Comparing social contact and group identification as predictors of mental health. more

1 British Journal of Social Psychology (2012) C 2012 The British Psychological Society The British Psychological Society www.wileyonlinelibrary.com Brief report Comparing social contact and group identification as predictors of mental health Fabio Sani1 ∗ , Marina Herrera2 , Juliet R. H. Wakefield1 , Olga Boroch1 and Csilla Gulyas1 1 2 University of Dundee, Scotland, UK University of Valencia, Spain Current research on social integration and mental health operationalizes social integration as frequency of interactions and participation in social activities (i.e., social contact). This neglects the subjective dimension of social integration, namely group identification. We present two studies comparing the effect exerted by social contact and group identification on mental health (e.g., depression, stress) across two different groups (family; army unit), demonstrating that group identification predicts mental health better than social contact. Research has revealed that greater social integration – generally defined as one’s active engagement in a range of interpersonal relationships and social activities (Brissette, Cohen, & Seeman, 2000) – is linked to better mental health (Kawachi & Berkman, 2001). For instance, Glass, Mendes de Leon, Bassuk, and Berkman (2006) found that social engagement predicted lower levels of depression over time in a large sample of community-dwelling older adults. According to researchers, greater social integration enhances mental health by providing people with normative expectations that afford a sense of meaning, purpose, belonging, and stability (Thoits, 1983), and by increasing their received or perceived social support (Cohen, 2004). While the importance of this literature is unquestionable, we believe that the way in which social integration has been measured is limited. Typically, measures of social integration ask participants either (1) to specify how often they see or talk to other individuals within specific groups (e.g., family, workplace, church) over a certain period of time (e.g., Cohen, Doyle, Skoner, Rabin, & Gwaltney, 1997), or (2) to list the number of social activities (e.g., doing charity work, playing cards) in which they are involved (e.g., Hanson, Isacsson, Janzon, & Lindell, 1989). That means that social integration is operationalized in terms of social contact. This neglects the fact that we do not just talk to and do things with other members of our groups; we also establish psychological links ∗ Correspondence should be addressed to Fabio Sani, School of Psychology, University of Dundee, Park Place, Dundee, Scotland DD1 4HN, UK (e-mail: f.sani@dundee.ac.uk). DOI:10.1111/j.2044-8309.2012.02101.x 2 Fabio Sani et al. with our groups (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). We ask ourselves questions such as: What does this group mean to me? Do I feel part of it? Do this group’s views reflect my own views? The answers we give to these questions determine the degree of our group identification (Tajfel & Turner, 1986): our subjective sense of belonging to the group and of commonality with other in-group members. Social psychologists endorsing a social identity approach to group processes (Haslam, 2004) have found that group identification has positive consequences for the quality of one’s interactions with others. For instance, Platow et al. (2007) showed that experimental participants undertaking a physically painful activity (i.e., immersing a hand into a bath of ice water) reported feeling less pain when they received support from a member of a group with which the participant identified than when support was received from a member of a group that was irrelevant to the participant. Social identity researchers have also found that group identification provides one with a sense of meaning, permanence, and stability, which protects against existential anxiety. For instance, Sani, Herrera, and Bowe (2009) found that people who are asked to reflect upon their own mortality (and therefore have to face existential anxiety) are more likely to emphasize the temporal persistence of, and identification with their national in-group, compared to people who are asked to think about a university exam (which also raises anxiety, but not of an existential type). Clearly, if stronger group identification predicts properties of social relationships such as stronger perceived support, which is known to exert positive effects on mental health (Lincoln & Chae, 2012), and shields people against negative mental states such as existential anxiety, then greater group identification should be associated with better mental health. Indeed, this is exactly what a growing body of literature is demonstrating (see Jetten, Haslam, & Haslam, 2012, for a recent collection). For instance, Reicher and Haslam (1996) had participants playing the role of either prisoners or guards within a purpose-built environment over 9 days. Results revealed that, over time, prisoners achieved high degrees of