Authors: Jordan E. Parker (1University of California, Los Angeles, Department of Psychology), Craig K. Enders (1University of California, Los Angeles, Department of Psychology), A. Janet Tomiyama (1University of California, Los Angeles, Department of Psychology), Jeffrey M. Hunger (2Miami University, Department of Psychology)
Categories: Article, Gendered racial microaggressions, disordered eating, self-silencing, Black women
Source: Body image
Authors: Jordan E. Parker, Craig K. Enders, A. Janet Tomiyama, Jeffrey M. Hunger
Gendered racial microaggressions are defined as the nuanced expressions of oppression that Black women can encounter at the intersection of their racial and gender identities. In the present study, we tested the hypothesis that greater frequency of gendered racial microaggressions will be associated with higher body dissatisfaction, drive for thinness, and binge eating and examined whether self-silencing mediated this association. Participants were 570 U.S.- based Black women (age M = 37.51 years) who completed an online survey on Prolific. Women completed self-reported measures of disordered eating, self-silencing, and the frequency of four specific dimensions of gendered racial (a) Assumptions of Beauty and Sexual Objectification, (b) Silenced and Marginalized, (c) Strong Black Woman Stereotype, and (d) Angry Black Woman Stereotype. Results indicated that greater frequency of gendered racial microaggressions in all four domains were associated with greater disordered eating. Self-silencing mediated this association across all domains except the Angry Black Woman stereotype. The results of the study not only highlight associations between gendered racial microaggressions and disordered eating, but additionally elucidate a potential mechanism through which this occurs—self-silencing—offering a promising avenue for future research oriented toward intervention among Black women, a population at elevated risk for disordered eating.
A robust body of literature links racial discrimination with adverse health outcomes among Black Americans in the United States. Across longitudinal studies, meta-analyses, and large-scale epidemiological research, experiences of racial discrimination have been linked to cardiovascular disease (Brondolo et al., 2011; Dolezsar et al., 2014; T. T. Lewis et al., 2014), diabetes (Whitaker et al., 2017), psychiatric disorders (T. T. Lewis et al., 2015; Pieterse et al., 2012), neurological conditions (Okeke et al., 2023), and overall poorer psychological well-being (Schmitt et al., 2014). Alongside theoretical models asserting that racism may also have powerful effects in the domain of eating behavior and body image (Exum et al., 2022; Kempa & Thomas, 2000), several recent studies have sought to test these associations empirically. Among Black women in particular, the literature concurs that discrimination is associated with several disordered eating cognitions and behaviors, including emotional eating, binge eating, and body dissatisfaction (Harrington et al., 2006; Hoggard et al., 2019, 2023; P. Johnson et al., 2012).
Although emerging research demonstrates an association between Black women’s experiences with racial discrimination and disordered eating, few studies have employed a framework that acknowledges the interplay between the multiple forms of oppression that Black women experience.^1^ As articulated by Black feminist scholars, intersectionality theory describes the tendency of law and other structural entities to consider discrimination through a single-axis lens, considering discrimination tied to singular social identities without acknowledging that particular identities are inextricable linked (Collins, 1989; Crenshaw, 1989). Indeed, Black women may not independently experience racism or sexism, but rather a merged or interactive type of oppression on the basis of their racialized gender identity. This includes not only overt forms of gendered racial discrimination, but also the “subtle everyday verbal, behavioral, and environmental expressions of oppression based on the intersection of one’s race and gender,” (J. A. Lewis et al., 2013), often referred to as gendered racial microaggressions (GRM). Through mixed-methods studies, scholars have identified the core features of GRM for Black women specifically (J. A. Lewis et al., 2013; J. A. Lewis & Neville, 2015) stereotyping of Black women, silencing and marginalizing of Black women, and expressing assumptions about Black women’s beauty and physical appearance. Studies to date have documented associations between GRM and depression, anxiety, traumatic stress symptoms, substance use, HIV viral load, and sleep disturbances (L. A. Burke et al., 2023; Dale & Safren, 2019; Erving et al., 2022, 2023; Shahid & Dale, 2023; Williams & Lewis, 2019).
