Journal of Adolescence 35 (2012) 1505–1514
Contents lists available at SciVerse ScienceDirect
Journal of Adolescence journal homepage: www.elsevier.com/locate/jado
Birds of an ethnic feather? Ethnic identity homophily among college-age friends Moin Syed*, Mary Joyce D. Juan Department of Psychology, University of Minnesota, 75 East River Road, Minneapolis, MN 55455, United States
a b s t r a c t Keywords: Ethnic identity Friendships Homophily
This study assessed the degree to which pairs of friends report similar levels of ethnic identity. College-age friends (n ¼ 107 pairs; N ¼ 214 overall) completed measures of ethnic identity exploration and commitment, identity synthesis, relationship closeness, and frequency of talking to friends and family about ethnicity-related issues. Participants were ethnically-diverse and constituted three types of dyads: both ethnic minorities (n ¼ 51), both White (n ¼ 24), or mixed ethnic minority and White (n ¼ 32). Overall, friends reported similar levels of ethnic identity exploration and commitment. The ethnic composition of the pair moderated similarity, with ethnic minority pairs showing greater similarity than the White and mixed pairs. Frequency of ethnicity-related discussions with friends and family, but not relationship closeness, mediated the observed similarity for ethnic minority pairs. These findings suggest a level of ethnic identity homophily between ethnic minority friends that can be explained by interactional mechanisms. Ó 2011 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.
Two decades of extensive research on ethnic identity have deepened our understanding of its development. In terms of developmental trajectories, we now know that that ethnic identity emerges at the onset of adolescence and continues to be of developmental significance well into young adulthood, waxing and waning in response to school transitions (French, Seidman, Allen, & Aber, 2006; Pahl & Way, 2006; Syed & Azmitia, 2009). Moreover, ethnic identity has been consistently linked to a number of positive outcomes, including greater well-being, academic engagement, and the ability to cope with discrimination (Quintana, 2007; Smith & Silva, 2011). Because of this accumulated knowledge, the current wave of research on ethnic identity is moving away from investigating the broad, universal, aspects of ethnic identity, and beginning to examine in depth the mechanisms that may contribute to its development. In this spirit, the purpose of the present study was to examine how ethnic identity develops in the context of friendships. Despite recent increased attention to how friendships may play an important role in ethnic identity development (e.g., Kiang & Fuligni, 2009), the majority of this research relies of individuals’ self-report about their friendships, rather than simultaneously assessing pairs of friends. Accordingly, in this study we investigated the degree to which pairs of friends reported similar levels of ethnic identity and tested for how ethnic identity similarity was moderated by the ethnic composition of the pair and mediated by indicators of relationship quality (i.e., relationships closeness, ethnicity-related discussions with friends and family). Social context of ethnic identity In the present study we adopted Phinney’s (1990) developmental model of ethnic identity. Under this model, ethnic identity is defined as the degree to which individuals identify with their ethnic group. Phinney’s model is an adaptation of * Corresponding author. Tel.: þ1 612 625 9501. E-mail address:
[email protected] (M. Syed). 0140-1971/$ – see front matter Ó 2011 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.adolescence.2011.10.012
1506
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
Marcia’s (1980) operationalization of Erikson’s (1968) theory of identity development. Accordingly, ethnic identity is believed to develop through two distinct yet interrelated processes: exploration, or the degree to which individuals participate in ethnicity-related activities or engage in a search for what their ethnic background means, and commitment, or the affective bond individuals have with their ethnic background and the degree to which they derive a sense of belongingness to their ethnicity. Numerous studies have investigated the social context of ethnic identity; however the studies have adopted different definitions of social context and have produced mixed results. For example, studies on how ethnic identity is related to ethnic density, or the degree of representation of individuals’ ethnic groups in the region in which they live, have not produced consistent findings (Juang, Nguyen, & Lin, 2006; Syed, Azmitia, & Phinney, 2007; Umaña-Taylor, 2004). Implied in the rationale for these studies, yet not explicitly measured, is that having exposure to and socializing with co-ethnics will lead to stronger ethnic identities. Thus, a more specific focus on the friendship patterns of youth may be a fruitful direction for understanding the social context of ethnic identity development. Research has demonstrated that adolescents’ closest friends tend to be of the same ethnicity (Kao & Joyner, 2004), but less is known about how these friendships might be associated with ethnic identity. A study relying on self-report data among a sample of Armenian-, Vietnamese-, and Mexican-American adolescents found that greater interactions with co-ethnics was associated with stronger ethnic identities (Phinney, Romero, Nava, & Huang, 2001). Moreover, co-ethnic interactions had stronger direct effects on ethnic identity than did parental cultural socialization, which is the degree to which parents transmit messages to their children about their cultural and ethnic background. Despite the findings of Phinney et al. (2001) more than 10 years ago, little research has followed up on how friends contribute to ethnic identity development. Recently, Kiang and colleagues have conducted a number of studies that have taken up this issue (Kiang & Fuligni, 2009; Kiang, Harter, & Whitesell, 2007; Kiang, Witkow, Baldelomar, & Fuligni, 2010). Extending the findings of Phinney et al. (2001), Kiang et al. (2010) demonstrated that ethnically-diverse adolescents’ increases over time in proportion of co-ethnic friends was associated with increases over time in ethnic identity exploration and commitment. Other studies have examined self-reports of how ethnic identity is expressed in different relational contexts, including co-ethnic peers, contra-ethnic peers (i.e., peers of different ethnicity), and family. For example, Kiang and Fuligni (2009) found that, on average, young adults reported incongruence in ethnic identity among the three relational contexts, with parents being the highest, followed by co-ethnic peers, and then contra-ethnic peers. This trend was the same for the four ethnic groups in the study: Latinos, Asian Americans, Filipino Americans, and Whites. One limitation of the past research on co-ethnics and ethnic identity is that they have relied on self-reports from a single individual, either about the number of co-ethnic friends they have or how their ethnic identities are expressed in different relational contexts. Implied in these studies is that friends may be similar in their levels of ethnic identity, but this has yet to be tested. The purpose of the present study is to do just that. Homophily, ethnicity, and ethnic identity Friendship homophily, or the similarities in attitudes and behaviors of friends, is believed to occur through the processes of selection and socialization (Kandel, 1978). That is, friends share similarities due to individual tendencies to befriend like-others and because the initial similarity tends to get stronger over time through successive interactions. Friendship homophily has been demonstrated in scores of studies across domains, including antisocial behavior, depressive symptoms, and prosocial behaviors (see Brechwald & Prinstein, 2011 for a review). Homophily is purported to serve an important identity-related function, as peers are a primary source through which individuals receive support and evaluation to determine who one is and who