Effects of gendered racial microaggressions on the mental health of black women

This study aims to evaluate the effects of gendered racial microaggressions on the mental health of black women, specifically the influence of the variables identity and self-esteem on the relationship between the frequency of gendered racial microaggressions and mental health. 76 women participated in the study. The mean age was 24.62 years (SD = 6.3). Participants answered four instruments in addition to sociodemographic questions: Gendered Racial Microaggression Scale, Goldberg General Health Questionnaire, Group Identification Scale, and Rosenberg Self-Esteem Scale. The results indicate that the high frequency of gendered racial microaggressions predicts worse levels of mental health and self-esteem. Self-esteem mediates the relationship between microaggressions and general health. It is a protective factor of mental health. Identity moderates this relationship, so that a high identification as a black woman is related to lower levels of mental health when faced with a high frequency of discriminatory events. Despite some limitations, the objectives were achieved. Future studies should contribute with explanations of the relationship between gendered racial microaggressions and mental health.


Introduction
Mental health is the result of multiple and complex relationships between biological, psychological and social factors that depends on a dynamic balance between the individual's interactions and the interaction of other people, taking into account organic characteristics and personal and family backgrounds. In this sense, exposure to traumatic and stressful life events, such as unstable family environments, experiences of gendered violence, discrimination, and social exclusion, in a single or systematic way, is a preponderant factor for negative emotional changes in the mental health of individuals and their lives in general 1,2 .
Stressors related to racial discrimination 3 may affect mental health by multiple ways. For example, asymmetrical power relations between discriminant and discriminated groups related to different socioeconomic status and access to goods and services, as well as the perception of discrimination experiences at an individual level, may culminate in high levels of psychological stress and physical and mental health problems 4 . This situation is aggravated when a large number of stressful events are perceived in individual or social contexts in a daily basis 5 , such as intentional or non-intentional verbal or behavioral verbal abuse, which communicate hostile, derogatory, or negative offensive racial offenses towards a person or a target group 6 .
Indeed, studies have suggested that exposure to certain types of discrimination frequently has negative repercussions on the mental health of black individuals 7 . Racist experiences, for example, are related to abusive use of substances 8 , low self-esteem 9 , mental disorders 4 , and general depressive symptoms 10 . Studies have also pointed out the effects of racism on mental health and, consequently, on physical health. These studies indicate strong correlations with stress, depression and decline of physical health, which leads to a high prevalence of cardiovascular diseases and obesity 11,12 .
Despite consistent findings relating skin color and mental health, little is known about how other dimensions of social categorization, such as gender and social class, interfere with this relationship. Research on racism has often treated black people as a monolithic group. Such treatment creates gaps with regard to potential gender differences in experiences of discrimination and the possible effects on mental health, for example. Specifically in the field of study of microaggressions, a frequent criticism is that most studies evaluate only how microaggressions are triggered by a single identity of an individual separately (for example, women or black people). Such studies do not seek to understand how intersections among multiple identities can affect an individual's microaggression experiences, such as being a woman and a black woman 13 .
Corroborating this perspective, studies have suggested that exposure to certain types of discrimination may vary according to gender 14 . For example, black men are often associated with crime. Thus, they are more likely to suffer physical and verbal assaults by police officers 14 . Black women, on the other hand, are more susceptible to rape and domestic violence 15 , as well as suffering indifference from the police as regards their individual and respective community protection 16 .
Lewis et al. 17 argue that experiences of discrimination of black women differ from the discrimination experienced by black men and white women. This is because black experience a hybrid phenomenon resulting from the combination of racism and sexism. Such experiences, called gendered racial microaggressions, also include processes of marginalization, silencing, and objectification 18 . They may impact interpersonal relationships, leading to suppression of emotions and negative effects on mental health that differ from those experienced by white women 14 . Data obtained by Stevens-Watkins et al. 19 corroborate this perspective. The authors note that African-American women, for example, report twenty times more psychological stress than white women do. In addition, the Center for Disease Control and Prevention in the USA 20 also points out higher rates of sadness and hopelessness among black women, which leads to suicide ideation and planning 21 .