Yet, few studies have explored the relationship between GRM and disordered eating among Black women. Whereas clinical eating disorders are defined by strict criteria set forth in the American Psychiatric Association’s Diagnostic and Statistical Manual (American Psychiatric Association, 2013), subclinical disordered eating comprises a broader set of abnormal eating-related cognitions and behaviors, such as loss of control over eating, excessive dieting and exercise, and preoccupation with one’s weight, shape, and/or size (Hirsch & Blomquist, 2020). These behaviors often precede and predict the onset of clinical diagnoses and are independently associated with adverse health outcomes (J. G. Johnson et al., 2002; McClelland et al., 2020; Stice et al., 2011). Studies conducted amongst Black women have found that GRM are associated with body image disturbances, such as lower body appreciation and higher body surveillance (Dunn et al., 2019; Stanton et al., 2022). Using a race-adapted version of the Schedule of Sexist Events (Klonoff & Landrine, 1995), one study also reported a significant zero-order correlation between gendered racial microaggressions and binge eating symptoms in Black women (Holly & Dickens, 2024). Finally, in related studies among Asian American women, frequency of GRM was associated with greater body shame and disordered eating (Le et al., 2020).
Two studies have examined and found support for the association between GRM and emotional eating, a maladaptive, eating-related coping behavior that appears to be uniquely relevant among Black women and that maintains similar associations with clinical eating disorder diagnoses (Hoggard et al., 2023; Lindeman & Stark, 2001; Pickett et al., 2020; Ricca et al., 2009). In a sample of young adult Black women, researchers found that greater frequency of GRM and the stress associated with experiences of GRM were positively associated with emotional eating (Volpe et al., 2023, 2024). Several factors mediated these associations, endorsement of the superwoman schema, which describes the racially-gendered pressures Black women feel to present an image of strength, put others’ needs first, and resist emotional vulnerability to avoid burdening others; food- and non-food related psychological inflexibility; and self-compassion. Studies have yet to examine how certain features of and/or specific domains of GRM may be associated with a broader spectrum of disordered eating cognitions and behaviors.
Moreover, extant evidence suggests that gendered racial microaggressions may not only lead directly to disordered eating but additionally contribute to a particular psychological state that explains this self-silencing. Both qualitative and quantitative studies find that, as a result of gendered racial microaggressions, Black women feel pressure to suppress, hide, or silence their feelings as well as refrain from burdening others with their emotional needs (Abrams et al., 2019; J. A. Lewis et al., 2013, 2016; Volpe et al., 2023; Watson-Singleton, 2017). Self-silencing is most often measured using the Silencing the Self subscale of the Silencing the Self Scale (Jack & Dill, 1992). Although it is measured in the context of intimate relationships, it is also thought to reflect the broader tendency to engage in emotional suppression to maintain relational harmony (Baeza, De Santis, & Cianelli, 2022; Jack & Dill, 1992). This interpretation is consistent with the use of this subscale in other studies wherein it is used to index general emotional suppression and inhibition of self-expression (e.g., Abrams et al., 2019; Avery et al., 2022). Self-silencing appears to be prevalent among Black women, theoretically owed to their unique social experience at the intersection of multiple forms of oppression (Abrams et al., 2019; Gratch et al., 1995). These findings have also been supported in empirical studies among Black women living with HIV, wherein gendered racial microaggressions are positively associated with self-silencing (Shahid & Dale, 2023).
Self-silencing may also be a mechanism linking gendered racial microaggressions and disordered eating among Black women. In the emotion regulation literature, empirical studies and meta-analyses reveal that the tendency to suppress emotions is associated with disordered eating cognitions and behaviors (Geller et al., 2000; Lavender et al., 2015; Leppanen et al., 2022; Prefit et al., 2019; Shouse & Nilsson, 2011). Studies have also found that self-silencing, in particular, predicts greater disordered eating cognitions and behaviors through mechanisms involving emotional dysregulation and suppression (e.g., (Litwin et al., 2017; Maji & Dixit, 2019; Norwood et al., 2011). Previous studies have found positive associations between self-silencing and drive for thinness, body dissatisfaction, bulimia, as well as composite measures of disordered eating symptomatology (Geller et al., 2000; Hambrook et al., 2011; Morrison & Sheahan, 2009; Piran & Cormier, 2005; Zaitsoff et al., 2002).
Although gendered racial microaggressions are associated with self-silencing and self-silencing predicts disordered eating, these associations have yet to be tested simultaneously amongst Black women. In a study conducted among Asian American women, gendered racial microaggressions were associated with greater emotional dysregulation, which in turn was associated with greater disordered eating (Le et al., 2020). These findings provide empirical support for our hypothesis that self-silencing may be a candidate mechanism linking experiences of gendered racial microaggressions and disordered eating cognitions and behaviors among Black women, specifically.
