Discriminatory Experiences and Depressive Symptoms among African American Women: Do Skin Tone and Mastery Matter?

Abstract

We apply structural equation modeling techniques to data from the National Survey of American Life to investigate the relationship between perceived discrimination and depressive symptoms among African American women ages 18–98 years (N=2,299). In addition, we evaluate whether or not personal mastery accounts for the intensity of African American women’s psychological response to discrimination and whether or not exposure to discrimination varies by skin complexion. Findings reveal that discrimination is a major threat to African American women’s mental health. They are vulnerable to discrimination, in part, because discrimination undermines their beliefs in mastery making them less psychologically resilient. Experiences of discrimination do not differ by complexion. We conclude that complexion does not matter, but mastery does.

Introduction

Numerous studies show that experiences of discrimination and unfair treatment are prevalent among African Americans and other racial minorities and that these experiences are positively associated with emotional distress, especially depressive symptoms (for reviews, see Paradies 2006; Williams et al. 2003; Williams and Mohammed 2009). The link between discrimination and emotional well-being may be especially salient for African American women because they face the conjoint effects of both racism and sexism in historically unique ways (Collins 2000). With the exception of the Pavalko et al. (2003) study of work discrimination, studies of interpersonal discrimination among African American women have been based on small and/or geographically limited samples (e.g., Borrell et al. 2006; Kwate et al. 2003; Schulz et al. 2006b; Siefert et al 2007; Vines et al 2006). The current study uses nationally representative data and draws from the social stress perspective to address three specific research questions regarding the relationship between experiences of discrimination and depressive symptoms among African American women. First, are discriminatory experiences associated with increased reporting of depressive symptoms among African American women? Second, are some African American women exposed to more discrimination than others? We focus on skin tone as a source of differential exposure to discrimination because issues surrounding lightness and darkness of skin complexion have been implicated in African American women’s self- concept and emotional well-being (Hunter 2005; Russell et al. 1992; Thompson and Keith 2001), and some research suggests that discrimination is more prevalent among dark-skinned African Americans than among their co-ethnics with lighter skin (Hughes and Hertel 1990; Klonoff and Landrine 2000). Third, what do African American women draw on to protect their emotional well-being in the face of discrimination? We focus on the sense of mastery, the belief that one can control important circumstances affecting one’s life, because studies show that mastery is a resource that promotes psychological resiliency (Avison and Cairney 2003; Broman et al. 2000). To better understand the heterogeneity among African American women, we employ structural equation modeling (SEM) to evaluate the linkages among complexion, discrimination, mastery, and depressive symptoms. Our focus is on African American women in the U.S. context, but skin complexion and mental health issues are found among other ethnic groups in the U.S. and worldwide (see Rondilla and Spickard 2007; Glenn 2009).

African American Women, Discriminatory Stress, and Depressive Symptoms

The stress process is an especially useful theoretical framework for understanding how discrimination on the basis of ascribed characteristics such as race and ethnicity impacts mental health (Aneshensel 1992; Pearlin 1999). This framework argues that individuals are likely to suffer emotionally when they are confronted with numerous and/or ongoing problematic life circumstances that overwhelm their ability to cope. Further, psychological responses to stressful conditions vary along race, gender, and class divisions owing to differences in exposure to stressors and/or to differences in reactions to stressors. Discriminatory stress arises from both institutionalized processes and from day-to-day interpersonal interactions (Williams and Mohammed 2009; Williams and Williams-Morris 2000). Institutionalized discrimination, which occurs largely through residential segregation, limits the economic opportunities of African Americans and other people of color, relegating them disproportionately to low socioeconomic status and residence in economically deprived neighborhoods (Massey and Denton 1993). The mental health of residents in such neighborhoods is threatened by exposure to stressors such as crime, safety concerns, and general feelings of disorder (Hill et al. 2005). African American women, particularly if they are poor, face additional stressors such as domestic violence and single motherhood (Barbee 1992; Cutrona et al. 2000; Jackson et al. 2000; Siefert et al. 2007).

Personal encounters with biased treatment are also heavily implicated in African American women’s lives. One study conducted in the Detroit area revealed that 82 percent of African American women reported at least one lifetime episode of everyday discrimination such as being treated discourteously or called derogatory names (Brown et al. 2003). Two other Detroit area studies found that Black women were significantly more likely to report discrimination than White women, and that changes in discrimination were positively associated with changes in depressive symptoms over time for African American women (Schulz et al. 2006a, 2006b). Black women report negative, racialized exchanges during medical visits, bank transactions, while shopping in retail stores, and in a host of other settings and public places (Lawson et al. 1999). This study seeks to better understand the implications of interpersonal discrimination for African American women’s mental health. We examine its impact on depressive symptoms and also explore two additional aspects of the stress process, differential exposure and differential reactivity to discrimination.