In Brazil, data from the Ministry of Health 22 indicate that, when the main causes of death among women are taken into account, black women die more than white women due to homicides, suicides, and ill-defined deaths. By analyzing common mental disorders (CMD) according to sociodemographic characteristics, Araújo et al. 23 observed a higher prevalence of CMD among black or brown women. These women had a low level of schooling, did not live with their companions, had an average income of up to one minimum wage, had children, and were heads of households.
Based on the above, it can be inferred that the knowledge on the social and personal determinants of health, as well as the way these variables are related, is fundamental for the establishment of a predictive perspective of mental health of black women. In addition to clarifying the relationship between the stress resulting from racial and gendered microaggressions and mental health, it is necessary to understand possible intervening factors in this relationship that act as a protector or potentiator of impacts of microaggressions on mental health. Individuals do not react to negative experiences in the same way (such as discrimination) due to personal, social and environmental variables 24 . For example, some studies have pointed out that the level of social identity 25 and self-esteem 26 may act as intervening variables in the effects of discrimination on mental health.
Individual aspects, consciousness of belonging to a social group, and the emotional value and meaning of such belonging define social identity 27 . In this sense, racial identity is a multidimensional construct that includes a sense of belonging or proximity to a group and the importance attributed to skin color regarding the self-concept and the self-image of individuals 28,29 . Indeed, identity may influence individual beliefs and the expectation that individuals have with regard to the treatment they will be given in different contexts 28,29 , which influences the perception of discrimination experiences 28,30,31 . Corroborating this view, Sellers and Shelton 32 have pointed to a significant association between racial identity and perceived racial discrimination, so that the greater the importance attributed to being black for the formation of self-concept, the greater the frequency with which experiences of discrimination are perceived.
In spite of contributions of studies on identity, there is a literature gap on individuals who identify themselves with multiple social categories since more than one social category may interact for the formation of self-concept. For example, Thomas et al. 33 conducted a study with African-Americans who stated that racial gender identity was more important to their self-concept than their racial and gender identity considered in isolation. According to this perspective, it is necessary to understand the implications of the intersection between both identities (being female and being black) for the experiences of discrimination and its consequences.
In addressing specifically the implications of the intersection between racial and gender identities, some studies have pointed out that such intersection may be both a risk factor (by increasing the perception of discrimination ex-periences) and a protective factor (lessening the negative impacts of mental health discrimination) 34,35 . In fact, the woman identifying herself as a black woman may raise awareness for discrimination experiences 36 . On the other hand, it may contribute with coping strategies to such situations, such as the ability to feel better by focusing on positive aspects of belonging to a group and less internalization of negative messages about themselves 34,37 .
In this sense, the centrality of racial and gender identity may also interact with self-esteem since there may be a lower probability of internalization of negative messages about being a black woman 38 . Self-esteem refers to global assessments of personal value we make about ourselves 39,40 . Some theorists recognize it as a central aspect of psychological well-being and a predictor of satisfaction with life 41 . In addition, self-esteem is strongly associated with how individuals relate to themselves and with society. Thus, individuals with high levels of self-esteem tend to have a more positive view of themselves and their abilities 42 .
Although individual aspects are important for the conformation of self-esteem, feelings of self-worth also seem to suffer the influence of collective aspects. In light of the theory of social identity 27 , this happens because there is a part of our identity that is influenced by participation in social groups and by the value and emotional importance attributed to them. It involves how the individual evaluates his/her belonging group, perceptions about the assessments that others make about it, and how much the individual identifies with that group. In fact, presenting a strong sense of belonging to a group may contribute to the psychological well-being of individuals, such as improving self-esteem 43 . However, some studies have pointed out that experiences of discrimination are associated with low self-esteem 30,36,44 . In addition, self-esteem has been pointed out as a possible mediator of the relationship between discrimination and depressive symptoms among people in minority groups 9,44,45 , i.e., self-esteem may lessen the negative effects of discrimination experiences on mental health.