one is not (Brechwald & Prinstein, 2011; Thorne, Korobov, & Morgan, 2007). Thus, it is odd that identity homophily, in terms of the degree to which individuals from the same peer group are similar in their developmental levels of identity, has not previously been examined. In their review, Brechwald and Prinstein (2011) advocate for three areas of research on friendship processes: 1) attention to cultural issues and ethnic minority populations, 2) moderators that strengthen or weaken similarities between friends, and 3) mediators that can help explain similarity. The present study addresses these three needs, as described below, in turn. With the focus on the processes involved in ethnic identity development, the current study clearly meets the call of needing to attend to cultural and ethnic minority issues in the study of friendships. Researchers have demonstrated connections between ethnic identity and a number of different life domains, including well-being, academic achievement, career choice, coping, and rejection of body image ideals (Chavous et al., 2003; Perron, Vondracek, Skorikov, Tremblay, & Corbière, 1998; Quintana, 2007; Rogers Wood & Petrie, 2010; Smith & Silva, 2011). Thus, understanding ethnic identity in the context of friendships would serve as a useful bridge for friendship researchers to increasingly incorporate constructs of relevance to ethnic minority youth. However, it is also important to investigate and understand how ethnic identity processes may be similar or different from other identity processes, including developing a general sense of identity (Schwartz, 2008; Syed & Azmitia, 2009). That is, it is possible that observed similarities between friends in ethnic identity is because friends have similar levels of clarity and synthesis of their identity in general. Thus, we also examined identity similarity on a domain general measure of identity synthesis. One of the most commonly tested moderators in homophily research is participant background characteristics, such as gender or age (Brechwald & Prinstein, 2011). In the present study, we examined how the ethnic composition of the friendship pair moderated similarity in ethnic identity. We focused on three different ethnic compositions: where both friends were
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
1507
ethnic minorities (EM–EM), where one friend was an ethnic minority and one was White (EM–White), and where both friends were White (White–White). The rationale for examining ethnic minorities as a group is supported by the literature on both theoretical and empirical grounds. Theoretically, ethnic identity is specifically a minority identity. That is, ethnic identity only has meaning in societies that are ethnically-diverse and involve unequal power and access to resources along ethnic lines (Phinney, 1990; Tajfel & Turner, 1986). Thus, while not identical, the circumstances that give rise to the meaningfulness of ethnic identity for American ethnic minority groups are similar. Furthermore, the current study was conducted at a university that is predominantly White (78%), and therefore ethnic minority students did not enjoy a large representation of their own ethnic group from which to make friends. It is also worth noting that many of the past studies on ethnic identity and friendships analyzed their data at the pan-ethnic level (e.g., Asian American, Latino), and therefore were not capturing ethnicspecific friendships processes (e.g., Kiang & Fuligni, 2009; Schwartz, Galliher, & Domenech-Rodríguez, 2011). The similarity of ethnic identity processes among American ethnic minority groups also derives empirical support. Indeed, research on ethnic identity has typically found that the correlates of ethnic identity do not vary appreciably among different ethnic groups. For example, a recent meta-analysis found no difference among ethnic minority groups in the association between ethnic identity and well-being (Smith & Silva, 2011). In terms of White youth, findings consistently show that Whites report weaker ethnic identities than do ethnic minorities (Roberts et al., 1999; Syed & Azmitia, 2009). However, research on the correlates of ethnic identity among White youth does not show as consistent differentiation from ethnic minorities. Thus, levels of ethnic identity may be lower for Whites compared to ethnic minorities, but how ethnic identity is correlated with other factors may be similar. However, given that ethnic identity tends to be more important for ethnic minority groups, the processes of friendship selection and socialization may be more strongly tied to ethnic identity for these groups than for others. Accordingly, we hypothesized that similarity in ethnic identity would be moderated by the ethnic composition of the friendship pair, with the strongest similarity observed for the EM–EM pairs. In asserting that friends are an important source for ethnic identity, it is implied that this is due, at least in part, to the quality of the friendship. Thus, various aspects of friendship quality may serve as mediators of friends’ similarity in ethnic identity. Indeed, in their analysis of the Add Health dataset, Kao and Joyner (2004) found that co-ethnic friends reported engaging in more shared friendship activities that indexed closeness than did contra-ethnic friends. Moreover, a study of Latino young adults found greater reports of personal disclosure with Latino friends than non-Latino friends (Schwartz et al., 2011). Accordingly, the most immediately evident mediator of identity similarity may be relationship closeness; friends are similar in their levels of ethnic identity because they have a close relationship based on shared interests and life pursuits. Thus, we hypothesized that friendship closeness would mediate the similarity in ethnic identity between friends. It is also possible that a specific aspect of the friendship may account for the similarity: having the ability to talk with them about ethnicity-related issues. Talking with friends about ethnicity-related issues may help clarify and deepen their thoughts about the role of their ethnicity for their own lives. Moreover, talking about ethnicity-related issues with some regularity may keep that identity active in one’s mind, forcing the identity to have cross-context relevance as opposed to situational salience. Thus, we hypothesized that frequency with which friends talk to each other about ethnicity-related issues would mediate ethnic identity similarity. Youth, however, also share a common ethnicity with their families. In their study of relational ethnic identity among ethnically-diverse young adults, Kiang and Fuligni (2009) found a consistent pattern across groups wherein relational ethnic identity was significantly higher with parents than with co-ethnic peers (both of which were higher than contra-ethnic peers). These findings suggest that family may be a prime context for the expression of ethnic identity, and therefore discussions with family about ethnicity-related issues may also account for the identity similarity of friends. However, a study of memory-telling about ethnicity-related events suggests a different possibility (Syed, in press). In this study, young adults were about twice as likely to report telling their ethnicity-related memories to peers rather than family. Further analysis indicated a more nuanced selection of audience based on the type of memory. Experiences of discrimination were more likely to be shared with peers, whereas experiences directly tied to leaning about one’s cultural heritage were more likely to be told to family. Thus, although peers and parents serve as a resource for talking about ethnicity-related issues, peers appear to be a more salient context for ethnicity-related discussions for young adults and therefore talking to them about ethnicity is more likely to account for the identity similarity between friends. In the present study we hypothesize that both talking with peers and family about ethnicity-related topics would mediate ethnic identity similarity, but we also hypothesized that talking with peers would be the primary mediator over talking with family.
