cultural differences' as 'cool conviviality' or 'light engagement'. Hence, while not talking about difference could be interpreted as a way to avoid tensions, it could also be explained with the existence of a general acceptance of people who are different and a sense that as long as people interact and are friendly, things are fine. Noble (2009) refers to the acceptance of people who are different as 'unpanicked multiculturalism', contrasting it with the 'panicked multiculturalism' which has dominated debates on cultural and religious diversity and which has focused on tensions and conflicts between different groups, reflected in the backlash against multiculturalism in public and policy discourses across Europe (Vertovec & Wessendorf 2010). Hackney is a prime example of such unpanicked multiculturalism. The lack of tensions I have observed in both public and parochial space could also ground in the fact of super-diversity itself, or in other words, that there is no one group dominating in the borough. A local councillor explained the reason why there was little resentment between different groups as follows: There are so many ethnicities in this borough that everybody is living on top of each other and knows all manner of ethnicities. There is not a mythical group out there that they can blame. He describes how 'it is impossible to have enemies' in this kind of mixed context. However, Hackney's unpanicked multiculturalism is manifest as a kind of superficial acceptance of different life-styles and cultures, sometimes paralleled by the appreciation of diversity, but it For Hackney's residents, many of whom are busy enough just getting by and dealing with their everyday concerns, paying particular attention to other people's cultural backgrounds and actively engaging with their differences might go beyond their capacity. If everybody around you is different, with whom do you start engaging about their difference? Van Leeuwen states that 'to argue that modern urbanites should be cosmopolitans by being "open" to cultural difference and by "celebrating" diversity might be too demanding given the state of many cities today' (van Leeuwen 2010: 635). In fact, it could be argued that 'indifference to ethnic or cultural differences might be a mode of "dealing with diversity"' (van Leeuwen 2010:639). Thus, not dealing with difference could also be interpreted as a way of avoiding conflict and tensions. This is also exemplified by the fact that parallel to numerous relations formed with people of other backgrounds in the parochial realm, these relations are seldom translated into private space. Especially the in-depth interviews, but also participant observation showed that despite positive relations across differences, people's private relations are often divided along ethnic and especially socio-economic lines. Despite an increasing number of people in Hackney who identify as'mixed' (6.4%, ONS 2013), this number is still relatively low compared with the rest of the population and the degree of ethnic diversity. While people live together in the public and parochial realm, they dwell apart when it comes to private relations (Wessendorf 2010;[forthcoming]). Importantly, none of my research participants have described this as a problem. Rather, it was described as normal that, as one of my informants put it,'similar people attract each other'. The example of Hackney shows how conviviality in public and parochial space can be paralleled by divisions in private space, but that these divisions do not necessarily present a problem. The existence of commonplace diversity and unpanicked multiculturalism shows how in a place like Hackney, cultural diversity is not the main issue of contestation. Rather, and as I show elsewhere, it is for example contestations over public space which can result in tensions between groups, be they defined by ethnicity, religion, life-style or other categories (Wessendorf 2013 [forthcoming]). Furthermore, and as demonstrated by the riots of August 2011, tensions in Hackney lie in social divides along generational and racial lines, with young black people forming the group against which the rest of the population, regardless of their own backgrounds or ethnicity, holds most prejudice. Rather than cultural differences, it is poverty and disadvantage which present the real challenges to a large part of Hackney's population. These challenges came to the fore during the riots, where all rules of civility broke down (Wessendorf [forthcoming]). Otherwise, however, they are rarely plaid out in public or parochial space where people treat each other with civility, independently of their backgrounds. The fact that among Hackney's residents, diversity has become commonplace points to the important role of processes of diversification over time and how people can get used to what is seen as exceptional demographic situation in other contexts. It also points to how people are able to adapt to their social surroundings, as exemplified by a Kurdish taxi driver who told me that when arriving in Hackney, he felt scared of Caribbean and African people, but over the years got used to their presence and now has very good relations with them. Although such positive talk about difference does not rule out the co-existence of privately held negative views, it shows that commonplace diversity and relaxed attitudes towards difference emerge over time and as a result of everyday lived conviviality.

The London Borough of Hackney is one of the most diverse places in the United Kingdom. It is characterized not only by a multiplicity of ethnic minorities, but also by differentiations in terms of migration histories, religions, and educational and economic backgrounds, both among long-term residents and newcomers. This paper attempts to describe how people negotiate social interactions in such a 'super-diverse' context. It develops the notion of 'commonplace diversity', referring to ethnic, religious, and linguistic diversity being experienced as a normal part of social life by local residents. This commonplace diversity has resulted in people acting with 'civility towards diversity'. While in public space, people mostly treat everybody universally the same without acknowledging differences, in semipublic spaces such as associations and local institutions, here conceptualized as 'parochial space' (Hunter 1985), people's different backgrounds are acknowledged and sometimes talked about. The article discusses how people negotiate their differences in these two different kinds of spaces. It shows how civility towards diversity is used as a strategy to both engage with difference as well as avoid deeper contact. Civility thus facilitates the negotiation of both positive relations and possible tensions. The London Borough of Hackney is one of the most diverse places in the UK. It is characterized not only by a multiplicity of ethnic minorities, but also by differentiations regarding migration histories, religions, and educational and economic backgrounds, both among long-term residents and newcomers. This article describes how people negotiate cultural differences in such a 'super-diverse' context (Vertovec 2007b). It describes how, due to the long history of diversification in Hackney, residents experience ethnic, religious and linguistic diversity as normal part of everyday life. I conceptualise this normalcy of diversity as 'commonplace diversity '. Gilroy (2004:xi) has described this process with the term

nightmares, appetite changes, and separation anxiety (Rossen & McKeever, 1996). Separation from the parents (primarily the mother), age of the child, and parental anxiety have been implicated as affecting the child's reaction after hospital discharge. However, most of this research was conducted prior to 1980 when maternal employment and, thus, experience with separation from the mother and alternate child care (routine care provided by adults other than the child's mother) at an early age was less common. According to Bowlby (1969Bowlby ( /1982)), children look to their primary attachment figures, generally their mothers, in times of stress, such as when they are frightened, tired, or ill. The stress of hospitalization for children is likely due to the fear caused by the child's separation from the mother and care by nonparental adults as well as the child's illness. Experience with separation from the mother and care by nonparental adults when the child is not ill may reduce the novelty and fear caused by hospitalization. Thus, experience with alternate child care could act as a buffer for the negative effects of hospitalization. Conversely, although research suggests that daily mother-child separations due to the mother's employment have no negative effects for children of any age (Hoffman, 1989), it also is possible that separation for employment is a low-level stressor that, when added to the major stressor of hospitalization, pushes the child's stress to an unmanageable level. In that case, return to alternate child care after hospitalization might prolong or exacerbate any negative effects of hospitalization. With 50% of single mothers with preschoolers in the workforce (U. S. Bureau of the Census, 1992) and the expected increase in this rate due to welfare reform efforts, more hospitalized children from single-parent families will have experience with nonmaternal care. However, the effects of experience with alternate child care on behavioral outcomes for hospitalized children are unknown. --- Relevant Literature Research on children's reactions after hospitalization clearly indicates that hospitalization is a stressful event for children and that negative effects can continue for months or years after discharge. In the seminal study by Prugh, Staub, Sands, Kirschbaum, and Lenihan (1953), a sizable proportion of preschool and school-age children who experienced "routine" ward conditions had "significant" negative reactions immediately after discharge (92%) and at 3 months (58%). Most subsequent studies have found posthospitalization upset at 1 to 4 weeks after discharge (Brain & Maclay, 1968;Couture, 1976;Goslin, 1979;Mahaffy, 1965;Melnyk, 1994;Rossen & McKeever, 1996;Sides, 1977;Sipowicz & Vernon, 1965;Wolfer & Visintainer, 1975, 1979). However, two other studies have found normal behavior in children after hospitalization (Carson, Council, & Gravley, 1991;Simons, Bradshaw, & Silva, 1980). Age of the child frequently has been related to posthospitalization upset, with younger children having more severe negative reactions than older children (Brain & Maclay, 1968;Sides, 1977;Simons et al., 1980). Wolfer andVisintainer (1975, 1979) found no age effects in their first experiment, but saw age-by-treatment effects in a later experiment. Tiedeman and Clatworthy (1990) found that older children (8 to 11 years old) experienced a decrease in anxiety through 7 to 14 days after hospital discharge, but 5-to 8-year-old children did not. Several investigators have found that posthospitalization behavior of preschool or school-age children was not related to the number of previous hospital admissions (Menke, 1981;Simons et al., 1980;Tiedeman & Clatworthy, 1990). However, Elander, Nilsson, and Lindberg (1986) found that 4-year-old children with a history of two or more hospitalizations displayed more somatic symptoms; oral tendencies; and dependent, aggressive, and hyperactive behaviors than children with only one hospitalization. Findings regarding the effect of length of hospital stay also vary. Although Menke and Simons et al. found no relationship between length of stay and child behavior, Tiedeman and Clatworthy found greater anxiety in 5-to 11year-old children with shorter lengths of stay. Two studies found significant differences in behavior and coping strategies based on experience with separation. Elander et al. (1986) found an interaction effect of daycare attendance and hospitalization on behavior of 4-year-old Swedish children. Daycare children who had been hospitalized sometime since birth displayed more aggressive, hyperactive/ impulsive, dependent, and regressive behaviors than daycare children who had not been hospitalized. Children had these reactions even though 116 (50.2%) of the 231 admissions had lengths of stay of 3 or fewer days; 35 (15.2%) of these were overnight stays. Children experience separation when they start school, around the age of 5 or 6 years. In a sample of 3-to 7-year-olds (N = 24), Corbo-Richert (1994) found that older children and those with school experience used more information-seeking coping strategies to deal with chest tube removal than younger children and those without school experience; previous experience with hospitalization did not influence choice of coping strategies. However, it is not clear whether these results are due to age of the child or to the effects of separation from having entered elementary school. In summary, the negative behavioral effects of hospitalization for children are well-known. However, whether experience with alternate child care and, therefore, routine separation of the child from the mother, has a moderating effect on the relationship between hospitalization and child behavior is not known. A sample of preterm and full-term preschoolers living in femaleheaded single-parent families from a larger, longitudinal study on the effects of maternal employment for preschoolers in single-and two-parent families (Youngblut, Singer, & Rodgers, 1997) was used to examine this question for several reasons. Single mothers are twice as likely to have a preterm birth than married mothers (National Center for Health Statistics, 1995). Research suggests that preterm children are more negatively affected by environmental stressors (Tobey & Schraeder, 1990). Preterm infants and children are more likely to experience hospitalization than full-terms. The research question for this secondary analysis was: Does experience with separation from the mother due to alternate child care moderate the negative effects of hospitalization on preschool children's behavior? Thus, the purpose of this study was to examine the interaction effects of alternate child care experience on the relationship between hospitalization and behavior of preschool children living in female-headed single-parent families. --- Method Sample The sample for this study consisted of 121 female-headed, single-parent families, with 60 preterm and 61 full-term preschoolers (3, 4, or 5 years old). Families with preterm preschoolers were identified from the admission records of three Level III neonatal intensive care units (NICUs); families with full-term preschoolers were identified from birth records of two normal newborn nurseries. A systematic random sample of families with full-terms and all families with preterms from four consecutive birth years were sent a letter that briefly described the study. Interviews were conducted within a month of the child's third, fourth, or fifth birthday. Single-parent families were eligible to participate if the mother had not lived with a man in the father role for at least 6 months prior to recruitment. Five index children were being raised by a single woman other than the birth mother (1 by adoptive mother, 2 by grandmothers, and 2 by foster mothers). In each case, the child had lived with the family for most of the child's life and the woman was performing the role of mother to the child. Thus, these women are referred to here as "mothers." Inclusion criteria for the preterm child in the family were: (a) born prior to 36 weeks gestation weighing less than 2,500 grams; (b) appropriate weight for gestational age; and (c) hospitalized for at least 1 week in a Level III NICU. Inclusion criteria for the full-term child in the family were: (a) birth between 38 and 42 weeks gestation; (b) discharged home with the mother after birth; and (c) no preterm siblings that were born within 10 years of the study child's birth. Children whose mothers were told the child had more than a 2-year developmental delay were excluded. None of the families contacted were excluded on this basis. Of the families contacted and eligible to participate in the larger study, 71% agreed to participate. Sample demographics are listed in Table 1. The sample had approximately equal numbers of male and female children. Birthweight and gestational age of the preterm and full-term groups reflected study inclusion criteria. In the preterm group, there were 25 (41.7%) low birth weight (LBW; 1,500-2,500 gms), 23 (38.3%) very LBW (1,000-1,499 gms), and 12 (20%) extremely LBW (<unk>1,000 gms) children. Average length of stay in the NICU at birth was 46.1 days (SD = 33.34; range, 4 to 128). Few had experienced complications of prematurity. Thirteen had had an intraventricular hemorrhage (10 with Grade I, and one child each with Grades II, III, and IV). Two children had been diagnosed with cerebral palsy and two with bronchopulmonary dysplasia (BPD). At the time of the study, one child with BPD was using an aerosol machine for asthma, and the other child with BPD did not require any special equipment. Most of the mothers were African American (66.1%), never married (70.2%), high school graduates (76.9%), and living in poverty (68.6%). Number of children in the family ranged from 1 to 8, including the study child (M = 2.5, SD = 1.44). --- Instruments Child behavior-The Child Behavior Checklist (CBCL) (Achenbach, 1991) was used to measure child behavior. It contains two behavior problems scales (internalizing, externalizing). The internalizing scale contains three subscales: anxious/depressed; somatic complaints; and withdrawn. The externalizing scale contains two subscales: aggressive and delinquent behavior. Mothers rated their child on each of 118 behaviors on a 3-point scale ranging from not true to often true. Scale and subscale scores were obtained by summing the mothers' ratings. Normed scores, based on child age and gender, then were obtained with the Revised Child Behavior Profile sheet. The CBCL has been widely used with a variety of populations, including LBW children and African American children, with documented evidence for its construct validity (Achenbach). Achenbach reports moderate to strong correlations between the CBCL's internalizing and externalizing scales and similar scales of the Quay-Peterson (. 52-.88). Achenbach reports test-retest reliabilities of.82 to.95 and internal consistency reliability estimates of.90 and.93 for the internalizing and externalizing scales, respectively. For this sample, the alpha coefficients were.83 and.91 for the internalizing and externalizing scales, respectively, and ranged from.58 to.89 for the subscales. Experience with hospitalization and alternate child care-Mothers indicated when the index child was in the hospital and the lengths of stay for each hospitalization on a Life History Calendar (LHC) (Freedman, Thornton, Camburn, Alwin, & Young-DeMarco, 1988). From that data, number of hospitalizations, total number of days hospitalized, and age at last hospitalization were derived. Additionally, mothers used the LHC to indicate the child's alternate child care experiences, including timing, duration, and types of alternate child care providers. Alternate child care experiences were defined as situations where routine child care was provided by adults other than the child's parent(s), primarily for the mother's employment. Episodic baby-sitting to allow the parent(s) to go out for a few hours were not considered alternate child care experiences. From these data, number of different child care arrangements and proportion of the child's life in child care were derived. In addition to these interval-level variables, two dichotomous variables also were created: experience with hospitalization (yes/ no) and experience with alternate child care (yes/no). Validity and reliability of the LHC is enhanced through its use of memory cues, relating one event to other events that occurred at about the same time. When Freedman et al. (1988) compared data obtained in 1980 about the respondent's current situation with data obtained retrospectively with the LHC in 1985 (N = 900), agreement ranged from 72% to 92%. The LHC constructed for the current study contained five segments (years), each with 12 blocks (months). Major life events, such as residential moves, births, and deaths also were recorded to aid the mother's memory. Recording began with the month and year of the study child's birth. --- Procedure The study was approved by the appropriate human subjects review committees at the university and each of the three hospitals. Interviewers were trained to administer a set of parent interview and self-complete instruments and developmental tests for the larger study. Within 2 weeks after an introductory letter was mailed, interviewers contacted each family to screen for inclusion criteria, answer questions, and schedule a home visit for data collection. Interviewers offered to read self-complete instruments to parents; parents and children were encouraged to take short breaks as necessary. --- Results More children experienced alternate child care (n = 67) than hospitalization (n = 35). Children with alternate child care experience had spent an average of 28.5 months (SD = 17.2) in up to four different arrangements (M = 1.37, SD =.62). Types of child care arrangements included care in the home by relatives (n = 30, 25.2%) and nonrelatives (n = 8, 6.7%), family daycare (n = 13, 10.8%), and an organized daycare facility (n = 39, 32.8%). Number of hospital admissions ranged from 1 to 18 (M = 1.9, SD = 2.90), days in hospital ranged from 1 to 355 (M = 30.8, SD = 66.78), and time since last admission ranged from 3.4 to 59.7 months (M = 33.7, SD = 18.57) for those who had been hospitalized. Children who had experienced only hospitalization did not differ from children who had both hospital and child care experiences on number of admissions or number of days hospitalized, t(33) =.79 and t(31) = 1.19, respectively, p = NS. However, children with only hospital experience had been hospitalized more recently than children with both hospital and alternate child care experience (months since last admission; M = 26.0, SD = 16.81 vs. M = 39.5, SD = 18.08, respectively), t(33) = 2.25, p =.03. Gestational status (preterm vs. full-term) was related to child care-by-hospital experience group (hospital only, child care only, neither or both hospital and child care), chi square (3, N = 119) = 15.27, p <unk>.01, primarily because more preterm than full-term preschoolers had been hospitalized (27 vs. 8). The numbers of preterm and full-term preschoolers who had experienced alternate child care were similar (34 vs. 32). Number of different child care arrangements and months spent in alternate child care were similar for preterms and full-terms, t(118) =.88 and.92, respectively, p = NS. However, preterm preschoolers experienced a greater number of hospitalizations and spent more days in the hospital than full-terms, t(60.5) = 2.63 and t(57.3) = 2.27, respectively, p <unk>.05. Separate variance estimates of the t values were used for these two comparisons due to unequal group variances. Time since last hospitalization was similar for preterm and full-term preschoolers, t(34) = 1.50, p = NS. Two-way ANOVA 1 was used to examine the interaction (moderating) effects of experience with hospitalization and child care on the subscales of the CBCL internalizing and externalizing behavior problems scales, controlling for gestational status and race. The interaction effect of experience with alternate child care and hospitalization was significant for somatic complaints, F(1, 105) = 8.40, p =.005, and aggressive behaviors, F(1, 105) = 5.70, p =.02. Children who had experienced hospitalization but not alternate child care had higher somatic complaints than other children (Figure 1). Children who had experienced both hospitalization and alternate child care had lower aggressive behavior scores than other children (Figure 2). To further explore the effects of hospitalization, additional analyses were conducted with data collected from the children who had experienced a hospital stay. Correlations between child behavior subscales and aspects of the child's experience with hospitalization (number of admissions, total number of days hospitalized, age at last admission, and months since last admission) and child care (proportion of child's life in alternate care) were examined (Table 2). More recent hospitalization and older age at last hospitalization were related to more aggressive behaviors. However, greater proportion of the child's life in alternate child care was associated with fewer aggressive behaviors and somatic complaints. --- Discussion In this study, experience with alternate child care and hospitalization had an interaction effect on behavior problems of preschool children regardless of the child's gestational status at birth. Elander et al. (1986) also found an interaction effect of these two factors. Children in our study who had a history of hospitalization had more somatic complaints. Since these children also had been hospitalized more recently than those with both hospital and child care experiences, this finding could reflect the effects of the child's more recent illness or an underlying chronic illness. However, because children with both experiences had scores similar to those who had not been hospitalized at all, this finding may reflect a protective effect of experience with alternate child care. That is, experience with child care and, therefore, experience with routine separations from the mother may decrease the negative effects of hospitalization on the child. Preschool children who had experienced both hospitalization and alternate child care displayed fewer aggressive behaviors than other children in our study. In contrast, Elander et al. (1986) found that 4-year-old daycare children with a history of hospitalization displayed more aggressive behaviors than daycare children without a history of hospitalization. The effects for aggressive behaviors and somatic complaints are in the same direction, although the mean for children with hospital experience only is similar to the mean for children without a history of hospitalization regardless of child care status. Thus, something about having both experiences makes the child less aggressive. Although effects of hospitalization on preschool children's behaviors were not related to number of hospitalizations or length of hospital stay, timing of the last hospitalization was important. Children whose last hospitalization was at a younger age and less recent displayed less aggressive behavior. Menke (1981), Simons et al. (1980), and Tiedeman and Clatworthy (1990) also found no effect of number of admissions or length of stay, although Elander et al. (1986) found more behavioral dysfunction in Swedish children with more hospitalizations. In our study, experience with alternate child care again appears to be protective. For children who had been hospitalized, amount of aggressive behavior and somatic complaints decreased as the proportion of the child's life in alternate child care increased. Thus, greater experience with 1 An alternate method for testing the interaction effects is to use multiple regression. However, the creation of interaction terms with variables that are not continuous for use in regression analysis is somewhat controversial. Violation of regression assumptions would invalidate the results; therefore, we have chosen to use two-way ANOVA because it assumes that the independent variables are categorical. alternate child care may ameliorate some of the negative psychological effects of hospitalization. A limitation of this study is that effects of different types of alternate child care situations as well as the quality of these child care arrangements were not considered. The emphasis of the programming in organized daycare has been found to improve cognitive performance, especially for children from low-income families (Caughy, DiPietro, Strobino, 1994). Quality of the child care arrangement has been found to affect cognitive development (Burchinal, Roberts, Nabors, & Bryant, 1996;Pierrehumbert, Ramstein, Karmaniola, & Halfon, 1996). However, research on the effects of quality of child care on the behavior of children from lowincome families is lacking. In summary, preterm preschoolers were more likely to have been hospitalized than full-term preschoolers; however, they were equally likely to have experienced alternate child care. Experience with alternate child care had a protective effect for preschoolers who also had been hospitalized, regardless of gestational status. For those with hospital experience, number of hospitalizations and duration of stay were not related to child behaviors; however, children showed less aggressive behavior as time since last hospitalization increased. Although more research in this area is needed, we speculate that experience with routine separation from the parent for child care diminishes the negative effect of separation from parents during hospitalization and perhaps other types of separation. Aggressive behaviors for children with and without child care experience.