Complexion, Discrimination, and Emotional Problems

The differential exposure hypothesis argues that some individuals are at greater risk for emotional difficulties because they experience more stressful conditions, and that greater or lesser exposure is linked to ascribed characteristics, status positions, and social roles (Turner et al. 1995). Skin complexion may be a major source of differential exposure to discrimination for African American women. Many scholars argue that one of the most enduring legacies of slavery and white supremacy in the United States is a racial stratification system that not only privileges Whites over Blacks, but also privileges African Americans with lighter complexions and a more Eurocentric appearance (e.g., straight hair texture, light eye color, and narrow lips and noses) over those with darker complexions and a more Afrocentric appearance (e.g., kinky hair, full lips, broad noses, and brown eyes) (Frazier 1957; Hunter 2005; Myrdal 1944). Researchers have documented a positive relationship between lighter skin tones and status achievement (e.g., Drake and Cayton 1945; Hill 2000; Hughes and Hertel 1990; Keith and Herring 1991), positive self-concept (Thompson and Keith 2001), and attributions of positive personality traits such as intelligence and competence (Anderson and Cromwell 1977). In a recent study Gullickson (2005) found that the salience of skin tone for status achievement, although not for marriage, largely disappeared for cohorts born after the mid 1950s, but other findings of an association between negative stereotypic traits (e.g., drug use, laziness) and Afrocentric physical characteristics (Maddox and Gray 2002) suggest the continuing significance of phenotypic stratification. Maddox (2004) theorizes that color gradations still matter because darker African Americans come closer to representing the cultural stereotypes, largely negative, that are associated with African Americans as a racial group. It is important to note that the argument that lighter skin color is more privileged than darker skin color is not synonymous with maintaining that light-skinned African Americans do not experience racism or discrimination. As Hunter (2005) points out, race and complexion represent two different, but overlapping, systems of oppression.

The hierarchical ranking of skin tone affects African Americans of both genders, but it has long been acknowledged that it plays a more essential role in women’s lives because it is inextricably tied to cultural values regarding physical attractiveness. In American society, idealized beauty and femininity are socially constructed to incorporate white or light skin, long hair, and European facial features as well as being thin. African American women who come closer to this ideal are considered more beautiful (Hill 2002a). Darker African American women have historically been deemed less attractive, less marriageable, and have been disadvantaged in terms of social mobility. Darker skinned African American women also report lower self-worth than their lighter skinned counterparts, although the relationship is strongest for those who are less affluent (Thompson and Keith 2001).

Only a few studies have explicitly investigated the relationship between skin tone and subjective experiences of discrimination, and these have yielded conflicting results. Herring (2004) found that darker African Americans reported more job discrimination than those of medium and light complexion. Klonoff and Landrine (2000) found that darker African Americans were 11 times more likely to be in the “high discrimination” group defined through cluster analysis. On the other hand, findings from the ongoing CARDIA study show no relationship between complexion and discrimination (Borrell et al. 2006). These inconsistent findings may reflect differences in sample composition as well as the measures of skin tone and discrimination employed. While the preponderance of evidence suggests that darker skinned African American women are at higher risk for unfair treatment, Hunter (2005) found that some light-skinned African American women experience social rejection and challenges to their racial authenticity from other African Americans. Whether darker skin is more disadvantaged than lighter skin or vice versa, the literature suggests that skin tone may be viewed as an additional status marker that exposes African American women to differing degrees of discrimination.

Mastery, Discrimination, and Depressive Symptoms

The differential vulnerability hypothesis suggests that some individuals have a higher risk of experiencing depression or depressive symptoms because they are more reactive to stressors. That is, when two individuals are exposed to equal levels of a given stressor, one may exhibit more or fewer symptoms than the other. We examine mastery as a mediator which might explain why African American women experiencing similar levels of discrimination vary in the number of symptoms reported. Mastery, along with other concepts (e.g., personal control, self-efficacy, and locus of control) is a social psychological construct that attempts to capture an individual’s sense of personal agency or the belief that one can control events or situations in one’s life that can serve as a source of resilience when stressors are operative (Avison and Cairney 2003). Individuals who believe that they have mastery over circumstances in their lives are less likely to be depressed, even when they are faced with persistent problems (Pearlin et al 1981). Those high in mastery are more likely to view themselves as being competent, to anticipate and avoid problems, possess skills that assist in resolving difficult issues, and are less likely to ruminate when problems do occur (Mirowsky and Ross 2003:195). Mastery appears to be especially protective of mental health when individuals face ongoing, intractable problems such as financial strain and role conflicts (Avison and Cairney 2003; Pearlin et al. 1981).