The present study The purpose of the present study is twofold: 1) to assess the degree to which pairs of friends report similar levels of ethnic identity, and 2) to test for moderators (ethnic composition of the pair) and mediators (relationships closeness, ethnicityrelated discussions with friends and family). In line with these three goals, we advanced the following hypotheses: 1. Friends will evidence similarity in ethnic identity exploration and commitment. We also assessed similarity between friends in general identity synthesis, but did not advance specific hypotheses about the association. 2. The degree of similarity will be moderated by ethnic composition of the friends, such that EM–EM pairs will show higher levels of similarity than EM–White or White–White pairs.
1508
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
3. Relationship closeness, ethnicity-related discussions with friends, and ethnicity-related discussions with family will each independently serve as mediators of the similarity in ethnic identity between friends. Subsequent analyses will indicate that the mediation is carried by discussions with friends. Method Sample Participants were recruited from a large public university in the U.S. Midwest through flyers on campus and online bulletin boards advertising the “Ethnicity and Friends Study.” Interested individuals contacted the researchers and completed a prescreen instrument to determine their eligibility for the study. Primary eligibility requirements included being at least age 18 and having a friend of similar age (two years) that they could bring with them for the study. Because of our particular interest in identity among ethnic minorities, we also established an approximate quota of no more than 25% of our target N (100 pairs) being from White backgrounds. Thus, once we had reached this number, we gave priority to ethnic minorities to ensure adequate representation. We only sought to regulate the ethnicity of the target participant (i.e., the one responding to the ad) and not the ethnicity of the friend they brought with them. The final sample comprised 107 pairs of friends (N ¼ 214 individuals; 56% women, 68% born in the U.S., M age ¼ 20.39, SD ¼ 2.25). Ethnicity of the target participants was 63% ethnic minority and 37% White, with the over-representation of ethnic minorities relative to the campus (78% White) being the result of oversampling. The specific ethnic composition was as follows: 37% White (most self-identified as White or Caucasian), 32% Asian American (mostly self-identified as Chinese American, Hmong American, Vietnamese American, Korean American, and Asian American), 11% Mixed-ethnic (mostly AsianWhite or Black-White), 9% South Asian (mostly Indian or Sri Lankan), 6% Black (self-identified East African, Black, and African American), 2% Arabic/Middle Eastern (Egyptian or Arab American), 1.4% Latino (two Latino and one Columbian American), and 1.4% Native American (one Navajo, one Objiwe, and one Native American). Because we did not specify the ethnicity of the friend that the targets brought, there were three naturally occurring pairs with respect to ethnic composition: ethnic minority–ethnic minority (EM–EM; n ¼ 51), ethnic minority–White1 (EM–White; n ¼ 32), and White–White (n ¼ 24). These three groups were used for the primary analysis reported below. Target participants were not required to bring a same-gender friend. As a result, the gender composition of the groups was 41% female–female, 28% male–male, and 31% female–male. Gender composition did not vary significantly with ethnic composition, and the three gender pair groups reported similar levels of relationship closeness. Moreover, despite repeated tests, there is almost no evidence for gender differences in the processes of ethnic identity exploration and commitment (see Smith & Silva, 2011 for a meta-analysis on links to well-being). Procedure After eligibility was determined, the target participant and friend were scheduled for a one-hour session in the PI’s laboratory on campus. Each friend was situated individually at a computer to take the survey. There was a large barrier separating the two computer stations to ensure that the participants could not see their friend or their friend’s computer screen. The research assistant was also in the room sitting in a chair facing the wall to be on hand for any questions and to ensure that the participants did not talk to each other. The survey was hosted by the online service http://www. surveymonkey.com, and was completely computer guided. The survey included numerous rating-scale and open-ended questions pertaining to ethnicity, identity, personality, well-being, and friendships. Participants took anywhere between 20 and 40 min to complete the survey. After both participants were finished, they engaged in a 10-min audio-recorded catch-up conversation. Those data are not included in the present study. Each participant was compensated $15.00 US for their time. Measures Ethnic identity We used the revised 12-item version of the Multigroup Ethnic Identity Measure (MEIM; Phinney, 1992; Roberts et al., 1999), which has shown strong reliability and validity (Roberts et al., 1999). The scale contains a 5-item exploration subscale and a 7-item commitment subscale. Participants responded on a 4-point Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). Items were averaged for each subscale so that higher values represent greater exploration or commitment. Sample items include “To learn more about my ethnic background, I have often talked to other people about my ethnic group” for the exploration subscale and, “I feel a strong attachment toward my own ethnic group” for the commitment subscale. Cronbach’s alpha for exploration was .78 and for commitment was .90. The scales were similarly reliable for ethnic 1 This group actually comprises two groups: ethnic minority targets who brought a White friend (n ¼ 19) and White targets who brought an ethnic minority friend (n ¼ 13). Because of the potential for differences between these two groups, we examined all similarity correlations separately and found that they were nearly identical. Thus, the two groups were combined into a single group for all analyses.