Background-With more single mothers entering the workforce due to welfare reform efforts, more hospitalized children from single-parent families will have experienced alternate child care arrangements where routine care is provided by adults other than the child's mother. Objectives-To investigate with secondary analysis of data whether experience with alternate child care has a moderating effect on the relationship between hospitalization and behavior of preschool children living in female-headed single-parent families. Method-A sample of 60 preterm and 61 full-term children who were 3,4, or 5 years old was recruited for the larger longitudinal study. Behavior problems were measured with the Child Behavior Checklist. History of hospitalization and alternate child care arrangements were measured with the Life History Calendar. Results-Preschool children who experienced hospitalization without alternate child care experience had more somatic complaints, but those with both hospital and alternate child care experience had fewer aggressive behaviors than other children. For children with a history of hospitalization, aggressive behaviors decreased as the proportion of the child's life in alternate child care increased.with alternate child care may ameliorate some of the negative effects of hospitalization, and potentially other novel and negative experiences, for preschool children. This could be due to child care providing positive experiences with separation from the mother, a peer group with which to talk about the novel experience, or actual instruction about the novel experience.child behavior; child care; hospitalization; single-parent families Hospitalization of a child is a stressful event for both parent and child. In his review of the research, Thompson (1985) concluded that children have negative reactions to hospitalization, including night terrors, feeding disturbances, loss of newly acquired developmental skills, and intolerance of separation from the mother, that persist for months after the child's discharge. Despite efforts to make the hospital more child-and family-focused, preschoolers in a recent study also demonstrated distress 1 week after hospital discharge, including irritability,

Résumé Alors que les savoirs autochtones et la biomédecine se rejoignent dans les soins de santé aujourd'hui, il est essentiel d'intégrer cette rencontre dans la formation des médecins afin d'améliorer les issues de santé et de renforcer l'identité culturelle des peuples autochtones. <unk> l'aide d'exemples tirés de la littérature et des méthodes pédagogiques et pratiques employées à l'Université de Toronto, la présente revue narrative vise à offrir aux éducateurs des conseils pratiques sur l'intégration de ce sujet dans leur programme. La méthodologie utilisée pour la recherche de littérature scientifique comprenait une consultation de la base de données informatique Medline. Quant aux informations sur le programme d'études de la Faculté de médecine de l'Université de Toronto, elles ont été obtenues par le biais du schéma tutoriel officiel et du Bureau de la formation médicale autochtone de l'université. Nous présentons six recommandations pour un enseignement réussi du croisement des savoirs autochtones et de la biomédecine dans le cadre d'un programme culturellement sécuritaire sur la santé des Autochtones. --- Introduction Indigenous healing practices have been an important part of Indigenous knowledges and ways of knowing that have existed for thousands of years. Colonization, forced assimilation, and racism have all contributed to these practices being split off from Indigenous healthcare. 1 Consequently, this has further impacted health disparities seen among some Indigenous communities today. 1 An important aspect of restoring these knowledges is the promotion and reclamation of Indigenous healing, which includes making space for Indigenous practices and knowledges alongside biomedicine. 2 Ways of making space take different forms in various programs -from including Elders in treatment care planning, to taking communitybased approaches that enable communities to set and lead their own health priorities. 3 Ultimately, these programs which integrate Indigenous knowledges with biomedicine improve health outcomes and strengthen cultural identity among Indigenous Peoples. 2 Therefore, instead of learning about these knowledge systems separately, healthcare providers will benefit from learning about the convergence of Indigenous knowledges together with biomedicine during their training. 2 In fact, there is a willingness and motivation to learn about Indigenous healing practices among healthcare professionals. 4 However, there are misguided practices to navigate in order to make these teachings successful, and to avoid furthering the racism or inequities that we seek to redress. In addition to the knowledge we seek to disseminate, we also need to consider the pedagogical approach, and ensure critical reflection on the history of harmful ideas and practices that can occur in teaching about the health of Indigenous populations. Some of these practices include: presenting only health disparities and risks, while omitting discussion of Indigenous community knowledge and strengths; creating curricula without Indigenous Peoples guiding, leading, and being educators in this process; 5 or taking a "one size fits all" approach that fails to see the heterogeneity of Indigenous Peoples. 6 Instead of striving to teach everything there is to know about complex and often privileged knowledge systems, it is far better to model cultural humility. 7 Moreover, we and others believe both Indigenous and non-Indigenous knowledge systems must be equally valued for the teachings to be meaningful. 8 With these considerations as a foundation, here are six recommendations to help introduce medical trainees to Indigenous knowledges, medicine and healing practices. --- Use multiple teaching methods to introduce Indigenous knowledges. According to various Indigenous leaders and patients, a way to develop respect and equal valuing of both Indigenous and non-Indigenous knowledge systems is by providing education and awareness to health professionals through training. 9 This education should be provided using multiple teaching methods. This means not only providing lectures or modules, but also using experiential methods to better understand Indigenous worldviews, cultural practices, and traditional ways of knowing. 10 In fact, a M<unk>ori study which looked at Indigenous teachings in the medical curriculum, found that students enjoyed multiple teaching methods since it helped them engage with the content more and encouraged high levels of interaction between each other. 11 These multiple methods included lecture style teaching, small group discussions, participating in simulated patient interviews, and workshops. 11 Furthermore, an example of how to use multiple teaching methods can be seen with what is done at the University of Toronto medical school. While there are lectures, modules, and case-based learning sessions related to Indigenous health, every student participates in a KAIROS Blanket Exercise as a way to experientially learn about Indigenous histories and explore the nation-tonation relationship between Indigenous and non-Indigenous Peoples in Canada. 12 --- Encourage self-reflection and cultural humility among learners Cultural humility is a process of self-reflection to understand systemic and personal conditioned biases. 7 It is a way for learners to acknowledge oneself as a perpetual learner when it comes to understanding another's experiences. Indigenous leaders have been encouraging the use of the concept of cultural humility, instead of 'cultural competency', because a person can never fully achieve cultural competency through training or a course. 7 Instead, learning about another person's culture is a life-long learning process. Healthcare providers should be encouraged to acknowledge the lifelong journey of learning and self-reflection, while being open to listening to what better care means for Indigenous Peoples. An example of encouraging the concept of cultural humility in a learning environment is what is done during the KAIROS Blanket exercise. This includes ensuring there is space for students to ask questions, to talk about difficult topics including racism, and have a moment during the teaching for self-reflection. 12 Ultimately, cultural humility is a vital part of creating a healthcare system which is culturally safe for all. 7 --- Invite Elders and other Indigenous community members to guide, facilitate, and/or teach Teachings involving the convergence of Indigenous knowledges and biomedicine, should involve Indigenous educators, as they are more knowledgeable about the cultures, histories, and worldviews of Indigenous Peoples. Even if it is not possible to have these educators directly teach, having an Elder or other Indigenous community member, oversee how and what is being taught is important to ensure the content is presented well without any negative bias. 13 In fact, the Office of Indigenous Medical Education at the University of Toronto created a Truth and Reconciliation Report (TRC) Reading Group, where an Elder was involved in the planning, initiation, and facilitation process of three small group sessions. With only an idea in mind, of having a TRC Reading Group, the medical student who helped organize the event, met and communicated over email with the Elder multiple times to aid in planning and initiating the group. The Elder provided guidance in what the sessions would entail, including the idea that there should be smudging practices done to commence each session. 14 Moreover, it is important when having an Elder partake in teachings, to ensure the honoraria provided to the Elder is respectful and appropriate for their knowledges provided. 14 Indeed, Elders, Traditional Teachers, and Knowledge Keepers are now a vital presence at UofT, and the university encourages staff, students and faculty to engage and grow in connection with them. 15 --- Consider the setting and learning environment Although lecture halls can be used as a teaching environment, learning about Indigenous histories and knowledges can be more meaningful in other environments. For example, if the teachings involve traditional practices like smudging, it is important to find a room that supports this. Or maybe you want to find a room in a building which is significant for Indigenous Peoples, like a sweat lodge or Indigenous Gathering Place or even a university building meant to support Indigenous students. With consultation of Indigenous educators, finding a learning environment different from a lecture hall is possible, and can positively impact the teachings for the students. In fact, the M<unk>ori study looking at the Indigenous health curriculum for medical students found learning at a marae (Indigenous meeting place) motivated the students to engage with the course content more deeply. 11 This learning environment also allowed students to experience specific traditional protocols, making their learning more meaningful. 11 Likewise, the TRC Reading Group done at the University of Toronto considered the learning environment when choosing where to conduct the sessions. In consultation with the Elder, two rooms were found which would allow for a smudging practice, which allowed for an enhanced learning environment. 14 The two rooms used were the Labyrinth room in Women's College Hospital which is meant as a quiet prayer/reflection room, and the lounge in First Nations House which is a place that provides culturally supportive student services to Indigenous students at the university. 14 --- Cultivate the perspective of holism, and teach a relationship-based approach to making space for Indigenous knowledges and practices within biomedicine While it is vital to think about how to best teach students from a learning environment and teaching methods perspective, it is also important to think about the content. After reviewing the literature on programs which converge Indigenous knowledges with biomedicine, here is what students should know about the key factors to successfully doing so: • Build upon holistic models of wellness in the creation of programs that bring together Indigenous knowledges with biomedicine, while working with local Indigenous organizations, leaders and clients. Although there is great diversity amongst Indigenous nations and cultures, many share an approach to health that is wholistic. This includes spiritual, emotional, physical, and mental dimensions of life, including connections to the land and other non-human aspects of the world. 16,17,18 • Ensure that communities who are the intended recipients of health care participate in all stages of the development and implementation of integrated Indigenous and biomedical programs. Recruit, mentor and support program leaders and practitioners from within local Indigenous communities. 19,20 • Build and nurture strong partnerships, between Indigenous and non-Indigenous peoples, as the foundation for creating and maintaining integrated programs. This process of 'two-way' learning allows for sharing of knowledge, while maintaining integrity and respect of individual expertise from each cultural domain. 8 • Build time and space when integrating Indigenous knowledge with biomedicine for individuals to focus on building trusting relationships. 19,21 While partnerships are important, the process of developing collaboration through relationshipbuilding is equally vital. Relationship building is an essential process to many Indigenous communities and is deeply embedded in history and context. 22 6. Teach the concept of wise practices as a way to bring Indigenous knowledges and practices alongside biomedicine in a respectful and meaningful way. Wise practices are locally-appropriate principles, actions, tools, or decisions that contribute significantly to the development of equitable and sustainable social conditions. 23 An approach which values wise practices is especially important in healthcare when considering Indigenous knowledges and practices in the context of biomedicine. 23 Therefore, this is an important concept for medical trainees to understand before going into practice. In contrast to the best practices approach, which creates a knowledge hierarchy and applies a proven method or process to many communities, wise practices acknowledges the heterogeneity of Indigenous peoples. 23 Failure to recognize this diversity and nation-specific knowledges and practices, can lead to pan-Indigenous approaches that over simplify and/or overgeneralize knowledge developed in one community to apply in other communities. 23 Indeed, each community has their own cultural background, knowledges, social practices, and unique experiences of colonization, and therefore generalizing practices into other contexts should not be the answer. 24 In fact, viewing Indigenous culture as a monoculture can reduce the effectiveness of Indigenous and mainstream health partnerships. 6 Ultimately, healthcare professionals need to understand the core of this approach as a relational covenant of respect, and honor the autonomy and sovereignty of Indigenous peoples and communities. --- Conclusion Since medical schools across the country and world have different curricula, class sizes, and resources, teaching on the convergence of Indigenous practices with biomedicine will vary and have its own challenges. For example, a school may not have the funding to support an Elder to facilitate teachings, or a learning environment which would make Indigenous teachings meaningful. However, an approach based on the principles of cultural humility, holism, and the existence of a diversity of knowledges and practices, will create a culturally-safe Indigenous health curriculum in which wise practices can flourish. This flourishing though cannot simply be imparted on an individual learner basis. Instead, it will only thrive in institutional and social settings that also value Indigenous knowledges, and in health profession faculties that recruit and sustain students and teachers from diverse Indigenous backgrounds. Ultimately, these recommendations are based within the specific social location of UofT, and future research can include ways in which different social locations successfully involve Indigenous knowledges with biomedicine in their medical school curriculum. --- Note from authors: As authors we recognize there are other examples of Indigenous Health frameworks in medical schools across Canada. However, we have chosen to focus on the UofT framework because of our specific social location of having developed and experienced the Indigenous health curriculum. Specifying social location is vital to the context of Indigenous Peoples, epistemology, and education. Therefore, we would like to specify our social location as a medical student, who experienced the curriculum, an Anishnawbe physician, and a non-Indigenous physician who has worked extensively in the field of Indigenous health. --- Conflicts of Interest: None Funding: From the Canadian Federation of Medical Students (CFMS) Indigenous Mental Wellness Summer Studentship. These funds were used to support the medical student for their research work commitment, and to aid with transport costs needed to complete the research

