Utilizing Latent Class Analysis to Assess the Association of Intersectional Stigma on Mental Health Outcomes Among Young Adult Black, Indigenous, and Sexual Minority Women of Color

Purpose: Discrimination has detrimental effects on mental health, particularly among Black, Indigenous, and people of color who are also sexual minority women (BIPOC SMW); however, measurement of multiple intersecting forms of discrimination (e.g., race, gender, and sexual identity discrimination among BIPOC SMW) poses methodological challenges. This analysis uses latent class analysis (LCA) to examine the influences of discrimination on mental health in a convenience sample of BIPOC SMW. Methods: Online survey data from BIPOC SMW aged 18–29 years (n = 324) were used to estimate latent classes for discrimination type (race, gender, and sexual identity). Data for this study were collected from July to October 2018. Adjusted linear regressions examined the influences of discrimination profiles on perceived stress and depressive symptoms. Results: Utilizing LCA, the following four classes emerged: (1) low discrimination; (2) mid-level discrimination; (3) high racial, medium gender, and low sexual identity discrimination; (4) high discrimination. Classes 3 and 4 were positively associated with perceived stress and depressive symptoms relative to Class 1 in adjusted models. Conclusion: This analysis highlights the importance of intersectionality and the adverse impact of multiple forms of discrimination on mental health outcomes for BIPOC SMW. Respondents reporting higher levels of racial or multiple forms of discrimination had poorer mental health outcomes. LCA is a promising analytical tool for investigating intersectional stigma and discrimination. There is an urgent need to develop tailored, culturally appropriate intersectional mental health interventions to address the multiple identities and oppressions faced by BIPOC SMW.


Introduction
B lack, Indigenous, and/or people of color who are also sexual minority women (BIPOC SMW) have multiple identities that hold unique positions, as defined by society. 1,23][4][5][6] BIPOC SMW do not simply exist as any one identity (e.g., sex, race, or sexual orientation) nor is their experience of mutually influential identities easily disentangled.Thus, it is insufficient to understand their experiences through a single identity. 2esearch has generated evidence supporting a link between intersectional stigma and mental health inequities. 7,8[12]

BIPOC SMW and discrimination
Intersectional stigma refers to the co-occurrence or convergence of multiple dimensions of discrimination (e.g., sexism, racism, heterosexism) or discrimination that can be attributed to intersecting marginalized social positions (e.g., Black lesbian woman). 1,4,13It is well known that sexual minority (SM) persons are exposed to heterosexist discrimination; 14,15 however, BIPOC SMW encounter gender-and race-based discrimination in addition to heterosexist discrimination. 1,7,8Notably, SMW are impacted by sexism, a component of intersectional stigma that is not often captured in research centering SM men or broader SM populations despite its documented impact on SMW's mental health. 16esearch needs to better document how sexism contributes to experiences of intersectional discrimination (such as gendered racism 17 ) or how SMW experiences likely vary from experiences from heterosexual women counterparts. 10,18Studies have found that BIPOC SMW report a higher frequency of discrimination experiences across multiple dimensions (i.e., sexism, racism, heterosexism) when compared with White SMW and BIPOC SM men. 8In addition, relative to their White counterparts, research suggests that BIPOC SMW are more likely to experience any discrimination, 7 multiple dimensions of discrimination, and a wider scope of discriminatory encounters (e.g., in more settings and circumstances). 7,8,14Overall, the literature indicates that intersecting systems of marginalization may function to expose BIPOC SMW to a disproportionate amount of discrimination relative to other SM subgroups. 5,19