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
1509
minorities and Whites, and these values are similar to those reported in prior research with ethnically-diverse college students (e.g., Syed & Azmitia, 2009). Identity synthesis Participants completed an 11-item version of the identity synthesis subscale of the Erikson Psychosocial Inventory Scale (EPSI), which has demonstrated adequate reliability and validity (Rosenthal, Gurney, & Moore, 1981; Schwartz, 2007). The EPSI identity scale assesses global identity synthesis and is not domain specific. Previous factor analytic work has suggested that one of the original 12 items does not fit with a one-factor solution, so that item was not used in the present study (see Reis & Youniss, 2004). The scale includes items that reflect both successful and unsuccessful identity synthesis pertaining to clarity, authenticity, and satisfaction with the self. Sample items include, “I’ve got a clear idea of what I want to be” and “I change my opinion about myself a lot” (reverse coded). Items were measured on a 5-point Likert scale ranging from 1 (hardly ever true) to 5 (almost always true). Negatively worded items were reverse coded and all items were averaged so that higher values represent a greater degree of identity synthesis. Cronbach’s alpha in the present study was .87 which is similar to the alphas obtained in the aforementioned studies. Relationship closeness Relationship closeness was assessed using a composite of the Subjective Closeness Index and the Inclusion of Other in the Self scale. The Subjective Closeness Index (Berscheid, Snyder, & Omoto, 1989) is a two-item measure designed to measure individuals own sense of subjective closeness with a particular friend. The items are “Relative to all your other relationships (both same and opposite sex), how would you characterize your relationship with this person?” and “Relative to what you know about other people’s close relationships, how would you characterize your relationship with this person?” Participants rated the items on a seven point scale with 1 ¼ not at all close and 7 ¼ extremely close. The Inclusion of Other in the Self (IOS) scale (Aron, Aron, & Smollan, 1992) is a single pictorial measure of closeness that comprises a series of seven progressively interlocking circles that represent the “self” and “other.” A score of 1 is given to the first picture that shows the circles as completely separate and a score of 7 is given to the final picture in which the circles almost fully overlap. Participants were asked to select the picture that best describes the relationship with their friend. In developing the IOS, Aron et al. (1992) found that the Subjective Closeness Index taps into a latent factor “feeling close” and the IOS taps into both feeling close and “behaving close.” Thus, the two scales were averaged into a single index of relationships closeness in order to assess both the “feeling” and “behaving” aspects of closeness. Cronbach’s alpha for the three-item measure was .88. Ethnicity-related discussions with others We assessed the frequency with which participants talk with their friends and family about ethnicity-related issues using a single item for each. The measure of discussions with friends is not specific to the friend who co-participated in the study, but rather is the general frequency with which they talk to friends about ethnicity-related issues. The two items are, “How frequently do you discuss issues of race, ethnicity, or culture with your friends?” and “How frequently do you discuss issues of race, ethnicity, or culture with your family (including parents, siblings, aunts/uncles, etc.)?” Both questions were assessed on a four-point scale ranging from 1 ¼ not at all to 4 ¼ a lot. As described below, the items were examined separately in all analyses due to potential differences in how identity similarity is associated with talking to friends vs. family. Results We carried out the analysis in two steps. First, we calculated pairwise intraclass similarity correlations between friends on ethnic identity exploration, ethnic identity commitment, and identity synthesis. The pairwise interclass correlation coefficient (denoted as r when reporting the results) is a method for assessing similarity among non-distinguishable dyads (Griffin & Gonzalez, 1995; Kenny, Kashy, & Cook, 2006). Second, using partial pairwise intraclass similarity correlations (denoted as rp) we tested for three mediators of the observed identity similarities: relationship closeness and discussions with friends and discussion with family about ethnicity. All primary analyses were carried out split by the three ethnic composition pairs; however, we also provide the correlations for the total sample. We used pairwise intraclass similarity correlations rather than more sophisticated modeling techniques (e.g., regression, path analysis) because such techniques imply a causal ordering. We do not wish to specify that one friend’s level of ethnic identity is uni-directionally causing the other friend’s level of ethnic identity. Thus, rather than specifying independent and dependent variables, our analysis focuses on the bivariate similarity between friends. Identity similarity between friends For our first hypothesis we specified that friends would demonstrate similarity in ethnic identity exploration, commitment, and identity synthesis. This hypothesis was supported for the two ethnic identity processes, r ¼ .22, p ¼ .03 for exploration and r ¼ .34, p < .001 for commitment, but not for identity synthesis, r ¼ .12, p ¼ .20 (Table 1). In our second hypothesis we predicted that the ethnic composition of the pair would moderate the similarity in identity, with the EM–EM pairs showing the highest similarity in ethnic identity exploration and commitment. This hypothesis was supported. EM–EM pairs’ scores on exploration and commitment were significantly correlated (rs ¼ .35, p ¼ .01 and .35,
1510
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
Table 1 Pairwise intraclass similarity correlations between friends for all study variables.
Overall EM–EM pairs EM–White pairs White–White pairs
Ethnic identity exploration
Ethnic identity commitment
Identity synthesis
Friendship closeness
Ethnicity-related discussions – friends
Ethnicity-related discussions – family
.22* .35** .16 .31
.34*** .35** .10 .16
.12 .22 .15 .22
.67*** .67*** .67*** .58**
.49*** .62*** .31 .17
.20* .34* .16 .32
Note. EM ¼ ethnic minority. *p .05, **p .01, ***p .001.