As Indigenous knowledges and biomedicine come together in healthcare today, to improve health outcomes and strengthen cultural identity among Indigenous Peoples, it is vital for physicians to learn about this convergence during their training. This narrative review article aims to provide practical advice for educators when implementing teaching regarding this topic, using examples from the research literature, and pedagogical and practice-based methods used at the University of Toronto (UofT). The methodology on obtaining the research literature included a search of a computer database called Medline. Moreover, the medical school curriculum information specific to UofT, was obtained through the formal curriculum map and UofT's Office of Indigenous Medical Education. The following six recommendations provide a way to successfully implement the teachings on Indigenous knowledges and biomedicine, within a culturally-safe Indigenous health curriculum.

Background --- Information and Communication Technology Use by Older Adults Using information and communication technologies (ICTs), for example, smartphone, may offer older adult users numerous advantages. The use of a software application that provides access to information and the opportunity for communication can reduce social isolation and loneliness (Czaja, Boot, Charness, Rogers, & Sharit, 2017). People in need of care can benefit, as ICT can support them in their everyday lives (Mollenkopf, Kloé, Olbermann, Klumpp, & Redecker, 2010). However, there are some prerequisites for using such technologies. The perceived usefulness and ease of use of these ICTs are significant predictors of the intention to use them (Davis, 1989;Melenhorst, 2002). The willingness to use them may also depend on the anticipated effort (Cornet, 2014). Internet access is required, together with Internet-ready devices; relevant ICT knowledge is also necessary (Morrell, Mayhorn, & Bennett, 2000). Financial resources, education, and a supportive social network, all play an important role in ensuring access to ICT and acquiring the knowledge necessary for dealing with it. As numerous studies have demonstrated, digital inequalities exist (e.g., Dutton, Blank, & Groselj, 2013; U.S. Department of Commerce, 2016); people who are able to participate more confidently and productively in Internetbased activities benefit from them, whereas others are left behind (Zillien & Hargittai, 2009). The latter group are likely to have a low income, a lower level of education, be older, have limitations on their (instrumental) activities of daily living, and are likely to belong to an ethnic minority (Choi & Dinitto, 2013;Werner, Carlson, Jordan-Marsh, & Clark, 2011). This also applies to the group of older adults (Gell, Rosenberg, Demiris, LaCroix, & Patel, 2015). --- Needs-Based Technology Design In addition to improving access to ICT and implementing measures for increasing media competences, a form of technology that meets users' demands and offers users a recognizable added value is required. This is particularly important for older cohorts, who generally have less experience of ICT (Melenhorst, 2002;Venkatesh, Morris, Davis, & Davis, 2003). As a result, standardized, user-centered design approaches in technology development processes (ISO, 2010) have been established together with the "persona" method. A persona refers to a fictitious character who represents the typical features of a user group (Pruitt & Adlin, 2006). However, the use of personas does not guarantee that the designer will meet the target group's demands. Although personas are sometimes based on representative survey data (e.g., Wöckl et al., 2012), they contradict the participatory design approach that has become more significant during recent decades, including in the field of gerontechnology (e.g., Swallow, Petrie, Power, Lewis, & Edwards, 2016;Verhoeven, Cremers, Schoone, & van Dijk, 2016). The participatory design approach allows potential users to be specifically involved in the design process, with the goal of designing a needs-based and easy-to-operate technical solution. Technical developers and potential users meet on an equal footing, and users hold a significant decision-making power (Muller, 2003). Even the participatory design approach has its limits. Users who participate in the development of (geron-)technology are not representative of older adults' heterogeneity; the participants are frequently younger seniors, who enjoy a higher socioeconomic status and who, because of their previous technological experience, are more open-minded toward modern technology (Künemund & Tanschus, 2014). As a consequence, there is a risk that the real needs of the target group will be overlooked, which exacerbates digital inequalities: people who are not technophiles or who are socially disadvantaged will not recognize any added value in the technology being developed. Ultimately the people benefiting from this technological development are those older adults who are already privileged and interested. Precisely this problem arose during the first months of a digital district platform's participatory development process. The platform is part of a project funded by the German Federal Ministry of Education and Research, "Older adults as co-producers of district networks in the Ruhr region (QuartiersNETZ)," which is being cooperatively developed with older residents of a town in the Ruhr area. The district platform was supposed to provide networking, information, and support services for older citizens. The development was organized using open working group meetings. The working groups' participants were recruited in community meetings held every 3 months per district, in which people with different sociodemographic characteristics took part and where the idea and the opportunity of participating were promoted. However, as shown by the first evaluation results, some groups that could have benefited the most from such a digital district platform (e.g., people with limited mobility, low socioeconomic status, migration background, or very old single people) did not participate or hardly participated in the platform development working groups. To prevent the dangers arising from selection and exclusion mechanisms, and to design the digital district platform in such a way that residents with all kinds of living situations are able to recognize benefits for themselves, we have developed and implemented an interdisciplinary, methodical approach to complement the other participatory design formats. In doing so, the digital platform requirements of as many as possible different groups of older residents were surveyed and included in the development. Our approach can be assigned to the broad interdisciplinary area of User Experience Research (Law, Roto, Hassenzahl, Vermeeren, & Kort, 2009), which has become increasingly important in recent years (Hassenzahl, Diefenbach, & Göritz, 2010). --- Theoretical Framework In our investigation, we combine the concept of the life situation (regarding it as a significant context) with the perspective of a subject-oriented sociology of everyday life (Jurczyk, Voß, & Weihrich, 2016). This causes us to focus on how older adults go about their respective everyday lives, while simultaneously taking into account how the life situation's structures may have an impact on the different individual lifestyles (Barkholdt, 2001). In our opinion, the combination of aforementioned concept and perspective ensures a higher, analytical potential in respect of those patterns of perception and action, which are considered less in the life situation concept. At the same time, we would like to point out that a purely subject-oriented view would lead to sociopolitical implications being lost. The term "life situation" refers to the set of conditions (multidimensionality) under which people live, in material, nonmaterial, and objective as well as subjective terms (Voges, Jürgens, Mauer, & Meyer, 2003). These can be demonstrated by the following dimensions: education, income and wealth, employment, health, habitation, social networks, and family constellation (Voges et al., 2003). The life situation concept allows social inequalities to be considered, and, as a consequence, the available scopes of action to be widened or narrowed (Nahnsen, 1975). In terms of the use of technology, this means that some groups lack the necessary financial means for installing and covering the running costs of an Internet connection, or for purchasing Internet-ready devices, and they lack the support of a social network and the experiential scope for acquiring the necessary knowledge about how to deal with ICT. In addition, education is necessary so that information sources can be properly evaluated (Kim & Kim, 2001). Moreover, we follow the sociological understanding of everyday life (Projektgruppe "Alltägliche Lebensführung," 1995) for the qualitative deepening of life situations. In this case, everyday life means "common or normal" practices and "usual and routine" activities, as opposed to conscious actions. On the one hand, our particular interest concerns the challenges confronting older people in everyday life, for which they have developed their own specific procedures over time and thereby have come to rely on certain resources. On the other hand, we are also interested in those problems that cannot be overcome and therefore lead to a, subjectively perceived, limitation of everyday life. Subject-oriented sociology also refers to problem areas created by societal structures (e.g., loss of opportunities for encounters in the neighborhood) or due to events (e.g., wars and resulting traumas; cf. Projektgruppe "Alltägliche Lebensführung," 1995). The person-environment interaction and the adaptation processes associated with aging are further orientation points in our study (Mollenkopf, Oswald, Wahl, & Zimber, 2004). Thus, barriers and environmental enabling structures have a crucial impact on how older persons organize their everyday lives. In summary, the study combines aspects of the life situation, the interaction between people and the structures and barriers resulting from their residential environment, as well as their everyday life in their homes, in the data collection and analysis of problems and resources and for tailoring specific solutions to the demands. This is all the more important because social inequalities appear to be intensified through digital transformation (Robinson et al., 2015). In order not to exacerbate these inequalities, it is necessary for technology development to take into account both the person's life situation (e.g., financial circumstances, health conditions) and how they organize their everyday lives. We are aware that technical applications alone, regardless of how well they fit into a person's habits, cannot solve inequalities (Damodaran & Sandhu, 2016). But technology development should at least ensure that they are not exacerbated further. --- Study Design We chose a sequential design to optimize the needs-based technical design, by combining quantitative and qualitative methods (see Figure 1). In this article, we refer to the first three steps of the design that are presented here: First, a cluster analysis was performed to identify the types of life situations in the model region. Second, we performed an interview study with participants who represented these life situations, focusing on their everyday lives. Third, we developed suggestions for possible technical solutions, within the contexts of both the life situation and everyday life. All the stages of the qualitative interview study, from planning to implementation to analysis, were carried out with close interdisciplinary cooperation between social science, gerontology, and information science. The two last steps are still pending and will be reported in the future. In the fourth step, we will present the proposals for technical solutions to the participants who will have the opportunity of evaluating them. This will also happen in interdisciplinary cooperation. The solutions that are considered useful will be technically implemented into the district platform in the fifth step. There will be several iteration loops (ISO, 2010). We consider that this a strong approach, as we have incorporated the routines, resources, challenges, and preferences of older adults' everyday lives, with as many different life situations as possible, whereby the life situations are based on representative data. --- Cluster Analysis We performed a cluster analysis to identify different types of life situations in the model region. The function of a cluster analysis is to classify cases according to specific criteria, thereby generating types (Everitt, Landau, Leese, & Stahl, 2011). The database was created using a representative, standardized postal survey in the model region of a random sample of 4,000 inhabitants, aged 50 and older (response rate: 29.7%, N = 1,186). A total of 962 respondents had valid values for all the variables relevant to the cluster analysis. The median age of this sample was 66 years (Q1 = 57, Q3 = 75, age range: 50-101 years). The proportion of women was 54%. For over half (52%), their highest general education level was completion of Hauptschule/ Volksschule (lower secondary school in Germany). The sample is comparable to the 50+ population in terms of the distribution of gender, age, and other characteristics. People with a migration background were underrepresented. The operationalization was carried out according to Voges and colleagues (2003, p. 56 ff.), and is shown in Table 1. The following variables were used in the cluster analysis: socioeconomic status based on education and income, health assessment, the frequency of personal contact with people not living in the same household, marital status, and an evaluation of the quality of the apartment/house. A more detailed explanation of the operationalization can be found in the Supplementary Material. We performed an agglomerative hierarchical clustering using IBM SPSS Statistics 23. This includes different methods, of which our choice was the complete linkage method (furthest neighbor). The aim of this method was a high homogeneity within the cluster (Everitt et al., 2011). The determination of the optimal number of cluster groups and the assessment of the validity of the cluster solution were made using multiple channels (among others, via interpretability of the content-see Supplementary Material). --- Interview Study Semi-structured interviews Following the cluster analysis, and to provide a further qualitative, in-depth view of the life situations and the everyday lifestyles, "representatives" of the respective life situations were canvassed and interviewed using a semistructured interview guide. Because of their broad knowledge of the district, the project's practice partners were able to communicate the importance of this interview study to the participants. The practice partners were social workers from a nonprofit association and are responsible for community organization, as well as qualified nurses from a mobile nursing service. They have daily contact with various people in different life situations. The practice partners used the life situation profiles we had developed, which listed the required characteristic values typical for a particular life situation, in their search for suitable participants. As people with migration backgrounds have been underrepresented in the postal survey and in order to reach them, we required that at least one participant has a migration background. The practice partners asked potential participants whether they were willing to take part in an interview about their everyday lives. If the person was interested, they arranged the contact with the research team for the appointment. --- Table 1. Operationalization of the Life Situations Dimensions for the Cluster Analysis --- Dimension Variable Dichotomization Education and income Socioeconomic status (SES) index a 2,0-8,5 = "low" (1) 8,6-14,0 = "high" (2) Health Health self-assessment b 4-6 = "bad" (1) 1-3 = "good" (2) Social networks Frequency of personal contacts to persons not living in the same household (index of personal contact) c 0-7 = "low" (1) 8-12 = "high" (2) Family constellation Marital status single and living alone, widowed/civil partner deceased, divorced/annulment of civil union, or married/civil union and separated = "single" (1) married/civil union and living together, or unmarried and living in partnership = "living in partnership" (2) Habitation Self-evaluation of the quality of the apartment/house b 3-6 = "bad" (1) 1-2 = "good" (2) a Possible range = 2-14; b Possible range = 1-6; c Possible range = 0-12. In preparing the interview guideline, we first followed the concept of everyday life (Projektgruppe "Alltägliche Lebensführung," 1995). Because daily life is rarely lived consciously, we also assumed that the problems and challenges of everyday life are not explicit and present in all areas of life (Voß, 2000); therefore, the issue of daily life was not the most important topic of conversation. In fact, the discussion centered on the daily routine and its significance for the participants, thereby following their specific relevance system. We not only discussed body-related activities or other work based on performance-based tasks, but also activities connected with the social environment or with social functions, for example shopping, and meeting other people. We also discussed neighborhood issues in this context, by focusing on "the relationship between the aging person and his/her physical-social environment" (Wahl & Oswald, 2010, p. 112). This allowed us to obtain an impression of the subjectively felt barriers and resources that are a part of daily activities (cf. environmental gerontology, e.g., Carp & Carp, 1984;Oswald, Jopp, Rott, & Wahl, 2011). We complemented the guideline with questions regarding information behavior and demands, and the role of social contacts in this respect. Being informed or knowing how and where to find relevant information is, in fact, one of the key factors for being able to participate in social life, and therefore a significant factor for health (Berkman, Glass, Brissette, & Seeman, 2000). We also addressed the issue of technology in our conversations, to explore attitudes toward technological systems and (the likelihood of) their integration into everyday life. A short questionnaire at the end of the interview verified the extent to which the participants and their life situations matched our life situation profiles. In almost all cases, the interview took place in the participant's home and was carried out by two researchers: a social scientist who was in charge of the topical introduction and the dialogue during the interview and a computer scientist who would ask detailed questions about informatics issues, if necessary. The participants were advised that their participation was voluntary and that the interview could be stopped at any time. We assured them that all information would be treated anonymously. Ethical approvals were not required. --- Analysis strategy The qualitative analysis strategy is divided into several steps, which are shown in Figure 2. We chose