In the present study, we examined the association between GRM and body dissatisfaction, drive for thinness, and binge eating among Black women in the U.S. (Figure 1). These outcomes were selected based on prior literature demonstrating that these disordered eating cognitions and behaviors are highly prevalent amongst Black women, as well as present across the lifespan (Awad et al., 2015; Gilbert et al., 2009; Gordon et al., 2010; Lowy et al., 2021; Parker et al., 2023; Pike et al., 2014; Rogers Wood & Petrie, 2010; Simone et al., 2022; Taylor et al., 2007, 2013). Moreover, we sought to determine whether self-silencing mediated the association between GRM and disordered eating. We hypothesized that greater frequency of GRM will be associated with higher body dissatisfaction, drive for thinness, and binge eating symptoms overall and indirectly, via self-silencing. The present cross-sectional mediation analyses should be interpreted with caution and aim to provide a foundation for future longitudinal studies.
We aimed to recruit U.S.-based participants who self-identified as Black women, were over the age of 18, and fluent in English (verbal and written proficiency). The target sample size was 600 participants, determined based on funding availability. Of the 600 eligible participants who completed the study, 30 people failed or did not complete attention checks, resulting in an analytic sample of 570 participants. This sample size was deemed sufficient based on simulation studies demonstrating that a sample of 558 participants provides 80% power to detect a small indirect effect in accordance with Cohen’s (1988) benchmarks (Fritz & MacKinnon, 2007).
The focal predictor for the present analyses was gendered racial microaggressions, indexed using the corresponding scale (J. A. Lewis & Neville, 2015). The Gendered Racial Microaggressions Scale was developed to capture the nuanced experiences of oppression that Black women contend with due to their racialized gender identity. The scale contains 26 items indexing four unique dimensions of gendered racial (a) Assumptions of Beauty and Sexual Objectification, (b) Silenced and Marginalized, (c) Strong Black Woman Stereotype, and (d) Angry Black Woman Stereotype. Sample items for each of the four dimensions, respectively, “I have received negative comments about my hair when I wear it in a natural hairstyle,” “I have felt unheard in a work, school, or other professional setting,” “I have been told that I am too independent,” and “Someone accused me of being angry when I was speaking in a calm manner.”
Each of the items is captured on a frequency dimension, indexing how often these events occur over the course of an individual’s life. Response options range from 0 = never to 5 = once a week or more. Total and subscale scores are obtained by calculating the sum of the respective items and then dividing by the number of items in the scale, such that higher scores index greater frequency of gendered racial microaggressions. The present analyses focus on the frequency subscale scores rather than the total frequency score, given initial findings suggesting that the four-factor model was a better fit to the data than the unidimensional model (J. A. Lewis & Neville, 2015). Analyses with the total frequency score as the focal predictor are presented in Supplementary Materials. The scale was developed for use in Black women and has been shown to have high internal consistency in prior studies (Dunn et al., 2019, 2023; J. A. Lewis & Neville, 2015; Stanton et al., 2022; Volpe et al., 2023). In the present sample, Cronbach’s alpha was .93 for the total score and ranged from .70–.89 for subscale scores. McDonald’s omega was .94 for the total score and ranged from .71–.90 for subscale scores. Convergent validity is supported based on its relationship with other measures of race- and gender-based microaggressions, as well as psychological distress (J. A. Lewis & Neville, 2015).
Disordered eating was measured using the Body Dissatisfaction and Binge Eating subscales from the Eating Pathology Symptoms Inventory (EPSI; (Forbush et al., 2013) and the Drive for Thinness Subscale from the Eating Disorder Inventory Version 3 (EDI-3; (Garner, 2004). The EPSI Body Dissatisfaction subscale contains seven items indexing preoccupation and dissatisfaction with one’s weight, shape, and size (e.g., “I did not like how clothes fit the shape of my body”), and the EPSI Binge Eating subscale contains eight items indexing the tendency to engage in uncontrollable eating or experience loss of control over eating (e.g., “I stuffed myself with food to the point of feeling sick.”). For both the Body Dissatisfaction and Binge Eating subscales, each item is scored on a 5-point Likert scale. Total subscale scores are derived by summing responses across items. The EDI-3 Drive for Thinness subscale contains seven items scored on a 6-point Likert scale, indexing excessive concern with weight and desire for a thinner figure (e.g., “I am preoccupied with the desire to be thinner.”). The total subscale score is obtained by summing all responses in the subscale. Across all subscales, higher scores represent a greater number/severity of disordered eating cognitions and behaviors. And though not developed for Black women, the scales demonstrate acceptable internal consistency and model-based reliability in the current data (Cronbach’s alpha = .85, .78, and .90 and McDonalds’s omega = .86, .89, and .90 for Body Dissatisfaction, Drive for Thinness, and Binge Eating, respectively). Internal consistency is also supported in other studies (Bardone-Cone & Boyd, 2007; Jo et al., 2024; L. A. Jones & Cook-Cottone, 2013; Kelly et al., 2012; Parker et al., 2023; Patarinski et al., 2023; Yeboah et al., 2024). Validity of the Drive for Thinness scale is supported by its relationship to other measures of disordered eating, body image, self-esteem, appearance idealization, and discrepancy between perceived current and ideal body shape, in samples exclusive to and including Black women (Gilbert et al., 2009; Gordon et al., 2010; James et al., 2001; Lang et al., 2019). Evidence for the validity of the present EPSI subscales is supported by their correlation with other measures of disordered eating, weight concerns, and psychological affect in samples that include Black women (Exum et al., 2022; Forbush et al., 2013, 2014; Martin & Racine, 2017; Soulliard & Vander Wal, 2022); however, extensive validation within our target population has not been established.
