Introduction

A large body of research has documented that sexual minority women (SMW) are more likely to report health-risk behaviors and negative health outcomes compared to their heterosexual peers (Goldbach, Tanner-Smith, Bagwell, & Dunlap, 2014; Hughes, 2011; Marshal et al., 2011). The minority stress framework (Meyer, 2003), which emphasizes the negative health effects of stigma, is often used to explain these disparities (Denton, Rostosky, & Danner, 2014; Figueroa & Zoccola, 2015; Newcomb & Mustanski, 2010; Walch, Ngamake, Bovornusvakool, & Walker, 2016). One under-examined facet of the minority stress model is the role of gender and its relationship to the diverse stress processes highlighted in the minority stress framework. These include both “distal” and “proximal” sources of minority stress. Meyer (2003) characterized distal forms of minority stress as those that occur externally to the individual, such as discrimination and victimization, while proximal sources of stress are those that have been internalized, such as internalized homophobia or the expectation of discrimination from others via stigma consciousness.

Heteronormativity, conceptualized as a social structure that exerts powerful influence in the everyday lives of men and women, carries a set of normative expectations for appropriate and idealized gendered and sexual behavior (Butler, 2011; Schilt & Westbrook, 2009). The conflict between heteronormative expectations and individual violations of these expectations has been characterized as a chronic and unique source of stress, rooted in stigma (Meyer, 1995, 2003). While hegemonic gender-role expectations are unattainable for most men and women, these norms are often in direct conflict with sexual minorities’ lives, particularly those in same-sex relationships and those whose appearance or behavior does not conform to traditional gender norms. Gender-role norms, as reflected in appearance (e.g. dress, hairstyle), behaviors, and personality characteristics, serve to structure interactions between individuals, signaling to others multiple meanings, including those related to sexual orientation. Studies have shown that sexual minorities are more likely to engage in gender-nonconforming dress and behavior than heterosexuals (Rieger & Savin-Williams, 2012). Because “doing” gender (West & Zimmerman, 1987) is such a highly prescriptive process, violations of gender-role norms may serve as an external signal or flag of a heteronormative violation, increasing the likelihood that an individual will be targeted for discrimination and victimization (Martin-Storey & August, 2016; Rieger & Savin-Williams, 2012; Roberts, Rosario, Corliss, Koenen, & Austin, 2012a, 2012b; Skidmore, Linsenmeier, & Bailey, 2006).

Gender nonconformity, however, may not have uniformly negative implications for SMW’s well-being. In fact, some research suggests that it may be protective against minority stress to the extent that it allows SMW to dress or behave according to their own in-group norms and to resist larger heteronormative structures (Frith & Gleeson, 2003; Hayfield, Clarke, Halliwell, & Malson, 2013; Riley & Cahill, 2005). Indeed, the minority stress framework emphasizes the role of minority group membership as a buffer against stigma (Meyer, 2003). Studies on variation in gender roles among lesbians have found a set of within group norms that are sometimes in contrast with heterosexual norms, but nonetheless represent prescriptive expectations for group participation (Rothblum, 1994). Thus, gender nonconformity can function as a way for SMW to signal to others that they are members of a specific group, where they may otherwise be rendered invisible.

Gender nonconformity can also be a powerful form of resistance to heteronormative expectations that serves a protective function against internalized homophobia and stigma (Frith & Gleeson, 2003; Hayfield et al., 2013; Riley & Cahill, 2005). Studies that have examined gender nonconformity among SMW have found it to be associated with feelings of personal freedom and authenticity (Clarke & Spence, 2013; Hutson, 2010). Further, recent research on the relationship between gender nonconformity and depression suggests that the negative effects of gender nonconformity during adolescence decreases over time, such that it may be protective against depression in adulthood (Li, Pollitt, & Russell, 2015). Taken together, findings from the studies reviewed above challenge simple causal assumptions that gender nonconformity leads to increased minority stress. Rather, the relationship between gender nonconformity (operationalized as self-perceptions of masculinity in this study) and minority stress likely vary between distal and proximal measures of minority stress.