p ¼ .01, respectively), whereas this was not the case for the other two pair types (Table 1). Identity synthesis was not significant for any of the three groups. We then compared the magnitude of the correlations using the Fischer z-test, one-tailed due to our directional hypothesis. The similarity correlation for exploration was significantly larger for EM–EM pairs compared with both EM–White pairs, z ¼ 2.28, p ¼ .02, and White–White pairs, z ¼ 3.82, p ¼ .008. The correlations for commitment, however, did not significantly differ. Testing for mediators: closeness and talking Our third and final hypothesis was concerned with mediators of the similarity in ethnic identity between friends. To this end, we predicted that relationship closeness, engaging in ethnicity-related discussions with friends, and ethnicity-related discussions with family, would each independently account for the similarity between friends. Mediation is often assessed using a “causal steps” approach in multiple regression (e.g., Baron & Kenny, 1986). However, that method for testing mediation presumes a specific causal sequence among the variables of interest. Such a method is not consistent with the goals of our study. For example, we do not wish to test whether one friend’s levels of ethnic identity causes how much they talk with their friends about ethnicity, which in turn causes the other friend’s level of ethnic identity. Rather, we began with non-directional similarity correlations and wished to test for alternative constructs that may explain the observed similarity. Thus, for these analyses we computed partial similarity correlations, controlling for the mediator of interest. Although using partial correlations to test mediation has been discouraged by some researchers (e.g., Preacher & Hayes, 2008), again we adopted this approach due to the unique situation of testing for what accounts for similarity between friends, rather than specifying a causal chain of effects (IV / mediator / DV). The approach that we used in the analysis is similar to that used by other researchers investigating similarity between friends and couples (e.g., Gonzaga, Campos, & Bradbury, 2007). Because we are not aware of any method for testing the statistical difference between a partial correlation and its corresponding zero-order correlation, we relied on the statistical significance of each for our interpretation of mediation. Due to the limitations of this approach, we also calculated a reduction of the proportion in variance explained when moving from the zero-order to partial correlations so as to give an indicator of the size of the effect (Singer & Willet, 2003). Relationship closeness On average participants rated being very close with one another (M ¼ 5.21, SD ¼ 1.04 for target, M ¼ 5.08, SD ¼ 1.14 for friends), and friends rated closeness similarly (r ¼ .67). These values were then partialled out of the similarity correlations reported above. For the sample overall, partialling out relationship closeness did not change the magnitude of the correlations in any meaningful way (Table 2). Similarly, partialling closeness for each pair type also did not alter the correlations. Indeed, the similarity correlations for ethnic identity exploration and commitment among the EM–EM pairs remained significant. Thus, relationship closeness cannot explain the observed identity similarity for this group. Ethnicity-related discussions with friends and family We had two items about the frequency with which the participants talk to other people about ethnicity and culture, one pertaining to friends and one to family. For the sample overall, participants rated having discussions with friends (for target, Table 2 Bivariate and partial pairwise intraclass similarity correlations for ethnic identity exploration and commitment. Ethnic identity exploration
Overall EM–EM pairs EM–White pairs White–White pairs
Ethnic identity commitment
Raw
Closeness partialled
Discussions with friends partialled
Discussions with family partialled
Raw
Closeness partialled
Discussions with friends partialled
Discussions with Family Partialled
.22* .35** .16 .31
.21* .31* .16 .35
.10 .19 .15 .33
.14 .26 .13 .37
.34*** .35** .10 .16
.35*** .32* .13 .20
.29** .21 .08 .05
.29** .24 .10 .17
Note. EM ¼ ethnic minority. *p .05, **p .01, ***p .001.
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
1511
M ¼ 2.46, SD ¼ .70; for friend, M ¼ 2.34, SD ¼ .79) more often than family (for target, M ¼ 2.22, SD ¼ .73; for friend, M ¼ 2.00, SD ¼ .77; t(107) ¼ 3.41 and 4.83, ps < .001, respectively). Means for both interlocutors indicate that, on average, participants reported engaging between “a little bit” and “a fair amount” of conversations with both. At the individual level, these two items were moderately correlated, r ¼ .47 for targets, r ¼ .55 for friends. At the overall pair level, the correlation between friends for talking to friends was r ¼ .49 and talking to family was r ¼ .20. We conducted the analysis separately by talking to friends and talking to family. For the sample overall, partialling out talking to friends attenuated the similarity correlation for exploration, rp ¼ .10, p ¼ .32 (vs. r ¼ .22; 79% reduction in variance explained), but only slightly for commitment, rp ¼ .29, p ¼ .003 (vs. r ¼ .34; 27% reduction in variance explained). Looking within the different ethnic pairs, the similarity correlation for exploration for the EM–EM pairs attenuated to non-significance, rp ¼ .19, p ¼ .17 (vs. r ¼ .35; 71% reduction in variance explained), but the correlations for the other two groups did not change (Table 2). The findings were similar for commitment, where the correlations attenuated to rp ¼ .21, p ¼ .14 (vs. r ¼ .35; 64% reduction in variance explained) for the EM–EM pairs only. The results were nearly identical for talking to family. Overall, the similarity correlations were rp ¼ .14, p ¼ .15 (60% reduction in variance explained) for exploration and rp ¼ .29, p ¼ .002 (27% reduction in variance explained) for commitment. Looking within ethnic pairs, the correlations for exploration lessened to rp ¼ .26, p ¼ .06 (45% reduction in variance explained) and commitment lessened to rp ¼ .24, p ¼ .09 (53% reduction in variance explained) among the EM–EM pairs, but the correlations were unchanged for the other two groups. Finally, due to the overlap in discussions with friends and family we conducted an analysis to attempt to discern the relative contribution of each for ethnic identity. For this test we conducted multiple regression analyses separately by target and friend, using discussions with friends and family as predictors of ethnic identity exploration and commitment. Because we hypothesized that talking with friends would be the primary predictor of ethnic identity, we conducted the analysis in two steps: first entering talking with family and second entering talking to friends. All four of these analyses (exploration and commitment for each pair member) indicated that discussions with friends was a significant predictor of ethnic identity and that discussions with family was not after accounting for the overlap with discussions with friends (Table 3). These analyses provide indirect evidence that discussions with friends about ethnicity-related issues, and not family, serves as the mediator of ethnic identity similarity. Discussion The purpose of the present study was to assess the degree to which pairs of friends report similar levels of ethnic identity and to test for moderators and mediators of the similarity. To this end we advanced three hypotheses: 1) friends would report similar levels of ethnic identity, 2) the similarity would be the strongest for EM–EM pairs of friends, and 3) relationship closeness and discussions with friends and family would mediate the similarity. In general, the pattern of findings supports the three hypotheses and contributes to a growing literature on friendships as a context for ethnic identity development. Below we discuss findings pertaining to each hypothesis, in turn. Looking at the sample overall, we found support for our first hypothesis that pairs of friends are similar in their levels of ethnic identity exploration and commitment. In conjunction with other findings on the self-reported relational expression of ethnic identities (e.g., Kiang & Fuligni, 2009; Kiang et al., 2007), our finding indicates that ethnic identity is a shared experience among friends. This finding, however, can only be fully understood in the context of our second hypothesis, that the ethnic composition of the friendship pair would moderate the similarity. Indeed, the similarity correlations for ethnic identity exploration and commitment were only significant for the EM–EM pairs, and not for the EM–White or White–White pairs. Moreover, it was only the exploration correlations that significantly differed among pairs. Thus, it appears that the pronounced variations in similarity by ethnicity are in regards to the process of ethnic identity exploration. Table 3 Relative contribution of talking to friends and family for ethnic identity exploration and commitment, separated by target and friend. Target b
SE
DR2
b
SE
b
DR2
Dependent variable: ethnic identity exploration Step 1 Talking to family .18
.08
.22*
.05*
.29
.08
.34
.12***
Step 2 Talking to family Talking to friends
.09 .09
.09 .29**
.07**
.17 .21
.09 .09
.21 .25*
.05*
.07
.15
.02
.22
.08
.27**
.07**
.08
.05 .21*
.04*
.10 .17
.09 .09
.13 .20y
.01
.07 .24
Dependent variable: ethnic identity commitment Step 1 Talking to family .11 Step 2 Talking to family Talking to friends y
Friend
b
.04 .17
p .10, *p .05, **p .01, ***p .001.