Self-silencing was indexed using the 9-item Silencing the Self subscale from the larger 31-item Silencing the Self Scale (Jack & Dill, 1992). The Silencing the Self subscale indexes an individual’s tendency to inhibit self-expression to avoid conflict in the context of relationships with close others. Although measured in a relational context, the tendency to inhibit self-expression with intimate others is thought to reflect a broader tendency to engage in emotional suppression due to gendered expectations for women in their everyday life (Baeza et al., 2022; Maji & Dixit, 2019). Example items “I don’t speak my feelings in an intimate relationship when I know they will cause disagreement,” and “I think it’s better to keep my feelings to myself when they do conflict with my partner’s.” Each of the nine items are measured on a 5-point Likert scale. A total score is obtained by summing responses across items with higher scores indexing greater self-silencing. The Silencing the Self subscale had adequate internal consistency and model-based reliability among Black women in the present study (Cronbach’s alpha = .86; McDonald’s omega = .87) and in other studies (Abrams et al., 2019; Avery et al., 2022; Stokes & Brody, 2019). Evidence for the construct validity of the Silencing the Self Scale in Black women has been supported by studies indicating its relationship to depression and the Strong Black Woman schema, a distinct but related race- and gender-specific schema that describes the tendency of Black women to avoid expressing their needs due to a perceived obligation to project strength (Abrams et al., 2019; Avery et al., 2022; Gratch et al., 1995).
All analyses included self-reported age, education, and income as covariates, given prior literature demonstrating their relationship to disordered eating among Black individuals (N. L. Burke et al., 2023; Mikhail & Klump, 2021; Taylor et al., 2007). Education was dummy-coded into a 4-point categorical scale corresponding high school graduate or less (reference group), some college or trade school, bachelor’s degree, and any degree higher than a bachelor’s. Income was captured on a 15-point scale, ranging from 0 = less than 150,000 or more, in $10,000 increments and was treated as a continuous variable.
Data for the present analyses were collected from June to September 2023 via Prolific, an online participant recruitment platform designed for social science research. Prolific has been shown to have higher-quality data than similar online platforms like MTurk, Qualtrics, and Sona (Douglas et al., 2023; Palan & Schitter, 2018; Peer et al., 2022).
Participants were invited to participate in a larger study about their eating behaviors, attitudes toward food and weight, and their racialized experiences. The opportunity to enroll in the study was made available to individuals who identified as Black women (including trans women) and who were fluent in English, over the age of 18, located/based in the United States, and had a Prolific approval rating of 95 or higher (a ratio of the number of researcher approved/accepted submissions to a participant’s total number of submissions, indicating that the participant is likely to submit a valid response). Study participation was estimated to take 30 minutes and participants were paid 0.25. Those who were eligible then completed demographic information about their highest level of education, household income, relationship status, and gender identity. Finally, participants completed focal questionnaires in randomized order, including the present measures of disordered eating, gendered racial microaggressions, and self-silencing. Given that the study content related to possible eating disorder psychopathology, participants were provided with a series of mental health resources at the end of the survey. Study participation then concluded, and participants were paid for their time. All participants provided consent to participate in the study, which was approved by the UCLA Institutional Review Board (IRB#22–001606).
Data quality was verified using a series of agreement-based attention checks embedded in the study. These questions asked participants to select a specific answer choice on a given question. For instance, an example item “Please select ‘4 – Very Often,’” for a Likert scale-based questionnaire. Participants were excluded from the analyses if they answered incorrectly or failed to answer the question altogether.