group identification that in turn led to a decrease in depressive symptoms, while guards’ group identification remained relatively low, which was associated with higher levels of depression. Consistent with this, Sani, Magrin, Scrignaro, & McCollum (2010) studied a sample of guards in an Italian prison and found that higher levels of identification with the group of prison guards led to lower levels of psychiatric disturbance and perceived stress. Interestingly, the separate effects of group identification and social contact on mental health have never been assessed in the same study, and therefore we do not know the extent to which the effects of group identification are independent from the effects of social contact. This is a problem and a current limitation of the social identity approach to health, because it could be legitimately argued that group identification is a mere epiphenomenon of social contact. That is, it could be that group identification is a byproduct of social contact, and that as such group identification is not a determinant of mental health in its own right, but only appears to be when its effects on mental health are not controlled for the effects of social contact. Therefore, it is important to explore the unique effects of group identification and social contact on mental health. Finding that group identification has implications for mental health that are beyond mere amount of interaction and participation in the life of a group would refine the social identity approach to health and advance our general understanding of the interplay between social ties and health. To explore this issue, we started with a study assessing the unique effects of both contact with individual members of the family group and family identification on mental Social contact and group identification as predictors of mental health 3 health. The family group was considered ideal for this type of investigation because the impact of families on people’s mental health is well documented (Elliott & Umberson, 2004). STUDY 1 Method Participants and procedure A sample of 194 Polish people (85 males, 109 females; M age = 34 years, range: 18–70 years) were recruited from the general public, using a snowball sampling approach. Of these, 100 lived in the United Kingdom and 94 in Poland. Participants completed a questionnaire with the aid of a researcher. Measures We measured family identification using the 4-item scale of group identification by Doosje, Ellemers, & Spears, (1995). Items (e.g. “I identify with other members of my family”) were rated using a scale ranging from 1 (I totally disagree) to 7 (I totally agree). To measure social contact with family members, we used a social network index (see Cohen et al., 1997 for a similar instrument). Specifically, we asked participants if they were married, and how many children, parents, in-laws, and relatives they were seeing or talking to on the telephone at least once every 2 weeks. We assigned one point when the participant was married and one point for every child, parent, and in-law that the participant declared to see or talk to on the telephone. Concerning relatives, we assigned one point if the participant specified that he or she saw or talked to one relative, two points for two to four relatives, and three points for five or more relatives. To assess mental health, we used three different indicators. First, we assessed depression using the 20-item Centre for Epidemiological Studies Depression Scale (CESD) (Radloff, 1977). Respondents specify how often they had felt or behaved in a certain way during the previous week (e.g., ’I had crying spells’), using a scale ranging from 1 (rarely or none of the time) to 4 (most of the time). Second, we used the Satisfaction with Life (SWL) scale by Diener, Emmons, Larsen, and Griffin, (1985). This is a 5item scale of global cognitive judgment about one’s life (e.g., ’In most ways my life is close to my ideal’). Respondents specify their degree of disagreement or agreement (1 = strongly disagree, 7 = strongly agree) with each item. Our third mental health indicator was the Perceived Stress Scale (PSS) (Cohen, Kamarck, & Mermelstein, 1983). This is a 10-item instrument tapping into one’s perceived ability to feel in control of emotions and everyday problems during the previous month (e.g., ’In the last month, how often have you felt that you were unable to control the important things in your life?’). Participants respond using a scale ranging from 0 (never) to 4 (very often). Finally, we assessed two variables to be used as controls in the statistical analyses: age and level of education. Concerning education, we distinguished between participants with up to a high school degree, and participants pursuing or having achieved a higher degree. These two categories were assigned the score of 0 and 1, respectively, so that they could be used as dummy variables in the analysis. 