Through this contextualization, the main objective of this article is to evaluate the impacts of gendered racial microaggressions on the mental health of black women. This study also tests the moderating role of social identity and the mediator role of self-esteem in this relation. Based on the theoretical review, four hypotheses were proposed. The Hypothesis 1 (H1) states that a high frequency of microaggressions leads to a worse mental health. The Hypothesis 2 (H2) proposes that a high frequency of microaggressions leads to low levels of self-esteem. In fact, some international studies have indicated that the most frequent perception of discrimination is related to worse mental health levels 3 and lower levels of self-esteem 46 .
In addition, this study proposes that the effects of the frequency of microaggressions on general health are mediated by self-esteem (Hypothesis 3, H3) and moderated by social identity (Hypothesis 4, H4). Some studies have already indicated that a lowering of self-esteem may be the explanatory factor of the impacts of microaggressions on the health of individuals 47 . On the other hand, the results on the moderating role of social identity are conflicting. Some studies have pointed out that a greater social identification with the belonging group is a protective factor 48 , thus lessening the effects of discrimination on mental health. Other studies suggest that a greater identification has an amplifying effect, increasing the perception of discrimination situations and the consequent negative effects on mental health 25 .

method sample
The criterion of inclusion was only women who identified themselves as black, excluding those who did not fit this profile. The sample consisted of 76 black women, aged between 18 and 56 years (M = 24.62; SD = 6.23). Among women reporting socioeconomic data, the majority reported being heterosexual (69.7%), single (53.9%), with incomplete higher education (47.4%), and having a family income above four minimum wages (36.8%). Because it is a convenience sample (non-probabilistic), only women who agreed to contribute to the study participated in it.

Instruments
In addition to reporting data on sociodemographic variables, participants answered the following instruments: Gendered Racial Microaggression Scale (GRMS). Originally developed by Lewis and Neville 18 , it comprises 39 items. Participants report frequencies (0 = Never, 5 = Once a week or more) and how much the event was experienced as stressful (0 = This never happened to me; 5 = Extremely stressful). For this study, this instrument was translated and adapted to Portuguese. Initially, two English-speaking researchers performed the translation, and a synthesis of the translated versions was then conducted. This synthesis has been translated back into English to evaluate whether original contents were preserved. To test its construct validity, a factorial analysis using the main axes method indicated the extraction of a general factor, which explained 50.2% of the variance. The factorial loads varied between 0.90 and 0.49. An analysis of the internal consistency of the measurement pointed out that it can be used as a unifactorial instrument (Cronbach's alpha = 0.97).
Goldberg General Health Questionnaire (GHQ-12). The instrument comprises 12 items that measure mental health and social dysfunction. It was developed by Goldberg 49 and adapted for Brazil by Gouveia et al. 50 . Due to the focus of this study, only items referring to mental health were used. Participants answered using a fourpoint Likert scale. For the inverted items, the response alternatives ranged from 1 = Absolutely not, and 4 = Much more than usual. In case of affirmative items, responses ranged from 1 = More than usual to 4 = Much less than usual. For practical purposes, the scores obtained were reversed, so that higher scores are the indication of a better level of mental health. The mental health factor had an α of 0.85.
Group Identification Scale. Originally produced for the Brazilian context by Wachelke 51 , this scale is a one-dimensional measure of group identification. It can be used to assess identification with different social groups. It comprises six items that can be adapted to any reference group. The following is an example of an item: "Being a ___ is an important part of how I see myself ". For this study, being a "black woman" was used as reference group (for example: Being a black woman is an important part of how I see myself). The items were answered using a Likert scale ranging from 1 = strongly disagree to 7 = strongly agree. The higher the score, the higher the level of identification with the group. The scale had a Cronbach's alpha of 0.86.