All analyses were pre-registered at https://osf.io/bxhn4/. Data, code, and materials are all available on the OSF page for this project, https://osf.io/jq7cb/.
Hypotheses and the analytic plan were formulated after data collection but prior to any analysis and were pre-registered at https://osf.io/bxhn4/. First, we hypothesized that greater frequency of gendered racial microaggressions would be associated with higher body dissatisfaction, drive for thinness, and binge eating symptoms (Hypothesis 1, total effect = c; see Figure 1). We also hypothesized that the association between gendered racial microaggressions and disordered eating would be mediated by self-silencing, such that greater frequency of gendered racial microaggressions would be associated with higher body dissatisfaction, drive for thinness, and binge eating symptoms indirectly, via positive associations between gendered racial microaggressions and self-silencing, and self-silencing and disordered eating outcomes (Hypothesis 2, indirect effect = ab*; see Figure 1).
To test the hypothesis that GRM (X) is associated with disordered eating (Y) and that self-silencing (M) mediates this association, we fit one mediation model for each of the GRM subscales. Associations between GRM and each of the three disordered eating measures were estimated jointly to allow for residual correlations amongst the outcome variables. The key test of Hypothesis 1 was the significance of the total effect (c) of GRM on each of the disordered eating outcomes. The key test of Hypothesis 2 was the significance of the indirect effect (a**b*) for each disordered eating outcome. All inference and analyses were conducted using the Bayesian estimation routines in the Blimp software (Keller & Enders, 2021). Significance of an effect was determined by the absence of a null value of zero from the 95% credible interval.
Effect size selection was guided by Preacher & Kelley (2011). We present completely standardized indirect effects for all models as well as Rmed2, which indexes the proportion of variance in Y that both X and M have in common, but that can be attributed to neither independently (Fairchild et al., 2009; MacKinnon, 2008).
Missing data for the present analyses ranged from 0 − 4.74%. Thus, we used a Bayesian model-based imputation procedure to fit each of the mediation models (Yuan & MacKinnon, 2009). This approach imputes missing data that is logically compatible with each model of interest, based on a conditionally missing at random assumption. This approach is particularly well-suited for the proposed mediation analyses as this method provides appropriate tests of the indirect effect (Enders et al., 2020). Our general imputation approach using Bayesian model-based imputation in mediation analyses aligns with and is further detailed in several prior studies (Chiu et al., 2024; Parker et al., 2022, 2023).
Outlier analyses were performed in accordance with our pre-registration, on the imputed datasets. We compared estimates and inference from models with and without standardized residuals greater than an absolute value of three excluded. Across all analyses, results were robust to the exclusion of potential outliers and thus all results presented leverage the full analytic sample.
Sensitivity analyses were performed to examine the assumptions of mediation analyses. We examined the impact of unmeasured confounders, potential interactions between our predictor and mediator, and assumptions surrounding linearity and correct functional form (Fairchild & MacKinnon, 2009; Tofighi et al., 2019; Valente et al., 2017).
Table 1 presents the descriptive statistics for the sample. We report the average sum score or average mean score across participants for all continuous analysis variables, as well as the percentage of participants in each category for categorical analysis variables. Compared to 2023 U.S. Census data, the present sample was similar in age (sample mean age = 37.51 vs. census median age = 37.2 years old), but of higher socioeconomic status (sample mean income = 69,999 vs. census median income = $33,355) than Black women on average in the U.S. Table 2 displays zero-order correlations between continuous study variables, which were obtained using Markov Chain Monte Carlo estimation with iterative missing data imputation. Results indicated that, across subscales, GRM were positively correlated with all three disordered eating outcome variables. GRM in the Silenced and Marginalized domain as well as in the Assumptions of Beauty and Sexual Objectification domain were positively correlated with self-silencing. For the Strong Black Woman and Angry Black Woman subscales, GRM were negatively and uncorrelated with self-silencing, respectively. We additionally observed positive correlations between self-silencing and each of the three disordered eating outcomes, as well as amongst the disordered eating variables themselves. The associations in the table are mostly weak; fourteen of the 45 associations were below Cohen’s (1988) small effect size benchmark (.10), nineteen were in the small range (.10 to .30), and only three were in the moderate range (.30 to .50). The GRM subscales and the three disordered eating outcomes exhibited correlations amongst themselves that would be considered large effects (.50), which is not surprising given their strong conceptual linkages. Observed data were normally distributed and this was maintained in the imputation (Tables S1 and S2).
