Effects of Friendship among Same-Sex Attracted Youth on Sexual Minority Identity Development in Young Adulthood

Abstract

Using data from Waves 1 and 3 of the National Longitudinal Study of Adolescent to Adult Health, this study investigated the interactive impact of adolescent friendship network characteristics and same-sex attraction (SSA) on the development of sexual minority (SM) identity in young adulthood. Results indicate that SSA youth who identified a best friend that also identified them as their best friend were 5.3 times more likely to identify as a SM compared to those who did not experience reciprocity. For each additional classmate who listed them as a friend, SSA youth were 1.2 times more likely to identify as a SM. Increased centrality in friendship network increased the likelihood that SSA youth would identify as a SM by 1.8 times. Sex-stratified results reveal that particular network characteristics impact boys and girls differently. These findings point to the importance of further investigating relationships between friendship characteristics and identity formation among adolescents.

Research demonstrates a consistently strong link between social integration, friendship, and health and well-being (Gariépy et al., 2016; Ge et al., 2017; Kawachi & Berkman, 2001). This link is even more pronounced among individuals who identify as lesbian, gay, bisexual, or queer (LGBQ) (Gillespie et al., 2015). Research suggests that LGBQ adolescents experience mental and behavioral health issues at higher rates than their peers (Di Giacomo et al., 2018; Ross et al., 2018), despite trends toward greater inclusion and growth in positive attitudes regarding sexual minority identity (Savin-Williams, 2005). These disparities are often attributed to the stigma and associated social isolation that sexual minority youth often face (Hatzenbuehler, 2014; Mustanski et al., 2014). Although much quantitative research has investigated the relationships between social isolation/integration and health for LGBQ youth, there is much research to be done that investigates the connections between social integration and identity formation for LGBQ youth. Particularly among adolescents, theorists have proposed that friendship and peer association are pivotal aspects of growth and identity development in this period. In the past, research has neglected empirically testing these theories among adolescents navigating sexual identity development processes.

Sociologists have long argued that identity is an aspect of human experience that is rarely biologically determined or immutable, and is often socially and interpersonally constructed. Despite having a rich history in sociological thought, this approach has not gained much traction in popular understandings of sexuality and sexual orientation (Morandini et al., 2015; Sánchez & Pankey, 2017). Supporters of LGBQ rights frequently challenge the notion of homosexuality as a chosen status, by arguing that it should be understood as biological or genetic in origin, and thus an uncontrollable aspect of one’s identity (Haider-Markel & Joslyn, 2008; Sánchez & Pankey, 2017). In constructing this dichotomy regarding the sources of sexual minority sexuality, popular understandings have failed to account for the role of social integration and social support in enabling or inhibiting the development of a sexual minority identity.

This study utilizes data from the National Longitudinal Study of Adolescent to Adult Health (hereafter, Add Health) to investigate the role of high school friendship networks in shaping probability of LGBQ-identification, especially among same-sex attracted (SSA) youth. I propose viewing sexual minority identity development as influenced by particular characteristics of an individual’s social network. Elaborating this connection between peer relationships and identity development has the potential to greatly expand our understanding of the interconnected ways in which social support and social integration inform developmental and health outcomes among LGBQ youth. In exploring the social determinants of the ways in which LGBQ identity is facilitated or stifled among youth, researchers can inform the efforts of those working directly with LGBQ adolescents and young adults.

Identity Formation in Adolescence

Scholarship on identity processes points to adolescence as a time of significant developmental importance, generally, but in terms of the exploration of sexual identity in particular. Sexual identity, here, is conceived of as the labels that an individual uses to describe their sexual orientation or sexuality, which, in turn, encompasses both romantic attraction as well as sexual behavior (Katz-Wise, 2015). Although the exploration of any type of sexuality among adolescents carries with it a certain degree of cultural discomfort, adolescents negotiating sexual minority identity development can face stigma on both a structural and an interpersonal level (Hatzenbuehler, 2014; Kosciw et al., 2018; Lyons et al., 2011).

The stigma that these individuals endure is frequently compounded by the overlapping contexts of high school and adolescence. In high school, LGBQ youth are victimized and discriminated against at high rates compared to their heterosexual peers (Chesir-Teran & Hughes, 2009; Kosciw et al., 2018). This atmosphere of hostility can represent a source of profound disconnection for adolescents questioning their sexuality. Judith Jordan (2001:2) describes the structural roots of disconnection, writing that “forces [such as] racism, sexism, heterosexism, ageism, and classism; all of the judgments which render groups marginalized, denigrated, or objectified contribute to an experience of disconnection and isolation.”

Other scholars have further posited that these forces produce disconnection between marginalized communities and larger society by perpetuating stigma that erodes social support and the resources that flow from that support (Link & Phelan, 2001; Williams et al., 2005). For LGBQ adolescents, schools can be an especially difficult context in which to explore sexual minority identities not just because of active stigma or discrimination, but due to the dearth of LGBQ narratives available. Scholars of identity have long recognized the importance of narrative as a template that individuals use to construct their own identities (Somers, 1994). When LGBQ narratives and experiences are not available to LGBQ youth, youth are limited in their ability to engage in these sexual identity formation processes. When heteronormative ideas of sexuality are hegemonic, LGBQ adolescents’ own sexual minority identity development can be suppressed (Pearson and Wilkinson 2013). Without friends and role models who have gone through sexual minority identity development, adolescents are left with limited resources for identity development.

The Dual Role of Friendship in Identity Formation

On the other hand, access to friendships and peer networks can offer support in these identity development processes. Scholars have posited that friendships are of profound significance to adolescent development to the extent that they act both as a source of support and as a source of behavior and identity modeling (Lyons et al., 2011). The support of significant others can facilitate the maintenance of an emotional space that is affirming and has positive effects on an individual’s sense of self. Friendships are also impactful to the extent that friends provide models that an adolescent can use to help shape their own beliefs and behaviors and construct their own identity (Rowe, 2014).

Much research has focused on the latter function. Social learning theorists have demonstrated that individuals learn from and model themselves on the people with whom they interact and respect the most (Bandura, 1977; Lyons et al., 2011; Rowe, 2014). Since most LGBQ individuals do not learn about gay culture from their parents, they must seek out LGBQ peers and mentors from whom to learn that culture (Rowe, 2014). For individuals navigating the construction of a stigmatized identity, some of the most important and influential relationships one can have are with others upon whom one can model one’s beliefs and behaviors and learn strategies of coping with stigma (Kaufman & Johnson, 2004). Despite scholarly focus on the importance of friendship in terms of identity modeling, friendship can facilitate the development of a sexual minority identity beyond their capacity to model an identity. It is through the provision of a supportive and personally affirming relational space that friendships are able to nurture identity exploration and development (Morris, 2018; Wright & Perry, 2006). As such, all significant and mutually fulfilling friendships can facilitate building of a sexual minority identity (Diamond & Dube, 2002; Morris, 2018).