Hypothesis 1

Masculinity will be associated with higher levels of distal minority stressors (i.e., discrimination and victimization), but lower levels of proximal minority stressors (i.e., internalized homophobia and stigma consciousness).

Less research has examined the role of gender conformity, or femininity, and its relationship to minority stressors among SMW. The few existing studies report conflicting results. Some have found that femininity is associated with greater internalized homophobia and sexual identity concealment (Lehavot & Simoni, 2011). Other research, however, has found that feminine SMW have no desire to conceal their identities, but rather struggle with issues of authenticity within the sexual minority community (Blair & Hoskin, 2015; Clarke & Spence, 2013; Levitt & Hiestand, 2005). For many lesbian and bisexual women, sexual identity recognition is an important part of sexual orientation development. As such, many SMW do not wish to “pass” as heterosexual, but rather to be recognized as sexual minority in their everyday lives and relationships (Pfeffer, 2014). Research by Blair and Hoskin (2015) showed that many feminine SMW report additional stressors associated with “coming out” and accepting their femme identity, which they perceive to be at odds with gender-role norms in the sexual minority community. In another study, Boyle and Omoto (2014) found that lesbians who reported a mismatch between their lesbian identity and lesbian community norms reported higher levels of depressive symptoms and anxiety. Thus, while feminine appearance and behavior may reduce exposure to victimization and discrimination, it may also increase stress associated with lack of adherence to within group norms and/or gender expectations.

Hypothesis 2

Femininity will be associated with lower levels of distal minority stressors (i.e., discrimination and victimization), but higher levels of proximal minority stressors (i.e., internalized homophobia and stigma consciousness).

Intersectionality scholars have called for research that provides a more nuanced understanding of how multiple minority identities work together to shape the lived experiences of individuals (Bowleg, 2008, 2012; Hankivsky, 2012). Gender, race, sexual orientation, and other sociodemographic characteristics are unique axes along which different sets of privilege and disadvantage are unevenly distributed across the population. Further, identities or marginalized statuses do not function independently, but rather are co-constitutive and interactive (Bowleg, 2012; McCall, 2005). To date, research on the relationship between femininity/masculinity, health, and health-risk behaviors has focused exclusively on White samples (Pfeffer, 2014) or makes no mention of race/ethnicity (Blair & Hoskin, 2015). Samples that lack racial/ethnic diversity may therefore lead to conclusions about the function of gender that are specific to one group (e.g., White sexual minorities), but generalized to all sexual minority people. There are several reasons, however, to believe that such variation exists.

First, the meaning and value attached to gender roles among sexual minorities has been shown to vary across race/ethnic groups, in part due to what some have argued was a “white-washing” of the early lesbian rights movement. In the 1970s, the emerging lesbian rights movement saw gender nonconformity as a way of rejecting heteronormative expectations for dress and behavior, including the rejection of cosmetics and the traditional feminine style of dressing (Moore, 2006). Strong masculine representations, however, were also discouraged within White lesbian communities, with a preference for androgynous gender presentations (Loulan, 1990).

Lesbians of color who were marginalized in the lesbian rights movement, however, often resisted the rejection of traditional gender norms. Within the Black lesbian community today, there are a variety of normative gender presentations, including ultra-feminine (Wilson, 2009). Some researchers have found distinct gender roles and presentation, particularly masculine–feminine pairings, are an important feature of Black lesbian’s social and romantic lives (Moore, 2011; Wilson, 2009). Further, although many Black lesbians reject butch-femme labels, they tend to be accepting of gendered presentations that are highly feminine or masculine (Moore, 2006; Wilson, 2009).

