A comparison of skin tone discrimination among African American men: 1995 and 2003

Abstract

This study investigated perceptions of skin tone discrimination among adult African American men. Research suggests that through negative African American stereotypes, out-group members (Whites) perceive light-skinned African Americans favorably and dark-skinned African Americans unfavorably. However, it is unclear how treatment by in-group members (other African Americans) uniquely affects men. Using data from the 1995 Detroit Area Study and the 2003 National Survey of American Life, we investigated these relationships among African American men representing a wide range of socioeconomic groups. We found that African American men’s perceptions of out-group and in-group treatment, respectively, were similar across time. Light-skinned men perceived the least out-group discrimination while dark-skinned men perceived the most out-group discrimination. In appraisals of skin tone discrimination from in-group members, medium-skinned men perceived the least discrimination while both light- and dark-skinned men perceived more in-group discrimination. Additionally, men of lower social economic groups were more affected by skin tone bias than others. Future research should explore the influence of these out- and in-group experiences of skin tone discrimination on social and psychological functioning of African American men.

Among African Americans, skin tone is an important physical characteristic that creates divisions in the community and affects quality of life. Like gender, a person’s skin tone is a visible physical trait that others immediately notice during social interactions and use to form judgments (Maddox & Gray, 2002). Throughout American history, variations in skin tone have contributed to socioeconomic status (SES) differences among African Americans (Herring, Keith, & Horton, 2004; Keith & Herring, 1991). Furthermore, African Americans use skin tone variations to distinguish themselves from each other, reflecting social status hierarchies (Celious & Oyserman, 2001; Hill, 2002b). Given the importance of gender in colorism (e.g. Hunter, 2002), it is possible that African American men could be affected by skin tone bias in ways that are not yet clearly understood. Importantly, reflecting different social prescriptions of masculinity, African American men’s perceptions of skin tone discrimination may vary across social classes. In this paper, we explored this possibility for men in their perceptions of skin tone discrimination from both within and outside of their racial group.

Historical context and empirical findings on skin tone

In American history, slavery constituted a strict caste system that distinguished Black slaves by their skin tones. Lighter-skinned slaves were usually mixed-raced and favored by White slave-owners. These lighter-skinned slaves were frequently fathered by White slave-owners (typically from nonconsensual sexual relations with female slaves) and were, therefore, privileged (Brown, Ward, Lightbourn, & Jackson, 1999; Keith & Herring, 1991); unlike dark slaves, lighter-skinned slaves were spared physically strenuous, outdoor work and instead held domestic indoor jobs like housekeeping in closer contact to Whites. Over time, these privileges in the antebellum period allowed lighter-skinned Blacks to become more educated (Wirth & Goldhamer, 1944) and to own more property (Frazier, 1957). Furthermore, to maintain their elite status and privileges, lighter-skinned men engaged in social practices to exclude darker-skinned Blacks from entering their social circles; these practices included the “Paper Bag Test,” (which banned Blacks from joining fraternities if their skin tones were darker than a brown paper bag), the “Comb test,” (which banned Blacks with coarse, nappy African hair if combs could not glide through it) and the “Blue veins” society (which banned Blacks whose skin tones were too dark to see the blue veins on their arms) (Bond & Cash, 1992). These findings consistently indicated that light skin tone resulted in clear social and economic advantages.

It is not surprising that early research conducted following the emancipation of slaves found that African Americans were influenced by colorism. Later research suggested that African Americans usually preferred skin tones that were similar to their own shade (Hall, 1992) while other studies indicated that they strongly preferred lighter skin tones regardless of their own shade (Porter, 1991). For the most part, psychologically, African Americans who physically appeared close to the average, medium-brown skin tone (i.e. a “prototypical” shade) seemed to be protected in their racial identity and were the least stigmatized by African Americans (Hall, 1992; Holtzman, 1973). In contrast, those with skin tones on the extreme ends of the spectrum (i.e., very dark and very light) experienced a lowered sense of mastery (Holtzman, 1973; Thompson & Keith, 2001) or felt less attached to other African Americans (Celious & Oyserman, 2001). Overall, results of those studies were variable; although light skin tone may have been idealized, medium-skinned persons may have been more protected in their identity and attachment to African Americans. Thus, skin tone bias can be complex—simultaneously serving as advantageous or disadvantageous depending on the social context.

Theoretical framework: African American masculinity and desires to belong

Although past studies have contributed to understanding skin tone influences on African Americans, they have overlooked the unique ways that skin tone may have affected African American masculinity. Culture plays an important role in constructing masculinity expectations (Kimmel & Messner, 1992) and it is these expectations that serve to shape the role of skin tone in African American men’s self-concept. Cultural psychology suggests that the thoughts and perceptions of groups are primarily oriented as either interdependent/collectivistic, emphasizing connectedness and similarity to their group, or independent/individualistic, emphasizing attention to the individual self and uniqueness from others (Markus & Kitayama, 1991).

Some research has suggested that interdependence and collectivism may be an important part of African American culture (e.g. Constantine, Gainor, Ahluwalia, & Berkel, 2003). Indeed this interdependent orientation may extend to African American constructions of masculinity ideals as well. For example, masculinity research suggests that African American manhood is often constructed relationally. Hammond and Mattis (2005) described it as being developed within an interdependent process in connection to others. This paralleled Hunter and Davis’ (1992, 1994) findings, Mattis, Hearn, and Jagers’ (2002) research on communalism, and Wade’s (1998) relational theory of African American masculinity. These relational constructions may extend to men’s need to belong to in-groups.