Rosenberg Self-Esteem Scale (RSES). This scale was originally developed by Rosenberg 40 and validated for the Brazilian context by Hutz 52 . It comprises ten statements that evaluate global self-esteem, that is, the set of thoughts and feelings referring to oneself (for example: I feel that I am a person of value, at least as much as other people; taking everything into account, I think I am a failure). The items were answered using a Likert scale ranging from 1 = strongly disagree to 7 = strongly agree. In this sense, the higher the score, the higher the level of overall self-esteem. The RSES had a Cronbach's alpha of 0.86.

Procedure
The instruments were applied electronically by online dissemination, and also using printed questionnaires. For the online application, the questionnaire link was sent to specific groups of black women, murals, and message boxes from Facebook users. Upon accessing the questionnaire in this version, the respondent was requested to access the Informed Consent that presented the objectives of the study and emphasized the anonymous and voluntary nature of the participation. By agreeing to participate, the respondent should click "Proceed" to be directed to the questionnaire. On average, it took thirty minutes to complete the filling of the form. The printed questionnaires were distributed in public places in the city of Fortaleza (CE), such as squares and shopping malls, where groups of women gather for specific events related to Afro-Brazilian culture. A prior appointment was arranged with organizers of events through social networks. Additionally, the questionnaires were distributed at a public university located in a municipality in the countryside of Ceará.

Data analysis
Initially, Pearson correlations were performed to evaluate the correlation between the variables identity, self-esteem, depression, and the frequency of gendered racial microaggressions. Then, a regression analysis (Enter method) was performed to test the relations between the variables of Hypotheses 1 and 2. For the mediation and moderation analyses (H3 and H4), the PROCESS macro (version 2.16) developed for SPSS by Hayes 53 was used. To evaluate the significance of mediation, specifically the reliability of indirect effects, the bootstrapping and the Sobel 54 test were used. Bootstrapping is a computational method involving repeated samplings drawn from a dataset to estimate the indirect effects on each of them 55 . The mediation hypothesis is accepted if the confidence intervals obtained through bootstrapping do not contain the value zero, that is, the indirect effects would have a val-ue other than zero 56 . One thousand samples of bootstrapping were computed, and the adopted confidence interval was 90%, which was calculated through the method bias corrected and accelerated (BCa). To test the existence of moderation, a significant interaction effect must be obtained between the moderating variable and the independent variable, indicating that the effects of the independent variable on the dependent variable vary according to the moderator levels 53 .

results
Initially, correlation analyses were performed between the study variables. As can be seen in Table  1, the frequency of microaggressions correlated positively with the stress level and negatively with mental health and self-esteem (p < 0.01). In addition, the stress of microaggressions had a negative correlation (p < 0.01) with mental health and self-esteem, and a positive correlation with the identity of women. There was also a negative correlation (p < 0.01) between mental health and identification as a black woman, and a positive correlation (p < 0.01) between mental health and self-esteem.
Subsequently, regression analyses were carried out in order to understand the predictive power of the frequency of gendered racial microaggressions to explain the mental health indicators (H1). As can be seen in Table 2, in the first step, mental health was regressed in relation to frequency of microaggressions. The frequency of microaggressions significantly explains 6% of the mental health variance (F(1.74) = 5.07, R = 0.26, R² adjusted = 0.06). Regression weights allow verifying that the variable frequency of microaggressions has an inverse and significant relation with mental health (b = -0.14; p < 0.05), so that a higher frequency of microaggressions predicts worse levels of mental health, corroborating H1. In the second step, with the objective of knowing the predictive power of the frequency of gendered racial microaggressions on the self-esteem (H2), another analysis was performed in which self-esteem was regressed in relation to microaggressions. Considering the self-esteem, the frequency of microaggressions explained 10% of the variance (F(1.74) = 8.08, p <0.01, R = 0.31, R² adjusted = 0.10). Similarly, a higher frequency of microaggressions predicts lower levels of self-esteem (b = -0.35, p < 0.01), corroborating H2. Considering that microaggressions significantly predict both self-esteem and mental health, it was possible to test the hypothesis of self-esteem mediation (H3) in the third step, in which mental health was reduced in relation to microaggression and self-esteem.