Empirical research with Latina lesbians is very limited (García & Torres, 2009). In fact, some scholars suggest that Latina lesbians are one of the least researched population groups (Calvo & Esquibel, 2011). Existing research, however, suggests that Latina SMW’s successful performance and embodiment of femininity reflects positively on their families’ social status, while non-normative gender performance may increase stigma for themselves and their families (Acosta, 2010; Asencio, 2009). As a result, many Latina lesbians may present differently when with their families of origin than in their everyday lives (Asencio, 2009). According to Calvo and Esquibel (2011), Latina SMW embrace a range of gendered presentations, including highly feminine. However, Acosta (2013) found that many Latina lesbian women express disapproval of overtly masculine gender presentations.

Because of their doubly marginalized statuses as both women of color and sexual minorities, sexual minority women of color face heightened risk of disapproval and discrimination associated with transgressing gender-presentation norms, and this may deter them from adopting a more masculine identity. Indeed, gender nonconformity may be most threatening or detrimental for Latina SMW who are already marginalized on the basis of their race, nationality, and sexuality (Acosta, 2013). In Meyer’s (2012) study of sexual minority experiences of violence, SMW of color reported additional pressure to represent their racial/ethnic minority community. This was particularly true of butch-identified women of color who experienced homophobic victimization; they reported feeling targeted for poorly representing their community (Meyer, 2012). Similar perceived pressures were not reported by White lesbians. Other work that has explored sexual orientation and race/ethnic disparities in health behaviors found that SMW of color experience greater discrimination than all other groups (e.g., White woman, sexual minority men) (Calabrese, Meyer, Overstreet, Haile, & Hansen, 2014). Taken together, this research suggests that masculinity and femininity may operate differently across race/ethnic groups, and that more gendered presentations will generally be associated with better outcomes for Black and Latina SMW.

Hypothesis 3

Higher levels of self-perceived femininity will be associated with lower levels of minority stressors (both distal and proximal) among Black and Latina SMW than among White SMW.

Hypothesis 4

Higher levels of masculinity will be associated lower levels of minority stressors (both distal and proximal) among White SMW than among Black and Latina SMW.

Current Study

No studies to date have used quantitative approaches to examine the relationships among masculinity, femininity, race/ethnicity, and minority stress among SMW. In this study, we used data from a diverse sample of SMW recruited from the greater Chicago metropolitan area to examine how self-perceived masculinity and femininity and race/ethnicity were associated with four indicators minority stress: discrimination, victimization, internalized homophobia, and stigma consciousness.

Method

Participants

Data were from the Chicago Health and Life Experiences of Women Study, an 18-year, 3-wave longitudinal study of adult SMW. Data collection began in the greater Chicago metropolitan area in 2000–2001, using a broad range of recruitment sources and strategies to recruit a diverse sample of 447 English-speaking women, aged 18 and older, who self-identified as lesbian. Concerted efforts were made to maximize sample representativeness by including subgroups of lesbians underrepresented in most studies of lesbian health (those aged under 25 and over 50, those with a high school education or less, those from racial/ethnic minority groups). The study was advertised in local newspapers, on Internet listservs, and on flyers posted in churches and bookstores; information about the study was also distributed to individuals and organizations via formal and informal social events and social networks. Other recruitment sources included clusters of social networks (e.g., formal community-based organizations and informal community social groups) and individual social networks, including those of women who participated in the study. Interested women were invited to call the project office to complete a short telephone-screening interview. Although respondents who reported being heterosexual, mostly heterosexual, bisexual or transgender at the initial screening interview were not enrolled in the study, 11 women identified as bisexual in the actual Wave I (baseline) interview. The baseline sample included a broad age range (18–82 years old), and less than one-half of the sample was White. The sample was re-interviewed in 2003–2005 (Wave II; 86% response rate), and in 2010–2012 (Wave III; 79% response rate).

In Wave III, an additional sample of 373 women was added to the existing longitudinal sample. Recruitment of the new study panel, using an adaptation of respondent driven sampling, was designed to oversample Black, Latina, and younger lesbians (ages 18–25) as well as women who identified as bisexual to increase the diversity of the sample in regard to sexual orientation identity, socioeconomic status, and race/ethnicity.