a case-oriented analysis, which allows for traceability back to the case at any time during the investigation (Yin, 2014). Initially, the interview participants' life situations and sociodemographic data of the nine interview partners were summarized in portraits, which is similar to the persona method. From the subsequent problem analysis of the transcribed interviews, we extracted-in line with our conceptual positioning of everyday life-the subject-oriented challenges, needs, and demands, as well as everyday coping strategies concerning the maintenance of the current lifestyle. The subsequent step called for suggestions for possible technical solutions, which were developed jointly with the computer scientists and based on the previous steps. The solutions had to be easy to integrate into the participants' daily routines, have a subjectively meaningful and supportive effect on everyday living, and therefore be as realistic as possible. The orientation toward the life situation's distinctive features and biographical factors provided a constant reflection and verification of whether or not the solutions developed were accepted and also implemented in daily life. These efforts included generating user stories that helped to describe the qualities of the technical solutions. They refer to the digital platform and follow the scheme: "As a (role) I want (something) so that (benefit)" (Cohn, 2013). The user-story method was also chosen for finding a common linguistic consensus between the two disciplines. The computer scientists derived technological requirements for the digital district platform from the user stories. --- Results --- Cluster Analysis Types of life situations identified On the basis of cluster analysis, nine clusters were identified (see Figure 3). Four of the nine clusters (life situation types A, C, F, and H) proved to be completely homogeneous regarding the five variables adopted (F values <unk> 1), three clusters (life situation types D, E, and I) were heterogeneous in one variable, and two (life situation types B and G) clusters were heterogeneous regarding two characteristics. The characteristics of the life situation types are set out in Table 2. --- Interview study The interview study's sample consisted of nine participants, each of whom represented one type of life situation. Seven of the nine participants were female and two were male. Their ages ranged from 57 to 81 years. Table 3 gives an overview of the participants. The analysis and results of the interviews will be presented here in the form of an example and in a condensed form, based on the case of life situation type H, together with selected ideas for technical solutions. The case of life situation type H was chosen for the following reasons: It represents the group with the most precarious life situation. People in this life situation are normally rarely or never reached in participative technology development settings. Therefore, it represents a special challenge for technology development. --- Case example Mrs. H: Overall, a rather precarious life situation Mrs. H was born in Germany in 1936, is widowed, and lives alone in a senior-friendly rental apartment. She did not complete her school education and has been a housewife ever since she got married. In the course of her life, she has also been the unofficial, primary caregiver for family members. For this reason, she was never able to maintain social contacts outside her family: "I <unk>ve got no friends, there was no time for that. I <unk>ve had to care for all [...] at first my aunt for ten years [...] then my husband became ill and I looked after my father-in-law for twenty years." She also experienced the distress of war and displacement, as well as various life crises, in particular the death of her son. Within her family, she only maintains regular contact with her grandchild. She states that she rarely has contact with neighbors, and an exchange of neighborhood support is also (too) rare. Mrs. H views her health as very poor. When asked to rate it on a six-point scale, one indicating very good health and six indicating very bad health, she answered: "Then you can write down a ten." She is cared for by a daily nursing service. She must even rely on assistance for keeping house. She is content with her housing situation despite some shortcomings. She has a low net income of less than €1,000 per month and is therefore at risk of poverty. As far as technology is concerned, Mrs. H considers herself a nonuser. Although she has had some experience of more modern technical devices, such as a cell phone, she only uses familiar technologies, such as the television and the telephone, regularly. --- Problem analysis Mrs. H suffers from severe health restrictions that have an impact on her daily life. She is especially limited in terms of housekeeping, which she is only partly capable of accomplishing, and then only with the assistance of certain tools, such as a reach extender. An individual cleaning aide, provided by the nursing service, does most of the cleaning. In terms of mobility, Mrs. H is restricted both in her home as well as in her broader environment. As she has shoulder problems, she is unable to use her walker all the time, and the use of public transportation is also difficult for her, because of the height of the boarding platforms. When asked whether she could imagine having a volunteer accompany her with her shopping, she replied: "Yes, but I have to do something. [...] Whatever I can still take care of myself, I'll do it slowly; if need be, I'll go (shopping) twice." The only person whom she asks for help with housework or technical devices, is her grandchild. On the other hand, she sees no chance of receiving assistance from her social environment: I mean, who would help us? There is nobody available. We're all "cripples" here. [...] Upstairs, she is 92 by now, I am 81, my neighbor is also 81, and the other one, I have no idea how old she is, but she's also gasping for breath. Nor can Mrs. H depend on friendships: "I don't know anybody here. I've never had time for friendships." Although she lives in "housing for senior citizens," which advertises contact opportunities for its inhabitants and support with everyday problems, she does not know the contact person provided by the housing association. She gets her day-today information from the daily newspaper, which is something she cannot afford herself but receives regularly from her neighbor, from the television, her grandchild and from regular meetings in the district's community center. Mrs. H often feels lonely in her daily life and has no close contact with other people in a similar situation. She describes her relationships with the neighbors as follows: "There's no establishing contact with people here. You might have a quick exchange, 'hello, how are you?', 'have a nice day,' well, yes, that's possible. But doing something together, no way." Mrs. H always visits the community meetings that are held as part of the district project, which emphasizes how important they are for her: "Even if I'm sick, I'll go anyway." Although this demonstrates how important these encounters with other people are to her and how she benefits from them, she is nonetheless unable to establish close relationships: --- SES index a Mdn (Q1, Q3) 10.7 (9.6, 13.0) 6.6 (5.7, 8.0) 10.5 (9.5, 11.8) 7.2 (6.0, 7.9) 7.5 (6.2, 9.5) 7.5 (6.0, 9.0) 6.4 (4.5, 8.0) 7.6 (6.0, 8.7) 6.0 (5.0, 7.2) 8.5 (6.5, 10,1) Health b Mdn (Q1, Q3) 2.0 (2.0, 3.0) 3.0 (2.0, 3.0) 2.0 (2.0, 3.0) 3.0 (2.0, 3.0) 4.0 (4.0, 5.0) 4.0 (4.0, 5.0) 4.0 (4.0, 5.0) 4.0 (4.0, 5.0) 3.0 (2.0, 4.0) 3.0 (2.0, 4.0) Index of personal contact c Mdn (Q1, Q3) 8.0 (6.0, 9.0) 5.0 (4.0, 6.0) 8.0 (6.0, 9.0) 9.0 (8.0, 10.0) 7.0 (5.0, 9.0) 6.0 (4.0, 8.0) 7.0 (5.0, 9.0) 6.0 (4.0, 7.0) 9.0 (8.0, 10.0) 7.0 (5.0, 9.0) Liv. in partnership, % and "what can I still become?" It's not like it used to be, now it's every man for himself. It's not like it used to be. Another problem is that Mrs. H is very limited in her activities, first because of her poor health mentioned earlier and second due to insecurity concerning particular locations in her environment, which is aggravated by a lack of friendships: I go into town quite a bit, but that's also silly to just stand around by myself, you see so many people walking through town alone, but I don't like to be alone. I go to bed as early as 8.30. Elsewhere she describes: Now they've made the "kissing path" nice [author's note: a footpath located in a nearby wooded area which is also an important connection between central contact points in the district]. However, it's still not ok, because there are high bushes on the left side when you walk down the path. They built everything up there, so if someone's hiding in the bushes, who's going to help me if an arm reaches out and grabs me by the throat? On the other hand, Mrs. H is well aware of the assistance that she could use, such as the volunteer cemetery escort service: Well, I haven't been to my son's grave for over half a year. I am always afraid that if I trip and fall, I won't be able to get up again. [...] Yes, sure, I could call the Red Cross, they even have a transport service, but well, [...] I'll go, when it's nice again [author's note: the weather], then there will be more people out and about. [...] No, I've never tried that service, but I have read in the newspaper several times that volunteers are there to accompany you. --- Interpreting the situation Despite her many limitations, Mrs. H would like to lead her everyday life as independently as possible. She evaluates the advantages and risks associated with various activities, an example of which is her mobility outside her home; it became clear that although with her health problems and without the use of her walker, she does not feel able to deal with the distances and unsafe roads, going into town and shopping unassisted are more important to her. Therefore, she spreads her resources during the week in order to gain time to pursue as many daily activities as possible independently. To maintain her independence, she accepts, in addition to the risk of falling as she describes it, that she must forego other specific activities, such as visiting the cemetery during wintertime. Another significant aspect is that Mrs. H would like to have more social contacts, with whom she could pursue joint activities. Considering the example of the companion service, of which she was well aware, it became clear that Mrs. H has a general tendency to reject offers of assistance that she has not yet taken up. The situation with the nursing and housekeeping service with which she has become quite familiar in her daily life is different. Mrs. H also mentioned specific support requirements that are not covered and for which she does not know who to contact (e.g., support for domestic tasks not covered by the nursing service). --- Suggestions for technical solutions The following paragraph describes three possible digital solutions, which were developed based on the previous steps. The corresponding user stories are presented in Table 4. 1. A "digital nudge": Mrs. H would benefit from meeting with other residents to establish social contacts and to prevent social isolation. Because Mrs. H is not used to initiating and maintaining social connections by herself, she is unlikely to take the initiative but instead would react to existing "offers," as can be seen by her regular attendance at the district's community meetings. A digital nudge on the digital district platform would help by explicitly informing her about activities and events in her neighborhood and motivating her to participate. 2. Matching of activities: Mrs. H would also benefit from carrying out everyday activities together with other residents, for example, shopping or cemetery visits. With the help of user profiles on the platform listing interests, daily activities, (time) preferences, resources, assistance needs, mobility information, etc., the system should be able to match activities. A digital nudge would then make users aware that other potential users share the same interests in activities. 3. A map with suggestions for safe routes: Mrs. H has limited scope for activities. Due to her physical limitations and the accompanying fear of being mugged or having to deal with the barriers of public transportation, etc., she avoids specific paths (e.g., a practical walkway) and activities (e.g., visits to the cemetery). A city map integrated into the digital district platform could show her safe routes for her individual itinerary. The route recommendations would be based on several parameters, for example, a path at ground level (barrier-free), lighting, frequented by other pedestrians, and emergency call facilities. --- Reflections on the proposed technical solutions Mrs. H requires access to the digital district platform. Her low retirement benefits are not sufficient to cover the running costs of an Internet connection and the purchase of the necessary hardware. She lives opposite a community center with a "technology meeting place." There she would As a user of the platform, safe and barrier-free routes to her activities' locations would be suggested to Mrs. H, so that she can arrange her daily life independently and feel safe be able to use the digital district platform free of charge and volunteer ICT assistants could teach her how to use the platform. She could also take her tech-savvy grandchild along, who could help her to use the platform. It would make sense for her grandchild to also have a registered account with the digital district platform-so that he could also inform his grandmother about events and activities in her neighborhood, in addition to the digital nudge, as Mrs. H is limited to the community center opening hours and cannot use these services around the clock. For this solution, it would be necessary to draft user stories and technical requirements. Cooperation with further stakeholders would be necessary for implementing the solution of the maps giving suggestions for safe routes; for example, the city administration would have to supply more information or install a system for emergency calls, or other services, such as providing public toilets. Generally, it is important to pass all the proposed solutions back to Mrs. H and other potential users, for evaluation. In order to counteract digital inequalities, the design and functions of the digital platform are fundamental. An intuitive usability that requires less effort (e.g., using the digital platform on the TV with which Mrs. H is familiar instead of a computer) is important as well as user-specific functions (e.g., for Mrs. H, news in the form of videos instead of long articles). The system must be modifiable in order not to exclude other groups by the design being too barrier poor or low pitched. --- Discussion --- Limitations of the Cluster Analysis and the Interview Study The dimensions of the life situations were only identified by a few variables and therefore were not completely represented. Moreover, the dichotomization of the variables was accompanied by a substantial loss of information (Bortz & Schuster, 2010). Further limitations of the cluster analysis are discussed in the Supplementary Material. Due to time constraints, only one representative was interviewed for each type of life situation. In future research, it would be necessary to interview more than one person per cluster, because a single person cannot represent the perspectives of a whole life situation group. --- Challenges of the Approach Compared to more common forms of participatory designs (e.g., focus groups), the interdisciplinary, methodical approach is very time consuming. This has to do, not least, with the interdisciplinary communication necessary: Cooperation between computer scientists and social scientists/gerontologists has shown that the negotiation processes (regarding the topics of the interview guidelines), the discussion of life situation-related and socially produced inequalities, as well as the fundamental problem of finding a common language ultimately take a great deal of time. We also needed to agree that technological development cannot necessarily offer an answer for every daily problem situation, and that certain problems need to be handled differently. Recruiting suitable participants for the interview study also proved to be challenging, because it was not always possible to find someone who matched a particular type of life situation 100%. It was a funded project; even before the project started, it was clear that a digital platform would be developed. Therefore, the technical solutions had to be tailored to it, even though other solutions may have been better. --- Opportunities Presented by the Approach The opportunities presented by this holistic approach are clear. The potential user groups in the community were identified using the cluster analysis, which raised the problem of awareness of the heterogeneous (disadvantaged) life situations' challenges and led to groups of people participating in interviews, who had not previously been available for such a project. This allowed the views of people in different-including disadvantaged-life situations, to be incorporated. This would not have been possible in the working groups, because people in precarious situations could not be reached in this setting. In addition, the interview situation allowed for more personal and in-depth information about everyday life. As a result, a better understanding of their everyday lives, their coping strategies, and preferences within the context of their life situation was obtained. For both the technical developers and the social scientists, the experience contributed to an increased sensitization, which may ultimately lead to realistic and workable solutions for the residents in the community. The close interdisciplinary cooperation is also important because technical solutions do not always meet expectations and needs correctly (Fozard & Wahl, 2012). Linking the knowledge of sociological, psychological, and gerontological theories will result in a sound assessment of whether or not a potential technical solution will find acceptance. Shopping, for example, does not simply serve the purpose of providing groceries but also has social and mobility-retaining functions. Thus, the opportunity of ordering groceries online and having them delivered may be rejected by some people. A more detailed analysis demonstrates that problems that seem similar at first glance may actually be entirely different. This can be demonstrated using the example of social contacts, which in Mrs. H's case were never really present, whereas for others, they diminished more and more over time. Therefore, different technical solutions must be individually developed: support for establishing new social contacts and support for maintaining them. The approach presented has the advantage that it can be employed not only for the initial investigation of the requirement, but also halfway through the process, to evaluate the prototypes already developed, that is, partially functional versions of a technical solution (Boehm, 1988). --- Implications Technical development projects should