Tied to this desire to “relate” and feel accepted are two theoretical frameworks that may be relevant to understanding skin tone discrimination: a) the “need to belong,” and, b) masculine discrepancy-strain. The first theory, the “need to belong,” was introduced by Baumeister and Sommer (1997) and suggested that men desire to feel connected to social groups in “… a broad group with multiple people, particularly by competing for a good position in a status hierarchy” thus, men will “…care quite seriously about how strangers perceive them” (Baumeister & Sommer, 1997, p. 39) to gauge their status and belonging in the group. Moreover, research on minority men has suggested that ethnic belonging is a robust predictor of endorsement of traditional masculinity ideology (Abreu, Goodyear, Campos, & Newcomb, 2000). Therefore, minority men may be particularly vulnerable to threats related to belonging.

The second theoretical framework, masculine discrepancy strain, is also tied to men’s desire to feel accepted. Masculinity research has suggested that men hold an idealized view of what it means to be a man, as well as features that they believe are possessed by the respected, “ideal men” (Levant & Pollack, 1995; Liu, Rochlen, & Mohr, 2005; O’Neil, Helms, Gable, David, & Wrightsman, 1986). For example, if a man perceives that he has failed to live up to idealized physical appearance standards, according to Pleck’s (1995) gender role strain paradigm, he may experience discrepancy-strain—a stressor that can negatively affect his self-esteem (O’Neil et al., 1986). Since skin tone biases are tied to cultural physical appearance ideals, it is plausible that discrepancy-strain can be applied to understand it. For example, feeling that one is “too dark” or “too light” to “fit in” may threaten an African American man’s sense of belonging, and, consequently, his self-esteem.

Masculinity and skin tone biases: Complexities with income and self-esteem

The discrepancy-strain suggested by Pleck (1995) may be relevant to the unique role of skin tone in African American masculinity in various ways. Some research suggested that among African American males, dark-skinned men may be idealized as “alpha” males, possessing heightened masculine characteristics, such as dominance (Hall, 1995), strength, virility and confidence in their physical attractiveness (Wade, 1996). Thus, it is possible that African American men may idealize darker skin tone as one indicator of “maleness.”

It has also been found that dark-skinned men have been stereotyped by African Americans and Whites as being “bad boys” and very dangerous (Kahn & Davies, 2011). Dark-skinned men have self-reported more frequent occurrences of racial discrimination than medium- and light-skinned men (Klonoff & Landrine, 2000). Experimental studies have corroborated this finding; dark-skinned men have been the most common targets of racial profiling and police harassment (e.g. Eberhardt, Davies, Purdie-Vaughns, & Johnson, 2006). Once arrested, dark-skinned African American men are ordered to serve longer and harsher prison sentences than their lighter-skinned counterparts for similar crimes (Blair, Judd, & Chapleau, 2004; Gyimah-Brempong & Price, 2006). Thus, all aspects of dark skin tone may not be idealized by African American men because of the associations with poor treatment in society.

Research has also found that darker-skinned African American men may receive less endorsement to advance their careers, earn significantly lower wages than light-skinned men (Goldsmith, Hamilton, & Darity, 2006) and consider their complexion as a barrier to career success (Sánchez, Liu, Leathers, Goins, & Vilain, 2011). For example, during the 2008 presidential election season, one study tracked skin tone perceptions of then-Senator Barack Obama (an African American Democratic candidate) and found that conservatives and persons that did not vote for Obama estimated his skin tone to be significantly darker than his true complexion (Caruso, Mead, & Balcetis, 2009); in contrast, liberals and persons that voted for Obama estimated that he was significantly more light-skinned than his true complexion.

Dark men may also be disadvantaged during job interviews. One study found that employers were especially reluctant to hire dark-skinned African American men over light-skinned men because they stereotyped dark men as being arrogant and dishonest employees who had poor work ethics (Kirschenman & Neckerman, 1991). In a similar vein, Acura, a major car company, recently apologized for language used in a casting call to hire an African American male actor who was “nice looking, friendly…not too dark” for a 2012 Super Bowl commercial (Duke, 2012 ¶5). This information leaked after a dark-skinned African American actor, who was rejected for the job, stated that the company feared that clients would feel threatened by his presence in the commercial. These preferences may block darker men’s access to economic rewards and undermine their ability to fulfill masculine breadwinner and protector roles (e.g. father or husband)—a central aspect of demonstrating traditional masculinity (Doyle, 1983; Thébaud, 2010).

Keith and Herring (1991) found that dark-skinned African American men were the most likely to be employed in blue-collar jobs, usually as laborers (in contrast to light-skinned men who were the most likely to be employed in professional white-collar jobs); this mirrored occupational trends since slavery when light-skinned slaves had less physically strenuous occupations. Moreover, among African Americans, dark-skinned men have been the most likely to be unemployed (Brown et al., 1999) and live in low-income, segregated inner-city neighborhoods (Hochschild & Weaver, 2007). These findings illustrate the SES variations across skin tone that have lingered long after slavery ended.