Although the connection between supportive friendships and identity construction is undertheorized, support provided in friendship has long been viewed as a strong contributor to adolescents’ capacity to be resilient in the face of structural and interpersonal stigma. Some researchers propose that resilience actually flows from one’s relationships and connections with other people (Fredriksen-Goldsen et al., 2017; Garrett-Walker & Longmire-Avital, 2018; Southwick et al., 2016). Social support scholars argue that the supportive character of friendship can influence individuals’ capacity for resilience in two distinct ways. Friendship can provide both structural and functional support (Hakulinen et al., 2016; Thoits, 1995). Structural support captures an individual’s degree of social integration within a larger network of friends. Focus on the type of support that flows from aspects of network structure, “rests on the testable assumption that the social structure of the network itself is largely responsible for determining individual behavior and attitudes by shaping the flow of resources which determine access to opportunities and constraints on behavior” (Berkman et al. 2000: 845). This type of support, which is derived from the larger network, is often operationalized using number of friends or position occupied within the network.

The significance of friendship can also be understood in terms of the functional support it can provide. The most consistent and powerful measure of functional support is emotional support—whether or not a person has one primary confiding relationship (Thoits, 1995). During adolescence, youth increasingly turn to their peers for this type of relationship and emotional support (Giordano, 2003; Papini et al., 1990). In a period of life when youth are looking outside of their families for emotional support, “close friendships support basic needs for belonging, empathy, and mutual engagement” and provide “companionship, help, intimacy, reliable alliance, self-validation, and emotional security” (Graber, Turner, and Madill 2016:2). Researchers have often operationalized these types of relationships using measures of friendship reciprocity. These measures capture the extent to which an individual’s friend also considers that individual a friend (Abbott & Freeth, 2008; Vaquera & Kao, 2008). Reciprocal friendships have been found to be a superior resource compared to unreciprocated friendships in that they are characterized by a greater degree of trust and emotional support (Vaquera & Kao, 2008). To the extent that both structural and functional support are sources of resiliency for adolescents, they both are able to not only facilitate positive health outcomes, but also positive identity development.

Gendered Nature of Social Support

Gender shapes experiences of friendship and social support as well as sexual identity development. Research has consistently shown that gender can have profound impacts on the experience of friendship among youth. Scholars find that boys tend to have larger social networks than girls, but girls tend to be more invested and have greater intimacy in their relationships (Thoits, 1995). In a study of seven and ten year old children, Bryant (1985) found that having a greater number of casual friends was more likely to encourage optimal social-emotional functioning in boys. While girls seemed to benefit more socio-emotionally from having fewer relationships of greater emotional intensity (Diamond & Dube, 2002). Importantly, these gendered differences are seen as a product of gendered socialization and thus societally, not biologically, contingent.

Similarly, in a context of extreme adversity, researchers have found that increased psychological resilience was only facilitated by close, intimate friendships among girls, whereas boys experienced no added benefits to resilience when they possessed close friendships (Graber et al., 2016). Some literature suggests that boys do not reap the benefits of close friendships because boy’s peer group culture does not prioritize intimacy, so boys are less adept at building, navigating, and deriving benefit from close relationships (Giordano, 2003; Vaquera & Kao, 2008). As such, peer culture that prioritizes normative masculinity can hinder the development of intimate relationships (Diamond & Dube, 2002). More recent research has complicated some of these gendered assumptions, showing that increasing cultural acceptance of homosexuality has resulted in friendships between men that are defined by more “inclusive masculinities” (Morris, 2018).

Norms of masculinity might further impact the role of friendship for boys in that their status within a social network is likely to be a more important predictor of friendship satisfaction and overall well-being for boys compared to girls. By occupying a central position in a social network, men are most able to exercise particularly masculine traits (Cornwell & Laumann, 2011). Centrality within a network indicates that one possesses a large number of weak ties that bridge multiple networks. Possession of these types of ties creates an impression of one as independent but influential (Cornwell & Laumann, 2011). As such, being central in a network is more conducive to the fulfilling of male gendered focal concerns and the attainment of masculine ideals associated with status and power (Steffensmeier et al., 2013).

Conversely, girls and women are socialized to prioritize relationships with others over status concerns (Chodorow, 1978; Gilligan, 1982; Steffensmeier et al., 2013). Empirical studies among adolescents have continued to substantiate the salience of these findings (Bearman & Moody, 2004; Soller, 2014). Women also experience more of the negative effects of social network ties as a result of investing a greater level of emotion in their relationships and the well-being of their friends (Kawachi & Berkman, 2001; Rook, 1984). Accordingly, women derive the greatest benefit derived from those relationships characterized by the most reciprocity (Vaquera & Kao, 2008). Girls are most well-served, in terms of mental health and overall well-being, by reciprocated close friendships, rather than number of friends or status within the larger network (Belle et al., 1987; Graber et al., 2016; Vaquera & Kao, 2008).

Friendship in the Context of Sexual Minority Identity Development

Symbolic interactionism proposes that individual identity development occurs in the context of connection with others (Graber et al., 2016; Papini et al., 1990; Stryker, 2001). It is in emotionally connected relationships with others that “we grow, learn, expand, and gain a sense of meaning” (Jordan 2001: 97). Although these arguments have not been extensively tested among LGB communities, individual-level connection is especially significant in the structural contexts of disconnection created by heterosexism and homophobia. Identity development can be fostered as a function of embeddedness within larger social networks or within the context of closer, more intimate relationships (Hirsch and Rapkin 1986).

LGB individuals can reap particular benefit from social network embeddedness in constructing a stigmatized sexual identity. Having a wider network of relatively close friends can help to affirm identity and offset the effects of minority stress (Gillespie et al., 2015). Minority stress in the context of sexual identity stigma is considered that stress that LGB youth and adults experience as a product of negative interactions with prejudiced individuals, on the basis of their sexuality (Meyer, 2003). Repeated experience of stress related to marginalized status and the internalization of stigmatizing beliefs together create a level of psychological distress that can delay or forestall sexual minority identity formation (Wright & Perry, 2006). On the other hand, for individuals in the process of LGB identity development, possession of significant friendships or integration in an accepting social network provide a relational context that is emotionally supportive and affirming, and thus more likely to nurture the development of a sexual minority identity (Morris, 2018; Wright & Perry, 2006).