Background and Objectives: Technologies can enable older adults to participate in social life. These technologies need to be designed so that their benefits are obvious and older users are willing and able to use them. Although user participation has gained importance, participants are frequently not representative of the diversity of potential user groups, so that actual requirements are often not met. Thus, digital inequalities may increase. This problem arose at the beginning of an interdisciplinary research project to design a digital district platform that was to be codesigned with older residents. Here we report an approach to participatory development that takes into account the requirements of older adults, in diverse life situations. Research Design and Methods: We performed a cluster analysis using data (N = 962) from a representative survey in the case study region. We then conducted semi-structured interviews with residents, as representatives of the different clusters, to explore their life situations and everyday lifestyles. We applied a case-related analysis. Results: We identified nine life-situation clusters and conducted nine interviews with residents representing them. In an interdisciplinary team, we subsequently developed suggestions for possible technical solutions, based on the respective life situation, everyday life, habits, and challenges of each case. Discussion and Implications: Although the process was time consuming, we were able to incorporate perspectives of heterogeneous life situations into the digital platform development. This approach, which raises awareness of user-centered requirements, could lead to more apposite and beneficial solutions-especially for those who are underprivileged. Nevertheless, additional strategies to overcome digital inequalities are essential.

community. The close interdisciplinary cooperation is also important because technical solutions do not always meet expectations and needs correctly (Fozard & Wahl, 2012). Linking the knowledge of sociological, psychological, and gerontological theories will result in a sound assessment of whether or not a potential technical solution will find acceptance. Shopping, for example, does not simply serve the purpose of providing groceries but also has social and mobility-retaining functions. Thus, the opportunity of ordering groceries online and having them delivered may be rejected by some people. A more detailed analysis demonstrates that problems that seem similar at first glance may actually be entirely different. This can be demonstrated using the example of social contacts, which in Mrs. H's case were never really present, whereas for others, they diminished more and more over time. Therefore, different technical solutions must be individually developed: support for establishing new social contacts and support for maintaining them. The approach presented has the advantage that it can be employed not only for the initial investigation of the requirement, but also halfway through the process, to evaluate the prototypes already developed, that is, partially functional versions of a technical solution (Boehm, 1988). --- Implications Technical development projects should start with a comprehensive analysis of the potential users' everyday lives (Künemund & Tanschus, 2014). The development of possible technical solutions by an interdisciplinary team of social scientists/gerontologists and computer scientists is particularly productive because the former have a theorybased view of resources and problem situations, whereas the latter are more familiar with the potential of technical innovation. From our perspective, it makes sense to link the life situation with everyday life activities because orienting the technical development solely toward the life situation does not address the concrete, performed actions, and conversely, orienting only toward the lifestyle would negate the impacts of societal structures on the residents. Further strategies are absolutely necessary for permanently counteracting digital inequalities. Support structures appear to be essential (Damodaran & Sandhu, 2016). Especially now, in times of digital transformation, digital inequalities are becoming ever more strongly apparent, even in gerontology, as industry has discovered the potential of the "silver market" (Cornet, 2014). Political and social parameters must be created in order to ensure that access to digital technologies and opportunities for acquiring digital skills are available to all, even those in precarious life situations. In our view, social scientists/gerontologists bear a big responsibility for highlighting the societal sources of the digital inequalities and for exerting political influence against these circumstances. And gerontologists should also be responsible for steering technology developers toward the complexity of digital inequalities, and contributing their expertise in the development of technical solutions. --- Supplementary Material Supplementary data are available at The Gerontologist online. --- Conflict of Interest None to declare.

Background and Objectives: Technologies can enable older adults to participate in social life. These technologies need to be designed so that their benefits are obvious and older users are willing and able to use them. Although user participation has gained importance, participants are frequently not representative of the diversity of potential user groups, so that actual requirements are often not met. Thus, digital inequalities may increase. This problem arose at the beginning of an interdisciplinary research project to design a digital district platform that was to be codesigned with older residents. Here we report an approach to participatory development that takes into account the requirements of older adults, in diverse life situations. Research Design and Methods: We performed a cluster analysis using data (N = 962) from a representative survey in the case study region. We then conducted semi-structured interviews with residents, as representatives of the different clusters, to explore their life situations and everyday lifestyles. We applied a case-related analysis. Results: We identified nine life-situation clusters and conducted nine interviews with residents representing them. In an interdisciplinary team, we subsequently developed suggestions for possible technical solutions, based on the respective life situation, everyday life, habits, and challenges of each case. Discussion and Implications: Although the process was time consuming, we were able to incorporate perspectives of heterogeneous life situations into the digital platform development. This approach, which raises awareness of user-centered requirements, could lead to more apposite and beneficial solutions-especially for those who are underprivileged. Nevertheless, additional strategies to overcome digital inequalities are essential.