Research shows that mastery mediates the relationship between stress and depressive symptoms for African Americans generally (Lincoln et al. 2003), and that mastery is compromised for African American and other women when they experience high levels of discrimination (Ryff et al. 2003). Collectively, the literature shows that stressors damage mental health by eroding perceptions of mastery and undermining resilience, and points to mastery as a key factor in understanding how discrimination increases African American women’s risk for depressive symptoms and general psychological distress. Variations in complexion do not appear to be directly related to African American women’s perceptions of mastery (Thompson and Keith 2001), but this relationship may be mediated by discrimination.

The Socio-Demographic Context

Consistent with previous community-based research on the stress process (Pearlin 1999; Turner et al. 1995), discriminatory experiences, mastery, and depressive symptoms are likely to vary by African American women’s status positions. Both age and education are status positions that entail inequalities in the possession of power and privilege and both can shape one’s social experiences in complex ways. Prior research generally shows that depressive symptoms are most prevalent in younger African American women (Brown and Keith 2003; Rikert et al. 2000), although there is some evidence that symptoms increase in extreme old age (George and Lynch 2003). These findings are thought to reflect younger women’s heavy involvement in balancing the demands of intimate relationships, bearing and raising children, and establishing stable employment and careers. On the other hand, feelings of mastery appear to decline with age in the general U.S. population (Mirowsky and Ross 2003) and among African Americans (Broman et al. 2000), possibly due to increasing physical dependency and loss of meaningful relationships. Although findings are inconsistent (Paradies 2006), younger African Americans tend to report higher levels of discrimination (Banks et al. 2006; Broman et al. 2000; Kessler et al. 1999). A more active lifestyle is likely to increase younger African Americans’ interactions with others, creating spaces where discrimination is encountered.

The association between education and depressive symptoms is well-established in the mental health literature. As one indicator of socioeconomic status, formal education exposes individuals to problem-solving skills that are useful in managing life problems and avoiding emotional upset (Mirowsky and Ross 2003:77–84; Ross and Sastry 1999). Among African American women, education is inversely related to depressive symptoms and psychological distress (Gazmararian et al. 1995; Jonas and Wilson 1997). Studies also show that a large part, although not all, of the relationship between socioeconomic status and depressive symptoms is mediated by mastery in the U.S. population (e.g., Pearlin et al. 1981) and among African Americans (William et al. 1997). That is, levels of mastery tend to be higher for higher status individuals and this resource acts to offset the potentially devastating impact of stressors that often accompany resource deficits. Findings from studies examining the relationship between education and exposure to discrimination are mixed. In some studies, African American women who are better educated report more unfair treatment than their less educated counterparts (Borrell et al. 2006; Brown and Keith 2003), while other studies report no relationship (Klonoff and Landrine 1999). We expect better-educated African American women to report more discriminatory experiences in part because they are more likely to live and work in racially diverse settings where exposure to interpersonal racism is much more probable.

Conceptual Model and Hypotheses

The present investigation has three major advantages that will advance the literature on African American women’s mental health. First, we use a nationally representative sample that will allow us to confirm or disconfirm findings that have come from small and geographically specific samples. Second, we address the issue of differential exposure to discrimination based on skin tone and its implications for African American women’s mental health, a topic that has not received much empirical attention in the literature but has been widely speculated upon. We also examine mastery as a possible mediating mechanism that explains how discrimination affects psychological reactions to discrimination. Third, we use structural equation modeling (SEM) to test all hypotheses and to evaluate direct, indirect, and total effects of skin tone, discrimination, and mastery on depressive symptoms, and the direct effects of socio-demographic variables on discrimination, mastery, and depressive symptoms. Figure 1 presents the conceptual model and outlines the hypotheses. Age and education are treated as exogenous variables and all other constructs as endogenous variables.

Fig. 1.

Conceptual model and hypotheses.