BIPOC SMW and mental health
Mental health is a serious concern among BIPOC SMW.Young SMW experience higher rates of depressive symptoms, anxiety symptoms, and suicidality relative to heterosexual peers. 20,21As much as 48% of Black SMW and 60% of Latina SMW met the diagnostic criteria for depression in a community sample. 9Moreover, BIPOC SMW face many barriers to accessing mental health services, including discouraging cultural norms, intersectional stigma enacted by providers, and structural disadvantage. 22Given the prevalence of depression among BIPOC SMW and lack of access to culturally appropriate care, it is critical to better understand the influence of intersectional factors on mental health outcomes of BIPOC SMW.
][25][26][27][28] Moreover, study findings indicate that experiencing intersectional stigma may further increase vulnerability to poor mental health outcomes above and beyond a single dimension. 10,29For example, greater exposure to discrimination, totaled across multiple dimensions, has been linked to lower psychological well-being, depression, chronic strain, and greater stressful life events. 8,29urthermore, studies have found that attributing experienced discrimination to more than one identity (e.g., sexual orientation, race, gender) is predictive of worse mental health outcomes. 7,8However, findings are complex.Although BIPOC SMW are more likely to experience both a higher frequency of discrimination and multiple dimensions of discrimination, overwhelmingly, studies have found no differences [30][31][32][33] and at times better mental health outcomes among BIPOC SMW relative to White counterparts 9,30,34 These seemingly paradoxical findings in a yet limited literature highlight the need to better understand intersectional stigma and mental health outcomes for BIPOC SMW.

Intersectional stigma and methodological challenges
Intersectionality scholars have long argued that to empirically explain the effect of intersectional stigma on outcomes, differing yet mutually constitutive dimensions of discrimination must be accounted for in a holistic way. 5,6,12To date, empirical investigations on the effects of intersectional stigma have often relied on analytical approaches that have been critiqued in relation to the intersectional framework, namely additive and multiplicative models. 10,12,35The exclusive use of these approaches has been critiqued based on their perceived incompatibility with intersectional frameworks, which challenges disentanglement of discrimination experiences at the intersection of identities. 2,4,6,10,12In relation to BIPOC SMW, the use of non-intersectional analytical models may be insufficient in capturing the complexity of intersectional stigma, thus inaccurately capturing experiences of discrimination at the intersection of identities and their influences on mental health. 4,6,36Intersectional theorists and methodologists continue to grapple with various approaches to measuring and analyzing intersectional stigma. 2,35e current study Latent class analysis (LCA) has emerged as a promising analytical tool for investigating intersectional stigma and discrimination. 4,6LCA is a person-centered analytical technique, useful for examining complex social phenomena, such as the effects of intersectional stigma. 4,6,37,38Researchers have utilized LCA to identify discrimination profiles among samples of multiply marginalized individuals and examine associations between discrimination profiles and various outcomes; 4,6,37 however, to the best of our knowledge, no study has used LCA to examine intersectional stigma in a sample of BIPOC SMW.
Given that few studies have investigated the effect of intersectional stigma on the mental health of BIPOC SMW, 10 and even fewer have done so using analytical methods consistent with intersectional perspectives, 6 the purpose of the current study was to utilize LCA to (a) identify distinct discrimination profiles among a racially and ethnically diverse sample of young SMW, based on their reported experiences of enacted discrimination, and (b) examine associations between discrimination profiles and two mental health related outcomes: (1) perceived stress and (2) depressive symptoms.

Procedures
In this secondary data analysis, we utilized survey data from an existing study 39 conducted with a racially and ethnically diverse sample of BIPOC SMW.Participants were recruited in July-October 2018 through paid advertisements on three social media platforms and unpaid online advertisements in various public forums.Advertisements described the study as a ''women's health study'' and provided a link to a landing page where participants screened for eligibility and provided informed consent electronically.Anyone who clicked on the advertisement was able to enter a raffle for one of nine $25 gift cards.The self-administered electronic survey took an average of 11 minutes to complete.Eligibility requirements imposed by the parent study included: (1) English-speaking, (2) identifying as a woman, (3) residence in the U.S. Southern Census Region, and (4) being between the ages of 18-29 years.Additional inclusion criteria for this analytic sample included: (1) identifying as BIPOC (Black, Hispanic/Latina, Asian/Pacific Islander, or Multiracial).The Emory University Institutional Review Board approved this study and provided ethical oversight.

Sample
Of the 1403 people who consented to participate, 324 (23.0%) met the inclusion criteria for the current analyses.Sample characteristics are presented in Table 1.