We found consistent support for Hypothesis 1 (total effect) across disordered eating outcomes and for each of the GRM subscales (see Tables 3–6; see Table S3 for the GRM total frequency model). Greater frequency of gendered racial microaggressions within each of the four specific domains was associated with greater binge eating, body dissatisfaction, and drive for thinness, controlling for age, income, and education.
Results indicated partial support for Hypothesis 2, examining the indirect effect of gendered racial microaggressions on disordered eating via self-silencing. For the Silenced and Marginalized and Assumptions of Beauty and Sexual Objectification subscales, greater frequency of gendered racial microaggressions had a significant indirect effect on body dissatisfaction, drive for thinness, and binge eating via positive relationships between gendered racial microaggressions and self-silencing (a path) and between self-silencing and each of the disordered eating outcomes (b path; see Tables 3 and 4). Observed effect sizes were standardized indirect effects ranged from .02 to .03 and the variance in disordered eating explained together, but not separately, by GRM and self-silencing ranged from 3–7%.
The indirect effect of the frequency of gendered racial microaggressions in the Strong Black Woman domain was in the negative direction—greater frequency of GRM was associated with lower disordered eating symptoms, due to the negative association between GRM in this domain and self-silencing (see Table 5). Nonetheless, the direct effect of GRM in the Strong Black Woman domain on disordered eating remained positive, in line with the previous total effect findings for Hypothesis 1. Effect sizes for the mediation were again the standardized indirect effect ranged from –.02 to –.03 and the variance in disordered eating explained together, but not separately, by GRM and self-silencing ranged from 2–4%.
Finally, the indirect effect of GRM in the Angry Black Woman domain on disordered eating was non-significant (Table 6). In this particular model, significant positive associations were maintained between self-silencing and disordered eating; however, the association between Angry Black Woman Stereotype GRM and self-silencing was non-significant, thus contributing to the non-significant indirect effect. Of note, the direct effect of GRM in the Angry Black Woman domain on disordered eating remained significant. Mediation effect sizes remained standardized indirect effects did not exceed .003 and Rmed2 ranged from 2–4%.
Sensitivity analyses examining the assumptions of mediation analyses are presented in Supplementary Materials. First, we estimated fixed residual correlations (ranging from −.30 to .30) between self-silencing and disordered eating outcome variables to represent the effect of a possible omitted variables (Tofighi et al., 2019; Valente et al., 2017). Residual correlations > .15 (comparable in magnitude to the sizes of our observed a and b paths) resulted in non-significant indirect effects. Negative residual correlations did not impact significance. This indicates that the influence of an unmodeled confounder on the mediator and outcome would need to be positive and at least as strong as estimated pathways in order to change the inference about the indirect effect from significant to non-significant. Addition of quadratic and interaction terms did not substantially alter model fit (see Table S4).
Our results provide greater insight into how specific domains of GRM are uniquely related to disordered eating. Previous studies have found associations between GRM with emotional eating and body image disturbances (Dunn et al., 2019; Stanton et al., 2022; Volpe et al., 2023, 2024), and the present findings extend this relationship to include binge eating and drive for thinness. Our results also provide novel insight into one mechanism through which GRM are associated with disordered eating cognitions and behaviors. Although greater frequency of GRM was consistently associated with a range of disordered eating outcomes, when examining whether self-silencing mediates this association, the pattern of indirect effects observed suggests that this association may depend on the nature or content of the microaggressions themselves.
Consistent with our hypotheses, self-silencing mediated the association between GRM in the Assumptions of Beauty and Sexual Objectification and in the Silenced and Marginalized domains with disordered eating. Results diverged from our hypotheses when examining associations in the Angry Black Woman and Strong Black Woman domains. For Black women who experienced GRM related to the Angry Black Woman stereotype specifically, they also reported greater disordered eating cognitions and behaviors, but this association was not mediated by self-silencing. Other dimensions of emotional regulation, specifically anger itself, may better explain this association. That is, when Black women experience microaggressions related to the Angry Black Woman Stereotype, rather than engaging in emotional suppression, they may instead lean further into the stereotype, reclaiming anger as a form of active resistance (Griffin et al., 2021; Griffin et al., 2012). Popular media articles encourage Black women to embrace rather than avoid anger when they encounter it, noting the historical context in which Black women did not have the autonomy or opportunity to freely express their response to living under conditions of oppression (Williams, 2024). In a speech aptly titled “The Uses of Anger: Women Responding to Racism,” Audre Lord argues that, in response to racism, anger is a powerful and necessary tool to combat complacency and to fuel change, resistance, and empowerment (Lorde, 1997). Given empirical studies indicating that both self-silencing and anger regulation differentiate subtypes of disordered eating from healthy controls (Norwood et al., 2011), future studies should also examine anger as a mediator of the association between GRM and disordered eating. Taken together with the present findings, it may be that emotional expression (i.e., lack of self-silencing and/or expression of anger), is particularly relevant for disrupting the link between GRM and disordered eating among Black women.