• H1Women who report higher levels of mastery will report fewer depressive symptoms.
• H2Women who report more experiences of discrimination will report lower levels of mastery and more depressive symptoms.
• H3Lighter women will report fewer experiences of discrimination, higher levels of mastery, and fewer depressive symptoms.
• H4The relationship between complexion and depressive symptoms will be mediated by discrimination and mastery.
• H5The relationship between discrimination and depressive symptoms will be mediated by mastery.
• H6Older women are expected to report less discrimination, lower mastery, and fewer depressive symptoms than younger women.
• H7Education is expected to be positively associated with discrimination, positively associated with mastery, and negatively associated with depressive symptoms.

Method

Data

The data used in this investigation are taken from the National Survey of American Life: Coping with Stress in the 21^st Century (NSAL). The field work for the study was completed by the University of Michigan’s Institute for Social Research’s Survey Research Center, in cooperation with the Program for Research on Black Americans. The NSAL sample has a national multi-stage probability design which consists of 64 primary sampling units (PSU’s). Fifty-six of these primary areas overlap substantially with existing Survey Research Center National Sample primary areas. The remaining eight primary areas were chosen from the South in order for the sample to represent African Americans in the proportion in which they are distributed nationally. The data collection was conducted from February 2001 to June 2003. The interviews were administered face-to-face and conducted within respondents’ homes; respondents were compensated for their time.

A total of 6,082 face-to-face interviews were conducted with persons aged 18 or older, including 3,570 African Americans, 891 non-Hispanic Whites, and 1,621 Blacks of Caribbean descent. The overall response rate of 72.3% is excellent given that African Americans (especially lower income African Americans) are more likely to reside in major urban areas which are more difficult and expensive with respect to survey fieldwork and data collection. Final response rates for the NSAL two-phase sample design were computed using the American Association of Public Opinion Research guidelines for Response Rate 3 samples (AAPOR 2006). A more detailed discussion of the NSAL sample is found in Jackson et al. (2004) and Heeringa et al. (2004). The NSAL has been approved by the University of Michigan Institutional Review Board. The African American sample is nationally representative of Black households in the 48 coterminous states with one adult aged 18 and over (Jackson et al. 2004). The present analyses are confined to African American women (N=2,229). The mean age of respondents is 42.93 years (SD=16.40) and years of schooling completed averages 12.34 (SD=2.52).

Measure

The depressive symptoms measure consists of 12 items from the Center for Epidemiological Studies-Depression scale (CES-D) (Radloff 1977). This short version of the scale has acceptable reliability and the factor structure is similar to that of the original scale. Item responses are coded from 0 (“rarely or none of the time”) to 3 (“most of the time”) with high scores indicating more symptoms. Given the difficulty of fitting structural equation models that incorporate scales with many items, we reduced the number of items by randomly assigning the 12 CES-D items to four parcels or subcategories and created four new variables (Kishton and Widaman 1994; MacCallum and Austin 2000). We then performed a confirmatory factor analysis within structural equation modeling to verify the selection. Thus, the latent construct that represents depressive symptoms consists of four parcels or subsets of the 12 items. The Depression1 parcel consists of “felt depressed,” “was happy,” and “felt people disliked me.” Items in Depression2 are “everything an effort,” “hopeful about the future,” and “sleep restless.” The Depression3 items are: “just as good as others,” “trouble keeping mind on tasks,” and “had crying spells.” Depression4 consists of “people unfriendly,” “couldn’t get going,” and “enjoyed life.” The alpha coefficient for all 12 items is .78. The procedures used in creating parcels were repeated for mastery and discrimination.

Mastery is measured using five items from Pearlin et al. (1981) and is categorized into two parcels. Responses to these items range from 1 (“strongly agree”) to 4 (“strongly disagree”) with high scores indicating higher levels of mastery. The Mastery1 items are: “no way I can solve problems,” “I have little control over what happens,” and “helpless in dealing with problems.” Mastery2 consists of “pushed around in life” and “little I can do to change things.” The alpha coefficient is .77 for all five items.