When friendships are reciprocated, best friends, in particular, “allow the adolescent to create a better understanding of the self, and they provide a support system to help the adolescent work through daily stresses” (Bagwell et al., 2001). Best friendships can also take on a particularly important role for LGBQ adolescents. These types of close, mutually fulfilling relationships can provide a safe space to foster a stigmatized identity that might be inhibited or entirely suppressed in a less supportive context. Gillespie et al. (2015) found that friendship mutuality was far more beneficial, in terms of overall life satisfaction, among LGBQ respondents compared to heterosexual respondents. In the face of discrimination and stigma, LGBQ individuals stand to gain significant benefit from close and supportive relationships as they navigate processes of identity construction.

Current Study

Existing literature, on the whole, is primarily focused on the health benefits of supportive relationships for LGBQ youth. I seek to illuminate the ways that supportive friendships and social integration are also important for LGBQ identity formation itself. I build upon existing literatures to suggest that certain aspects of social relationships shape sexuality development to the extent that they provide a source of resilience. In recognition of the ways in which experiences of friendship and sexuality are profoundly gendered, I also investigate the extent to which these processes vary for adolescent boys and girls. Based on previous literature on social integration and identity, I hypothesize that same-sex attracted (SSA) adolescents who possess certain markers of peer network integration are more likely to adopt a sexual minority identity in adulthood. Further, I expect that the identity processes of SSA girls will be more influenced by their integration in intimate relationships, while integration in broader peer networks will be more important for the identity processes of SSA boys.

Data and Methods

Sample

This study uses the public-use sample of two waves of data collected as part of the National Longitudinal Study of Adolescent Health, a longitudinal survey concerned with issues related to adolescent health and development (Harris et al., 2009). Wave 1 data consist of responses to the parent, adolescent in-school, and adolescent in-home questionnaires administered between 1994 and 1995. Respondents were interviewed again in 2001 and 2002 for Add Health’s third wave of data collection. At Wave 1, the mean age of the sample was 15.9 years. In Wave 3, respondents’ mean age was 22.2 years.

The public-use Add Health sample was randomly selected from a clustered random sample of high schools across the United States. The sample is clustered insofar as school enrollment determined eligibility to take the survey. The sample is implicitly stratified in that certain schools had a higher chance of being selected based on the size and type of school, its census region, level of urbanity, and its racial and ethnic composition. All those students who were interviewed at Wave 1 were eligible to be interviewed at Wave 3. Public-use data included 3,359 individuals who responded to both Waves 1 and 3, including the sexual behavior module at Wave 3.

Dependent Variable

The dependent variable, sexual identity at Wave 3, was measured based on self-report. The Wave 3 measure of sexual identity was chosen for this study because Wave 3 corresponded to late adolescence/young adulthood. Only by this period in development do most adolescents report romantic attraction and sexual relationships (Saewyc, 2011). As such, Wave 2 would have been administered in a period that would have been too early, developmentally, and Wave 4 (administered in 2008) would have been too far removed. The options available to respondents were based on a Kinsey-inspired scale. Respondents could report that they were 100% heterosexual, mostly heterosexual but still attracted to members of the same sex, bisexual, mostly homosexual but still attracted to members of the opposite sex, 100% homosexual, or that they were not interested, sexually, in men or women (asexual). Although asexuality represents one type of sexual minority status, asexuality represents a perspective on sexuality that is qualitatively different from that of other sexual minority groups (Walton et al., 2016). Due to this and the relatively small number of respondents who reported being asexual at Wave 3 (N=12), these respondents were excluded.

This scale was condensed into a binary variable, which reflects whether an individual classified their sexual identity as “Heterosexual” or not. Only those individuals who reported that they identified as “100% heterosexual” were considered to be heterosexual. Although research supports the distinctiveness of each of the following identities (Savin-Williams & Vrangalova, 2013; Vrangalova & Savin-Williams, 2012), those respondents who identified as “Mostly heterosexual,” “Bisexual,” “Mostly homosexual,” or “100% homosexual,” were grouped together to reflect identification with a sexual minority identity to maximize statistical power. I have included even those respondents who identified as “mostly heterosexual” in the sexual minority category because using “mostly” as a qualifier to describe one’s sexual identity indicates some understanding of one’s sexuality as a fluid phenomenon. Empirical work has shown that women who characterize their sexuality as “mostly heterosexual” shared more in common with bisexual women and lesbians, compared to their exclusively straight counterparts (Thompson & Morgan, 2008). As a result of this coding strategy, a total of 291 respondents were considered to have a sexual minority identity at Wave 3.

Independent Variables

One of the main independent variables of focus is respondents’ report of SSA when they were interviewed at Wave 1. This variable was constructed based on a comparison of respondent’s sex versus the sex to which they reported being attracted. Thus, this variable captures all male youth who reported being romantically attracted to other male youth, and all female youth who reported being romantically attracted to other female youth—including those who simultaneously reported being attracted to the opposite sex. This represents a better measure of SSA in high school, compared to actual dating or sexual behavior, because of the difficulties of openly acting on something as stigmatized as same-sex feeling in a high school context (Kosciw et al., 2018).

The other main independent variables used in this analysis consist of a series of social network variables, most of which were pre-constructed by the designers of Add Health based on the network data collected during the administration of the survey’s first wave. The first network variable employed was a measure of reciprocated best friendship. I constructed this variable based on a number of the pre-constructed variables in the network data— (1) whether or not the respondent nominated a male or female best friend, (2) whether or not the male and female individuals who were nominated as best friends reciprocated any friendship ties, (3) whether or not the male and female individuals who were nominated as best friends reciprocated the tie as a best friend. The categorical variable I created based on these four measures is intended to capture an increasing level of reciprocity in these relationships. This measure of reciprocated friendship is ideal because it is based on actual tie nominations by respondents within the surveyed school, rather than the respondents’ perceptions of reciprocity within their friendships.

An individual’s centrality in the network was measured using Add Health’s pre-constructed Bonacich centrality and in-degree measures. Bonacich centrality captures the extent to which an individual is tied to other individuals within a large or small network. This type of centrality implies that being connected to others who are highly connected confers a greater amount of power than being connected to an equal number of others who do not have many connections (Hanneman & Riddle, 2005). In-degree measures centrality by counting the number of people within the school that nominated the respondent as a friend. Whereas Bonacich centrality represents a nuanced measure of the extent to which a person’s centrality is associated with power in a network, in-degree simply measures centrality through sheer number of direct connections.