Some research also suggested that low income environments may place dark-skinned African American men at a psychological disadvantage. Thompson and Keith’s (2001) analysis of the 1980 National Survey of Black Americans (NSBA), the first nationally representative sample of African Americans (Jackson & Gurin, 1987), found that among low-income men, interviewer-rated skin tone was not associated with lowered self-esteem but was associated with lowered self-efficacy. They concluded,

Traditional definitions of masculinity demand men specialize in achievement outside the home, dominate in interpersonal relationships, and remain rational…consistent with gendered characteristics…skin color is important in self-domains that are central to masculinity (i.e., competence) (Thompson & Keith, 2001, p.351).

Although they concluded that self-esteem was unrelated to skin tone for men, Thompson and Keith (2001) could not examine self-reported skin tone relationships; this measure was not included in the 1980 NSBA. The NSBA only included interviewer ratings of skin tone, precluding past researchers from exploring the possibility that discrepancies in self- and interviewer-ratings may be linked to self-esteem, as Pleck’s (1995) discrepancy-strain predicts.

Furthermore, the relationships of skin tone with income illustrate the importance of social context in shaping perceptions of African Americans. Some research has shown that skin tone can be especially consequential in low income environments because those settings are the most racially segregated (Massey, 2004). Skin tone stigma may be more prevalent in intra-racial contexts (i.e., all Black) rather than in inter-racial (e.g., Black-and-White) contexts because African Americans are particularly motivated to distinguish themselves from each other in the absence of Whites (Harvey, LaBeach, Pridgen, & Gocial, 2005). Succinctly, African Americans distinguish themselves from Whites in inter-racial settings and distinguish themselves from Blacks (using skin tone) in intra-racial settings.

Additionally, in these low income intra-racial contexts, light-skinned African American males may also be at a psychological disadvantage. Males may be especially concerned about their skin tones in these racially homogenous settings because it can serve as an indicator of “belonging” to their racial group. One study of low income African American adolescent boys found that being light-skinned was a risk factor for perceiving academic success as a non-masculine aspiration (Oyserman, Brickman, Bybee, & Celious, 2006). Culturally, since African American masculinity has a collectivistic construction, light-skinned boys may have felt threatened, and, thus, motivated to overcompensate for their feelings of physical dissimilarity from their racial in-group. They may have attempted to “prove” their “maleness” through negative behaviors (e.g. receiving poor grades) that could make them appear more similar to their under-achieving African American male peers.

Self-rated versus interviewer-rated skin tone: Discrepancies and psychological implications

This struggle over real versus ideal skin tone has been examined in children and adolescents (Erkut, Fields, Coll, Szalacha, & Alarcon, 2000). Empirical studies on skin tone usually include only one of the following measures: interviewer-rated skin tone, reflectance meter recordings of skin tone, or self-rated skin tone. Occasionally, some studies include two of these measures and examine them for corroboration. One might intuitively suspect that misalignment (discrepancy) of self- and interviewer-ratings could possibly be linked to insecurities and self-esteem. Skin tone research on Latino children (males and females) found that discrepancies, where the children perceived themselves as darker than the Latino interviewer rated them, was a predictor of poor self-esteem (Erkut et al., 2000). However, research on adult minority groups (such as African American men) has not carefully investigated the possibility that misalignment of skin tone ratings may manifest low self-esteem or threats of not feeling masculine enough (Pleck, 1995).

Hypotheses

In sum, research on inter-racial (i.e. White-and-Black) social interactions consistently indicated that dark-skinned African American men have been treated poorly by out-groups (namely, Whites) while their light-skinned counterparts have been treated favorably by out-groups historically (Klonoff & Landrine, 2000). Literature on intra-racial (i.e. Black-and-Black) relationships also suggested that both dark-skinned males (e.g. Kahn & Davies, 2011) as well as light-skinned males (e.g. Oyserman et al., 2006) have been stereotyped negatively by in-group members, particularly in low SES contexts; in contrast, medium-skinned men may be protected from in-group discrimination Although previous research revealed that interviewer-rated skin tone was not associated with self-esteem for African American men (Thompson & Keith, 2001), discrepancies in self-rated versus interviewer-rated skin tone were not examined and could be related to feelings of strain (Pleck, 1995) and self-esteem (Erkut et al., 2000).

We explore these relationships among African American men across two cross-sectional datasets—an older, regional sample from Detroit and a more recent, nationally representative sample. Our objective in using two samples is twofold: First, we expect to find that general patterns in African American men’s responses to skin tone discrimination will replicate across time because colorism has been deeply ingrained in the culture; second, since research indicates that skin tone bias is more prevalent in intra-racial, low SES contexts, we expect men’s responses in the Detroit sample to be more pronounced because Detroit reflects a decidedly impoverished and racially segregated environment (Eisinger, 2003).

First, we hypothesize that African American men of dark-, medium-, and light-skin tone will report skin tone discrimination from their in-group (the “in-group appraisal”) and their out-groups (the “out-group appraisal”) in consistent patterns at two cross-sectional time points (i.e., 1995 and 2003); this prediction is informed by the aforementioned literature and the ingrained nature of colorism culturally. Moreover, these appraisals should be robust across time because the social factors related to men’s “need to belong” (Baumeister & Sommer, 1997) should motivate them to be aware of their position in largely unchanged status hierarchies. Second, when self and interviewer skin tone reports differ, we hypothesize that men will self-rate closer to the skin tone that reported the least discrimination in the in-group appraisal (i.e. medium brown). This is informed by findings on the collectivistic nature of African American masculinity, suggesting that they will desire to appear more similar to the most respected skin tone group. Third, we hypothesize that discrepancies in self and interviewer ratings will be associated with lower self-esteem, as Pleck’s (1995) discrepancy-strain construct predicts.