The current study focuses on data from women interviewed in Wave III, which allowed us to maximize the sample size to permit examination of interactions between femininity, masculinity, and race/ethnicity. In addition, the primary independent variables, perceived masculinity and perceived femininity, and all but one of our dependent variables, were not assessed in previous waves of the study. We excluded women who had missing data on the key dependent variables (N = 77), women who identified as “other” race/ethnicity (N = 24), and women who identified as mostly or exclusively heterosexual (N = 13), resulting in a final sample size of 612 respondents.

Measures

Self-perceived masculinity and femininity were measured by asking respondents “In general, how masculine do you think you are?” and “in general, how feminine do you think you are?” Possible responses ranged from 1 (not at all) to 7 (extremely). We used continuous measures of masculinity and femininity, which are preferable to terms such as butch/femme as many SMW avoid and resist such labels (Hutson, 2010; Moore, 2006).

Victimization included reported experiences of victimization during adulthood. The measure was created using five items: (1) experience of rape, unwanted or forced sexual activity or any other type of sexual assault, (2) experience being “shot or stabbed or attacked with a gun, knife, other weapon,” (3) being “mugged, held up, or threatened with a weapon,” (4) being “been attacked with the intent to kill” and (5) being “badly beaten up.” Responses to each question were a dichotomous yes (1) or no (0) and were combined to create a count variable ranging from 0 to 5.

Discrimination was measured using six questions that asked about personal experiences with discrimination in various settings. Following the Experiences of Discrimination scale (Krieger, Smith, Naishadham, Hartman, & Barbeau, 2005) (see Appendix section). Responses options were never (0), once (1), two to three times (2), and four or more times (3). All six items were combined into a count variable of experiences with discrimination ranging from 0 to 18 with a scale reliability of alpha = 0.71.

Stigma consciousness was measured using a series of items to capture respondents’ sensitivity to, and awareness of, stigma. Because lesbian and bisexual women report differences in their experiences of stigma (Bostwick, 2012), the questions used to construct the scale were tailored to separately capture the unique dimensions of stigma consciousness for lesbian and bisexual respondents. The Lesbian Stigma-Consciousness scale follows Pinel (1999) and asked respondents how strongly they agree with 10 statements (see Appendix section). The scale has an alpha of 0.69, indicating adequate internal consistency reliability. Bisexual women were asked how strongly they agree with 11 statements (alpha = 0.83). Response options for both scales range from “strongly disagree” (1) to “strongly agree” (5). When appropriate, responses were reverse-coded so that higher scores indicate higher levels of stigma consciousness. Because the scales for lesbian and bisexual women included different number of items, both scales were standardized and each respondent was assigned the standardized score for the scale that corresponded with their reported identity. The final scale ranged from − 1.4 to 2.0, with a mean of 0.

Internalized homophobia was measured using responses from a 10-item scale (see “Appendix” section) that asked respondents how strongly they agreed with a series of statements. Possible responses ranged from strongly disagree (0) to strongly agree (4). The combined scale ranged from 0 to 40 with a scale alpha of 0.83.

Control Variables

We controlled for sexual identity, race/ethnicity, age, education, and income. Sexual identity was coded as a categorical variable of exclusively lesbian (referent), mostly lesbian, or bisexual. Race was a self-identified categorical variable assessed by asking respondents “Which of these categories best describes your race?” and “Are you of Hispanic or Latina origin or descent?” Responses were categorized into White (referent), Black, and Latina. Age was treated as continuous and ranged from 18 to 82 years. Level of education was measured by asking respondents their highest grade or year of school completed. Responses were coded as high school graduate or less (referent), some college, and bachelor’s degree or higher. A measure of self-reported household income for the last tax year was used as an income measure and coded categorically into five income brackets of less than $10,000 (referent), $10–$29,999, $30–$49,999, $50–$74,999, more than $75,000 per year, and missing.