Introduction Patients with End-Stage Renal Disease (ESRD) face many challenges, including increased burden of comorbidities, reduced life expectancy and impaired quality of life, compared to general population [1,2]. Renal Replacement Therapy (RRT) are lifesaving treatments which include dialysis and kidney transplantation. Kidney transplantation is generally regarded as the preferred treatment choice compared to dialysis as it not only extends patients' life expectancy but also improves quality of life [3,4]. However, the available supply of kidneys cannot meet the demand, which leads to both long waiting lists and prolonged waiting times [5]. Previous studies have shown that inequities exist in access to kidney transplantation. Evidence, mainly from the USA, indicate that there is an association between kidney transplantation and race/ethnicity [6], gender [7], socio-economic status (SES) [8][9][10], marital status [11], and patient awareness [12]. The role of SES is complex because SES affects service along the pathway to transplantation [13], such as higher SES patients may have good communication with health care providers [14]. In order to explore the role of SES on access to kidney transplantation, the first thing is the definition of SES. Three classic defining indicators of SES are income, occupation, and education [14]. However, previous studies mostly used the ZIP code or residential postcode of patients to proxy SES measures [9,[15][16][17]. Based on national data from Australia, Grace et al. [9] reported that patients with higher SES were more likely to receive pre-emptive and living-donor kidney transplantation, although no association could be found for deceased-donor kidney transplantation. Studies from the US have shown that high SES increased access to transplantation for both living [15] and deceased donor kidneys [15,16]. A study from the UK found that socioeconomic disadvantaged patients were less likely to be placed on the waiting list, although had equal chance of transplantation with socioeconomic advantaged patients once listed [17]. Two studies have investigated the effect of education on access to kidney transplantation in the US, however the effect of education was inconsistent. Schaeffiner et al. [14] showed that college graduates had three times greater chance to be waitlisted or receive kidney transplantation compared to patients without a high school degree. The international Dialysis Outcomes and Practice Patterns Study (DOPPS) controlled for income and education at the same time and found that only income was (positively) associated with access to kidney transplantation [18]. The studies above vary with regard to study design, patient selection, sample size, time period, statistical methodology, outcomes of interest, SES measures, and availability of potential confounders. Beyond this, the health care systems are also different across countries. For SES measures, previous studies generally only used one of the three classic defining indicators of SES, and (or) used the ZIP code or residential postcode data as the only SES indicator, which potentially lead to less accurate estimates of individual level SES [9,15,17]. Furthermore, earlier studies are also limited due to lack of major potential confounders (e.g., comorbidities [9,14]), only including a subsample of the population [8,9], or being single center studies [16]. Sweden is a country with universal health care system and has no waiting lists for ESRD care except for kidney transplantation, as assessment and allocation of kidneys are based on numerous factors [19]. Studying equity in access to kidney transplantation is especially of interest in the context of the health care system and the outspoken egalitarian welfare system in the Nordic countries. It has been a long tradition in these countries to reduce or even try to eradicate social inequality, both in health care where treatment should be given irrespective of socio-economic status and in the society at large. Unfortunately, the association between individual SES and access to register-based kidney transplantation has not previously been studied in Sweden or any other Nordic country. Therefore, the aim of this study is to assess the impact of SES on access to kidney transplantation using a Swedish population sample and register data. The main contributions of the present study are that we used longitudinal register-based individual SES measured as two of the main indicators of SES before start of RRT; individual disposable income and education. In addition, we also control extensively for both medical and non-medical factors. --- Materials and Methods --- Material The Swedish Renal Register (SRR) [20] was linked to the Longitudinal Integration Database for Health Insurance and Labor Market Studies (LISA by Swedish acronym) [21] and the Register of the Total Population (RTB) from Statistics Sweden using the national personal identification numbers. The SRR is a national high quality register for patients undergoing RRT starting in 1991, with almost 100% coverage and a data reporting incidence of 95% [22]. The register includes rich information related to the patient, disease and treatment. The LISA data set combines information from several demographic and socioeconomic population registers [21], such as income, education, and employment status. These data are registered yearly for all older than 15 living in Sweden. Marital status and citizenship information comes from the RTB. --- Study Population The study population was defined as all patients aged 18 years and older who started RRT between 1 January 1995, and 31 December 2013, as recorded in the SRR. During this period there have been no changes to the Swedish kidney allocation policy. Patients whose current treatment modality are unknown and patients who recovered or died within 91 days after start of RRT were excluded, in order to only include patients with chronic conditions. We lack information on counter-indication, i.e., we cannot identify patients that were never suitable for kidney transplantation. We therefore assumed that all patients starting renal replacement therapy were suitable for kidney transplantation. --- (i) Study outcome and SES indicators This study investigated the association between SES and access to kidney transplantation, in terms of probability of receiving kidney transplantation. Access to kidney transplantation was defined as receiving a first living-or deceased donor kidney transplant during the study period. SES was measured in terms of income and education before the patient started RRT. Income was defined as the individual disposable income the year before the patient started RRT, derived from the household disposable income using consumption weights [23]. We adjusted disposable income to 2012 year's price level using the Consumer Price Index (CPI) from Statistics Sweden [24]. The disposable income was divided into quintiles, with quintile 1 (0-99,998 SEK) being the most disadvantaged and quintile 5 (188,751-6,685,735 SEK) being the most advantaged. Education was divided into three categories based on the Swedish educational system: mandatory education (<unk>9 years), high school education (9-12 years), and college education (>12 years). --- (ii) Control (potentially confounding) variables We identified several baseline demographic and clinical variables before the patient started RRT, in order to adjust for potential confounding based on prior studies. These factors included age at start of RRT [8,9], gender [8,9], year of first RRT [16], marital status [11], residence area (county) [9], citizenship [25], primary renal disease [8,9], and comorbidities [8,9]. Based on the distribution of transplantation events [26], age was divided into four age groups: 18-39, 40-49, 50-59, <unk>60 years old. Year of first RRT was handled as a continuous covariate (per year). The marital status included married, single, divorced, and widow. In Sweden, there are four kidney transplantation centers, located in the four largest cities (Stockholm, Gothenburg, Malmo, and Uppsala). We created the binary variable home county to capture any potential advantage of living in the same administrative area that performed the transplantation and/or living close to the performing hospital. Citizenship was measured as Swedish or non-Swedish. The comorbidities in the Swedish renal register database had eight categories, but were re-categorized due to small samples in certain groups. Therefore, comorbidities were categorized as hypertension, diabetes mellitus, cancer (blood-, skin-, and other cancer) and heart disease (cerebrovascular-, peripheral vascular-, Ischemic-, and other heart disease). Primary renal diseases were grouped into seven categories: glomerulonephritis, adult polycystic kidney disease, diabetes mellitus, hypertension, pyelonephritis, unspecified kidney disease, and others kidney diseases. The employment status was defined as employed or not, according to the employment status the year before the patient started RRT. --- (iii) Statistical analysis We described the frequency and distribution of patient characteristics in Table 1 based on treatment (transplantation and dialysis). We expressed continuous variables as mean and Standard Deviation (SD) and categorical variables as percentages. Between-group comparisons of continuous and categorical variables were done using t-and chi-square statistics, respectively. We used conventional logit models to study the association between SES and access to kidney transplantation, first in univariate models of the association between kidney transplantation and education and income respectively (model 1 in Tables 2 and3). For income, three step-wise multivariate models (models 2-4 in Table 2) were then estimated controlling for both income and education (model 2), demographic variables, i.e., age, gender, year of first RRT, marital status, home county, and citizenship (model 3) and clinical variables, i.e., primary renal disease and comorbidities (model 4). We included income and education simultaneously because education can be seen as a factor underlying the association between income and access to kidney transplantation. Education is generally defined early in life, and income is partly the result of educational achievements. Moreover, we can also explore which was the stronger independent factor which might inform inferences about mechanisms. For education, we did similar analysis without controlling for income in two step-wise multivariate models (models 2 and 3 in Table 3). In order to assess if the association between treatment and socioeconomic factors are different depending on time in RRT we conducted a sensitivity analysis where the sample was divided based on pre-emptive or non-pre-emptive transplantation (patients received kidney transplantation before or after day 91 of RRT). For non-pre-emptive transplantation patients, we also used multivariate Cox proportional hazards regression analysis to assess the association between treatment and socioeconomic factors. We also conducted a sensitivity analysis where the sample were limited to the working aged patients (20-65 years of age) as there are concerns that employment status determines treatment choice [27]. Statistical significance was assumed for p-values <unk> 0.05. All the statistical analyses were performed using Stata software, version 14.0 (College Station, TX, USA). The study has been approved by Lund Regional Ethical Review Board (Dnr: 2014/144). --- Results --- Baseline Characteristics Patients were excluded if they had missing information for the main variables of interest, i.e., income (340 patients; 2.0%), education (525 patients; 3.1%), marital status (344 patients; 2.0%). We also excluded patients with negative income (22 patients; 0.1%) and extreme income (1 patient, 1.08 <unk> 10 7 SEK). The final sample constituted of 16,215 adult patients (age <unk> 18 years) with RRT, of whom 4392 (27.1%) received a living or deceased donor kidney transplantation by 31 December 2013. The frequency and distribution of patient baseline characteristics were displayed in Table 1. The mean age at start of RRT was 63.7 years old (SD: 15.1). Kidney transplantation patients were younger, healthier, higher educated, lived closer to a transplantation center, and had higher income compared to dialysis patients (p <unk> 0.001). For example, 25% of the patients in the transplantation group had only mandatory education while this was the case for over 50% of patients in dialysis. A small but significant difference in terms of citizenship was also noted where, against expectation, the proportion with Swedish citizenship was higher in the dialysis group. RRT-renal replacement therapy; Tx-kidney transplantation; APKD-adult polycystic kidney disease; SEK-Swedish krona. Pre-emptive Tx can be either living-or deceased-donor Tx. Disposable income was divided into quintiles, where quintile 1 was the most disadvantaged and quintile 5 was the most advantaged. Continuous variables presented as mean (and standard deviations), categorical variables presented as percent of total. Groups (kidney transplantation vs. dialysis) were compared by t-test for continuous variables and by chi-square for categorical variables. ** p <unk> 0.01; *** p <unk> 0.001. --- Effects of SES on Access to Kidney Transplantation For income, four logistic models were shown assessing the likelihood of receiving kidney transplantation (Table 2). The highest income group (quintile 5) had more than two times the likelihood of access to kidney transplantation compared with patients in the lowest group (quintile 1; reference group), as shown in the univariate model 1. The effect of high income decreased when simultaneously adjusting for education (model 2). Also adjusting for demographic variables (model 3) and clinical factors (model 4) increases the effect size and shows a clear positive association between income and likelihood of kidney transplantation. For education, three logistic models assessing the likelihood of receiving kidney transplantation were shown (Table 3). In the univariate analysis (model 1) patients with college education had more than five times greater chance of receiving kidney transplantation compared to patients with mandatory education. Although adjusting for other covariates reduced the effect of high education, higher education remains clearly positively associated with the likelihood of transplantation in the fully adjusted model (model 3). Overall, the likelihood of access to kidney transplantation increased across both income and education. In both full models, a positive effect of younger age, being married, and having a Swedish citizenship was also noted. In the income model, neither gender nor living in the county of the transplantation center was shown to have any effect on access to kidney transplantation. However, in the education model, a positive effect of male gender was shown. --- Sensitive Analyses The sensitive analyses are presented in Tables 4567. Focusing on pre-emptive transplantations (model 1 in Tables 4 and5), the effect of high income and education is even higher than in the fully adjusted model which is also the case for the effect of citizenship. Interestingly, living in the county of a transplantation center now increased the likelihood of transplantation. Non-pre-emptive transplantation however (model 2 in Tables 4 and5) showed similar results as in the fully adjusted model, except that living in the county which has kidney transplantation center was now negatively associated with the chance of having non pre-emptive transplantation in the income model. Using Cox proportional hazards for non-pre-emptive transplantation patients, the results were similar as in model 2 in Tables 4 and5. We assessed the probability of access to kidney transplantation for working aged patients which reduced the positive effect of income and education (model 3 in Tables 4 and5). Also, adjusting for employment status (model 4 in Tables 4 and5) substantially reduced the effects of income and education although remaining significantly associated. Being employed the year before start of RRT was strongly positively associated with the likelihood of receiving a kidney in both analysis for income and education. In the income models, no gender differences were found in any of the sensitivity analysis while younger age, being married, and a Swedish citizen remained positively associated with the likelihood of receiving kidney transplantation. The results of the sensitivity analysis of the education models were similar to those in the income models, except that male gender had a positive effect for non-pre-emptive transplantation patients (model 2 in Table 5) and working aged patients without adjustment for employment status (model 3 in Table 5). --- Discussion The results showed a strong association between SES and access to kidney transplantation. After multivariate analysis, patients in the most advantaged income quintile was associated with more than three times greater chance of receiving kidney transplantation compared with patients in the most disadvantaged quintile. An equally large effect was found for patients with higher education compared to patients with mandatory education only. The international Dialysis Outcomes and Practice Patterns Study (DOPPS) [18] found that income was highly positively associated with access to kidney transplantation although education was not when controlling for both variables, which is in contrast to our study. This difference may be due to DOPPS included patients from several countries, where education system are different and educational levels may have different meaning [14]. Patients in the DOPPS study were 18 to 65 years old, i.e., comparable with our sensitivity analysis on working aged patient which still showed a strong positive association between education and access to kidney transplantation. Axelrod et al. [15] found that socioeconomic advantaged (mainly used post-codes as SES indices) were highly positively associated with both living-and deceased donor kidney transplantation while Grace et al. [9] found that socioeconomic advantaged were only highly positively associated with living-donor kidney transplantation, not with deceased-donor kidney transplantation. In order to compare with other studies, we also separated kidney transplantation to living-and deceased donor transplantation and our results were consistent with the study by Axelrod et al. [15] (results not shown). Schold et al. [16] found that higher income was associated with increased likelihood to receive a transplant, which was consistent with our results which did not separate living-and deceased donor kidney transplantation. Schaeffiner et al. [14] found higher educated patients had greater likelihood of access to kidney transplantation in the USA, which was consistent with our results. Thus, different studies using different definition of SES performed in different countries with different health careand educational systems, all find that high socioeconomic status increases the chances of access to kidney transplantation. Access to kidney transplantation probably depends on both patients and physicians. In Sweden, the decision of RRT was made by nephrologists, based on the Swedish guidelines [28] and the corresponding European guidelines [29]. In spite of this, in identifying potential transplant candidates, the possibility of physician bias cannot be ruled out [30,31]. However, it certainly is difficult to capture and therefore the reasons discussed below are only from the aspects of patients. One suggested explanation for shown income and education discrepancies is communication barriers between patients and health care provider, which might be easier to overcome for higher SES patients [14]. This theory is possibly supported by our sensitive analysis results, where the effect size between income and education and access to pre-emptive transplantation (where human decision plays a great role) is substantially higher than that between income and education and later kidney transplantation. Another suggested explanation is that higher SES patients are better prepared for transplantation due to a known association between SES and medication and health advice compliance [32]. A third reason may be that higher SES patients actively seek living donors, while lower SES patients may be hindered in this due to a lack of awareness or means [10]. Moreover, there may be donor barriers to live donation in lower SES groups. Inferior health and high-risk behaviors are more common in lower SES population and social support networks may be poorer in these individuals. Therefore, prospective donors from lower SES populations may frequently be found medically unsuitable to donate a kidney [9]. Sweden is a high-income country with universal health care system with a goal to achieve equity in health. Sweden has relatively strong income equity and low degree of inequity in terms of education [33,34]. However, even in such a context SES are related to the likelihood of kidney transplantation. Other studies have found similar inequities in Sweden, for example related to drug utilization [35]. Interesting to note is that in the income models, there is no significant association between sex and access to kidney transplantation in our study, which is inconsistent with most other studies. This could potentially be explained by the large number of covariates included in the current study, including socioeconomic status. Further, our study confirmed the results of a previous study [11] where married patients had higher likelihood of kidney transplantation. The potential mechanism may be spousal kidney donation, more developed social networking, and more qualified for kidney transplantation duo to less depression [36] and positive health behaviors in terms of alcohol and drug abuse [37,38]. We also found that patients with Swedish citizenship had higher chance of access to kidney transplantation compared to non-Swedish patients, which was consistent with a French study [25]. The suggested reason may be that patients of non-Swedish were less likely to have an adequate Human Leukocyte Antigen (HLA)-and blood group-matched kidney than patients of European origin [25]. Socioeconomic and culture factors may also limit access to kidney transplantation [39], however Sweden provides healthcare coverage for other citizenship similar to Swedish citizenship so financial obstacles to kidney transplantation should be much smaller than in other countries. --- Strengths and Limitations We conducted multifaceted analyses of SES-related variables in relation to access to kidney transplantation by using individually linked national register data, which provided us vast and detailed information. As many confounding factors as possible were adjusted for in order to minimize systematic errors. Furthermore, the Swedish Renal Register has an excellent coverage which gives the study high power and excellent generalizability for end-stage renal disease patients. We used individual SES-related data, which is expected to give more accurate effect estimates compared to geographically defined SES. The measures of SES were also measured before start of RRT, thanks to the longitudinal nature of the data. We also took into account all major comorbidities before start of RRT which is better when controlling for confounding, compared to studies in USA [14], Australia [9] and Scotland [17]. Nonetheless, there are some potential limitations to our study. The present study was in absence of biochemical data (e.g., blood type, serum albumin level, levels of parathyroid hormone) which are known factors influencing access to kidney transplantation. However, these biochemical covariates are not expected to be strongly correlated with SES [40]. The outcome being defined as probability of access to kidney transplantation during the study period could influence the interpretation of inequity in access to kidney transplantation. Patients might die before having the chance to receive kidney transplantation, which is more likely to occur for the sicker and older patients. We have handled this risk by adjusting for age groups, primary renal disease, as well as comorbidities in the regression models. Moreover, using Cox proportional hazards for non-pre-emptive transplantation patients, the results were similar as what we got from logistic models. In addition, we could not identify factors influencing access to waiting list. However, studies have showed that for factors such as age and primary renal disease, differences persist at both stages. Finally, although we have controlled for important confounding factors, we lack information on other unobserved factors (e.g., other comorbidities, severity of comorbidities, adherence, race/ethnicity and physician bias). --- Conclusions Access to kidney transplantation among Swedish end-stage renal disease patients is associated with common measures of socioeconomic status. Low income and education both reduced the chance of having kidney transplantation. However, reasons for these socioeconomic differences in the Swedish health care are unknown. The principal goal of Swedish healthcare system is to provide good and equal healthcare for all Swedish citizens. To this end, further studies are needed to identify the mechanisms of these inequities in order to construct effective interventions and policies. Moreover, since waiting list process is a key intermediate step to kidney transplantation, further studies can break down inequity of wait listing vs. inequity of kidney allocation, if data on the waiting list is available. --- Author Contributions: Ye Zhang is the first author and was primarily responsible for data analysis, interpretation and writing of the paper. Johan Jarl and Ulf-G Gerdtham contributed in study design, interpretation of the results and the writing of the paper. All authors have approved the final manuscript and its submission to your journal. --- Conflicts of Interest: The authors declare no conflict of interest.

Socioeconomic status-related factors have been associated with access to kidney transplantation, yet few studies have investigated both individual income and education as determinates of access to kidney transplantation. Therefore, this study aims to explore the effects of both individual income and education on access to kidney transplantation, controlling for both medical and non-medical factors. We linked the Swedish Renal Register to national registers for a sample of adult patients who started Renal Replacement Therapy (RRT) in Sweden between 1 January 1995, and 31 December 2013. Using uni-and multivariate logistic models, we studied the association between pre-RRT income and education and likelihood of receiving kidney transplantation. For non-pre-emptive transplantation patients, we also used multivariate Cox proportional hazards regression analysis to assess the association between treatment and socioeconomic factors. Among the 16,215 patients in the sample, 27% had received kidney transplantation by the end of 2013. After adjusting for covariates, the highest income group had more than three times the chance of accessing kidney transplantation compared with patients in the lowest income group (odds ratio (OR): 3.22; 95% confidence interval (CI): 2.73-3.80). Patients with college education had more than three times higher chance of access to kidney transplantation compared with patients with mandatory education (OR: 3.18; 95% CI: 2.77-3.66). Neither living in the county of the transplantation center nor gender was shown to have any effect on the likelihood of receiving kidney transplantation. For non-pre-emptive transplantation patients, the results from Cox models were similar with what we got from logistic models. Sensitive analyses showed that results were not sensitive to different conditions. Overall, socioeconomic status-related inequities exist in access to kidney transplantation in Sweden. Additional studies are needed to explore the possible mechanisms and strategies to mitigate these inequities.