Discrimination was measured by asking respondents about 10 episodes of unfair treatment experienced in their day-to-day lives and were grouped into three parcels. Discrimiantion1 items are: “called names or insulted,” “people act afraid of you,” “people act as if they are better,” and “threatened or harassed.” Discrimination2 includes “treated with less courtesy,” “people act as if you are not smart,” and “followed around in stores.” Discrimination3 includes “people act as if you’re dishonest,” “treated with less respect,” and “receive poorer service.” Response values ranged from 1 (“almost everyday”) to 6 (“never”) with variables coded so that high scores reflect more discrimination. This measure of discrimination captures perception of chronic daily discrimination rather than major forms of unfair treatment such as being denied a job. Prior research finds that chronic discrimination is more closely linked to depressive symptoms and general distress (Kessler et al. 1999; William et al. 1997). The alpha coefficient is .88 for all 10 items.

In the NSAL, complexion was assessed by asking respondents to evaluate their skin tone using five categories: 1 (“very dark brown”), 2 (“dark brown”), 3 (“medium”) 4 (“light brown”) and 5 (“very light brown”), and by asking the interviewer to rate the respondent’s skin tone using seven categories: 1 (“very dark”), 2 (“dark”), 3 (“somewhat dark”), 4 (“medium”) 5 (“somewhat light”), 6 (“light”), and 7 (“very light”). While the response formats differ for the two indicators of complexion, a major strength of structural equation modeling is that it accommodates varying response formats by directly incorporating measurement error into model estimation, eliminating the need for transformations (Bollen and Long 1993). The correlation between the two measures of complexion is .80.

Socio-demographic Controls

All analyses include controls for age and education. Age is measured as continuous years and education is measured as years of schooling completed.

Data Analysis

Structural equation modeling (SEM) is used to examine paths that link the primary model constructs as well as their relationships with the socio-demographic controls. SEM permits testing of the entire model rather than testing coefficients individually (Garson 2009). A coefficient depicting a relationship between two model constructs is interpreted in the same way that a standardized beta coefficient is interpreted in ordinary least squares regression. We allow for correlated error terms between the socio-demographic variables (Bollen and Long 1993) which are treated as exogenous. All analyses were performed using Mplus 5.0. To obtain results that can be generalized to the population of African American women, all of the analyses utilize analytic weights. Additionally, standard error estimates corrected for unequal probabilities of selection, nonresponse, post-stratification, and the sample’s complex design (i.e., clustering and stratification) are utilized.

Results

Table 1 presents the standardized factor loadings and measurement error associated with items used to measure the latent constructs (Fig. 1). The factor loadings provide some information on the psychometric properties of the items. Although there are no firm rules with respect to cutoff values, factor loadings at or above .400 are generally deemed acceptable. Factor loadings in Table 1, which range from .592 to .912, surpass the conventional value. An estimate of the true score variance can also be obtained by squaring the factor loadings. The higher the proportion of true score variance, the higher the reliability. Squared factor loadings indicate that the measures account for between 37% and 89% of the variance in the latent constructs.

Table 1.

Factor loadings and residual error terms^ab.

Item	Descriptions	Factor Loadings	Error Terms
η1	Complexion
	y1 R’s rating of skin tone	.868	1.0
	y2 IW’s rating of R’s skin tone	.880	1.0
η2	Discrimination
	y3 Discrimination1	.912	.142
	y4 Discrimination2	.759	.423
	y5 Discrimination3	.760	.357
η3	Mastery
	y6 Mastery1	.851	.189
	y7 Mastery2	.657	.467
η4	Depressive Symptoms
	y8 Depression1	.861	.111
	y9 Depression2	.592	.331
	y10 Depression3	.701	.183
	y11 Depression4	.760	.132

^aStandardized factor loadings and error terms.

^bThe residual variance is fixed to 1 with the theta parameterization.

Correlations, means, standard deviations, and ranges for all variables are included in Table 2. In keeping with the goal of being able to generalize our results to the population of African American women in the U.S., these descriptive data reflect adjustments to the sample after application of analytic weights and corrections for clustering and stratification. Mean complexion is 3.6 (SD=1.0) on a scale that ranges from 1–6, indicating that as in previous studies (e.g., Thompson and Keith 2001; Udry et al. 1969) medium complexion is the most prevalent category. The mean score on the discrimination scale is 2.15 (SD=.86; range=1–5.80). Although 88% of respondents reported some discrimination and 27% reported each type of experience, the results suggest that respondents were on average exposed to quite modest levels of unfair treatment. The descriptive statistics for mastery (M=3.14; SD=.77; range =1–4) and depressive symptoms (M=.60; SD=.51; range 00–2.75) reveal that on average African American women have high levels of mastery but not high levels of emotional distress. However notable is that a significant minority of women (15%) scored two standard deviations above the mean (M= 1.62 or higher) on the CES-D scale and, on the summed version of the scale, 34.8% and 29.3% reported a raw score of 9 or higher or 10 or higher, respectively, out of a possible score of 36. Some research suggests that raw scores of 9–10 or above on the short CES-D scale signal that clinical depression should at least be suspected (see Kohout et al. 1993; Pascoe et al. 2006). Table 2 also shows that the correlations between variables are not unduly high. The largest correlation is between mastery and depressive symptoms (r=−.492) and is within standard limits (Pearlin et al. 1981). To further ensure that multicollinearity did not affect our results, we conducted collinearity diagnostic tests. Results of from the diagnostic tests show that in all cases tolerance is greater than .20 and the VIF is less than .10, indicating that there is no problem with multicollinearity. Correlations for individual items in the complexion, discrimination, mastery, and depressive symptoms scales also indicate no multicollinearity issues and are available from the authors.