Analytic Plan

First, we present descriptive statistics for the entire sample and by race/ethnicity. We used one-way ANOVA to examine bivariate associations between race/ethnicity and the continuous variables including masculinity, femininity, age, and all indicators of minority stress. We used chi-square to test bivariate associations between race/ethnicity and the categorical variables included in our analyses (sexual identity, education, income). We next present a correlation matrix for our independent measures (self-reported masculinity and femininity) and all dependent variables. We used multivariate generalized linear models to test the relationships between perceived masculinity and femininity and both distal (victimization and discrimination) and proximal (internalized homophobia and stigma consciousness) minority stressors. We used Poisson regression to analyze victimization, discrimination, and internalized homophobia outcomes because of the skewed distribution of these variables. We used ordinary least squares OLS regression to analyze stigma consciousness because responses were normally distributed. For all outcomes, Model 1 controlled for sociodemographic characteristics. Model 2 included interactions between masculinity and race/ethnicity and femininity and masculinity. For all analyses using Poisson regression, results are presented as incident risk ratios (IRRs), the exponentiated form of the beta. All models were tested using Stata 14, and figures were produced using the “margins” commands.

Results

Descriptive Statistics

Table 1 summarizes the sample characteristics. Almost 75% of the sample identified as exclusively or mostly lesbian and 25% identified as bisexual. The mean age of the sample was 39 years. Although nearly half (47%) of respondents reported having a Bachelor’s degree or higher level of education, income levels were well distributed with between 15 and 22% of the sample falling within each income category. The sample was racially diverse with 38% identifying as White, 37% as Black, and 25% as Latina. On a scale from 1 to 7, women reported mean scores of 3.7 and 4.3 for masculinity and femininity, respectively, indicating slightly higher self-perceived femininity in the sample.

Table 1 Descriptive statistics for the total study sample and by race/ethnicity.
Full size table

Results from one-way ANOVA tests between race/ethnicity and perceived gender presentation showed that White women reported lower levels of masculinity (M = 3.44) than both Black (M = 3.96) and Latina (M = 3.79) women (p < .01). White women also reported lower levels of femininity than Black women (M = 4.04 and M = 4.46, respectively, p < .05). Results from one-way ANOVA tests between race/ethnicity and minority stress measures showed that both Black (M = 1.74) and Latina (M = 1.56) women had higher mean victimization scores than White women (M = 1.13, p < .001). White women reported lower levels of internalized homophobia (M = 3.99) than both Black (M = 5.18) and Latina (M = 4.94) women (p < .05), but higher levels of stigma consciousness (M = 0.04, p < .01) than Black (M = 0.01) and Latina (M = − 0.13). No differences in discrimination by race/ethnicity were found.

Chi-square tests between race/ethnicity showed significant differences in education and income across racial/ethnic groups. Education and income were therefore included as controls in all multivariate analyses.

Correlation Matrix

Table 2 presents a correlation matrix for our measures of masculinity, femininity, and our dependent variables (df = 610). There was a negative correlation between masculinity and femininity (r = − .44, p < .001). Masculinity was not significantly correlated with any of the minority stress indicators; however, femininity was positively correlated with internalized homophobia (r = .17, p < .001). There were correlations between several of the minority stress indicators: discrimination and victimization were positively correlated (r = .20, p < .001), as were discrimination and stigma consciousness (r = .34, p < .001) and stigma consciousness and internalized homophobia (r = .15, p < .001).

Table 2 Correlation matrix for masculinity, femininity, and minority stressors.
Full size table

Multivariate Analysis

Distal Minority Stressors

Table 2 presents results for distal minority stressors, victimization, and discrimination. Panel A shows that there was no significant association between masculinity or femininity and victimization in Model 1, but both Black (IRR = 1.20, p < .05) and Latina (IRR = 1.23, p < .05) respondents were more likely to report victimization than were White respondents. The interactions in Model 2 show that as masculinity scores increased for White women, victimization scores also significantly increased (IRR = 1.10, p < .05). However, higher levels of masculinity were associated with lower victimization scores for Black (IRR = 0.87, p < .05) and Latina (IRR = 0.89, p < .10) women (see Fig. 1). The interaction between femininity and race/ethnicity was not significant.