The results observed in Table 2 indicate that, by adding the self-esteem to the model, microaggressions fail to predict mental health significantly. The total effect of microaggressions on mental health, which was b = -0.14 (p < 0.05), was composed of a direct, non-significant effect (b = 0.05 (p = 0.40)) and self-esteem was composed of an indirect effect (b = -0.09 (p < 0.01)). This indirect effect presents a confidence interval that does not include the zero value (90%CI [-0.17; -0.04]), indicating a significant mediation effect, corroborated by Sobel's test (Z = -2.47, p < 0.01). As the direct effect of microaggressions on mental health became insignificant after considering self-esteem, we can state that mediation is complete, corroborating the H3. That is, the effects of microaggressions on mental health were completely mediated by self-esteem.
Having been corroborated by the mediation effect of self-esteem, the H4 was tested in sequence (i.e., identity moderates the effects of microaggressions on mental health). To do so, in the fourth step, we regressed mental health in relation to microaggressions, self-esteem, identity, and the interaction term between microaggressions and identity. To calculate the interaction term, the variables microaggression and identity were centered (subtracted from their mean). The results indicate that the main effects of microaggression and identity were not significant, while the effects of self-esteem remained significant. When we interpret the main effects of self-esteem, we observe that a higher self-esteem is related to better levels of mental health (b = 0.27; p < 0.001; η 2 p = 0.34). The most relevant for the moderation analysis was the effects of microaggression interaction term and identity. Thus, they were a significant prediction of mental health (b = -0.13; p < 0.001; η 2 p = 0.22). This result points out that the effects  of microaggressions on mental health are moderated by identity. By decomposing the effects and evaluating the impacts of conditioned microaggressions on identity levels, we observed that in participants with low identity (1 SD below the mean), microaggressions do not predict mental health (b = 0.09; p = 0.12). On the other hand, for participants with a high identity (1 SD above the mean), microaggressions predict mental health significantly (b = -0.19; p < 0.001). Therefore, the effects of microaggressions on mental health only occur among participants who present a higher identity value, which configures a conditioned effect to the values of the identity, thus corroborating Hypothesis 4.

Discussion
The main objective of this study was to evaluate the impacts of gendered racial microaggressions on the mental health of black women, as well as to test the moderating role of social identity and the mediator of self-esteem in this relation. The results obtained provided satisfactory evidence to corroborate the hypotheses, obtaining consistent results with the literature in the area. Next, the results observed for each hypothesis will be discussed.
The Hypothesis 1 states that a higher frequency of gendered racial microaggressions predicts a lower level of mental health. The results observed in this study corroborate the findings of the meta-analysis performed by Paradies et al. 4 of the impacts of discrimination on physical and mental health. The authors reported correlations between experiences of discrimination, mental disorders, stress, and general depressive symptoms. In addition, Stevens-Watkins et al. 19 and Teng et al. 21 pointed to the negative consequences of a high frequency of discriminatory experiences on mental health, suggesting that, in fact, coexistence with chronic environmental stressors predicts worse rates of mental health. It should be noted that the low variance observed between microaggression frequency and mental health may be related to the central role of other variables, such as self-esteem and social identity, in the explanation of mental health.
The Hypothesis 2 states that the frequency of microaggressions negatively predicts self-esteem. The results confirm the hypothesis, corroborating the findings of the literature on the negative effects of discrimination on self-esteem 9,30,36,44 . In fact, being exposed to a greater frequency of gen-der-based racial discrimination negatively affects the self-esteem of black women. This decrease in self-esteem happens because this concept is related to how others see us. Therefore, people who have socially devalued characteristics tend to internalize these opinions in a way that they have a self-esteem lower than individuals belonging to most valued groups 26 .