Table 2.

Correlations, means, standard deviations, and ranges.

	1	2	3	4	5	Mean	SD	Range
1. Age						42.93	16.40	18.00–93.00
2. Education	−.202**					12.34	2.52	.00–17.00
3. Complexion	.106**	−.084**				3.6	1.00	1.00–6.00
4. Discrimination	−.252**	.024	.030			2.15	.86	1.00–5.80
5. Mastery	−.084**	.260**	−.049*	−.218**		3.14	.77	1.00–4.00
6. Depressive Symptoms	−.145**	−.227**	.021	.360**	−.492**	.60	.51	.00–2.75

^*p<.05,

^**p < .01.

The linkages between the theoretical constructs are depicted in Fig. 2. The coefficients are also presented in Table 3 along with statistics that indicate how well the hypothesized model actually fit the data. Consistent with large sample sizes, the chi-square is large and significant (1,411.515, p=.000), but the alternative statistics indicate a good fit between the data and proposed model (Hancock and Mueller 2006; Tucker and Lewis 1973). The Comparative Fit Index (CFI=.955) and Tucker-Lewis Index (TLI = .970) are each above .90, the acceptable level. The Root Mean Square Error of Approximation (RMSEA=.026) is less than .05, the conventional level, although the Weighted Root Mean Residual, (WRMR=1.007) does exceed .90.

Fig. 2.

Model coefficients.

Table 3.

Complexion, discrimination, mastery and depressive symptoms^ab.

	Discrimination	Mastery	Depressive Symptoms
Age	−.302***	−.180***	−.208***
Education	.015	.276***	−.124***
Complexion	.066	.005	−.013
Discrimination		−.339***	.246***
Mastery			−.544***

^aModel Fit: Χ² (10, 2,298)=1,411.515; CFI=.955; TLI=.970; RMSEA=.026; WRMR=1.007.

^bStandardized regression coefficients.

^***p<.001.

Turning to the substantive findings, there is support for Hypotheses 1 and 2. African American women who view themselves as being able to exercise some control over their life circumstances report fewer depressive symptoms (β=−.544, p ≤ .001). As predicted, discrimination experiences tend to undermine mastery (β=− .339, p ≤ .001), but is associated with a greater number of depressive symptoms (β=.246, p ≤ .001). We find no support for Hypothesis 3. Discrimination is not less prevalent among women with lighter complexions. Instead, the coefficient is positive, although not significant (β=.067). Complexion does not have a significant association with either mastery (β = .005) or depressive symptoms (β=−.013).

An examination of the effects of social-demographic variables in Table 3 and Fig. 2 is also instructive for understanding the social patterning of discrimination, mastery, and depressive symptomatology. Hypothesis 6 is supported by findings showing that older African American women report significantly fewer experiences of discrimination (β=−.302, p ≤ .001), lower levels of mastery (β= −.180, p ≤ .001), and fewer depressive symptoms (β=−.208, p ≤ .001) compared to their younger counterparts. Thus, older women maintain better mental health overall because their lower levels of mastery are offset by an even lower likelihood of experiencing discrimination. The results also show partial support for the effects of education, Hypothesis 7. Higher levels of education are associated with higher levels of mastery (β=.276, p ≤ .001) and fewer depressive symptoms (β=−.124, p ≤ .01). The relationships between education and discrimination is not significant (β=.015)

To evaluate Hypotheses 4 and 5, Table 4 presents direct, indirect, and total effects for the major model constructs. Hypothesis 4 is not supported. Complexion is not associated with African American women’s depressive symptoms either directly or indirectly through discrimination or mastery. The results, however, do show that mastery is a critical link between discrimination and depressive symptoms and is a source of differential reactivity, confirming Hypothesis 5. Indeed, mastery accounts for about 43% of the relationship between discrimination and depressive symptoms (.185/.431=.426). African American women who are subjected to higher levels of unfair treatment experience more depressive symptoms, in part, because day-to-day discrimination undermines overall confidence in their ability to manage life challenges; leaving them feeling powerless and depressed.