Fig. 1
figure 1

Victimization experience by race

Full size image

The results for discrimination (Panel B) in Model 1 show that both masculinity (IRR = 0.95, p < .05) and femininity (IRR = 0.93, p < .01) were negatively associated with self-reported discrimination and that both Black (IRR = 1.20, p < .01) and Latina (IRR = 2.45, p < .05) respondents reported more discriminatory experiences than White women. Interactions in Model 2, however, revealed that masculinity was negatively associated with discriminatory experiences only for Black (IRR = 0.82, p < .01) and Latina (IRR = 0.86, p < .05) women. For White women, as self-perceived masculinity scores increased so did the risk of discriminatory experiences (IRR = 1.09, p < .10). The interaction between femininity and race/ethnicity was not significant (Fig. 2).

Fig. 2
figure 2

Discrimination experience by race

Full size image

Proximal Minority Stress

Table 3 presents the multivariate results for internalized homophobia (Panel A) and stigma consciousness (Panel B). As shown in Panel A, Model 1 the relationship between masculinity and internalized homophobia was not significant. However, we found a positive relationship between femininity and internalized homophobia (IRR = 1.05, p < .001), such that as femininity scores increased so did internalized homophobia. Both Black (IRR = 1.21, p < .001) and Latina (IRR = 1.22, p < .001) respondents were significantly more likely to report internalized homophobia than White respondents. The interactions in Model 2 show that the negative effects of femininity on internalized homophobia were concentrated among Black (IRR = 1.07, p < .05) and Latina (IRR = 1.06, p < .10) respondents (see Fig. 3). The interactions in Model 2 show that greater masculinity was associated with lower levels of internalized homophobia for Black SMW (IRR = 1.08, p < .01) (see Fig. 4).

Table 3 Indicators of victimization and discrimination incidence among sexual minority women IRR’s from Poisson regression (N = 612).
Full size table
Fig. 3
figure 3

Internalized homophobia and femininity by race

Full size image
Fig. 4
figure 4

Internalized homophobia and masculinity by race

Full size image

The results in Panel B, Models 1 and 2 showed no direct relationship between masculinity, femininity, and stigma consciousness. However, Latina SMW reported lower levels of stigma consciousness than White SMW (b = − 0.11, p < .10). The interaction in Model 3 reveals that higher femininity scores were associated with lower levels of stigma consciousness for Black (b = − 0.07, p < .05) and Latina (b = − 0.09, p < .05) women, but had no effect for White respondents (see Fig. 5). Higher scores on the masculinity scale were also associated with higher levels of stigma consciousness for White SMW, but lower levels of stigma consciousness for Black and Latina SMW (see Fig. 6).

Fig. 5
figure 5

Stigma consciousness and femininity by race

Full size image
Fig. 6
figure 6

Stigma consciousness and masculinity by race

Full size image

Discussion

To date, the majority of US-based research on gender and minority stress has focused on the negative effects of gender-role nonconformity among sexual minorities (Gordon & Meyer, 2008; Martin-Storey & August, 2016; Rieger & Savin-Williams, 2012; Roberts et al., 2012a, 2012b). Research using a sample of SMW from Canada, the USA, and the UK, however, challenged prevailing assumptions by suggesting that masculinity may have benefits for SMW via conformity with in-group norms and that femininity may be associated with some minority stressors (Blair & Hoskin, 2015). However, research has focused predominately on White SMW or has not taken an intersectional analytical approach. Inferences about the relationship between gender and minority stress drawn from White samples may not be generalizable to women of color given that the meaning and value of gender presentation varies across race/ethnic groups (Moore, 2006). Our results showed that race/ethnicity was an important moderating factor for the relationships between masculinity, femininity, and minority stress and, in many cases, these relationships functioned in opposite directions for White SMW and SMW of color (Table 4).