09
1512
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
It could be that the differences are evident for exploration and not commitment due to the more behavioral nature of exploration. EM–EM pairs may actually engage in exploratory behaviors together, such as joining organizations, attending events, or talking about ethnicity-related issues (more on this below). In contrast, the EM–White and White–White pairs may not have shared experience, but for different reasons. For EM–White pairs there is a clear disconnect between the types of exploratory behaviors that each may engage in; not only do they not share their ethnic background, but they also do not share in the minority experience. For White–White pairs exploration may not be congruent because it is a more singular experience. That is, due to their majority status in the U.S., Whites are not inclined to learn about, or even think about, their ethnic background. Some White youth even worry that being prideful and engaged with their ethnicities might give the impression that they are White supremacists (Santos, Ortiz, Morales, & Rosales, 2007). Thus, ethnic identity exploration among White youth may not occur in as public a manner as for ethnic minorities. A contrasting perspective comes from a study by Sidanius, Van Laar, Levin, and Sinclair (2004), who found that participation in Greek organizations among White college students was associated with greater ethnic identity commitment and ethnic activism. These findings suggest that there may in fact be some public contexts for White youth to explore their ethnicities. Additionally, it is possible that White identity exploration occurs more frequently in the context of family than friends. Indeed, a study of ethnically-diverse college students found that parental cultural socialization had a particularly strong impact on White students’ levels of ethnic identity explorationdmore so than for ethnic minority groups (Juang & Syed, 2010). An important finding is that the similarities observed for ethnic identity were not also observed for identity synthesis. This finding suggests that there is not a general “identity homophily” that occurs between friends, but rather a more specific “ethnic identity homophily.” The construct of identity synthesis corresponds to the degree to which individuals have developed a clear, coherent, consistent sense of identity (Erikson, 1968; Rosenthal et al., 1981; Schwartz, 2007). Accordingly, it does not necessarily correspond to any specific behavioral or attitudinal manifestations. It would be valuable for future research to assess friends’ similarity in other domains of identity, such as occupational identity or political identity, in order to understand whether identity homophily is restricted to ethnic identity or can be seen with any specific identity domain. Our third hypothesis was concerned with testing for other variables that may account for the similarity in ethnic identity between friends. We found support for two of three mediators that we tested. Contrary to the hypothesis, relationship closeness could not account for the observed ethnic identity homophily, suggesting the similarities in identity between friends is not simply a matter of feeling especially close to them. Indeed, taking a page from Erikson (1968), we would expect that sharing an identity with another person would facilitate feelings of intimacy, rather than vice versa (see Beyers & SeiffgeKrenke, 2010 for recent empirical support of this sequence). We did, however, find that engaging in discussions about ethnicity-related topics with both friends and family served as a mediator of ethnic identity homophily. Further analyses indicated that it was discussions with friends in particular that accounted for the similarity in levels of ethnic identity exploration and commitment. This suggests that the similarity in ethnic identity is not simply limited to like-folks being drawn to each other (i.e., selection), but rather the similarity is due to ongoing shared interactions, particularly in terms of talking about ethnicity-related issues (i.e., socialization; Kandel, 1978). Our measure of discussions was based on a single item and did not query about how frequently they talked with that particular friend, so these results should be interpreted with some degree of caution. Nevertheless they hold promise for developing a deeper understanding of how ethnic identity develops in the context of close friendships. Indeed, our findings dovetail with those of Kao and Joyner (2004), who found that co-ethnic friends report engaging in more shared activities, as well as Schwartz et al.’s (2011) findings of greater personal disclosure with co-ethnics. Taken together, the accumulated evidence suggests that friendships are an important social context for ethnic identity development. Implications for theory and research on identity and friendships The findings in the present study have implications for theory and research on both identity and friendships. Identity research has largely focused on individuals. Even though there has been a recent uptick in studies on identity development within social contexts (e.g., Juang et al., 2006; Way, Santos, Niwa, & Kim-Gervey, 2008), such studies are still focused on single individuals within those contexts. Few studies have examined friendship pairs or parent–child dyads to understand how identities may be intertwined with close personal relationships. A notable exception is the work by Thorne and colleagues (e.g., Korobov & Thorne, 2006; Thorne et al., 2007), who have conducted a number of studies demonstrating how identities are expressed and negotiated in social interactions between friends. Such an approach is congruent with many of the influential grand theories of identity development (e.g., Erikson, 1968; Goffman, 1959), which highlighted the importance of social interaction and reflected appraisals for developing a sense of identity. In terms of ethnic identity specifically, understanding social and interactional processes are endemic to the construct given the role of minority status for the meaningfulness of the ethnic identity (Tajfel & Turner, 1986). Indeed, experimental social psychology has generally done a good job of unearthing the situations and contexts that activate ethnic identities (e.g., Cheryan & Monin, 2005), but how such experiences might be related to the development of ethnic identity remains unclear. The current study, which was unique in having independent reports of ethnic identity from pairs of friends, is a small step in this direction. In terms of contributions to friendship research, the current study suggests the need for considering ethnic identity when studying the friendship processes of ethnic minority youth. The findings indicate that ethnic identity similarity is not redundant with relationship closeness, and thus may have unique contributions to other aspects of the friendship. For example, both ethnic identity and peer interaction processes have been independently linked to well-being, academic