Introduction With the proliferation of social network services, an increasing number of individuals are involved in multiple social networks at the same time. This trend has been witnessed by a recent survey: multi-platform use is on the rise, and 52% of online adults now use multiple social media services (Cohen 2014). Following this trend, multi-platform applications have attracted many researchers' attention due to the fact that multiple social networks can characterize the same user from different perspectives. For example, Twitter reflects users' casual activities and personal opinions; Facebook exposes users' social connections and daily events explicitly; and Linkedin uncovers users' professional skills and career paths. The heterogeneous information distributed across those diverse social networks is usually complementary rather than conflicting. Hence, as compared to single social network, appropriate aggregation of multiple social networks could provide a better way to understand users comprehensively, and consequently facilitate many applications, such as the inference of users' age, gender, race, occupation, personality and political orientation. With the help of those multiple social networks, in this paper, we scientifically study an ancient problem, the fortune teller, specifically in the field of career path prediction. A user's career path, in this work, refers to the user's occupational growth in his or her career life. It comprises sev- Copyright c 2016, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. eral distinct career stages, and each stage contains a set of equivalent occupational titles. The objective of this work is to predict the future career stages of a given user, the socalled career path modeling, which can provide potential benefits for employees, employers and headhunters. For employees, they can get information about their current career stages, the time point for their next job-hopping, as well as the whole picture of their own career paths. For employers, they will be informed of the career progressions of their employees and decide what would be the best time to promote their employees or increase their salaries. When it comes to headhunters, they could be advised of the appropriate time to talk to their target customers as well as the proper job positions for their customers. These efficient and accurate job hunting and recommendation processes will greatly facilitate headhunting and reduce their efforts considerably. As a consequence, career path modeling is a research topic with high potential and has many real-world applications. Despite its significant value, career path modeling from multiple social networks is a non-trivial task due to the following reasons. (1) Source Fusion. The information from multiple social networks of the same user describes his or her characteristics from various views, but it should reflect his or her career progression consistently. Thus, how to seamlessly and effectively fuse such heterogeneous information is a tough challenge. (2) Temporal Relatedness Modeling. A user's career path normally comprises a sequence of occupations. Instead of mutual independence, they are correlated with each other in chronological order. Therefore, how to temporally characterize such relatedness poses another challenge. (3) Influential Factors Identification. Different career paths have different influential factors. For example, education background may play an important role in the academic career path, but it might not be so crucial to an acting career. Furthermore, even within the same career path, the influential factors for different career stages also vary. For instance, publications might be key factors to research fellows, while research community services may be a more significant consideration for full professors. Hence, learning the stage-sharing and stage-specific features in each career path presents another crucial challenge. To tackle the above challenges, we present a multi-source learning framework with a fused lasso penalty (MSLFL). It co-regularizes the following factors: (1) Source Consis-tency. In particular, the predicted results from individual sources should be the same or similar. Thus their disagreements should be penalized. (2) Temporal Smoothness. The career path is equally split into multiple time intervals, and each of them is treated as a task. A career path is generally a gradual process, and hence sudden changes of career stages between neighboring time points should be penalized. For instance, it is much more smooth for a research fellow to become an assistant professor rather than a full professor in their next position. (3) Feature Learning. Features extracted from multiple sources are in high-dimension spaces. We employ a fused lasso to control sparsity and identify the task-sharing and task-specific features, which will identify the influential factors that affect a user's career progression at different stages. This enhances the interpretation of influential factors. We summarize the contributions as follows: --- Related Work It is worth mentioning that several research efforts have been dedicated to occupation analysis from social networks (Preot <unk>iuc-Pietro, Lampos, and Aletras 2015; Filatova and Prager 2012; Sloan et al. 2015). For example, Preot <unk>iuc-Pietro et al. (2015) inferred the occupation of a user based upon user profiles and social contents. However, they mainly used the information from one single source, which makes it difficult to comprehensively characterize a user's personality from various aspects. Instead of learning from a single source, multi-source learning has been proposed and has demonstrated its success in user modeling, profiling and behavior analysis with the assumption that information extracted about the same user from different sources may complement one another (Abel et al. 2011;2013;Meo et al. 2013;Xiang et al. 2013;Huang et al. 2014;Song et al. 2015a;2015b). For instance, Huang et al. (2014) proposed a multi-source integration framework to infer a user's occupation by combining both content and network information from Sina Weibo. However, as compared to occupation inference, career path modeling is much more complex, since it exhibits dual heterogeneities. In particular, besides comprehensive user description, a career path comprises a sequence of occupations and progressively develops from junior to senior stages. Multi-source learning fails to consider these progressions, and thus their performance is far from satisfactory. Multi-task learning is a learning paradigm that jointly learns multiple related tasks and has demonstrated its advantages in handling dynamic progression problems in many domains, such as medical science and transportation (Zhang and Yeung 2010;Chang and Yang 2014;Zhou et al. 2011;2012;Liu et al. 2015;Zheng and Ni 2013). In this framework, the prediction at each time point is treated as a task, and the intrinsic correlations among different time points are automatically learned, which could capture the dynamics effectively. For example, Zhou et al. (2011) proposed a multitask learning model to capture the intrinsic temporal relatedness for disease progression prediction. However, most exiting efforts on multi-task learning failed to consider the appropriate source fusion, which usually leads to suboptimal performance. Thus, multi-view multi-task learning is proposed to explore both source relatedness and task relatedness simultaneously (Zhang and Huan 2012;He and Lawrence 2011;Jin et al. 2013). For instance, He et al. (2011) proposed a graph-based iterative framework (GraM2 ) for multi-view multi-task learning and obtained impressive results in text categorization applications. However, as far as we know, the literature on multi-view multi-task learning is relatively sparse, and very few efforts have been applied to career path modeling. In contrast, our MSLFL model provides a natural way to fuse information from different sources by penalizing their disagreements. Moreover, MSLFL can better capture the dynamic progressions of career paths and learn the stage-sharing and stage-specific features via fused lasso penalty. --- Data Collection and Preprocessing Social Accounts Alignment. To the best of our knowledge, there is no available benchmark dataset suitable for career path modeling. We thus created new datasets by crawling four popular career paths, namely software engineer, sales, consultant and marketing. Each career path is an individual dataset. Social accounts assignment is the key challenge in this data collection, which aims to build the links among different social network accounts of the same user (Abel et al. 2011;2013;Meo et al. 2013). Towards this end, we employed the social service About.me1, which encourages its users to list their multiple social accounts explicitly in their personal profiles. We collected the data from About.me by searching keywords corresponding to the career paths. Considering the software engineer dataset as an example, we used "software engineer", "programmer" and "program developer" as the keywords to search from About.me and got 6, 284 candidates. Then we retained only those candidates who provided their Twitter (Tw), Facebook (Fb) and Linkedin (Lk) accounts as the software engineer dataset. The dataset statistics 2 are presented in Table 1. Career Stages and Occupation Variants. Based on prior knowledge, we roughly split each career path in our collected datasets into four stages, where each stage represents a milestone within the whole career path. Take software engineer as an example. We define it to compose four stages, namely software developer, senior software developer, manager and CEO. It is worth noting that due to vocabulary vari- Ground Truth Construction. To simplify the computation, we quantized the four career stages in different career paths from junior to senior as 1 to 4. For instance, software developer, senior software developer, manager and CEO in the software engineer path are mapped to 1, 2, 3 and 4, respectively. In addition, for each career path, we defined several time stamps with equivalent time period. In particular, we chose the start time of a user's first job as the first time stamp. For example, for a given user u, his/her first work is a software developer; we thus labeled his/her career stage at time t 0 as 1. Assuming that the gap between two neighboring time stamps in the software engineer path is three years, then four years later the occupational title in his/her Linkedin is senior software developer. We thus label his/her career stages at time t 1 as 2. On the other hand, from users' Linkedin profiles, we can obtain their working experiences, including occupational titles and corresponding time periods (i.e., the career stage and time stamp). Figure 1 depicts the ground truth construction for the career datasets. --- Career Path Modeling Notation We first define some notation. In particular, we use bold capital letters (e.g., X) and bold lowercase letters (e.g., x) to denote matrices and vectors, respectively. We employ non-bold letters (e.g., x) to represent scalars, and Greek letters (e.g., <unk>) as parameters. Unless stated, otherwise, all vectors are in column form. Let us assume that we have N labeled users for a given career path. This career path can be split into M time points and each time point is aligned with a task. Meanwhile, each user is described by S <unk> 2 sources. X s <unk> R N <unk>Ds denotes the feature matrix extracted from the s-th source, where D s is the feature dimension of the corresponding source. The whole data matrix can be written as X = <unk>X 1, X 2,..., X S <unk> <unk> R N <unk>D, where D = S s=1 D s. The label matrix is denoted as Y = <unk>y 1, y 2,..., y M <unk> <unk> R N <unk>M, and y m = (y 1 m, y 2 m,..., y N m ) T <unk> R N is the label vector of the m-th task. --- Problem Formulation The career status of users at the m-th time point can be linearly predicted from the s-the source as follows: f s m (Xs) = Xsw s m,(1) where w s m <unk> R Ds denotes the linear mapping function for the task m with source s. Without prior knowledge on the contributions of different sources, we assume that all sources contribute equally. Thus, the final prediction model of all sources for task m is obtained by the following late fusion: fm(X) = 1 S S s=1 f s m (Xs).(2) Information distributed in various sources in fact describes the inherent characteristics of the same user from various views, and hence their predicted results should be forced to be similar. In a sense, we can reinforce the learning performance of individual sources. Considering the leastsquares loss function, we can define the following objective function: min w s m 1 2 M m=1 y m - 1 S S s=1 Xsw s m 2 2 + <unk> 2 M m=1 S s,s =1 Xsw s m -X s w s m 2 2. (3) To consider the temporal smoothness of career progressions and learn descriptive features, we expand the model in Eqn. (3) to incorporate a fused lasso penalty. This penalty ensures a small deviation between two tasks at successive time stamps and automatically selects task-specific and tasksharing features for career path modeling. In particular, the fused lasso penalty comprises a temporal lasso and a regular lasso. Let W = <unk>w 1, w 2,..., w M <unk> <unk> R D<unk>M denote the overall weight matrix, where w m = <unk>w 1 m, w 2 m,..., w S m <unk> T <unk> R D. The overall objective function can be restated as min W s m 1 2 M m=1 y m - 1 S S s=1 Xsw s m 2 2 + <unk> M -1 m=1 wm+1 -wm 1 + <unk> 2 M m=1 S s,s =1 Xsw s m -X s w s m 2 2 + <unk> W 1,(4) where <unk>, <unk> and <unk> are regularization parameters. • 1 denotes the entry-wise matrix 1 -norm. --- Optimization The optimization of our overall objective function is not easy due to the two non-smooth terms: temporal lasso and regular lasso. To solve this problem, we first rewrite the second term in Eqn. (4) as, T L = M -1 m=1 wm+1 -wm 1 = WH 1 <unk> max A <unk>1 A, WH,(5) where H <unk> R M <unk>(M -1) is defined as follows: H ij = -1 if i = j, H ij = 1 if i = j + 1, and H ij = 0 otherwise. U, V <unk> tr(U T V) denotes a matrix inner product; -1) <unk> is an auxiliary matrix associated with WH 1 ; and • <unk> is the matrix entry-wise <unk> -norm. Eqn. ( 5) is still a non-smooth term. We approximate it by the following smooth one (Chen et al. 2011), A = <unk>A| A <unk> <unk> 1, A <unk> R D<unk>(M T L<unk>(W) = max A <unk>1 A, WH -<unk>d(A),(6) where d(A) is defined as d(A) = 1 2 A 2 F. We can obtain an analytical solution of A in Eqn. (6). It is obvious that T L <unk> in Eqn. ( 6) is a lower bound of T L in Eqn. ( 5), and the parameter <unk> controls the gap between the two: G = 1 <unk> max A <unk> <unk>1 d(A) = 1 2<unk> A 2 F = 1 2<unk> D(M -1). (7) The gradient of T L <unk> (W) is computed by <unk>T L<unk>(W) = A * H T, (8 ) where A * is the optimal solution of Eqn. ( 6) and is computed by A * = <unk>( WH <unk> ),(9) where <unk> is a defined as follows: For x <unk> R, <unk>(x) = x if -1 <unk> x <unk> 1; <unk>(x) = 1, if x <unk> 1; and <unk>(x) = -1, if x <unk> -1. For matrix A, <unk>(A) applies <unk> on each entry of A. Then the overall objective function can be approximated by <unk> = L + <unk>C + <unk>T L<unk> + <unk>L1. (10 ) With the discussion above, it is easy to show that Eqn. ( 10) is a convex function with three smooth terms and one nonsmooth lasso penalty. We define h(W) = L + <unk>C + <unk>T L<unk>, g(W) = <unk>L1. (11 ) We can thus use the Fast Iterative Shrinkage-Thresholding Algorithm (FISTA) (Beck and Teboulle 2009) to solve it. One of the key steps in using FISTA is to solve the proximal step: W (k) = argmin W <unk>g(W) + L k 2 W -(V (k) - 1 L k <unk>h(V (k) )) 2 F <unk>, (12 ) where V (k) is the search point and is defined by the affine combination of W (k-1) and W (k-2) ; and L k is a scalar that can be determined by the line search method (Beck and Teboulle 2009). The gradient <unk>h(W) in Eqn. ( 12) can be computed as <unk>h(W) = 1 S X T ( 1 S XW -Y) + PW + <unk>A * H T, (13 ) where P <unk> R D<unk>D is a sparse block matrix with S<unk>S blocks, and its entries are defined as, Pss = <unk>(S -1)X T s Xs, P ss = -<unk>X T s X s. (14 ) As Eqn.( 12) is computed in every FISTA iteration, it needs to be solved efficiently. Specifically, it can be reformulated as W (k) = argmin W <unk> 1 2 W -B 2 F + <unk> W 1<unk>, (15 ) where k) ) and <unk> = <unk> L k. It is easy to show that Eqn.( 15) has a closed-form solution (Wright, Nowak, and Figueiredo 2009), B = V (k) -1 L k <unk>h(V ( W (k) = max(0, 1 - <unk> B 1 )B. (16 ) We thus can solve Eqn.( 12) quite efficiently. --- Experiments Experimental Settings To validate our model, we first need to define the time stamps. In particular, we treated the start time of a user's first job as the first time stamp (time t 0 ), and set three years as the time window between two neighboring time stamps, since three years appears to be a typical period of time between transitions in a person's career path. In this work, we examined four successive time stamps for each user, since we believe that nine years should be long enough to reflect a person's career progressions effectively (Veiga 1981). Meanwhile, each time stamp is aligned with a task. We kept the same settings for all careers we considered. In addition, we employed the average classification accuracy over these four tasks in each career as our performance metric. The experimental results reported in this paper are based on 10-fold cross-validation. The parameters were selected using grid search on each career dataset. --- Feature Extraction We extracted some career-oriented features from multiple social networks to informatively represent each user. • Demographic features. As reported previously (Stacy 2003), some demographic characteristics, such as education and participation, are important factors that influence the speed of users' career progression. We thus extracted a rich set of demographic features including gender, education level, relationship status, and the number of social connections from users' Linkedin and Facebook profiles. • LIWC features. LIWC is a widely-used psycholinguistic analysis tool for investigating the relationship between word use and psychological variables, and it has been successfully applied to identify personality and social traits of users in many social applications (Paek et al. 2010). It has been reported that personality and social traits are strongly correlated to a user's career path (Kafetsios et al. 2009;Judge and Kammeyer-Mueller 2007). The main component of LIWC is a dictionary that contains mapping from words to 72 psychologically meaningful categories. 3 For a given document, LIWC calculates the frequency of words related to the particular category and represents the document as a 72-dimensional feature vector. • User topic features. According to our observation, the topics discussed by users are strong indicators of their career stages. For example, software engineers may frequently talk about programming skills, while CEOs may be interested in the topic of company management or business. This drives us to explore the topic distributions of their social posts. In particular, we employed Latent Dirichlet Allocation (LDA) (Blei, Ng, and Jordan 2003) to generate topic distributions, which has been proven to be effective in latent topic modeling (Hu et al. 2012;Li, Huang, and Zhu 2010). With the assistance of this tool 4, we ultimately obtained 85-, 45and 130-dimensional topic-level features from users' Twitter, Facebook and Linkedin profiles, respectively. The number of topics over each source were separately tuned by optimizing the perplexity metric (Blei, Ng, and Jordan 2003). --- Learning Model Comparison To demonstrate the effectiveness of our MSLFL model, we compared our model with the following five baselines: • SVM: The first baseline is the Support Vector Machine (SVM), which is a mono-source mono-task learning method that concatenates the feature vectors from all sources to form a single feature vector and then learns each task individually. We selected a linear kernel and implemented this method based on LIBSVM (Chang and Lin 2011). • RLS: Regularized least square (RLS) is a multi-source mono-task learning method, which learns each task separately by minimizing 1 2N y m -1 S S s=1 X s w s m 2 2 + <unk> 2 w m 2 2. 3 http://www.liwc.net/ 4 http://nlp.stanford.edu/software/tmt/tmt-0.4/ • MTL: Multi-task learning (MTL) is a mono-source multitask learning approach (Zhang and Yeung 2010), and it is able to capture pairwise task relatedness. • FL: Fused Lasso (FL) (Tibshirani et al. 2005) is a monosource multi-task approach, which aims to capture dynamic progressions by minimizing the objective function 1 2 Y -XW 2 F + <unk> M -1 m=1 w m+1 -w m 1 + <unk> W 1. • regMVMT: The regularized multi-view multi-task learning model (regMVMT) (Zhang and Huan 2012) jointly regularizes source consistency and uniform task relatedness. The experimental results are summarized in Table 3. From this table, we have the following observations: 1) The last four multi-task learning methods stably outperform the first two mono-task learning methods, which verifies that the tasks are not independent and that jointly learning them can boost learning performance. Moreover, it is not unexpected that SVM achieves the worst performance. This may be due to two reasons. First, appending the features from different sources directly may lead to suboptimal performance, since they may belong to different feature spaces. Second, certain tasks may hold insufficient training samples. 2) As compared to MTL, our model and regMVMT achieve higher accuracies due to the fact that MSLFL and regMVMT can incorporate heterogeneous information from different sources, which may help to improve overall performance. 3) FL performs better than MTL and regMVMT. This is because MTL and regMVMT assume uniform task relatedness, while FL captures the temporal relatedness between neighboring time points and is more suitable for progressive applications. In addition, FL automatically discovers task-sharing and taskspecific features for a better representation. This further confirms the correctness of our hypothesis that there exist certain temporal patterns in career path progression and that the fused lasso penalty is effective for capturing these dynamic progressions. 4) The MSLFL model significantly outperforms regMVMT in sales and marketing and shows superiority in software engineer and consultant, which underscores the complex career progressions in various careers paths. For instance, sales and marketing may exhibit stronger progressive trends than software engineer and consultant. In addition, our model outperforms FL. This demonstrates that our model is able to leverage source relatedness and further improve performance. --- Source Comparison To demonstrate the descriptiveness of multiple social networks, we compared the performance of our model over individual social networks and their various combinations. The results are presented in Table 4. From this table, we observe that the combinations of two distinct social networks outperform each individual one, and that the "Twit-ter+Facebook+Linkedin" combination achieves the best performance over the others. This observation reveals that the more sources fed to our model, the better the performance that can be achieved. This verifies that information from various sources are complementary to each other instead of conflicting, and that information fusion from multiple sources can comprehensively capture users' characteristics. In addition, we observed that the performance of MSLFL on Twitter and Facebook is better than that on Linkedin. This may be caused by the data sparsity in Linkedin, since most of the users update their Linkedin profiles less frequently. Moreover, even on the same source, the predictive performance of the model over different career paths is completely different, which implies that the capability of source description varies from career to career. --- Computational Complexity Analysis In this section, we discuss the computational complexity for solving the MSLFL model. For the optimization of W, the main computational cost comes from calculating the gradient <unk>h(W) in each FISTA iteration. In particular, as X T X, X T Y and P remain unchanged in every FISTA iteration, we can pre-compute and store them with time complexity of O(ND 2 +NMD). For each FISTA iteration, computing the gradient in Eqn. ), where is the desired accuracy. Thus, the MSLFL model can be solved efficiently. Moreover, the per-iteration complexity of FISTA for solving MSLFL is independent of the sample size N, which demonstrates that our model has the potential to scale to large-scale Web data. --- Conclusion and Future Work This work has presented a scientific fortune teller for career path prediction. The algorithm behind it is a novel multisource multi-task learning model. This model fuses information distributed over multiple social networks to characterize users from multiple views. Meanwhile, it jointly penalizes disagreements among sources and the sudden changes between tasks on two neighboring time points. In addition, it is able to learn the task-sharing and task-specific features simultaneously. Extensive experimental results have demonstrated its superiority over the state-of-the-arts competitors. In the future, we will extend our model to consider the source descriptiveness and learn the source confidence adaptively. Moreover, we will explore model-based approaches for title standardization, which may help our approach scale.