Table 4 Indicators of internalized homophobia and stigma consciousness among sexual minority women (N = 612) IRR’s from Poisson regression and Beta’s from ordinary least squared regression.
Full size table

We hypothesized that masculinity would be associated with higher levels of distal minority stressors (discrimination and victimization), but with lower levels of proximal minority stressors (internalized homophobia and stigma consciousness) and that masculinity would be associated with lower levels of both distal and proximal minority stressors among White SMW than SMW of color. We found that masculinity was positively associated with victimization, discrimination, and stigma consciousness among White SMW and was negatively associated with these outcomes among SMW of color. The findings for White women are in line with previous work that suggests gender nonconformity is associated with increased exposure to victimization and discrimination (Martin-Storey & August, 2016; Roberts et al., 2012a, 2012b). However, our results showed that gender conformity/nonconformity operated in different ways for SMW of color.

Although less research has examined the ways masculinity functions within bisexual and lesbian communities of color, existing research suggests that masculinity is an important organizing feature of the social and romantic life of SMW in these communities (Moore, 2011; Reed, Miller, Valenti, & Timm, 2011; Wilson, 2009). Masculine gender presentations would likely not persist if they were only associated with negative outcomes. Some research has found that Black masculine-identifying women (i.e., “studs” or “aggressives”) have access to some aspects of privilege traditionally associated with men (Lane-Steele, 2011). Within the Black community, “stud” or “aggressive” identified women may be at times accepted as “one of the boys” and complimented by other men on their style (Lane-Steele, 2011; Moore, 2011) A more nuanced understanding of the links between masculinity and minority stress among SMW of color is needed. The sample in this study originated in Chicago, which is a highly segregated city. It may be that in the case of these findings, Black communities are more accepting of gender nonconformity than previously thought. It is also possible that results reflect the fact that SMW of color are an already multiply marginalized group that face high levels of victimization and discrimination (Grollman, 2014). Thus, perceived masculinity may not be as robust in influencing women of color’s experiences of victimization or discrimination. This would not necessarily explain, however, why masculinity would be protective against distal stressors.

Our results related to internalized homophobia, a proximal source of minority stress, however, were slightly different: For both Black and White SMW, masculinity was associated with lower levels of internalized homophobia. This result suggests that White and Black SMW who perceive themselves as being more masculine may be more self-assured in their minority sexual orientation. Given the historical preference for androgynous and masculine gender presentations, White women may also reap psychological benefits from conformity to in-group norms. The association between masculinity and lower levels of internalized homophobia among Black SMW may be linked to duration of sexual minority identification. Previous research has shown that, at least for African Americans, women with more masculine gender presentations tend to be aware of their sexual identity earlier and have maintained their identity for longer periods of time (Moore, 2011). Earlier disclosure may be indicative of more supportive home and neighborhood environments or higher levels of self-esteem, but also longer periods of time to embed themselves within the sexual minority community.

Masculinity was associated with higher levels of internalized homophobia among Latina SMW. Given the lack of research with this population group, it is difficult to interpret these results. However, extant research suggests that Latino families are more concerned with their daughters’ sexuality than their sons (Garcia, 2012). As a result, girls who violate gender and sexuality norms may experience more negative messages about their sexual orientation. Consistent with this, Acosta (2013) showed that Latina gender-nonconforming lesbian girls experienced more family rejection, which may result in greater levels of internalized homophobia.

We hypothesized that femininity would be associated with lower levels of distal minority stressors (discrimination and victimization), but higher levels of proximal minority stressors (internalized homophobia and stigma consciousness), and that femininity would be associated with lower levels of minority stressors (both distal and proximal) among SMW of color compared to White SMW. We found that femininity was associated with lower levels of reported discrimination for all three racial/ethnic groups, suggesting that femininity may protect against discrimination. Sexual minorities are often targeted for discrimination not because their sexual orientation is immediately recognizable, but because of gender-nonconforming presentations. Thus, SMW who report higher levels of perceived femininity may be less likely targets for discrimination.