We hypothesize the following associations for African American men:

• H₁In the out-group appraisal, light-skinned men will report the least discrimination. In the in-group appraisal, medium-skinned men will report the least discrimination.
• H_2aIn the comparison of interviewer-rated and self-rated skin tone, when skin tone ratings misalign, men will most often self-rate closer to a medium-brown—the group that will report the least discrimination in the in-group appraisal.
• H_2bSince skin tone bias is most salient and impactful in intra-racial settings (Harvey et al., 2005), there will be more discrepancies in skin tone ratings among men of lower SES since they are more likely to reside in intra-racial settings.
• H₃Consistent with past research, men’s interviewer-rated skin tone alone will not be associated with self-esteem (Thompson & Keith, 2001). But men’s skin tone misalignment (self-rating with a darker or lighter bias than the interviewer) will indicate discrepancy-strain (Pleck, 1995) and associate with lower self-esteem.

Method

Participants and design

Two data sources were used in this paper: The 1995 Detroit Area Study—Social Influence on Health (1995 DAS) and the 2001–2003 National Survey of American Life: Coping with Stress in the 21^st Century (2003 NSAL). Both of these surveys were conducted through the University of Michigan’s Institute for Social Research and include similar measures on SES, discrimination, psychosocial influences on well-being, and skin tone.

Specifically, the DAS is a multi-stage, area probability sample that represented the adult population in three Detroit-area counties. The complete 1995 DAS survey sample included 1,139 adults with 586 African Americans (overall response rate 70%). Most interviewers who surveyed the African American sample (90.2%) were Black (i.e. race-matched). These trained interviewers from the Survey Research Center of the University of Michigan completed the fieldwork in respondents’ homes in 1995. In this paper, analyses were limited to race-matched African American men of ages 18–95 years (M_age=41.55, SD=16.65) with complete data on covariates (173 un-weighted). For additional information, see Jackson and Williams (2002) or the study website: http://dx.doi.org/10.3886/ICPSR03272.

The 2001–2003 NSAL (commonly referred to as the 2003 NSAL) is the most comprehensive and most recent nationally representative household sample of 5,191 non-institutionalized Black Americans, African Americans, and Black Caribbean’s in the United States. The response rate for NSAL African Americans was 70.7%. The full NSAL sample includes 1,217 African American men of which 85.8% were race-matched to Black interviewers. Because of the importance of culture, in our final analysis sample, we excluded Black men that were not African Americans (i.e. Black Caribbean’s, West Indians, Africans, and Blacks of other cultural heritages). Analyses in the NSAL were also restricted to race-matched African American men (M_age=41.72, SE=.72, range of 18–91) with complete data on interviewer- and self-rated skin tone and covariates (N=944 un-weighted in final sample). For more information, see http://www.rcgd.isr.umich.edu/prba/nsal (Jackson, Neighbors, Nesse, Trierweiler, & Torres, 2004).

The DAS was a multi-stage area design based upon the demographics of the greater Detroit area in 1995. Data were statistically weighted to account for the different probabilities of selection and analyses were conducted using IBM SPSS v19.0 (SPSS, 2010). Because of the complex sample design of the NSAL, analyses were conducted using the svy commands of STATA 12.0 (StataCorp., 2011) to properly handle the clustering, stratification, and weights.

Measures

Interviewer-rated skin tone

In the DAS and NSAL, interviewers privately rated respondents’ skin tone at the end of the interview. Interviewer-rated skin tone was used for the main appraisal analyses, consistent with previous skin tone studies (e.g. Hughes & Hertel, 1990; Keith & Herring, 1991; Thompson & Keith, 2001) that attempted to simulate community judgments. In the DAS, trained interviewers were members of the tri-county Detroit community, and, for the most part, trained interviewers in the NSAL were also drawn from local communities.

One question in the DAS was used to capture how light or dark the interviewer assessed the respondent: “R’s skin color is – 1. Very dark brown (3.6%); 2. Dark brown (29.3%); 3. Medium brown (47.3%); 4. Light brown (15.7%); 5. Very light brown (4.1%).” The responses for this ordinal measure were re-coded into a three-level ordinal measure (Dark, Medium, Light); “Very dark” and “dark” collapsed to “Dark,” medium remained the same, and “light” and “very light” collapsed to “Light.” This three-category measurement scheme is consistent with previous skin tone studies (e.g. Bond & Cash, 1992; Hughes & Hertel, 1990; Keith & Herring, 1991) since ratings for skin tone in the United States historically peak at descriptions of “dark,” “medium,” and “light,” perhaps because African Americans have generally categorized each other in those ways. In the NSAL, a similar question was available on a 7-point scale: “1. Very dark (3.8%); 2. Dark (20.1%); 3. Somewhat dark (18.3%); 4. Medium (41.0%); 5. Somewhat light (9.9%); 6. Light (5.5%); 7. Very light (1.4%).” This measure was re-coded to “dark,” “medium,” and “light” by collapsing the darkest three categories as “dark” and the lightest three categories as “light” and leaving “medium” the same.