Control Variables

This study also included a number of control variables. Socio-demographic characteristics including age, race/ethnicity, sex, and parental education at Wave 1 were controlled for in this analysis. Sex was operationalized using an item which asked the Add Health interviewer to confirm with the respondent whether their sex was male or female. Although research suggests that a more nuanced operationalization of gender and gender identity is vital for disentangling the ways in which sex, gender, and sexual orientation intersect to shape youth experiences (Worthen, 2013), Add Health did not collect data in Waves 1 and 3 beyond biological sex.

Wave 1 age was calculated based on the respondent’s birth date and the date of the administration of the survey. Parental education was measured as the highest level of education attained by one of the respondent’s parents. Parental education was measured with 9 categories from 0, indicating never having attended school to 8, indicating graduate or professional training beyond 4-year college.

Respondents’ race/ethnicity were measured based on their responses to the in-school questionnaire as well as in-home questionnaire, and dummy variables were coded in accordance with Add Health researchers’ recommendations. The mutually exclusive race categories represent White (non-Hispanic), Black (non-Hispanic), Hispanic/Latino(a), Asian and other Pacific Islander (non-Hispanic), American Indian/Alaska Native (AIAN; non-Hispanic), and other race (non-Hispanic).

Analytic Strategy

To test this study’s hypotheses, logistic regression was performed with each of the theoretically relevant network variables: friendship reciprocity, in-degree, and Bonacich centrality. Bivariate analyses assessed whether SSA youth significantly differ from other-sex attracted youth on each network variable. Bivariate analyses were also performed to assess whether males and females differed significantly on each network variable. Then, adjusted logistic regressions were performed to investigate the main effect of SSA on young adult sexual identity and the interactive effects of same-sex attraction and each network variable. Across all tables, all models labeled Model 1 test the main and interaction effects of SSA and friendship reciprocity on Wave 3 sexual identity, controlling for demographic variables. Models 2 and 3 do the same, but instead test the effects of in-degree and Bonacich Centrality, respectively. Incorporating interactions between SSA and each network variable allows one to capture the extent to which different network characteristics moderate the relationship between adolescent SSA and adult sexual identity. The next stage of analyses stratifies these analyses by gender, separately testing the main and interactive effects of SSA and the network characteristics for males and females.

All analyses were performed in Stata 15 MP. To accommodate missing data, multiple imputation was performed using Stata’s mi estimate suite of commands. To account for the Add Health sampling strategy and the use of a Wave 3 measure as an outcome with models otherwise based in Wave 1 data, these analyses employed the Wave 3 cross-sectional sampling weight. The Wave 3 cross-sectional sampling weight is most appropriate to account for both the sampling strategy and proposed analytic design, particularly in the case of a dependent variable that is not modeled at multiple timepoints (Chen & Chantala, 2014). The final sample of 3,359 respondents were from 113 different schools.

As Stata does not allow for the incorporation of sampling weights with multiple imputation and mixed effects, random intercept models to account for clustering were performed separately as a robustness check and to assess the impact of school-based clustering. Model results using a hierarchical structure did not differ substantially from the analyses using the Add Health sampling and design weights

Results

First, means and standard deviations of each continuous variable and proportions of each binary variable were examined prior to imputation. Table 1 presents these results as well as means and proportions separately for SSA (N=183) and other-sex attracted (Non-SSA) (N=3,149) youth. Approximately 91% of respondents considered themselves “100% Heterosexual” at Wave 3, while 9% identified as “Mostly Heterosexual,” “Bisexual,” “Mostly Homosexual,” or “100% Homosexual.” For brevity, all those who received a “1” for the dichotomized Wave 3 sexual identity variable, will be referred to as having an LGBQ identity.

Table 1:

Means, Standard Deviations, and Proportions for Full Sample and By SSA at Wave 1

	Full Sample (N=3,359)		SSA (N=183)		Non-SSA (N=3,149)
	Mean (SD)/ Proportion (N)	Range	Mean (SD)/ Proportion (N)	Range	Mean (SD)/ Proportion (N)	Range
Outcome (Wave 3)
Sexual Orientation		1 – 6		1–5		1–5
100% Heterosexual	0.905 (3,051)		0.749 (137)		0.918 (2,890)
Mostly Heterosexual	0.059 (202)		0.126 (23)		0.056 (177)
Bisexual	0.015 (52)		0.066 (12)		0.013 (40)
Mostly Homosexual	0.006 (19)		0.033 (6)		0.004 (12)
100% Homosexual	0.005 (18)		0.027 (5)		0.004 (13)
Asexual	0.004 (12)		0.000 (0)		0.000 (0)
LGBQ (dichotomized sexual orientation)	0.087 (291)	0 – 1	0.251 (46)	0 – 1	0.077 (242)	0 – 1
Independent Variables (Wave 1)
Same-Sex Attraction (SSA)	0.055 (183)	0 – 1
Female	0.547 (1,837)	0 – 1	0.443 (81)	0 – 1	0.555 (1,746)	0 – 1
Race/Ethnicity
White (ref.)	0.550 (1,815)	0 – 1	0.536 (98)	0 – 1	0.552 (1,706)	0 – 1
Black	0.228 (752)	0 – 1	0.197 (36)	0 – 1	0.228 (706)	0 – 1
Asian/Pacific Islander	0.039 (130)	0 – 1	0.044 (8)	0 – 1	0.039 (122)	0 – 1
American Indian/Alaska Native	0.042 (140)	0 – 1	0.044 (8)	0 – 1	0.043 (132)	0 – 1
Hispanic/Latino(a)	0.111 (365)	0 – 1	0.164 (30)	0 – 1	0.107 (331)	0 – 1	-
Other Race	0.031 (101)	0 – 1	0.016 (3)	0 – 1	0.031 (96)	0 – 1
Age	15.861 (1.712)	11.4 – 20.6	16.139 (1.831)	11.4 – 20.6	15.846 (1.704)	12.4 – 20.6
Parent Education (maximum)	5.907 (1.671)	0 – 8	5.615 (1.851)	1 – 8	5.924 (1.662)	0 – 8
Network Variables
Friendship Reciprocity
No Best Friend	0.250 (785)	0 – 1	0.325 (55)	0 – 1	0.254 (724)	0 – 1
Best Friend Not Reciprocated	0.234 (735)	0 – 1	0.237 (40)	0 – 1	0.233 (686)	0 – 1
Best Friend Reciprocated	0.197 (619)	0 – 1	0.130 (22)	0 – 1	0.201 (592)	0 – 1
Best Friend Reciprocated as Best Friend	0.320 (1,005)	0 – 1	0.308 (52)	0 – 1	0.321 (948)	0 – 1
Indegree	4.678 (3.722)	0 – 30	4.312 (3.623)	0 – 20	4.703 (3.719)	0 – 30
Bonacich Centrality	0.815 (0.634)	0 – 4.288	0.719 (0.604)	0 – 2.973	0.821 (0.634)	0 – 4.288

Note: All descriptive results presented are derived from public use data, prior to imputation.