Table 4.

Decomposition of effects for the model of complexion, discrimination, mastery and depressive symptoms.

Dependent Variable/Independent Variable	Causal Effectsa
	Direct (A)	Indirect (B)	Total (A+B)
Discrimination (η2)/Complexion (η1)	.066	.000	.066
Mastery (η3)/Complexion (η1)	.005	−.023	−.017
Depressive Symptoms (η4)/Complexion (η1)	−.013	.026	.013
Mastery (η3)/Discrimination (η2)	−.339***	.000	−.339***
Depressive Symptoms (η4)/Discrimination (η2)	.246***	.185***	.431***
Depressive Symptoms (η4)/Mastery (η3)	−.544***	.000	−.544***

^aStandardized regression coefficients.

^***p<.001.

Discussion

This study drew upon the stress process framework to investigate the relationship between experiences of discrimination and depressive symptoms among African American women. We also sought to further understand the implications of discrimination for African American women’s mental health by evaluating whether or not women with darker complexions are treated more unfairly than women with lighter complexions, and by determining if some women have more intense psychological reactions to discrimination than others because it undermines their sense of mastery. Our results show that perceptions of unfair treatment, like other chronic stressors, are associated with more symptomatology among African American women, confirming findings previously reported in less representative studies (e.g., Borrell et al. 2006; Schulz et al. 2006b; Vines et al. 2006). Being treated with less courtesy, insulted or called names, and receiving poorer service on a persistent basis are psychologically burdensome to African American women. Discrimination can be viewed as a status strain, a stressor that arises from one’s position in a hierarchical social system with unequal access to resources and opportunities (Pearlin 1999). As women and as members of a racial minority, African American women experience the strain of discrimination at both structural and interpersonal levels. Racialized and gendered structural processes assign African American women disproportionately to the bottom of this hierarchy; to low socioeconomic status and poor living environments where they are exposed to many types of problems that threaten emotional stability. Within this stratified system, the research presented here indicates that African American women are further subjected to psychological assaults through disrespectful everyday interactions with others whether or not they are poor or live in poor neighborhoods.

Our findings confirm that mastery mediates the relationship between discrimination and depressive symptoms and plays a major role in explaining why some African American women are more emotionally responsive (i.e., vulnerable) to discrimination than others. Many women suffer emotionally because they are unable to view themselves as efficacious and competent actors when treated with suspicion and confronted with dehumanizing interactions. Ross and Sastry (1999) theorize that mastery is shaped over time by the successful performance of tasks and the ability to effectively solve problems, patterns that reinforce the belief that personal efforts are rewarded by goal attainment. These researchers argue that discrimination undermines personal agency because when people are treated on the basis of unalterable, ascribed characteristics such as race the relationship between effort and outcomes is broken. Broman et al. (2000) also observed that racism and discrimination instill a sense of powerlessness because discriminatory experiences are largely uncontrollable. Interpersonal discrimination involves other actors, making it difficult to anticipate or avoid such encounters. While African American women, as do other women, generally have lower levels of mastery than their male counterparts (Hughes and Demo 1989, Mirowsky and Ross 2003:188–189), it is instructive to note that our findings also acknowledge that many African American women find ways to cope with discrimination and retain beliefs in agency. Better educated and younger women in this study maintained a higher sense of mastery than their counterparts, although for younger women it was not enough to overcome the effects of discrimination. These results are consistent with arguments that educational advancement builds problem-solving skills, competencies, and confidence (Mirowsky and Ross 2003:182–183; Ross and Sastry 1999). In contrast, mastery appears to decline with age perhaps because increasing health problems that require care by others leads to dependency. Alternatively, older African American women may feel less masterful because they have experienced a life time of discrimination where efforts were not rewarded, and contemporaneous mastery is rooted in these past experiences (Pearlin et al. 2007). Yet, despite less mastery, older women reported fewer symptoms.