Self-rated skin tone

Additionally, in both the DAS and NSAL, self-rated skin tone was collected using a 5-point scale, “1. Very dark brown; 2. Dark brown; 3. Medium brown; 4. Light brown; 5. Very light brown.” The percentages in the DAS from “very dark brown” to “very light brown” were 5.1%, 26.7%, 44.0%, 20.2%, and 4.0%, respectively, while the NSAL’s were 7.5%, 30.6%, 45.4%, 13.4%, 3.2%, respectively. Unlike the DAS, the NSAL had two different scales for ratings of skin tone (one 7-point category for interviewers and one 5-point category for respondents). Self-rated skin tone has generally not been used in this type of research because interviewer judgments of skin tone can be thought of as more “objective” assessments representing community standards of variations in skin tones (Hill, 2002a).

Discrepancy in skin tone reports

Standardized scores (z-scores) were computed for self- and interviewer-rated skin tone, respectively. DAS z-scores of the self-rated (z_self) and interviewer-rated (z_iwr) skin tones were as follows: 1. Very dark brown (z_self = −2.13, z_iwr= −2.24); 2. Dark brown (z_self = −1.02, z_iwr= −1.06); 3. Medium brown (z_self =.09, z_iwr=.11); 4. Light brown (z_self =1.20, z_iwr=1.29); and 5. Very light brown (z_self =2.31, z_iwr=2.47). NSAL self-rated skin tones (the same 5-point scale) z-scores were as follows: z = −1.94, z = −.82, z =.30, z =1.42, z =2.53, respectively. NSAL interviewer-rated skin tone (7-point scale) z-scores were as follows: 1. Very dark (z = −2.03), Dark (z = −1.24), Somewhat dark (z = −.44), Medium (z = .36), Somewhat light (z = 1.16), Light (z = 1.95), Very light (z = 2.75). Discrepancies were calculated by subtracting the z-scores of interviewer ratings from the z-scores of self-ratings for each respondent; these values were then squared, eliminating negative values. In the DAS and NSAL, mean values of these squared standardized units were as follows: M_das= .68, SD_das= 1.18 (range 0.00–11.73) and M_nsal= .59, SD_nsal=1.19 (range 0.01 – 14.21).

In-group and out-group appraisals (perceived skin-tone based discrimination)

In the DAS, two questions assessed how respondents appraised their skin tone as either an advantage, a disadvantage, or irrelevant during interactions with African Americans/Blacks (in-group) as well as Whites (out-group). The out-group appraisal read: “Because of the shade of your skin tone do you think White people treat you: 1. A lot better; 2. Somewhat better; 3. No different; 4. Somewhat worse; or 5. A lot worse than other Blacks?” Similarly, they were also asked to appraise their perception of in-group (other African Americans/Blacks) treatment based on skin tone, using an identically-phrased question replacing “White” with “Black.” In the NSAL, the out-group appraisal was worded differently: “How often would you say that Whites treat you badly because of the shade of your skin color? 1. Very often; 2. Fairly often; 3. Not too often; 4. Hardly ever; 5. Never.” An identical question was asked for their in-group appraisal, replacing “Whites” with “Blacks.” These NSAL appraisals were reverse coded to be consistent with the DAS (higher scores indicate more discrimination).

Self-esteem

Since an important consequence of skin tone bias is its influence on self-evaluations (Thompson & Keith, 2001), self-esteem is investigated. In both samples, the self-esteem measure was taken from items in Rosenberg’s (1965) self-evaluative instrument on a Likert-type scale ranging from 1 (strongly agree) to 4 (strongly disagree). Only 4 items were available for the self-esteem measure for DAS African American men (M=3.79, SD=.37): “I feel that I am a person of worth, at least on an equal plane with others”; “All in all, I am inclined to feel that I am a failure”; “I am able to do things as well as most other people”; “I feel I do not have much to be proud of.” Internal consistency estimates for DAS men in the final sample varied by income: α= .48 overall, α=.64 for men earning less than $20,000. In the NSAL, all 10 items of the scale were available (M=3.62, SE=.02) (α= .77 for NSAL men in this final sample).

Respondent characteristics

Age, education, and income were collected. In both the DAS and NSAL, date of birth was used to calculate age, while highest education completed was captured as “1. Kindergarten-11^th grade; 2. High school/GED graduate; 3. Some college; 4. Bachelor’s degree or higher.” In the DAS, family income was categorized as “1. Under $10,000; 2. $10,000–$19,999; 3. $20,000–$39,999; 4. $40,000–$59,999; 5. $60,000 or higher.” Family income in the NSAL was categorized as “1. Less than $16,000; 2. $16,000–$24,999; 3. $25,000–$34,999; 4. $35,000–$49,999; 5. $50,000–$74,999; 6. $75,000–$99,999; 7. $100,000 or more.”

Socioeconomic status differences in samples

A premise of some arguments in this paper is that the DAS sample is proportionally more impoverished than the NSAL. Income differences reflect this—the middle 50–60% of this NSAL sample earned between $25,000 and $50,000 while DAS men, on average, earned around $20,000–40,000. NSAL men were more educated—nearly 20% attended college in contrast to 8% of DAS men. Similarly, nearly 30% of DAS men did not complete high school in contrast to 19% of NSAL men.