4 Fabio Sani et al. Table 1. Means, standard deviations, and reliabilities for variables, and intercorrelations in Study 1 Variable 1 2 3 4 5 6 7 1. Family identification (M = 6.00; SD = 1.13; – = .91) 2. Social contact (M = 5.06; SD = 1.95) .34∗∗ – – 3. Depression (M = 1.72; SD = .52; = .85) −.46∗∗ −.22∗∗ .26∗∗ −.61∗∗ – 4. Satisfaction with life (M = 4.62; SD = 1.16; .43∗∗ = .88) 5. Stress (M = 1.72; SD = .70; = .85) −.33∗∗ −.10 .69∗∗ −.46∗∗ – ∗∗ −.08 −.29∗∗ .28∗∗ −.29∗∗ – 6. Education .23 .39∗∗ −.07 .08 −.06 −.17∗ – 7. Age (M = 30.83; SD = 11.55) .27∗∗ Note. ∗∗ p .01; ∗ p .05. Results Only results for the overall sample are reported because the analyses conducted separately for subsamples (i.e., the two gender groups, and the Poland and UK groups) produced very similar results. Descriptives, reliabilities, and intercorrelations Means, standard deviations, and reliabilities (where applicable) for all the variables as well as the intercorrelations among the variables, are reported in Table 1. Concerning the relationship between education and other variables, we used point-biserial correlations. Measurement instruments had good reliabilities, with Cronbach alphas ranging from .85 to .91. Family identification was significantly correlated with depression (r = −.46), SWL (r = .43), and stress (r = −.33). Social contact was significantly correlated with depression (r = −.22) and SWL (r = .26). Also, family identification and social contact were significantly correlated with each other (r = .34). It should also be noted that education was significantly correlated with all three mental health indicators, with coefficients being in the high .20s. Regression analysis We performed three hierarchical multiple regressions. In each hierarchical regression, we entered the two control variables (education and age) at Step 1, while group identification and social contact were entered at Step 2. This enables an examination of the unique contribution of each variable in predicting mental health, as well as an assessment of the variance on mental health that group identification and social contact may explain in addition to the variance explained by age and education. Each regression used a particular mental health indicator (either depression, SWL, or stress) as the outcome. Table 2 reports the results. In Step 2, family identification was a significant predictor of all three health outcomes; betas were −.40, .37, and −.23, respectively, when depression, SWL, and stress were used as outcomes. On the other hand, social contact significantly predicted only SWL ( = .18). Among the control variables, education had a significant effect on all three health outcomes too, with betas being −.20, .18, and −.26, respectively, when Social contact and group identification as predictors of mental health 5 Table 2. Summary of hierarchical multiple regression analysis for variables predicting depression, satisfaction with life, and stress in Study 1 Depression B Step 1 Constant Education Age 2.08 −.36 −.00 SE .14 .09 .00 B 3.97 .70 .01 Satisfaction with life SE .30 .18 .01 B 2.18 .30∗∗ −.49 .09 .01 (R2 = .09) Stress SE .18 .11 .01 −.32∗∗ −1.00 (R2 = .10) −.33∗∗ .08 (R2 = .11) Step 2 Constant 3.10 .22 1.82 .46 2.93 .31 .42 .18 .18∗ −.38 .12 −.26∗∗ Education −.22 .08 −.20∗ Age .00 .00 .07 −.01 .01 −.09 .00 .01 .01 .39 .08 .37∗∗ −.15 .06 −.23∗∗ Family −.19 .04 −.40∗∗ identification .03 −.05 Social −.03 .02 −.13 .10 .05 .18∗ −.02 contact R2 = .27; R2 = .18∗∗ R2 = .16; R2 = .05∗ R2 = .27; R2 = .17∗∗ Note. ∗∗ p .01; ∗ p .05. depression, SWL, and stress were used as outcomes. Taken together, group identification and social contact explained a significant amount of variance in addition to the variance explained by age and education on each health outcome, especially depression ( R2 = .17) and SWL ( R2 = .18). Discussion These results suggest that family identification is more beneficial to mental health than mere contact with family members. It could be argued, however, that these findings are especially applicable to the family group because of the particularly high degree of meaning that people tend to attach to the family irrespective of how much contact people have with family members. In order to assess the extent to which these results might also apply to other social groups, we replicated this study using an army unit, a type of group that places great importance on working together and on contact between group members. STUDY 2 Method Participants and procedure One hundred and fifty members of an army unit (102 males, 48 females; M age : 39 years; age range: 25–56 years) from an Eastern European country volunteered to participate in the study. Participants completed a questionnaire individually at their leisure. 