The Hypothesis 3 seeks to assess whether the impacts of gendered racial microaggressions on mental health is mediated by self-esteem. The results indicate that the direct effects of microaggressions on general health are not significant, but that the indirect effects, mediated by self-esteem, are significant. These results allow inferring that the effects of microaggression on mental health are completely mediated by self-esteem, that is, self-esteem is an important protective factor for mental health in the context of discrimination. These results are consistent with those observed in other studies that also tested the mediating role of self-esteem 9,44,45 . Mereish et al. 44 , for example, obtained similar results in a study with American and Caribbean blacks in the United States, pointing out that self-esteem plays a moderating role in the impacts of daily discrimination on the indicators of depression in the study's sample.
Finally, the moderating role of social identity in the relationship between racial microaggression and mental health is tested in the Hypothesis 4. Data indicated that social identity acts as moderator in this relationship. By decomposing the effects, we observe that microaggressions do not predict significantly mental health of participants with a low identity, so that the effects of microaggressions on mental health only occurs among participants with a high identity. This result is consistent with other studies that seek to clarify the role of identity in discrimination experiences 28,30,57 . Crocker and Major 30 , for example, reported that identification with minority groups may facilitate the perception of negative treatments as discriminatory behavior depending on group belonging. Similarly, Shelton and Sellers 32 observed that black people who considered skin color a central component of their identity were more likely to perceive situations with ambiguous interpretations as a result of discrimination. Major et al. 58 , for example, also noted that high levels of racial identification are associated with a greater perception of vulnerability to discrimination, culminating in chronic stress that may lead to lower rates of mental health 59 . Therefore, a greater group identification has an amplifying effect, increasing the perception of situations of discrimination and the consequent negative effects on mental health 25 .
Although the results presented here provide satisfactory evidence for the hypotheses tested, some limitations may be pointed out. The cross-sectional design adopted does not contemplate the cumulative effects of gendered racial microaggressions on the mental health of black women, nor does it allow establishing causal relationships. Longitudinal studies could provide stronger evidence for the predictive relationships observed between discrimination, identity, self-esteem, and mental health. Social identity may function as a buffer in the relationship between racial and gender microaggressions. However, this result should be interpreted with caution, since a low social identity may prevent individuals from detecting and reporting gendered racism when they experience it. That does not mean, however, that such experiences cannot negatively affect psychological well-being at an auto-processing level 25 .
Another limitation concerns the reduced sample size, a fact that may have occurred because the women invited to answer the survey did not identify themselves as black, but as brunettes or brown, which may have led to the exclusion of many participants from the final sample. The final sample obtained also presents different sociodemographic characteristics of the majority of black women in Brazil, which reduces the possibility of generalization of the results obtained here. In addition, most participants were engaged in groups or activities that emphasized the identity of black women, which may imply a higher average group identification compared to other black women who do not participate in such groups. Future studies on gendered racism should include samples with a greater sociodemographic diversity, which could provide a greater variability in possible gendered racial experiences. Other variables may also include mediators/moderators of the relationship between gendered racial microaggressions and mental health, such as social support and involvement with social movements.

final considerations
This study provides empirical evidence of the effects of gendered racial microaggressions on the mental health of black women, as well as of the role of social identity and self-esteem as intervening variables in this relationship, which may aid in measures to promote mental health and well-being among this population. The findings emphasize the importance of incorporating general psychological processes in understanding the relations between discrimination and psychic suffering among people in socially marginalized groups.

Collaborations
TV Martins contributed to the elaboration of the introduction, data collection and discussion of the results. TJS Lima collaborated in the preparation of the introduction, data analysis and discussion of the results. WS Santos participated in the discussion of the results and final review of the manuscript.