In the domain of Strong Black Woman stereotype microaggressions, although greater frequency of Strong Black Woman-related GRM maintained a positive association with disordered eating cognitions and behaviors for the total effect, the indirect effect of GRM on disordered eating was in the negative direction. This finding suggests that the aspects of Strong Black Woman Stereotype indexed by the present subscale—specifically, being perceived as a strong Black woman, independent, and/or assertive—may not necessarily be universally negative for Black women. In particular, it is important to acknowledge that the Strong Black Woman stereotype has been conceptualized as both a possible positive psychological resource as well as a detriment to the health of Black women (N. N. Watson & Hunter, 2016; Woods-Giscombé, 2010). At face value, it appears to describe the resilience of Black women’s spirit; however, it also often places unrealistic expectations of strength upon Black women, existing as a barrier against their ability to address their own emotional needs (T. Nelson et al., 2016).
Furthermore, the present 3-item subscale of the “Strong Black Woman” may not fully capture the scope of negative tropes placed on Black women in this domain of GRM. The present Strong Black Woman stereotype subscale captures independence, assertiveness, and the general Strong Black Woman trope. However, it does not comprehensively capture a perceived obligation to help others, the need to appear unwaveringly strong, or the intense motivation to succeed (e.g., Superwoman Schema subscales; (Woods-Giscombe et al., 2019) thus omitting other negative features of this stereotype that have been associated with adverse health outcomes for Black women in previous studies (Erving, McKinnon, Thomas Tobin, et al., 2024; Erving, McKinnon, Van Dyke, et al., 2024; Erving, Zajdel, et al., 2024).
The results of this study overall convey that GRM are associated with disordered eating cognitions and behaviors; however, self-silencing was not always the mechanism through which this association occurred. Indeed, it was a significant mediator in some models, but effect sizes remained in the small range across all models. Standardized indirect effects did not exceed .03 and, together, GRM and self-silencing explained only 2–7% of the variance in disordered eating outcomes. In sensitivity analyses, we estimated fixed residual correlations between the mediator and outcome variables, representing the possible impact of unmeasured confounders; (Tofighi et al., 2019; Valente et al., 2017). We found that residual correlations > .15 resulted in non-significant indirect effects across models. Negative residual correlations did not impact significance. Given that our observed associations on the a and b paths were of similar magnitude, in order to change the inference about the indirect effect from significant to non-significant, the influence of an unmodeled confounder on the mediator and outcome would need to be positive and at least as strong as estimated pathways. Moreover, given that direct effects remain significant even as residual correlations increase above .15, we have further evidence that the relationship between GRM and disordered eating is robust to unmeasured confounders. Nonetheless, we acknowledge that myriad other identity-specific and identity-agnostic intervening mechanisms remain. For example, depression has been linked to both GRM and disordered eating in previous literature and may also be a candidate mediator of this association (Bradford & Petrie, 2008; Erving et al., 2022; Williams & Lewis, 2019). Similarly, GRM has also been shown to be associated with decreased self-compassion and self-esteem as well as traumatic stress symptoms, which also have been related to disordered eating (Fresnics et al., 2019; Pelc et al., 2023; Vanzhula et al., 2019; Volpe et al., 2023; L. B. Watson & Henderson, 2023). Our previous work also demonstrates that more negative attitudes toward one’s racial group are associated with greater disordered eating among Black women, another plausible identity-specific mediator of this association (Yeboah et al., 2024).
These pathways merit future investigation as they may represent meaningful points of intervention for ameliorating the negative effects of GRM on disordered eating through indirect pathways. However, it is also important to note that the direct effects of GRM on disordered eating remained significant across all models. Thus, it is not sufficient to solely target individual-level mechanisms. Research, policy, and prevention efforts aimed at combatting rising rates of disordered eating among Black women must dually consider structural-level change targeted at GRM in and of themselves. Studies examining structural sexism have found that sexism in state-level policies are associated with disordered eating outcomes among for participants who spent more time living in states with greater structural sexism—indexed by things like the gender distribution of voting patterns, elected officials, business ownership, poverty status, and reproductive legislation—there was excess risk of disordered eating (Beccia et al., 2022). Through the lens of GRM, policies that seek to promote equity for Black women in these spaces, such as increasing Black women’s voter turnout, promoting their bids for public office, enabling targeted small business loans, and protecting their reproductive freedom, might specifically decrease disordered eating risk among Black women. The extent to which these structural changes might influence both the frequency of GRM and disordered eating is a promising avenue of inquiry for future studies.