Source: Waves I and III of the National Longitudinal Study of Adolescent to Adult Health

Reported SSA at Wave 1 was less prevalent than a sexual minority identity at Wave 3, with only 5.5% of the sample having reported SSA at Wave 1. Of those who did report SSA at Wave 1, only 25.1% identified as LGBQ at Wave 3. Of those who did not report SSA at Wave 1, 7.7% identified as LGBQ at Wave 3.

Table 1 shows that the sample was majority white, with white respondents making up 55% of the sample. Black respondents accounted for 22.8% of the sample; Asian or Pacific Islander respondents accounted for 3.9% of the sample, American Indian/Alaska Natives were 4.2% of the sample, Hispanic/Latino/as were 11.1% of the sample, and those who identified as “other” race were 3.1% of the sample. The average age of respondents at Wave 1 was 15.9 years. The average age was slightly higher among the SSA sample, and maximum level of parental education was slightly lower in the SSA sample.

Overall, respondents were fairly evenly distributed between each of the friendship categories. Youth who reported SSA at Wave 1 were more likely to not have nominated a best friend (32.5% did not nominated a best friend) compared to non-SSA youth (25.4%). The average number of incoming friendship nominations for the whole sample was approximately 4.68, with a minimum of 0 and a maximum of 30. The average Bonacich Centrality score for the whole sample was 0.815, ranging from 0 to 4.3, and was slightly lower for SSA youth, whose mean Bonacich Centrality score was 0.719.

Bivariate Results

Results of bivariate analyses assessing differences among SSA and non-SSA youth and male and female youth on each of the network characteristics is presented in Table 2. Odds ratios are presented for the binary variables representing each category of friendship reciprocity, and Bs are presented for coefficients expressing the relationship between SSA, sex, indegree, and Bonacich Centrality. There was not a significant difference between SSA and non-SSA youth in the number of incoming friendship nominations that they received, but female respondents did receive significantly more friendship nominations (on average, about half a nomination more) than male respondents. Bivariate results show that non-SSA were significantly more central in their networks compared to SSA respondents in the sample, and females were significantly more central than males. As was also seen in Table 1, SSA respondents were significantly more likely to not have nominated a best friend, such that SSA respondents were 1.4 times as likely to not have nominated a best friend. Odds ratios also show that female respondents were less likely than male respondents to not have nominated a best friend. Females were 80% as likely to have not nominated a best friend. SSA and non-SSA youth were not significantly different in their likelihood of having a best friend not reciprocate their friendship nomination, nor were they significantly different in having their best friend nomination fully reciprocated (also nominated as a best friend). Female respondents were 30% less likely than males to not have their best friendship nomination reciprocated at all compared to males, and they were 1.6 times more likely to have their best friendship nomination fully reciprocated.

Table 2:

Bivariate Analyses of Network Characteristics by Same-Sex Attraction and Sex

	Indegree B (SE)		Bonacich Centrality B (SE)		No Best Friend OR (SE)		Best Friend Not Reciprocated OR (SE)		Best Friend Reciprocated OR (SE)		Best Friend Reciprocated as Best Friend OR (SE)
SSA	−0.390		−0.114**		1.385*		1.120		0.633*		0.924
	(0.310)		(0.054)		(0.192)		(0.222)		(0.261)		(0.191)
Female		0.457***		0.081***		0.806**		0.701***		0.950		1.644***
		(0.159)		(0.025)		(0.095)		(0.097)		(0.103)		(0.089)
Intercept	4.754***	4.502***	0.832***	0.785***	0.292***	0.331***	0.302***	0.360***	0.262***	0.263***	−0.689***	0.386***
	(0.082)	(0.122)	(0.013)	(0.019)	(0.049)	(0.067)	(0.050)	(0.069)	(0.053)	(0.077)	(0.045)	(0.069)
N	3,359	3,359	3,359	3,359	3,144	3,144	3,144	3,144	3,144	3,144	3,144	3,144

Note: Friendship reciprocity results reflect a lower number of observations (N) due to missingness on the dependent variable, which cannot be corrected with multiple imputation. Robust standard errors in parentheses

^***p<0.01

^**p<0.05

^*p<0.1

Main and Interaction Effects of Friendship Network Characteristics

The adjusted results demonstrating both the main and interaction effects of SSA and friendship network characteristics are presented in Table 3. Main effects of friendship reciprocity on Wave 3 sexual identity are presented in Model 1a. There is a positive and significant relationship between reporting SSA at Wave 1 and reporting a sexual minority identity at Wave 3. As expected, reporting SSA at Wave 1 increased the odds of adopting a sexual minority identity by 4.6 times. Reporting female sex is also significantly and positively associated with reporting a sexual minority identity at Wave 3. Female youth had 3 times greater odds compared to male youth of reporting a sexual minority identity at Wave 3. None of the friendship reciprocity dummies were found to be significantly different from the reference category in predicting sexual minority identity. In Model 1a, we also see that Black respondents had significantly lower odds of reporting sexual minority identity at Wave 3, compared to white respondents.