Measures
Discrimination.The 9-item Everyday Discrimination Scale was used to measure experiences of discrimination. 40he scale was modified to ask about discrimination relative to specific identities by adding a phrase such as ''due to your race or ethnicity'' to each prompt.Use of the modified scale has been published previously in a number of studies. 39,41The full scale was repeated for three identitiesrace/ethnicity, gender, and sexual identity.The scale assesses experiences of discrimination, such as ''You are treated with less respect than other people due to your sexual orientation'' during the preceding year.The six response options ranged from ''never'' (0) to ''almost every day'' (5).Scores on the summative scale ranged from 0 to 45, with higher scores reflecting more frequent experiences of discrimination.The Cronbach's alpha in the current sample was 0.96 for race, 0.91 for gender, and 0.93 for sexual identity.
Stress.We utilized the Perceived Stress Scale. 42The 10-item scale measures symptoms of stress, such as ''felt nervous or stressed,'' with a 5-level Likert response scale, ranging from ''Never'' to ''Very Often,'' for the past year. 42here were no missing data.The Cronbach's alpha in the sample was 0.85.
Depressive symptoms.Depressive symptoms were measured with the short form of the Center for Epidemiological Studies Depression Scale, which has 12 items 43 assessing depressive symptoms such as ''I could not get going'' during the preceding week.The four response options range from ''rarely or none of the time'' (0) to ''most or almost all of the time'' (3).Responses are summed across items and scores range from 0 to 36.A score of 0 to 11 represents minimal symptoms, 12 to 20 represents elevated symptoms, and 21 to 36 represents very elevated symptoms. 43The Cronbach's alpha in the current sample was 0.86.
Demographics.Age was measured in years (18-29 years).Self-identified race was measured as non-Hispanic Black, Latino or Hispanic, American Indian or Alaskan Native, Asian or Pacific Islander, and biracial/multiracial.For analyses, American Indian or Alaskan Native, Asian/Pacific Islander, and people who selected ''other racial identity'' were combined into one category due to low numbers.Education was measured as high school diploma or less, some college or an associate degree, or a bachelor's degree or a graduate degree.All women who are included in the analytic sample indicated that they identify as SMW.Participants were asked to indicate at least one sexual identity.
Sexual identity was then categorized as (1) Bisexual, (2) Pansexual, (3) Queer, (4) Gay/Lesbian, and (5) individuals who indicated multiple identities or less frequently indicated identities (e.g., demisexual).The fifth group was categorized as ''other sexual identity'' due to small numbers.Participants indicated their gender modality which in this sample included cisgender and transgender women.

Analyses
Univariate and bivariate analyses were used to examine the distributions of the variables included in this analysis and the assumptions for statistical tests.These procedures were conducted in SAS version 9.4.First, the latent classes were estimated with maximum likelihood estimation and bootstrapping methods in MPlus 8. Items used to estimate latent classes included the three discrimination scales.Decisions about class number selection were guided by the following seven criteria: (1) all class membership sizes being above 20 participants or roughly 5% of the sample, (2) entropy of 0.80 or higher, (3) Vuong-Lo-Mendell-Rubin likelihood ratio test, (5) Lo-Mendell-Rubin test, (6) parametric bootstrapped likelihood ratio test, and, importantly, (7) conceptual interpretability of latent classes. 44Resulting latent classes were then exported as a dependent variable and used in regressions for perceived stress and depressive symptom outcomes, controlling for demographic characteristics.

Univariates
The mean scale scores are reported in Table 1.The mean stress score was 25.4 indicating an average score in the moderate range (14-26). 45The mean depression score was 17.9 indicating elevated depressive symptoms. 43