Interviewer characteristics

Gender (male or female), age, and education level were collected on all interviewers. In both the DAS and NSAL, educational attainment for interviewers was collected on a consistent 7-point scale (1=1^st–8^th grade, 2=some high school, 3=high school grad, 4=some college, 5=college grad, 6=Master’s degree or equivalent, 7=PhD grad). Race was assessed and used to remove non-race matched pairs.

Results

H₁: In the out-group appraisal, light-skinned men will report the least discrimination; in the in-group appraisal, medium-skinned men will report the least discrimination

These hypothesized relationships are investigated before and after controlling for respondent characteristics (age, education, income) and interviewer characteristics (age, education, gender). Analysis of variance (ANOVA) and covariance (ANCOVA) were used in DAS analyses while bivariate and multiple regressions were used in NSAL analyses in order to properly handle the complex sample design of the NSAL in STATA software. In the DAS, African American men’s out-group appraisal of skin tone discrimination was significantly related to skin-tone, ANOVA F(2, 225) =24.74, p<.0005, η_p²= .18 and ANCOVA F(2, 219) =30.78, p<.0005, η_p²= .22. Dark-skinned men (M=3.29, SD=.49) reported the most out-group skin tone discrimination, followed by medium-skinned men (M=3.08, SD=.64), and light-skinned men (M=2.51, SD=.63) reported the least (see Figure 1). All DAS light-skinned men reported either neutral or favorable treatment by Whites (none reported worse treatment). Fisher’s Least Significant Difference (LSD) post hoc tests indicated that light-skinned men’s reports significantly differed from medium-skinned (p<.000) and dark-skinned men’s (p<.000). Medium-skinned men’s reports differed from dark-skinned men’s reports (p=.022).

Figure 1.

Out-group appraisal of African American men in the DAS and NSAL (centered). In general, scores above 0 indicate perceived discrimination while scores below 0 indicate perceived favorable treatment.

In-group appraisals in the DAS were also statistically significant before and after controlling for the same covariates, ANOVA F(2, 225) =8.63, p<.000, η_p²= .07 and ANCOVA, F(2, 219) =4.33, p=.014, η_p²= .04. Dark- (M=3.16, SD=.62) and light-skinned men (M=3.11, SD=.49) reported the most in-group skin tone discrimination, while medium-skinned men reported the least (M=2.86, SD=.44). An LSD post-hoc analysis suggested that medium-skinned men’s reports were significantly lower than dark- (p<.000) and light-skinned men (p=.007); however, dark- and light-skinned men did not differ significantly (see Figure 2).

Figure 2.

In-group appraisal of African American men in DAS and NSAL (centered). Scores in out-group and in-group appraisals were centered at the respective means for the DAS and NSAL samples (0 = mean score for the respective sample). In general, scores above 0 indicate perceived discrimination while scores below 0 indicate perceived favorable treatment.

Next, we examined the NSAL sample. The out-group appraisal of skin tone discrimination for African American men in the NSAL followed a similar pattern as the DAS sample. Light-skinned men reported less discrimination than dark-skinned (b=.40, SE_B=.13, p=.003) and medium-skinned men (b=.28, SE_B=.11, p=.019). The association was statistically significant, F(2, 914.23) = 4.96, p =.012, R² = .012. Mean reports for each group were as follows: dark-skinned (M=2.93, SE=.06), medium-skinned (M=2.81, SE=.08) and light-skinned men (M=2.53, SE=.11). After adjusting for covariates, the results were similar as dark- (b=.39, SE_B = .14, p=.009) and medium-skinned men (b=.30, SE_B=.12, p=.020) reported worse treatment than light-skinned men, F(8, 893.35) = 2.50, p=.031, R² = .025; however, there were no differences between dark- and medium-skinned men (b=.09, SE_B=.06, p=.17) (see Figure 1).

In-group appraisals of skin tone discrimination of NSAL African American men followed the same pattern as the DAS, although the association did not reach statistical significance: dark- (M=2.30, SE =.06), medium- (M=2.09, SE =.07), and light-skinned men (M=2.20, SE =.08). Medium-skinned men reported less in-group skin tone discrimination than dark-skinned men (b=.21, SE_B=.09, p=.022) and slightly less than light-skinned men, although it was not significant (b=.12, SE_B=.09, p =.231); the overall association was marginally significant, F(2, 914.23) = 2.86, p=.069, R² = .01. After adjusting for covariates, medium-skinned men still reported significantly less in-group skin tone discrimination than dark-skinned men (b=.22, SE_B=.08, p=.014), but not significantly less than light-skinned men (b=.02, SE_B=.10, p=.811), R² = .05. Across both the DAS and NSAL, the pattern of out-group appraisals suggested that light-skinned men were advantaged (see Figure 1) while results of in-group appraisals suggested that medium-skinned men were advantaged (see Figure 2).