The relationship between complexion and discrimination was not statistically significant, indicating that complexion is not a source of differential exposure to unfair treatment among African American women. This finding is rather puzzling in view of the preponderance of literature suggesting that the color hierarchy disadvantages darker African American women. There are several possible explanations for why discrimination does not vary by complexion. First, 88 percent of respondents reported some unfair treatment, regardless of skin tone. Discrimination on the basis of race alone and in combination with complexion may be operating to varying degrees depending on the racial composition of settings where discrimination is experienced. Thus, hypothetically a lighter skinned woman may experience higher levels of racial discrimination in an all white setting where she is the only Black person, and a darker skinned woman may experience skin tone discrimination in a setting where there are other African American women of all hues. Both might report similar levels of discrimination, but the sources or types (color versus raced-based) might be different. Second, it is possible that skin tone variations are not as relevant as they once were. Support for this supposition comes from Gullickson (2005) who found that the light complexion advantage in status attainment was no longer relevant for cohorts born after 1953, although complexion remained important for mate selection. He speculated that the significance of skin tone for achievement declined owing to improved educational and occupational opportunities for African Americans generally and because integration brought African Americans into contact with White gatekeepers for whom skin tone was less salient. Glullickson’s explanation is consistent with Hill’s (2002b) findings that Whites make fewer distinctions than Blacks when asked to judge black pigmentation. The NSAL did not collect information on race or other characteristics of perpetrators or the context in which unfair treatment was experienced, so we are unable to pursue this line of reasoning.

This study has several limitations. As with any cross-sectional analyses, we cannot say emphatically that discrimination is causally connected to depressive symptoms. It is possible that women with psychological difficulties are more likely to interpret unpleasant encounters as being discriminatory. However, the Schulz et al. (2006a) finding of a longitudinal relationship between discrimination and symptoms argues strongly that this is not the case. This study employed a general measure of unfair treatment and did not formally evaluate attributions of discrimination. Consequently, responses reflect experiences across a variety of ascribed and achieved statuses including race, gender, age, and body image. Such distinctions have important implications for research on discrimination and mental health. For example, Moradi and Subich (2003) found that sexist events, more than racist events, were predictive of African American women’s psychological distress, although both were highly correlated. As Moradi and Subich (2003) concluded, it is likely that these two identities are fused for African American women, and more research is needed on how discrimination based on intersecting statuses operate to undermine African American women’s emotional well-being. Our analyses were also limited to everyday-discrimination and did not explore the effects of major lifetime episodes of discrimination such as being passed over for a job. It is likely that the relationships reported here will differ for major events and should be explored in future research.

There remains much that we do not know about the many ways that discrimination compromises African American women’s mental health. Mastery is a critical mediating link between discrimination and psychological health, but it may also act as a moderator. Future research should investigate the stress buffering effects of mastery as well as other resources that may condition the impact of discrimination on African American women’s well-being, including social support, spirituality, and active involvement in organizations. Qualitative studies have documented that African American women use a variety of strategies to cope with discrimination and other stressors (Feagin and Sikes 1994), but more research is needed to determine which strategies are most successful and in which situations. For example, Jones and Shorter-Gooden (2003) found that African American women cope with racism by “shifting” their presentation of self to counter negative stereotypes; an interesting finding that should be pursued further together with long-established coping resources such as spirituality (Chatters et al 2008) and social support (Chatters et al 2002).

It will be important to investigate more thoroughly the relationship between discrimination and mental health in various settings including work (see Pavalko et al 2003), school, and neighborhoods. Work by Schulz and colleagues (2006a; 2006b) in Detroit and Cutrona et al. (2000) in Iowa and Georgia are excellent models for understanding how social environments, particularly the neighborhood context, impinge upon emotional well-being. Although we found no relationship between complexion and discrimination, this line of research should not be abandoned until we have examined issues such as race of the perpetrator, the setting in which unfair treatment occurs, and whether or not there are African American women of varying complexions in those settings. Without controls for the racial composition of the context, we cannot determine if African American women are being treated unfairly due to skin tone, to race alone, or to a combination of the two.

Despite the need for further research, this study makes several contributions to the race, gender, stress, and mental health literature. First, this study uses nationally representative data to confirm findings from smaller studies reporting that discrimination is a major threat to African American women’s emotional well-being. Second, it uses structural equation modeling to identify an important mechanism, the erosion of mastery, which mediates the relationship between discrimination and depressive symptoms. Third, we identify several high risk groups. Discrimination poses the greatest psychological risk to younger and less educated women. We conclude that skin tone does not matter, but mastery does.