In contrast to our hypothesized relationship, femininity was positively associated with internalized homophobia among Black and Latina SMW. Given the historical rejection of femininity among White lesbians and the broader acceptance of more feminine gender presentation among SMW of color, these results were the opposite of our hypothesis. One explanation for our results may be found in Moore’s (2011) qualitative work on Black lesbian families where femininity was associated with a specific sexual identity development trajectory: women who were more feminine generally disclosed their identity later in life and were less likely to see their lesbian identity as firm and essential as Black women with more masculine gender presentations. Moore also found that feminine lesbians were the least likely to have disclosed their identities to coworkers and family members. It is possible that femininity may be a proxy for other important aspects of sexual identity development, such as identity commitment or identity concealment, which in turn are associated with internalized homophobia (Frost & Meyer, 2009; Walch et al., 2016).

Findings related to femininity and stigma consciousness also differed between White SMW and SMW of color. That is, higher levels of femininity were associated with lower levels of stigma consciousness among sexual minority women of color but higher levels among White SMW. Again, this divergence may reflect different in-group norms. Traditional feminine gender presentation may result in lower expectations of stigma in the broader Latina and African American communities. For White SMW, higher levels of femininity may reflect deliberate attempts to avoid discrimination arising out of greater expectations of stigma.

Despite its strengths, there are several limitations that should be considered when evaluating results of this study. First, due to sample size, we were unable to reliably test three-way interactions among race, sexual identity, and masculinity/femininity. Inclusion of bisexual women may have biased some of our findings given that they tend to report a weaker set of gender-role norms (Rothblum, 2010). Second, our measure of stigma consciousness for lesbian respondents had a Cronbach’s alpha of .69, which is considered adequate or acceptable, but ideally would be higher. More work is needed to refine minority stress measures; however, given the lack of data sets that include such measures, we believe that our results using these measures provide important insights into the links between gender roles and minority stress. Third, our results were derived from a sample of women recruited in the Chicagoland area. Although the sample is diverse in its socioeconomic and racial/ethnic composition, it is not geographically diverse. Other research has shown that gendered behavior or appearance is sometimes tied to geographic characteristics and that SMW in rural spaces often weigh the safety risks of appearing out of place in their larger communities verses the benefits of being more identifiable as sexual minority (Gray, 2009).

Third, we were unable to assess level of connection to the sexual minority community as a potential mediating factor in the relationship between our gender measures and the dependent variables. It is possible that relationships between gender and minority stress are moderated by the extent to which an individual is connected to the sexual minority community and to the extent that they have adopted in-group norms. Our results were also limited in that we were unable to investigate racial/ethnic group differences beyond the three major groups included in the sample. Other factors not included in our analyses may have influenced findings such as sexual identity development milestones, disclosure to family and friends, and perceived peer group gender norms. Further, in this study, we used two single-item measures of perceived masculinity and femininity which likely do not capture the multifaceted nature of gender expression. Single-items measures, however, have been used in other studies (Martin-Storey & August, 2016; Steele, Everett, & Hughes, 2017). The broad nature of the survey questions may actually be beneficial as femininity and masculinity may mean different things to SMW of differing races/ethnicities (Harris & Crocker, 1997).

More research is needed to better understand the perceptions of gender roles and gender presentation among women in sexual minority communities of color. Finally, previous research has shown that SMW may “do” gender differently based upon context; that is, SMW may manage their gender presentations strategically to reduce potential conflict and discrimination within their families, workplaces, and communities. Qualitative and quantitative research is needed to understand how masculinity and femininity may vary by context.

Despite these limitations, our findings from a large, diverse sample of SMW provide insight into the complex relationships among masculinity, femininity, race/ethnicity, and minority stress. Adhering to gender-role norms is a complicated process of negotiating both in-group and out-group norms and expectations for both sexual minorities and heterosexuals. Although much more research is needed, our results highlight the limitations of drawing inferences from White samples of SMW and generalizing to all SMW. Future research should continue to investigate the complex relationships between gender, race, and minority stress in order to understand the factors that lead to sexual orientation-related health disparities across racial and ethnic groups.