6 Fabio Sani et al. Measures To assess army identification, we used the 14-item scale of group identification developed by Leach et al. (2008). Items tap on group-based self-investment and selfdefinition (e.g., ’I am glad to be a member of this Army’; ’I have a lot in common with the average member of this Army’). Participants rated the items using a scale ranging from 1 (I totally disagree) to 7 (I totally agree). To measure social contact with other members of the army unit, we created a new instrument that covers multiple aspects of what researchers have traditionally considered as ‘social integration’. This instrument – which we consider good at capturing the intensity of social contact within this specific type of group – was based on three questions, respectively, asking the participant about (1) the number of different members of the army with whom he or she has a face-to-face conversation on a single day, (2) the number of different members of the army with whom he or she has a telephone conversation on a single day, and (3) the number of social events related to the army (parties, reunions, sports events, dinners, picnics, excursions, gatherings, etc.) attended in an average month. As far as mental health is concerned, we used three indicators. First, we assessed depression using the Beck Depression Inventory II (BDI II) (Beck, Steer, & Brown, 1996). This is a 21-item measure that is commonly used as a screening instrument of depressive symptoms in normal populations. Items tap into several dimensions (e.g., sadness, loss of pleasure, worthlessness, pessimism, and suicidal thoughts) and are responded to on 0–3 scale where higher scores indicate higher severity of symptoms. Second, we assessed SWL using the scale by Diener et al. (1985) that we had also used in Study 1. Third, we assessed job satisfaction by asking participants to specify their disagreement or agreement with two items (e.g., ’I find my profession enjoyable’) on a 7-point scale ranging from 1 (I totally disagree) to 7 (I totally agree). Concerning the control variables, we assessed age and an indicator of status, namely rank within the Army. We opted for rank instead of a more traditional socio-economic status indicator because status within specific professions and organizations is known to have effects on health (Marmot, 2004). Participants were assigned a number ranging from 1 to 12, with higher numbers indicating higher ranks (e.g., 1 = Sergeant; 12 = Colonel). The intervals between ranks can be legitimately considered as roughly equivalent, and therefore this variable was treated as continuous in the analyses. Results Analyses conducted separately for the two gender groups produced very similar results; therefore, we report results only for the overall sample. Descriptives, reliabilities, and intercorrelations Table 3 shows the means, standard deviations, and reliabilities (where applicable) for all the variables as well as the intercorrelations among the variables. (Note that, to calculate the overall score for social contact, we standardized scores on the three items and then added these scores together.) The measures had good reliability, with Cronbach alphas ranging from .86 to .90, and the two job satisfaction items being substantially correlated (r = .51). Army identification was significantly correlated with all mental health indicators; correlation with SWL Social contact and group identification as predictors of mental health Table 3. Means, standard deviations, and reliabilities for variables, and intercorrelations in Study 2 Variable 1 2 3 4 5 6 7 7 1. Army identification (M = 5.39; SD = 0.74; – = .90) 2. Social contact (M = 0.03; SD = 2.17) .19∗ – 3. Depression (M = 3.80; SD = 4.58; = .86) −.18∗ −.05 .12 4. Satisfaction with life (M = 4.91; SD = 1.06; .47∗∗ = .86) 5. Job satisfaction (M = 5.68; SD = .82; = .51) −.59∗∗ .25∗∗ 6. Rank (M = 7.49; SD = 3.28) .06 .21∗ 7. Age (M = 39.11; SD = 6.97) −.06 .20∗ Note. ∗∗ p .01; ∗ p .05. – −.33∗∗ – −.25∗∗ .53∗∗ – −.01 .12 .00 – −.12 −.06 −.13 .48∗∗ – (r = .47) and job satisfaction (r = .59) were especially high. Social contact was significantly correlated only with job satisfaction (r = .25) among the three mental health indicators. However, social contact was correlated significantly with both rank (r = .21) and age (r = .20). Concerning the relationship between army identification and social contact, a significant but relatively modest correlation was observed (r = .19). Regression analysis As in the first study, we performed three hierarchical multiple regressions. In each regression, we entered rank and age in Step 1, and army identification and social contact in Step 2. Each regression used a different mental health indicator as outcome. Results are reported in Table 4. Table 4. Summary of hierarchical multiple regression analysis for variables predicting depression, satisfaction with life, and job satisfaction in Study 2 Depression B Step 1 Constant Rank Age SE B 5.40 .06 −.02 Satisfaction with life SE .51 .03 .02 B 6.37 .02 −.02 Job satisfaction SE .40 .02 .01 0.30 2.31 −.15 .15 .12 .07 −.11 .18 (R2 = .03) .18 −.16 (R2 = .03) .08 −.18 (R2 = .03) −.00 −.14 −.55∗∗ .16∗ R2 = .36∗∗ Step 2 Constant 6.63 3.98 Rank −.13 .15 −.09 Age .11 .07 .17 Army −1.13 .56 −.18∗ identification Social −.08 .20 −.03 contact R2 = .06; R2 = .04 Note. ∗∗ p .01; ∗ p .05. 