Table 3:

Adjusted Analyses of Same-Sex Attraction and Network Characteristics Predicting Sexual Identity

	Model 1a	Model 1b	Model 2a	Model 2b	Model 3a	Model 3b
SSA	4.581***	1.395	4.563***	2.389**	4.491***	2.927***
	(0.210)	(0.653)	(0.213)	(0.345)	(0.212)	(0.315)
Female	2.992***	2.948***	3.047***	3.065***	3.022***	3.056***
	(0.165)	(0.164)	(0.165)	(0.164)	(0.167)	(0.167)
Race/Ethnicity (ref: white)
Black	0.558***	0.557***	0.551***	0.542***	0.557***	0.548***
	(0.213)	(0.214)	(0.212)	(0.213)	(0.214)	(0.214)
Asian	1.181	1.183	1.135	1.126	1.195	1.207
	(0.349)	(0.352)	(0.344)	(0.346)	(0.347)	(0.345)
American Indian/Alaska Native	0.635	0.613	0.632	0.612	0.636	0.626
	(0.369)	(0.370)	(0.366)	(0.370)	(0.369)	(0.367)
Hispanic/Latino(a)	0.993	0.986	0.978	0.986	0.986	0.987
	(0.228)	(0.227)	(0.225)	(0.223)	(0.228)	(0.228)
Other Race	1.086	1.074	1.056	1.059	1.102	1.122
	(0.401)	(0.401)	(0.407)	(0.408)	(0.397)	(0.395)
Age	0.954	0.952	0.956	0.954	0.947	0.946
	(0.039)	(0.039)	(0.039)	(0.0393)	(0.040)	(0.040)
Parental Education	1.048	1.043	1.060	1.056	1.057	1.057
	(0.045)	(0.045)	(0.045)	(0.045)	(0.045)	(0.046)
Friendship Reciprocity (ref: no reciprocity)
No Best Friend	1.269	1.134
	(0.226)	(0.239)
Best Friend Reciprocates	1.063	0.932
	(0.245)	(0.253)
Best Friend Reciprocates as Best Friend	0.963	0.801
	(0.222)	(0.235)
SSA*No BF		3.225
		(0.745)
SSA*BF Reciprocates		3.908
		(0.862)
SSA*BF Reciprocates as BF		5.296**
		(0.737)
In-degree			0.945***	0.925**
			(0.022)	(0.024)
SSA*In-degree				1.157**
				(0.058)
Bonacich Centrality					0.748**	0.688***
					(0.122)	(0.134)
SSA*Bonacich Centrality						1.828*
						(0.320)
Constant	0.065***	0.078***	0.081***	0.093***	0.092***	0.099***
	(0.768)	(0.765)	(0.750)	(0.760)	(0.762)	(0.768)
N	3,359	3,359	3,359	3,359	3,359	3,359

Note: All estimates presented are odds ratios. Robust standard errors in parentheses.

^***p<0.01

^**p<0.05

^*p<0.1

Model 1b incorporates the interaction effects between SSA and the friendship reciprocity dummies. In Model 1b, incorporation of the interaction effect reduces the size of the effect of SSA, and the relationship loses statistical significance. In Model 1b, once can see a statistically significant interaction between reporting SSA and having their nominated best friend reciprocate that best friend tie at Wave 1 on sexual identity at Wave 3. Odds ratios in Model 1b show that those SSA respondents who experienced a fully reciprocated best friendship at Wave 1 were 5.3 times more likely than those in the reference category to identify with a sexual minority identity in Wave 3. The other categories of friendship reciprocity did not confer this advantage.

Models 2a and 2b show results for the main effects and interactions effects of in-degree on sexual identity at Wave 3. Again, the main effect of SSA was strong and statistically significant. When controlling for number of incoming friendship nominations, experiencing SSA at Wave 1 increased one’s odds of having a sexual minority identity at Wave 3 by 4.5 times. Identifying as female and identifying as Black at Wave 1 also significantly increased one’s odds of identifying with a sexual minority identity at Wave 3. In Model 2a, the odds ratio of 0.945 for in-degree shows that, for the whole sample, as number of incoming friendship nominations increased, the odds of a given respondent identifying as a sexual minority at Wave 3 decreased. In Model 2b, this relationship remained significant and negative for the main effect of in-degree, but the interaction effect with SSA trended in the opposite direction. For those respondents who reported SSA, having a greater number of incoming friendship nominations increased their likelihood of adopting a sexual minority identity at Wave 3 by 15%.

Models 3a and 3b show the main and interaction effects, respectively, of Bonacich centrality on sexual minority identity at Wave 3. There are consistently positive and significant main effects of SSA and female identification on adopting a sexual minority identity at Wave 3. The odds ratio for Bonacich centrality shows a significant negative effect of Bonacich centrality on sexual minority identity at Wave 3 such that as centrality within one’s network increased, one’s odds of identify as LGBQ decreased by 25.2%. When the interaction effect between SSA and Bonacich centrality is included in Model 3b, the relationship between SSA and Wave 3 sexual identity is attenuated slightly. Similar to the case with the in-degree and SSA interaction, the main effect of Bonacich centrality becomes even more strongly negative, but when interacted with SSA, the effect runs in the opposite direction. The odds ratio for the interaction of SSA and Bonacich centrality shows that as network centrality increases for respondents who reported SSA, their odds of identifying as LGBQ increased by 83%.

Gender-Stratified Effects of Friendship Network Characteristics on Sexual Identity

The next stage of analysis stratifies these results by sex. Table 4 presents the main and interaction effects of each of the friendship network characteristics for female youth and Table 5 presents these same results for male youth. Model 1c shows the main effects model for friendship reciprocity among female respondents. Model 1d shows in the interaction effects of the friendship reciprocity measures among female respondents. The main effects of SSA and identifying as Black remain similar to those results found for the whole sample. Among girls, those who did not nominate a best friend from their school were significantly more likely to identify with a sexual minority identity at Wave 3. As was true for the whole sample, the addition of the interaction terms in Model 1d attenuated the relationship between SSA at Wave 1 and sexual minority identity at Wave 3. In contrast to the results with the whole sample, interaction terms for each of the friendship reciprocity categories was significant.