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
1513
achievement, and disordered eating (Brechwald & Prinstein, 2011; Quintana, 2007; Smith & Silva, 2011), suggesting the possibility of a transactive relation between ethnic identity and closeness for these outcomes. Additionally, if ethnic identity is associated with the types of activities that individuals prefer to engage in, it may actually serve as a barrier to closeness with others who do not share such interests, regardless of their ethnicity. Indeed, understanding the role of constructs such as ethnic identity can be useful for understanding the very high rates of co-ethnic friends that people have (e.g., Kao & Joyner, 2004). In other words, for ethnic minorities it may be ethnic identitydnot ethnicitydthat plays a role in driving the tendency for preferring co-ethnics as friends. Limitations and future directions To our knowledge the current study is the first to assess friendship similarity in identity process using independent reports from friends. Despite this methodological strength, and the interesting findings it yielded, there are some limitations worth considering. First, collapsing all ethnic minority groups into a single category precluded our ability to fully understand the role of ethnicity and ethnic identity in the context of friendships. As we argued in the Introduction we believe that using the single group makes sense both theoretically and empirically, but nevertheless one cannot deny that more precise information would be gained through ethnic-specific analyses. Part of our justification for examining ethnic minorities as a group was because the context that they inhabited was predominantly White (78% at the university). Accordingly, the generalizability of the findings in the present study may be limited to similar settings. Indeed, investigating the role of the social context, in terms of ethnic group representation, for identity homophily would be a useful direction for future research. Much of the research on ethnic identity and local ethnic group representation has produced mixed findings (e.g., Juang et al., 2006; Syed et al., 2007; Umaña-Taylor, 2004). A recent study of African American adolescents at 10 high schools holds promise for future work (Yip, Seaton, & Sellers, 2010). They found that the school-level diversity on its own was not related to ethnic identity, but the interaction between diversity and the number of African Americans friends the adolescents had was related to identity. Thus, both objective (diversity) and subjective (friendships) aspects of social context are important to consider in future work (see also Bronfenbrenner, 1979). Our measures of talking about ethnicity with friends and family were based on a single item for each. Furthermore, the questions only assessed the frequency with which they talk to friends or family about ethnicity. Given the strength of the association between ethnic identity and talking to friends and family, and their mediating role in the similarity in ethnic identity between friends, it would be useful for future research to examine the content of what people talk about. Such an analysis could then lead to a multidimensional measure of talking about ethnicity with others, one that assesses both frequency and content. Finally, the cross-sectional nature of the study was a limitation, as it only afforded a static representation of how identity plays out in friendships. Future longitudinal research on ethnic identity homophilydand identity homophily more broadlydis much needed. In particular, research that examines how the two aspects of homophily, selection and socialization (Kandel, 1978), interact over time would be especially beneficial. Conclusion The purpose of this study was to assess the degree to which pairs of friends report similar levels of ethnic identity and to test for moderators and mediators of the similarity. The results provide evidence for an ethnic identity homophily, in terms of exploration and commitment, among ethnic minority friends. Furthermore, findings indicate that ethnic identity homophily is not simply a product of relationship closeness, but rather can be accounted for by individuals’ propensity to engage in conversations with their friends about ethnicity-related issues. This study builds upon the current movement in ethnic identity research to consider the social context of ethnic identity development, and helps to move it forward by highlighting the importance of friendship dynamics for the developmental process. References Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology, 63, 593–612. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Berscheid, E., Snyder, M., & Omoto, A. M. (1989). The relationship closeness inventory: assessing the closeness of interpersonal relationships. Journal of Personality and Social Psychology, 57, 792–807. Beyers, W., & Seiffge-Krenke, I. (2010). Does identity precede intimacy? Testing Erikson’s theory on romantic development in emerging adults of the 21st century. Journal of Adolescent Research, 25, 387–415. Brechwald, W. A., & Prinstein, M. J. (2011). Beyond homophily: a decade of advances in understanding peer influence processes. Journal of Research on Adolescence, 21(1), 166–179. Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard University Press. Chavous, T. M., Bernat, D. H., Schmeelk-Cone, K., Caldwell, C. H., Kohn-Wood, L., & Zimmerman, M. A. (2003). Racial identity and academic attainment among African American adolescents. Child Development, 74(4), 1076–1090. Cheryan, S., & Monin, B. (2005). “Where are you really from?”: Asian Americans and identity denial. Journal of Personality and Social Psychology, 89(5), 717–730. Erikson, E. H. (1968). Identity: Youth and crisis. New York: Norton. French, S. E., Seidman, E., Allen, L., & Aber, J. L. (2006). The development of ethnic identity during adolescence. Developmental Psychology, 42, 1–10.