H_2a: When self- and interviewer-skin tone ratings misalign, men will most often self-rate closer to a medium-brown skin tone

As shown in Table 3, interviewer-rated versus self-rated skin tone misalignment (using the 5-category measure) followed a distinct pattern in the DAS. As interviewer-assessed skin tones ratings became lighter, there was greater alignment with self-rated skin tone until a perfect 100% alignment among the “very light” men; specifically, the concordance of ratings across the five skin tones from darkest to lightest were 12.5%, 46.5%, 66.1%, 71.1%, 100% respectively. As predicted, in the DAS, whenever self- and interviewer- skin tone ratings differed, the most typical misaligned shade was located one shade closer to a medium-brown skin tone. Specifically, 87.5% of “very dark,” 29.6% of “dark,” and 23.7% of “light” men self-rated one shade closer to “medium-brown” (see Table 3). Since the NSAL’s skin tone ratings were on different scales, consistent descriptive results are not available without compromising (and potentially biasing) the NSAL sample.

H_2b: There will be more discrepancies in self versus interviewer skin tone ratings among men of lower social classes (using education level as a proxy for social economic status)

Table 3.

Overlap of interviewer-rated versus respondents’ self-rated skin tone—1995 Detroit Area Study African American men

	Very dark		Dark		Medium		Light		Very light		Total
Very dark	1		10		0		1		0		12
		12.5%		14.1%		0.0%		2.6%		0.0%		5.0%
Dark	7		33		23		1		0		64
		87.5%		46.5%		20.0%		2.6%		0.0%		26.4%
Medium	0		21		76		9		0		106
		0.0%		29.6%		66.1%		23.7%		0.0%		43.8%
Light	0		7		16		27		0		50
		0.0%		9.9%		13.9%		71.1%		0.0%		20.7%
Very light	0		0		0		0		10		10
		0.0%		0.0%		0.0%		0.0%		100.0%		4.1%
Total	8		71		115		38		10		242
		100.0%		100.0%	100.0%		100.0%		100.0%			100.0%

Note. The horizontal line is interviewer-rated skin tone and vertical line is self-rated skin tone (counts are statistically weighted).

In both the DAS and NSAL, there was less overall alignment in self- and interviewer-rated skin tone ratings among the less educated men (see Figure 3). Mean standardized discrepancies for the samples were .68 (SD=1.18) in the DAS and .59 (SD=1.19) in the NSAL. Across the four education levels, the mean standardized discrepancies were as follows: 1. Did not complete high school (M_das= .99, SD_das= 1.75; M_nsal= .79, SE_nsal=.08); 2. completed high school/GED (M_das= .69, SD_das= .88; M_nsal= .59, SE_nsal=.06); 3. attended some college (M_das= .40, SD_das= .61; M_nsal= .60, SE_nsal=.09); and 4. graduated with a Bachelor’s or higher (M_das= .33, SD_das= .60; M_nsal= .40, SE_nsal=.07). Using bivariate regression, these results were statistically significant in the DAS (b= −.21, SE_B=.08, p =.014), F(1, 219) = 6.10, p=.014, R² = .03 as well as NSAL (b= −.11, SE_B=.01, p =.012), design-based F(2, 914.23) = 83.26, p=.012, R² = .01.

H₃: Interviewer-rated skin tone will not be associated with self-esteem but standardized discrepancies in self- and interviewer-ratings will be associated with self-esteem

Figure 3.

Mean standardized discrepancies of interviewer vs. respondent skin tone ratings across education level for African American men. The standardized discrepancies were determined by subtracting the z-score of the interviewer’s skin tone rating from the z-score of the self-rated skin tone; this value was then squared. Larger discrepancies indicate greater misalignment in self versus interviewer judgments of skin tone.

ANOVA and ANCOVA were used for the DAS sample while bivariate and multiple regressions were used for the NSAL, controlling for respondent age and education as well as interviewers’ gender, age, and education. As predicted, interviewers’ judgment of DAS African American men’s skin tone was not significantly associated with self-esteem for dark- (M=3.75, SD=.36), medium (M=3.82, SD=.39), and light-skinned (M=3.81, SD=.35) men before adjusting for covariates, ANOVA, F(2,225) =0.81, ns, and after adjusting, ANCOVA, F(2, 219) =1.82, ns. Among NSAL African American men, self-esteem was also not associated with skin tone (interviewer-reported) for dark (M=3.61, SE=.03) medium (M=3.61, SE =.02) and light-skinned men (M=3.66, SE =.04). The overall relationship was not statistically significant, F(2, 883.84) = 0.93, p = .403 R² = .002. Adjusting for covariates improved the model (R² = .09) but men’s skin tone was still not significantly associated with self-esteem.

Although interviewer-rated skin tone alone did not predict self-esteem in the DAS, the standardized discrepancy measures of self- versus interviewer-rated skin tone (M=.68, SD=1.18) and respondents’ income (M=2.98, SD=1.35) (two between-subject variables) predicted self-esteem (M=3.79, SD=.37) (dependent variable). Using multiple regression analyses in the DAS, a significant model emerged: F(2, 218)=7.90, p<.000, R² =.07. The standardized discrepancy measure was a significant negative predictor of self-esteem (b= −.07, SE_B= .021, p=.001) while income was a significant positive predictor (b= .037, SE_B=.018, p=.040) of self-esteem. This suggests that larger discrepancies in skin tone ratings were associated with lower self-esteem in the DAS, supporting the hypothesis. Adjusting for covariates improved the model (R² = .14).