Discussion
This analysis is one of the first published studies to employ an LCA approach to measuring intersecting forms of discrimination among BIPOC SMW by estimating latent classes based on gender, racial, and sexual identity discrimination.These results suggest by using a participant-centered analytic approach (e.g., LCA), we can observe more nuanced qualitative differences in discrimination profiles among BIPOC SMW and their differential impact on health outcomes.Our findings align with the literature suggesting that higher levels of discrimination are associated with negative health outcomes among BIPOC SMW. 1,7,8,14,47However, LCA may provide a more nuanced understanding suggesting that classes with higher levels of discrimination in at least one category (Classes 3 and 4) had significantly higher negative mental health indicators, with those (Class 4) having the highest levels of discrimination across categories with the worst mental health outcomes.Moreover, in post hoc analyses with Class 3 as the referent category, we observed significantly higher stress and depression scores among participants in Class 4 relative to Class 3.This finding suggests that not only high discrimination, but high discrimination across multiple discrimination targets (race, gender, and sexual identity) has a detrimental impact on mental health in BIPOC SMW.Although race, gender, and sexual identity discrimination have been found to independently affect the health of BIPOC SMW, 48 not all three forms are consistently captured in intersectional research.For example, studies in which BIPOC SMW are included in broader samples or are characterized by a single identity such as SM status may not fully capture the impact of sexism or racism among BIPOC SMW.This analysis underlines the importance of considering the combined influence of race, gender, and sexual identity discrimination in the lives of BIPOC SMW.
This analysis is informative to approaches of addressing intersectional stigma in that LCA allows for parsing of identity-based attributions and frequency of attribution while also establishing qualitatively different latent classes based on discrimination profiles to examine the combined influences on health.Some have suggested that researchers should move toward intersectional measurement where the intersection of identity is viewed holistically (e.g., measuring discrimination uniquely targeting the combination of Black, queer, women identities as opposed to measuring discrimination targeting each specific identity and summing these measures). 2,4,47Although an LCA approach does not fully address this critique, it provides another tool in addition to existing additive and intersectional approaches in an everexpanding toolbox. 2 This analysis emphasizes that BIPOC SMW do not simply exist as a single identity (e.g., sex, race, or sexual orientation), but rather at the intersection of multiple marginalized FIG. 1. Four-class solution with standardized estimates by discrimination type.1][22] Moreover, BIPOC SMW reside in higher concentrations in states lacking legal protections, for example, the South has the largest concentration of states without legal protections against discrimination of SM populations. 49nterventions that seek to address a single type of discrimination may not be effective at diminishing the mental health impact of discrimination among BIPOC SMW. 4,50Recent literature reviews on racial and sexual identity stigma emphasize social support interventions as beneficial buffers from the mental health effects of discrimination; 51,52 however, given the nature of intersectional forms of discrimination, interventions need to be adapted to incorporate the perspectives and needs of BIPOC SMW. 50

Limitations
The findings of this study should be considered in light of several limitations.First, our data were cross-sectional, precluding causal interpretations.Second, the final latent classes had strong fit indices across the board; however, Class 4 had a limited size (n = 13).The general recommendation is to consider class size in balance with other criteria, including conceptual coherence. 46Future studies should seek to replicate the LCA in larger samples.Third, this sample was age and geographically restricted by the parent study inclusion criteria, which thus limited the generalizability of these findings to the geographies, age ranges, and calendar period for the present analysis.
Given that experiences of discrimination and outcomes related to perceived stress and depressive symptoms may vary over the life course, replicating this study in other age groups of BIPOC SMW and in longitudinal cohort studies will be important to deepen our understanding of how historical, socio-contextual, and age-related experiences of discrimination relate to mental health in BIPOC SMW.Fourth, a strength of LCA methods is the ability to consider the combined effect of multiple forms of discrimination; however, in the present analysis, we were unable to fully account for differences by race/ethnicity or sexual identity due to sample restrictions or specific types of overlap in discrimination (e.g., gendered racism) due to the way discrimination is measured.Future studies may consider examining differences in latent classes across identities and overlapping forms of discrimination.Finally, the small size of the subsample of transgender women prevented us from drawing conclusions specific to transgender women.

Conclusion
BIPOC SMW experience intersectional stigma, including but not limited to gender, sexual identity, and racial discrimination.The present analysis suggests that BIPOC SMW who have discrimination profiles characterized by elevated discrimination across multiple types may experience a greater degree of stress and depressive symptoms relative to those who report lower rates of discrimination.Multilevel intersec-tional interventions are urgently needed to address the intersectional stigma faced by BIPOC SMW to improve their mental health and well-being.

Table 1 .
Demographics of Young Adult Black, Indigenous, and Sexual Minority Women of Color (n = 324)

Table 2 .
Fit Statistics for Possible Latent Class Solutions by Number of Classes Lower BIC and adjusted BIC values indicate a better fitting model.Significant LMR and BLRT indicate a preference for the current model over the model with one less class.Bold text is the selected solution.BIC, Bayesian information criterion; BLRT, bootstrapped likelihood ratio test; LMR-LRT, Lo-Mendell-Rubin likelihood ratio test.

Table 3 .
Regressions for Perceived Stress and Depressive Symptoms Among Young Adult Black, Indigenous, and Sexual Minority Women of Color (n = 324) a Adjusting for age, racial identity, gender, and sexual identity.SE, standard error.INTERSECTIONAL STIGMA IN SEXUAL MINORITY WOMEN