1514
M. Syed, M.J.D. Juan / Journal of Adolescence 35 (2012) 1505–1514
Goffman, E. (1959). The presentation of self in everyday life. New York: Doubleday. Gonzaga, G. C., Campos, B., & Bradbury, T. (2007). Similarity, convergence, and relationship satisfaction in dating and married couples. Journal of Personality and Social Psychology, 93(1), 34–48. Griffin, D., & Gonzalez, R. (1995). Correlational analysis of dyad-level data in the exchangeable case. Psychological Bulletin, 118, 430–439. Juang, L. P., Nguyen, H. H., & Lin, Y. (2006). The ethnic identity, other-group attitudes, and psychosocial functioning of Asian American emerging adults from two contexts. Journal of Adolescent Research, 21, 542–568. Juang, L. P., & Syed, M. (2010). Family cultural socialization practices and ethnic identity among college-going emerging adults. Journal of Adolescence, 33(3), 347–354. Kandel, D. B. (1978). Homophily, selection, and socialization in adolescent friendships. American Journal of Sociology, 84(2), 427–436. Kao, G., & Joyner, K. (2004). Do race and ethnicity matter among friends? Activities among interracial, interethnic, and intraethnic adolescent friends. The Sociological Quarterly, 45(3), 557–573. Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York: Guilford. Kiang, L., & Fuligni, A. J. (2009). Ethnic identity in context: variations in ethnic exploration and belonging within parent, same-ethnic peer, and differentethnic peer relationships. Journal of Youth and Adolescence, 38(5), 732–743. Kiang, L., Harter, S., & Whitesell, N. R. (2007). Relational expression of ethnic identity in Chinese Americans. Journal of Social and Personal Relationships, 24(2), 277–296. Kiang, L., Witkow, M. R., Baldelomar, O. A., & Fuligni, A. J. (2010). Change in ethnic identity across the high school years among adolescents with Latin American, Asian, and European backgrounds. Journal of Youth and Adolescence, 39(6), 683–693. Korobov, N., & Thorne, A. (2006). Intimacy and distancing: young men’s conversations about romantic relationships. Journal of Adolescent Research, 21(1), 27–55. Marcia, J. E. (1980). Identity in adolescence. In J. Adelson (Ed.), Handbook of adolescent psychology (pp. 159–197). New York: Wiley. Pahl, K., & Way, N. (2006). Longitudinal trajectories of ethnic identity among urban Black and Latino adolescents. Child Development, 77(5), 1403–1415. Perron, J., Vondracek, F. W., Skorikov, V. B., Tremblay, C., & Corbière, M. (1998). A longitudinal study of vocational maturity and ethnic identity development. Journal of Vocational Behavior, 52, 409–424. Phinney, J. S. (1990). Ethnic identity in adolescents and adults: a review of research. Psychological Bulletin, 108, 499–514. Phinney, J. S. (1992). The multigroup ethnic identity measure: a new scale for use with diverse groups. Journal of Adolescent Research, 7(2), 156–176. Phinney, J. S., Romero, I., Nava, M., & Huang, D. (2001). The role of language, parents, and peers in ethnic identity among adolescents in immigrant families. Journal of Youth and Adolescence, 30(2), 135–153. Preacher, K. J., & Hayes, A. F. (2008). Contemporary approaches to assessing mediation in communication research. In A. F. Hayes, M. D. Slater, & L. B. Snyder (Eds.), The Sage sourcebook of advanced data analysis methods for communication research (pp. 13–54). Thousand Oaks, CA: Sage. Quintana, S. M. (2007). Racial and ethnic identity: developmental perspectives and research. Journal of Counseling Psychology, 54(3), 259–270. Reis, O., & Youniss, J. (2004). Patterns of identity change and development in relationships with mothers and friends. Journal of Adolescent Research, 19(1), 31–44. Roberts, R. E., Phinney, J. S., Masse, L. C., Chen, Y. R., Roberts, C. R., & Romero, A. (1999). The structure of ethnic identity of young adolescents from diverse ethnocultural groups. Journal of Early Adolescence, 19(3), 301–322. Rogers Wood, N. A., & Petrie, T. A. (2010). Body dissatisfaction, ethnic identity, and disordered eating among African American women. Journal of Counseling Psychology, 57(2), 141–153. Rosenthal, D. A., Gurney, R. M., & Moore, S. M. (1981). From trust to intimacy: a new inventory for examining Erikson’s stages of psychosocial development. Journal of Youth and Adolescence, 10(6), 525–537. Santos, S. J., Ortiz, A. M., Morales, A., & Rosales, M. (2007). The relationship between campus diversity, students’ ethnic identity and college adjustment: a qualitative study. Cultural Diversity and Ethnic Minority Psychology, 13(2), 104–114. Schwartz, A. L., Galliher, R. V., & Domenech Rodríguez, M. M. (2011). Self-disclosure in Latinos’ intercultural and intracultural friendships and acquaintanceships: links with collectivism, ethnic identity, and acculturation. Cultural Diversity and Ethnic Minority Psychology, 17(1), 116–121. Schwartz, S. J. (2007). The structure of identity consolidation: multiple correlated constructs or one superordinate construct? Identity: An International Journal of Theory and Research, 7(1), 27–49. Schwartz, S. J. (2008). Self and identity in early adolescence: some reflections and an introduction to the special issue. The Journal of Early Adolescence, 28(1), 5–15. Sidanius, J., Van Laar, C., Levin, S., & Sinclair, S. (2004). Ethnic enclaves and the dynamics of social identity on the college campus: the good, the bad, and the ugly. Journal of Personality and Social Psychology, 87(1), 96–110. Singer, J. D., & Willet, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press. Smith, T. B., & Silva, L. (2011). Ethnic identity and personal well-being of people of color: a meta-analysis. Journal of Counseling Psychology, 58(1), 42–60. Syed, M. College students’ storytelling of ethnicity-related events in the academic domain. Journal of Adolescent Research, in press. Syed, M., & Azmitia, M. (2009). Longitudinal trajectories of ethnic identity during the college years. Journal of Research on Adolescence, 19(4), 601–624. Syed, M., Azmitia, M., & Phinney, J. S. (2007). Stability and change in ethnic identity among Latino emerging adults in two contexts. Identity: An International Journal of Theory and Research, 7(2), 155–178. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel, & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago: Nelson-Hall. Thorne, A., Korobov, N., & Morgan, E. M. (2007). Channeling identity: a study of storytelling in conversations between introverted and extraverted friends. Journal of Research in Personality, 41(5), 1008–1031. Umaña-Taylor, A. J. (2004). Ethnic identity and self-esteem: examining the role of social context. Journal of Adolescence, 27, 139–146. Way, N., Santos, C., Niwa, E. Y., & Kim-Gervey, C. (2008). To be or not to be: an exploration of ethnic identity development in context. In M. Azmitia, M. Syed, & K. Radmacher (Eds.), New directions for child and adolescent development: Intersections between personal and social identity, 120, (pp. 61–79). San Francisco: Jossey-Bass. Yip, T., Seaton, E. K., & Sellers, R. M. (2010). Interracial and intraracial contact, school-level diversity, and change in racial identity status among African American adolescents. Child Development, 81(5), 1431–1444.