Table 4:

Adjusted Analyses of Same-Sex Attraction and Network Characteristics Predicting Sexual Identity among Females

	Model 1c	Model 1d	Model 2c	Model 2d	Model 3c	Model 3d
SSA	5.452***	0.477	5.601***	5.280***	5.366***	4.272***
	(0.275)	(1.184)	(0.282)	(0.484)	(0.284)	(0.428)
Race/Ethnicity
Black	0.538**	0.526***	0.540**	0.539**	0.530**	0.529**
	(0.247)	(0.248)	(0.246)	(0.246)	(0.249)	(0.249)
Asian	1.019	1.000	0.990	0.988	1.042	1.054
	(0.459)	(0.467)	(0.444)	(0.444)	(0.458)	(0.455)
American Indian/Alaska Native	0.626	0.610	0.627	0.626	0.624	0.618
	(0.451)	(0.462)	(0.443)	(0.443)	(0.447)	(0.447)
Hispanic/Latino(a)	0.876	0.870	0.872	0.873	0.855	0.863
	(0.298)	(0.297)	(0.291)	(0.291)	(0.296)	(0.295)
Other Race	1.026	1.013	1.010	1.009	1.065	1.069
	(0.481)	(0.481)	(0.487)	(0.487)	(0.480)	(0.480)
Age	0.992	0.988	0.998	0.998	0.982	0.981
	(0.048)	(0.048)	(0.047)	(0.0468)	(0.048)	(0.0478)
Parental Education	1.070	1.068	1.086	1.086	1.086*	1.087*
	(0.050)	(0.051)	(0.051)	(0.051)	(0.050)	(0.050)
Friendship Reciprocity (ref: no reciprocity)
No Best Friend	1.802**	1.565
	(0.282)	(0.284)
Best Friend Reciprocates	1.214	1.064
	(0.300)	(0.295)
Best Friend Reciprocates as Best Friend	1.185	1.005
	(0.270)	(0.270)
SSA*No BF		13.370**
		(1.267)
SSA*BF Reciprocates		13.092*
		(1.455)
SSA*BF Reciprocates as BF		15.658**
		(1.261)
In-degree			0.938**	0.938**
			(0.027)	(0.028)
SSA*In-degree				1.013
				(0.081)
Bonacich Centrality					0.608***	0.587***
					(0.151)	(0.161)
SSA*Bonacich Centrality						1.404
						(0.449)
Constant	0.077***	0.094***	0.111***	0.112***	0.159**	0.166**
	(0.849)	(0.840)	(0.837)	(0.841)	(0.859)	(0.865)
N	1,837	1,837	1,837	1,837	1,837	1,837

Note: All estimates presented are odds ratios. Robust standard errors in parentheses.

^***p<0.01

^**p<0.05

^*p<0.1

Table 5:

Adjusted Analyses of Same-Sex Attraction and Network Characteristics Predicting Sexual Identity among Males

	Model 1e	Model 1f	Model 2e	Model 2f	Model 3e	Model 3f
SSA	3.155***	1.616	3.136***	0.958	3.184***	1.763
	(0.361)	(0.709)	(0.365)	(0.589)	(0.358)	(0.536)
Race/Ethnicity
Black	0.598	0.605	0.600	0.600	0.623	0.598
	(0.430)	(0.426)	(0.424)	(0.428)	(0.424)	(0.426)
Asian/Pacific Islander	1.495	1.502	1.408	1.454	1.484	1.483
	(0.544)	(0.552)	(0.539)	(0.539)	(0.538)	(0.536)
American Indian/Alaska Native	0.630	0.609	0.601	0.560	0.623	0.618
	(0.706)	(0.674)	(0.697)	(0.706)	(0.692)	(0.673)
Hispanic/Latino(a)	1.293	1.245	1.229	1.229	1.297	1.259
	(0.357)	(0.364)	(0.360)	(0.349)	(0.362)	(0.370)
Other Race	1.296	1.318	1.182	1.270	1.278	1.334
	(0.723)	(0.729)	(0.744)	(0.727)	(0.722)	(0.708)
Age	0.871**	0.876*	0.857**	0.852**	0.862**	0.866*
	(0.069)	(0.069)	(0.070)	(0.071)	(0.0734)	(0.074)
Parental Education	1.004	0.997	1.010	1.006	0.995	0.994
	(0.091)	(0.090)	(0.090)	(0.090)	(0.090)	(0.090
Friendship Reciprocity (ref: no reciprocity)
No Best Friend	0.663	0.624
	(0.366)	(0.394)
Best Friend Reciprocates	0.944	0.820
	(0.405)	(0.440)
Best Friends Reciprocate as Best Friend	0.679	0.533
	(0.399)	(0.451)
SSA*No BF		1.498
		(1.064)
SSA*BF Reciprocates		2.601
		(1.077)
SSA*BF Reciprocates as BF		4.154
		(0.975)
In-degree			0.957	0.895***
			(0.038)	(0.042)
SSA*In-degree				1.302***
				(0.091)
Bonacich Centrality					1.125	0.995
					(0.200)	(0.229)
SSA*Bonacich Centrality						1.992
						(0.507)
Constant	0.451	0.480	0.558	0.795	0.416	0.436
	(1.305)	(1.289)	(1.326)	(1.323)	(1.361)	(1.368)
N	1,522	1,522	1,522	1,522	1,522	1,522

Note: All estimates presented are odds ratios. Robust standard errors in parentheses.

^***p<0.01

^**p<0.05

^*p<0.1

For girls who reported SSA at Wave 1, each category of friendship reciprocity increased the likelihood that they would identify with a sexual minority identity at Wave 3, compared to those who nominated a best friend that did not reciprocate that tie at all. This means that even those girls who experienced SSA at Wave 1 and did not nominate a best friend were more likely than SSA girls whose best friend did not reciprocate their nomination at all to identify as LGBQ at Wave 3.

Although being identified as having “no best friend” by the Add Health social network indicators might indicate an actual lack of best friend, it is also likely that this finding is an artifact of the Add Health data collection strategy for the network measures. To develop meaningful social network data, Add Health researchers had respondents nominate best friends from within their school. If a respondent considered their best friend to be someone who did not attend their school, it would appear as though they did not have a best friend at all. It might be particularly likely for sexual minority youth to have friends outside of school. As largely heterosexist spaces, schools can represent exclusionary environments for sexual minority youth (Pascoe, 2007). As a result, sexual minority youth withdraw from these environments and develop social connections elsewhere, or not at all (Perales et al., 2020). On the other hand, those SSA girls who did have a best friend who reciprocated that tie fully, were 15.7 times more likely than those who nominated a best friend that did not reciprocate that tie at all to develop a sexual minority identity by Wave 3. Experiencing reciprocity in friendship ties helped to facilitate the development of a sexual minority identity, whereas a lack of reciprocity in friendship seemed to stifle that identity development.

In the subsample of girls, there were significant negative main effects of in-degree and Bonacich centrality on sexual minority identity development in Wave 3, such that having more incoming friendship nominations and being more central in one’s friendship network decreased the likelihood of identifying as LGBQ at Wave 3, on average. Neither of the interaction terms for each of these friendship network characteristics showed a significant relationship. As such, number of friends and network centrality were not important predictors of sexual minority identity development among girls who reported SSA.