People go to fortune tellers in hopes of learning things about their future. A future career path is one of the topics most frequently discussed. But rather than rely on "black arts" to make predictions, in this work we scientifically and systematically study the feasibility of career path prediction from social network data. In particular, we seamlessly fuse information from multiple social networks to comprehensively describe a user and characterize progressive properties of his or her career path. This is accomplished via a multi-source learning framework with fused lasso penalty, which jointly regularizes the source and career-stage relatedness. Extensive experiments on real-world data confirm the accuracy of our model.

Seeing the child in context: Supporting gender diverse children and their families in multiple ways. An introduction to this special edition --- Trilby Langton, Bernadette Wren & Polly Carmichael This special issue of Clinical Child Psychology and Psychiatry includes a group of papers that originated in the work of the Gender Identity Development Service (GIDS) which is the nationally-designated NHS service in the UK to work with children and adolescents experiencing significant concern and distress about their gender. This short introduction to those papers aims to set the context and explain why certain areas have been given a focus. In publications on childhood gender diversity there can be a disproportionate focus on the physical interventions that are sometimes provided to young people and children. Although medical treatment is an important area of research and expertise, and raises particular anxieties and dilemmas for families and clinicians, at GIDS we support young people and families in multiple ways and it is this contextual, systemic and psychosocial support that we feel is worthy of fuller description. At the GIDS we have always made space to think and talk together reflexively about the nature of our work and the tensions we encounter. Some of us have been in the service many years while others are relatively new to the work. We share worries about the rapidly changing demands that are being placed on the service, the limited and inadequate evidence base available to clinicians to provide answers to the many questions that the work generates, as well as the known and also unknown implications of supporting early social transition and providing physical interventions to children and young people. These meetings are a place to explore and reconcile important theoretical ideas derived from our clinical trainings with the complex work we were undertaking. We also examine the ways in which the work affects us personally and professionally. In the media, amongst support groups and with service users there are many stories and ideas about gender diversity in children and what the GIDS stands for and advocates. These ideas can create for families many expectations and pre-conceived ideas about our process. We take time to consider the experience of being positioned in very particular ways and the effect that this can have on forming helpful relationships. It is clear that we have a responsibility to continue to articulate our approach in different formats, to better explain some of the ways in which our knowledge of child and adolescent difficulties and development informs our thinking, to help others understand how we try to make sense of the broader cultural influences at this particular time in the history of the service, to share more broadly the richness of the work and, finally, and perhaps most importantly, to illustrate our concerns and our dilemmas in ways that would not further marginalise the young people we support. Over time we have formed a commitment to bring together a collection of papers that highlighted the multiple ways that support is offered and how the work can unfold. Carmichael and de Graaf's introductory overview of the emerging trends helpfully introduces the reader to some of the broad clinical questions that are regularly discussed within our service at team meetings, in reflective spaces and in supervision. These include questions about why there has been a large increase in referrals and, more specifically, what factors have influenced a shift in the sex ratio so that we are now seeing many more young people, especially in their mid-teens, who were assigned female at birth. Carmichael and de Graaf also highlight the growing diversification of young people's gender identity and expression and the ways in which gender is increasingly thought about in more fluid ways. This is captured in more detail by Twist and de Graaf who provide this special issue with an up to date analysis of the diverse self-descriptions of young people who present to the service. Conversely de Graaf, Manjra, Hames and Zitz provide insights into the young people of whom we see much less. They highlight the fact that less than 10% of referrals to the service are from black and minority ethnic groups. This makes for a significant underrepresentation in referrals to the service as compared to both the national population and Child and Adolescent Mental Health service figures. In the paper they discuss the potential barriers for certain groups in accessing services and consider the possibility of cross-cultural variations in the conceptualisations of gender, gender roles and gender variance. Wheeler, Langton, Lidster and Dallos use qualitative research to provide an insight into young people's experience of having a gender diverse sibling. This paper reflects on the types of support that siblings might need as families adjust to a child's gender transition. Much of our work is conducted with whole families and this paper highlights the unique needs of siblings. Many young people and their families attending our service can have limited access to sources of community support. This can heighten feelings of isolation, shame and difference which in turn can contribute to a more hopeless perception of the options and the opportunities that are available. Bringing families and young people together to meet and share experiences through our family days, parent support groups and young people's groups is an important part of the psychosocial support that is offered at GIDS. Davidson, Morrison, Skagerberg, Hames and Russell provide a detailed description of the aims of the therapeutic young people's group that has been offered at Tavistock centre for the last six years and they illustrate how these groups help young people manage difficult experiences, develop coping skills and resilience and explore their identities through effective social solidarity and the accepting group environment. There can be a powerful narrative amongst certain groups and in the media that young gender diverse people unavoidably experience deep distress and are highly susceptible to suicidality and self-harm, requiring rapid medical intervention. In our experience the clinical picture is more complex and the limited nature of the research to date can lead us to premature conclusions. Mann, Taylor, Wren and de Graaf provide this special issue with an up to date review of the relevant UK-derived literature. They offer tentative support for the suggestion that trends in self-injurious thoughts and behaviours among genderdiverse young people may, to some extent, reflect trends in self-injury in the general population, and that these rates may be approximately comparable to rates reported among children and young people in contact with other mental health services. There are many important questions that are in the forefront of clinician's minds as we begin the process of making sense of an individual child's intimate and private feelings about their gender. Miller and Davidson illustrate some of the ways in which families at GIDS are invited to explore their beliefs and behaviours around gender, using systemic ideas. They show that an overarching aim of our work is to consider how children are situated within the multiple domains in which they live their lives. In our view the young people referred to the service need be supported to live well, through working with parents, siblings, schools, local mental health and social care professionals and support group networks. This is especially the case as we consider that clinicians are often required to understand complex situations in which young people might have multiple other difficulties which can significantly impact on a process of gender exploration and decision making. In the same way that the binary construction of gender can feel limited as a way of capturing the diverse and varied nature of our gendered lives, the polarised way in which gender care in this area has increasingly been constructed as either 'affirmative' or'reparative' seems equally unhelpful. The affirmative stance described in the paper by Ehrensaft, Giammattei, Storck, Tishelman & Keo-Meier, (2018) advocates clinicians following entirely the child's preferred gender both by supporting an early social transition and by being ready to recommend physical interventions -including surgery -at increasingly young ages, whereas the'reparative' approach is said to view gender diversity as pathology and therefore the aim of the clinician holding this stance would be to support or teach a child to not have these gender difficulties. From our clinical experience at GIDS, these ways of describing the work can seem very limited in relation to the complexity and the varied nature of how young people present at the clinic. Of course our work is 'affirmative' in many ways, in that we respect and accept completely children's sense of themselves, in many cases actively support them to live in the way they feel is most comfortable, work hard to reduce the social inequalities and discrimination within the systems around them, and in some cases, where clinically appropriate, we affirm in the most profound way by providing treatment to alter their bodies. However, our experience tells us that that there are varied outcomes for children who present as gender diverse. So whilst accepting and respecting a child we also hold in mind there can be no certainty about the pathway each individual will take. For many a longer period of thoughtful and supportive exploration is needed, and in many cases this allows them to arrive at a place of decreased distress and increased clarity about who they are and how they might live in their bodies in a settled way. Churcher Clarke and Spiliadis in this issue have highlighted two of the many examples in our service of young people who, over a period of time and with thoughtful care and support, make sense of their difficulties as not directly related to gender or gender change. This in our view is not'reparative' therapy: we do not aim to change someone's gender feelings or utilise any methods historically associated with the ideology of conversion therapy. However, the ultimate outcome for many is that pressing gender concerns pass as the child or adolescent develops, or as other difficulties are understood better. Putting words to some of these tensions can be problematic and brings with it considerable trepidation for clinicians, who fear being tarred with negative slurs. Our experience is that accounts of diverse outcomes are often refuted or are not experienced as helpful by some stakeholders, groups and particular families. It seems that there may be a fear that to acknowledge or highlight stories of diverse outcomes will in some way discredit or devalue those who do feel certain of their gender identity and who will continue to identify as transgender into adult life. In her paper, Wren provides a vital reminder of how difficult it is to adequately contemplate the many layers of ethical concern in this work with children and young people, while keeping in mind the historical and ongoing context of oppression and discrimination that has affected gender diverse individuals. The support offered at GIDS aims to understand the complex ways in which many young people present -often with significant associated difficulties -and to maximise their developmental opportunities, tolerate the distress of the perceived mismatch of body and gender feelings, whilst exploring the possible pathways available to them. There can be a powerful pull to reduce complexity and simplify children's experiences and this pull can be seen in many of the polarised stances from which these issues are often argued. We hope that this special issue of CCPP provides some insights into the richness of this work and ultimately complicates for the reader what can so often become a simple story. We continue to feel that children are done an injustice if we do not try to examine the grey, the nuance, the unsaid, and dance lightly around certainty when it comes to making statements about the fixity of children's gender feelings. We feel a responsibility to articulate our clinical experience, report our research findings, name some ethical concerns and convey some of the different therapeutic modes and approaches that are utilised by our team on a child-by-child basis.

Seeing the child in context: Supporting gender diverse children and their families in multiple ways -An introduction to this special edition. Clinical Child Psychology and Psychiatry.