1.57 .04 −.02 .68 .01 .80 .03 .01 .11 .04 .13 −.11 .46∗∗ .02 R2 = .21∗∗ 2.99 .00 −.02 .61 −.06 .56 .02 .01 .08 .03 R2 = .24; R2 = .38; 8 Fabio Sani et al. In Step 2, army identification had significant unique effects on all three mental health outcomes; impact on both SWL ( = .46) and job satisfaction ( = .55) was remarkable. On the other hand, social contact had a significant, albeit relatively modest impact only on job satisfaction ( = .16). Rank and age had no significant impact on any mental health outcome. Taken together, group identification and social contact explained a significant amount of variance in addition to the variance explained by rank and age, on both SWL ( R2 = .21) and job satisfaction ( R2 = .36). Discussion These results are consistent with those of the previous study, and confirm that subjective identification with a group has stronger implications for mental health than contact with other group members. GENERAL DISCUSSION Across two studies involving two different social groups (i.e., the family and an army unit), we found that measures of group identification and traditional measures of social integration based on social contact (i.e., intensity of interactions with group members and participation in group-related activities) were only modestly related, and that group identification was better than social contact at predicting mental health. These findings imply that the effects of group identification on mental health cannot be explained by social contact, and that therefore group identification cannot be seen as a mere epiphenomenon of social contact. In fact, group identification appears to play a central role in the process leading from social integration to health. In line with the social identity approach to health (Haslam, Jetten, Postmes, & Haslam, 2008), we can speculate that group identification (1) affords a sense of structure and meaning, and (2) constitutes a precondition for positive social relationships based on trust, support, and respect. This in turn paves the way for positive mental states and mental health. On the other hand, social contact seems to have relatively modest unique effects on mental health. This could be because the effects of social contact are partly explained by group identification. Also, it could be that the positive effects of social contact on mental health that some people enjoy are balanced out by the negative effects that other people may experience. It should be kept in mind that for some people social contact may be a source of negative affect (Rook, 1984), and therefore it may be detrimental to health (e.g., Hooley, 2004). A wider implication of our findings is that in order to understand how in-groups may improve our mental health we must take into account the fact that they are experienced as ‘we’ and ‘us’, thereby defining who is either similar (and therefore to be trusted) or different (and therefore to be seen with some suspicion) to self. Without acknowledging this, it is hard to explain how and under which circumstances contact with others and social support received from others is going to be beneficial. From a methodological point of view, our findings demonstrate that when investigating the effects of social integration on mental health, researchers should include measures of group identification among the indicators of social integration. Occasionally, researchers have acknowledged that the way in-groups are subjectively experienced Social contact and group identification as predictors of mental health 9 should be seen as an aspect of social integration (e.g., Brissette et al., 2000). Unfortunately, however, the subjective dimension of group life is almost always ignored in empirical research about social integration and mental health. Group identification is an ideal indicator of this subjective dimension, as it is based on how the group is perceived and experienced, especially in terms of its relationship to the self (Jetten et al., 2012). Our findings have practical implications too, in that they provide important suggestions on how to implement a social cure (Jetten et al., 2012). Specifically, we should not assume that by favouring interaction and active participation in group activities we will automatically improve people’s mental health. In fact, social contact per se might produce very modest mental health gains, if any, unless we make sure that the sort of social contact that is encouraged contributes to enhancing group identification. To conclude, it is important to stress the fact that as well as providing a response to some relevant questions, our findings raise further questions, particularly concerning the interplay between group identification and social contact. To what extent, and under which circumstances does higher social contact intensify group identification? Does higher group identification motivate people to get involved in more intense contact with group members? Might strong group identification be more detrimental to health than lack of identification, in cases where contact with group members is impossible (e.g., exiles and their families)? These are important and intriguing questions that, in our opinion, future research should address. References Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation. Brissette, I., Cohen, S., & Seeman, T. E. (2000). Measuring social integration and social networks. In S. Cohen, L. G. Underwood, & B. H. Gottlieb (Eds.), Social support measurement and intervention: A guide for health and social scientists (pp. 53–85). Oxford: Oxford University Press. Cohen, S. (2004). Social relationships and health. American Psychologist, 59, 676–684. doi:10.1037/0003-066X.59.8.676 Cohen, S., Doyle, W. J., Skoner, D. P., Rabin, B. S., & Gwaltney, Jr., J. M. (1997). Social ties and susceptibility to the common cold. Journal of the American Medical Association, 227, 1940–1944. doi:10.1001/jama.1997.03540480040036 Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396. Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71–75. doi:10.1207/s15327752jpa4901 13 Doosje, B., Ellemers, N., & Spears, R. (1995). Perceived intragroup variability as a function of group status and identification. Journal of Experimental Social Psychology, 31, 410–436. doi:10.1006/jesp.1995.1018 Elliott, S., & Umberson, D. (2004). Recent demographic trends in the US and implications for wellbeing. In J. Scott, J. Treas, & M. Richards (Eds.), The Blackwell companion to the sociology of families (pp. 34–53). Oxford: Blackwell Publishing. Glass, T. A., Mendes de Leon, C. F., Bassuk, S. S., & Berkman, L. F. (2006). Social engagement and depressive symptoms in late life. Journal of Aging and Health, 18, 604–628. doi:10.1177/0898264306291017 Hanson, B. S., Isacsson, S.-O., Janzon, L., & Lindell, S.-E. (1989). Social network and social support influence mortality in elderly men. American Journal of Epidemiology, 130, 100–111. Haslam, A. S. (2004). Psychology in organizations: The social identity approach (2nd ed.). London: Sage. 10 Fabio Sani et al. Haslam, S. A., Jetten, J., Postmes, T., & Haslam, C. (2008). Social identity, health, and well-being: An emerging agenda for applied psychology. Applied Psychology, 58, 1–23. doi:10.1111/j.14640597.2008.00379.x Hogg, M. A., & Hardie, E. A. (1991). Social attraction, personal attraction and selfcategorisation: A field study. Personality & Social Psychology Bulletin, 17, 175–80. doi:10.1177/014616729101700209 Hooley, J. M. (2004). Do psychiatric patients do better clinically if they live with certain kinds of families? Current Directions in Psychological Science, 13, 202–205. doi:10.1111/j.09637214.2004.00308.x Jetten, J., Haslam, S. A., & Haslam, C. (2012). The social cure: Identity, health, and well-being. New York: Psychology Press. Kawachi, I., & Berkman, L. F. (2001). Social ties and mental health. Journal of Hurban Health: Bulletin of the New York Academy of Medicine, 78, 458–467. doi:10.1093/jurban/78.3.458 Leach, C. W., van Zomeren, M., Zebel, S., Vliek, M. L. W., Pennekamp, S. F., Doosje, B., & Spears, R. (2008). Group-level self-definition and self-investment: A hierarchical (multicomponent) model of in-group identification. Journal of Personality & Social Psychology, 95, 144–165. doi:10.1037/0022-3514.95.1.144 Lincoln, K. D., & Chae, D. H. (2012). Emotional support, negative interaction, and major depressive disorder among African Americans. Social Psychiatry & Psychiatric Epidemiology, 47(3), 361– 372. doi:10.1007/s00127-011-0347-y Marmot, M. (2004). Status syndrome. London: Bloomsbury. Platow, M. J., Voudouris, N. J., Coulson, M., Gilford, N., Jamieson, R., Najdovski, L., . . . Terry, L. (2007). In-group reassurance in a pain setting produces lower levels of physiological arousal: Direct support for a self-categorization analysis of social influence. European Journal of Social Psychology, 37, 649–660. doi:10.1002/ejsp.381 Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. doi:10.1177/014662167700100306 Reicher, S., & Haslam, S. A. (1996). Rethinking the psychology of tyranny: The BBC prison study. British Journal of Social Psychology, 45, 1–40. doi:10.1348/014466605X48998 Rook, K. S. (1984). The negative side of social interaction: Impact on psychological wellbeing. Journal of Personality and Social Psychology, 46, 1097–1108. doi:10.1037/00223514.46.5.1097 Sani, F., Herrera, M., & Bowe, M. (2009). Perceived collective continuity and ingroup identification as defence against death awareness. Journal of Experimental Social Psychology, 45, 242–245. doi:10.1016/jesp.2008.07.019 Sani, F., Magrin, M. E., Scrignaro, M., & McCollum, R. (2010). Ingroup identification mediates the effects of subjective ingroup status on mental health. British Journal of Social Psychology, 49, 883–893. doi:10.1348/014466610X517414 Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behaviour. In S. Worchel & W. G. Austin (1986). Psychology of intergroup relations (pp. 2–24). Chicago: Nelson-Hall Publishers. Thoits, P. A. (1983). Multiple identities and psychological well-being: A reformulation and test of the social isolation hypothesis. American Sociological Review, 48, 174–187. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. (1987). Rediscovering the social group: A self-categorisation theory. Oxford and New York: Blackwell. Received 19 August 2011; revised version received 21 February 2012