Table 5 presents the main and interaction effects of each of the network characteristics examined among boys in the sample. Among boys, there was a significant main effect of SSA in Models 1e, 2e, and 3e. Those boys who reported SSA at Wave 1 were approximately 3 times more likely to identify with a sexual minority identity at Wave 3. The only other consistently significant relationship among boys was with age at Wave 1. Those boys who were older at the time of Wave 1 data collection were less likely to identify with a sexual minority identity at Wave 3.

Considering the friendship network characteristics, there were no significant main or interaction effects of friendship reciprocity or network centrality among boys. The only significant relationship evident in the subsample of boys was in-degree. In Model 2f, which incorporates an interaction term to account for the multiplicative effect of SSA at Wave 1 and number of in-coming friendship nominations, both the main effect and the interaction term are significantly associated with sexual minority identity at Wave 3. Overall, having more incoming friendship nominations was associated with 10.5% lower odds of having a sexual minority identity at Wave 3. For those youth who reported SSA, the relationship changes direction such that boys who reported SSA were 30% more likely to identify with a sexual minority identity for each additional incoming friendship nomination. As such, for boys who experienced SSA, reciprocity in individual relationships and centrality in networks were not as essential to identity development as size of friendship network.

Discussion

This study’s findings demonstrate the importance of investigating the mechanisms by which sexual identity can be socially and relationally constructed. Popular discourse around sexual identity is frequently dominated by essentialized ways of understanding sexual identity (Grzanka et al., 2016). Epstein (2007) notes that essentializing discourses such as these rest on the impulse to locate a biological basis for human identities. Such a philosophy might argue “that because homosexuality is innate (and rooted in a ‘gay gene’ or in distinctive brain structures), gay people should not be expected to change their sexuality and, indeed, are entitled to legal protection of their rights” (Epstein 2007:294). This narrow definition takes as given issues of identity building that should instead be empirical questions, open for investigation and debate (Epstein, 2007).

This study has shown the relevance of considering the ways in which adolescents are actively engaged in relationally constructing their sexual identities and in gendered ways. This research has shown that adolescents that experienced particular aspects of friendship support were more likely to translate SSA into a sexual minority identity than adolescents that did not experience those same supportive friendships. These results also supported literature on the gendered experience of friendship by identifying that certain measures of supportive friendship were more significant predictors for boys than girls, and vice versa. Girls who were romantically attracted to other girls were more likely to translate that attraction into an LGBQ identity if they had a highly reciprocal relationship with their best friend. Friendship reciprocity was inconsequential for boys. On the other hand, boys who were romantically attracted to other boys were more likely to translate that attraction into an LGBQ identity if they had many friends. Thus, aspects of friendship networks seem to support processes of sexual identity development in highly gendered ways (Diamond & Dube, 2002).

These findings are compelling, but should be considered in the context of some of the study’s limitations. In particular, the Add Health data were collected in 1994/1995 and 2001/2002 at a time when negative attitudes toward sexual minority communities were more widespread and legal protections less available than they are today. In 2013, the Supreme Court struck down the Defense of Marriage Act in United States v. Windsor and declared marriage and constitutionally-protected right for same-sex couples in Obergefell v. Hodges. The percentage of Americans that believe in the necessity of these protections has also dramatically increased. In 2003, only 32% of Americans supported same-sex marriage (Jones et al., 2014). In 2020, the percentage of Americans that believe same-sex marriages should be legally valid has more than doubled to 67% (McCarthy, 2020). This study’s findings should be considered in the context of these dramatic changes, and further research must investigate how this changing landscape impacts the relationship between friendship and sexual identity development for SSA youth.

Also, my analysis hinges on the question asked in the first administration of the survey regarding respondents’ experience of romantic attraction. Even though these questions were answered on a computer, rather than in the context of a conversation with the interviewer, it is possible that the stigmatized nature of same-sex romantic attraction led to underreporting of this phenomenon. Further, small sample sizes have necessitated comparing SSA youth to other-sex attracted youth. The more valid comparison for understanding these identity-building processes would be among SSA youth who adopted a sexual minority identity versus a heterosexual identity. Further research should utilize larger sample sizes to make these within-group comparisons.

Finally, this study was limited by its reliance on reported biological sex, with dichotomous response options, as a proxy for gender identity. Not only has the notion of dimorphic biological sex been obviated by acknowledgement of intersex populations (Fausto-Sterling, 2000), but dichotomous biological sex does not capture the immense diversity of gender identities adopted by youth. It is possible that this measure of sex misclassified youth who identify with gender(s) that do not coincide with their biological sex. This diversity of experience of gender and sex has profound implications for mental health, identity development, and friendship (McKay & Watson, 2020; Olson-Kennedy et al., 2016; Pulice-Farrow et al., 2017). As a result, this study failed to capture this nuance, and future research using data that can overcome this limitation is sorely needed.

This study raises a number of pressing questions for further research. Its findings made clear that there are significant connections between friendship characteristics and LGBQ identity development. Even though this study assumes the positive value of an LGBQ identity as an end in and of itself, it is likely that identity achievement can have other positive implications for young adults. For example, Gamson (1995) suggests that identity formation processes can help facilitate social movement mobilization. Epstein (1995) points out that a solidified identity, especially a politicized one, would increase people’s willingness to emotionally invest in a social movement predicated on that shared identity and their likelihood of taking risks on behalf of movement success. Thus, an achieved sexual minority identity, is likely a greater predictor of social movement involvement and political activism.

A potential further implication of this research is in the link between achieved identity and particular health outcomes or overall well-being. Early developmental research in psychology revealed the importance of a synthesized identity in allowing individuals to cope with daily stressors and experience greater levels of overall health over the course of their lives (Erikson, 1968). More recent research has shown that individuals with a more fully developed sense of their identity are more likely to experience greater levels of psychological well-being (Oles, 2016). When looking at more specific health outcomes, Hardy et al. (2013) found that “identity maturity” predicted lower levels of depression, anxiety, and hazardous alcohol use and predicted higher levels of self-esteem. Further research should work to determine how identity consolidation, friendship, and particular health outcomes might be associated for sexual minority youth.

Although this study did not establish causal links between friendship network characteristics and sexual identity development, knowledge in the field would benefit from future research that seeks to establish these connections. Further exploration of causal pathways in relation to the ways in which strongly supported identity building processes encourage well-being and positive mental health outcomes would represent another vital facet of this research. These types of investigations could produce concrete ideas and frameworks on which to base programs and services for sexual minority youth, especially ones that focus on fostering positive identity growth.