This study builds on existing literature demonstrating associations between discrimination and disordered eating among Black women, using an intersectional measure of discrimination and an operationalization of oppression tied to multiple, linked social identities. This is a particularly important contribution—although there is a large body of literature examining individual forms of oppression and disordered eating, quantitative intersectional perspectives are lacking.
We also note the use of the Gendered Racial Microaggressions scale as an important recent development in psychological measurement. As a measure specifically designed to capture Black women’s gendered racial experiences, this scale allowed us to capture participants’ intersectional experiences of oppression without needing to consider the effects of either race or gender, independently.
The use of cross-sectional data is a key limitation in the present study. Concurrent measurement of our predictor, mediator, and outcome variables precludes us from establishing causality in our mediation analyses and we acknowledge that the temporal ordering of our analysis variables could occur in alternate directions. When these criterion for mediation analyses are not met, the onus is subsequently on the researchers to provide compelling evidence for the hypothesized temporal order (Fairchild & McDaniel, 2017), which in this case, is well-supported by other empirical studies. Several qualitative studies indicate that Black women report that they respond to instances of GRM by engaging in emotional suppression behaviors (A. M. Jones, 2023; J. A. Lewis et al., 2013, 2016). This directionality is also based in historical context, in which self-silencing was a necessary part of survival for Black women, a legacy that has been inherited by current generations, who feel external pressure to engage in emotional suppression (Scott et al., 2023). Evidence for the temporal association between discrimination and disordered eating is also well-supported by daily diary, experimental, and ecological momentary assessment-based studies (Brown et al., 2022; J. D. Nelson et al., 2023; Pascoe & Richman, 2011) as well as the broader literature linking discrimination with adverse health outcomes and behaviors (Richman et al., 2017). Finally, with regard to the temporal ordering between self-silencing and disordered eating, a large body of evidence causally links emotional dysregulation and suppression with disordered eating (Bodell et al., 2019; Svaldi et al., 2014; Warne et al., 2023), providing support for our hypothesized model.
We also must consider that the observed association between GRM and disordered eating may be due to an unmeasured confounding variable that predicts both constructs. For example, individuals who are hypervigilant may report greater exposures to GRM and also engage in disordered as a coping mechanism. Future longitudinal studies will provide greater insight on the temporal ordering, causal mechanisms, and allow us to measure and control for relevant confounds.
We also note that the present study examined trends across Black women, without fully exploring the heterogeneity that exists within. We did not capture detailed information on participants’ cultural or ethnic identities which may be related to their body image and eating behaviors. In our sample, although three participants identified as transwomen, we were unable to examine whether observed associations differ across gender identities. Given studies suggesting that body image and disordered eating are distinct among gender and sexual minority individuals and that they experience greater burden of disordered eating (Nagata et al., 2020; Rasmussen et al., 2023), greater representation is needed in future studies.
Finally, the current measure of self-silencing captured participant’s tendency for emotional suppression in the context of intimate relationships. Although this measure is the most common measure of self-silencing used in the literature (Baeza et al., 2022), it has only been accepted but not established that this particular subscale indexes general propensity for emotional inhibition in everyday life. Although the scale is focused on the context of close relationships, the underlying psychological phenomenon—silencing one’s own feelings and emotional experiences in service of interpersonal harmony—likely also manifests with non-intimate others. The need to have close and secure relationships is fundamental to our socioemotional development and informs how we interact with the broader social world (Bowlby, 1982). That is, to the extent that individuals engage in self-silencing in order to maintain close relationships, it may also inform their general interpersonal tendencies thus explaining their subsequent impact on psychological and physical health (Baeza et al., 2022; Maji & Dixit, 2019). This warrants more direct investigation in future studies.
Black women who experience greater frequency of GRM also exhibit more severe disordered eating symptoms. Self-silencing appears to mediate this association, such that Black women who experience GRM are also more likely to engage in emotional suppression, which is subsequently associated with greater engagement in disordered eating cognitions and behaviors. In addition to contesting the greater systemic context in which microaggressions arise, self-silencing may be an intermediate, modifiable behavioral practice relevant to mitigating the negative sequelae of gendered racial microaggressions on disordered eating cognitions and behaviors among Black women.