Among African American men in the NSAL, a similar model emerged using multiple regression: design-based F(2, 883.84) = 1.49, p<.00), R² =.05. The standardized discrepancy measure of self- versus interviewer-skin tone ratings (M=.59, SE=.00) and income (M=3.33, SE=.05) predicted self-esteem (M=3.62, SE=.00). Consistent with DAS results, the NSAL standardized discrepancy measure was a significant negative predictor of self-esteem (b= −.023, SE_B=.005, p=.041) while income was a significant positive predictor (b= .053, SE_B=.012, p=.045), indicating that larger discrepancies were associated with lower self-esteem reports in the NSAL, supporting the hypothesis. Adjusting for covariates improved the model (R² = .08).

Discussion

This paper examined African American adult men in one regional sample and one nationally-representative sample about eight years later. Largely similar patterns were found in both samples in men’s reactions to skin tone discrimination. Men’s appraisals of negative treatment from in-group and out-group members persisted across both studies. Several key themes were replicated in results from both datasets related to in-group and out-group appraisals and the implications of discrepancies in self versus interviewer-rated skin tone.

First, results suggest that skin tone matters in contexts outside of their racial group. As predicted, light-skinned men consistently perceived the best treatment from Whites, while dark-skinned men consistently perceived the worst treatment, supporting the first hypothesis. These responses are consistent with the ways that Whites have treated lighter and darker African Americans since slavery. This also has contemporary implications for masculinity on various levels in inter-racial settings; economically, if dark-skinned men are less likely to be promoted at work, over time, they may feel emasculated and incapable of fulfilling masculine “provider” roles. Psychologically, dark-skinned men’s self-efficacy could suffer if they appraise that their hard work will not “pay off” similarly to their light-skinned counterparts. Additionally, there could be health consequences. Kahn’s (2010) experimental research suggested that dark-skinned African Americans are more vulnerable to “stereotype threat” (Steele & Aronson, 1995) than light-skinned counterparts; stereotype threat was shown to increase arterial high blood pressure (Blascovich, Spencer, Quinn, & Steele, 2001) and may account for the higher prevalence of hypertension among dark-skinned African American men (Sweet, McDade, Kiefe, & Liu, 2007).

Second, results indicate that skin tone matters within the racial group. In the in-group appraisals, it was hypothesized that medium-skinned men would report the least discrimination; this was supported in both datasets. The in-group appraisal finding has important implications for African American men who primarily reside in intra-racial settings, such as low SES neighborhoods. Although the psychological impact of racial discrimination from Whites against Blacks has been long researched, among lower SES African American men who have very little contact with Whites in their day-to-day lives, skin tone discrimination from fellow Blacks may be a more relevant source of discrimination and status threats. Medium-skinned men in these contexts may be advantaged over others.

Additionally, the results of both the out-group and in-group appraisals illustrate the complexity of skin tone bias that was alluded to earlier in this paper. As Celious and Oyserman (2001) indicated, a skin tone that is advantageous in one context could be disadvantageous in another. For example, being a light-skinned African American man can be advantageous when interacting with Whites, but disadvantageous when interacting with Blacks (i.e. a win-lose situation) if Blacks perceive him as being “racially impure” or an Uncle Tom (e.g. Herring et al., 2004; Hochschild & Weaver, 2007). Similarly, medium-skinned men may perceive substantial discrimination from Whites but could be protected in interactions with Blacks (i.e. lose-win situation). It is particularly striking that dark-skinned men were the only group that consistently appraised the worst treatment in both appraisals (i.e., a lose-lose situation); this high level of racial discrimination could be detrimental to their mental health as well as increase their endorsement of traditional male gender norms and masculine-typed behaviors, according to a recent experimental study (Goff, Di Leone, & Kahn, 2012). The results suggest that none of the skin tone groups were consistent “winners” in both out-group and in-group appraisals, illustrating that African American men of virtually all skin tones can be vulnerable to perceiving stigma in at least one racial context.

Third, discrepancy in self and interviewer skin tone reports and its variation across social economic groups illustrate how colorism may influence how they perceive themselves. Since medium-skinned men reported the least in-group skin tone discrimination, it was hypothesized that African American men may have internalized that complexion as “ideal.” Although similar descriptive results could not be computed using NSAL skin ratings that were on different scales, results from the DAS sample supported the second hypothesis and theoretical notion of brown skin as idealized. As described earlier, African American masculinity has been shown to be interdependent in its construction; thus, it is logical that interdependently-oriented men may desire to psychologically feel more similar to in-group members by self-rating (perceiving themselves) as more close to a respected, prototypical medium-brown skin tone. Furthermore, since standardized skin tone discrepancy scores across both datasets indicated that the less educated men had the largest discrepancies in ratings, the second hypothesis was supported; it may indicate that the climate of their social contexts may be intricately linked to how salient skin tone discrimination is in their lives.

Fourth, the results may illustrate that those discrepancies may be linked to self-esteem in certain SES levels. As discussed previously, African Americans of low SES are surrounded almost entirely by other low SES African Americans (Massey, 2004) and this is relevant since skin tone bias may be more salient in intra-racial contexts (Harvey et al., 2005). Moreover, the 1995 DAS sample was based in Detroit, one of the most low-resourced cities in the United States; approximately 85% of Detroit residents were Black in the 1990s (i.e. an intra-racial context) (Eisinger, 2003). For these reasons, it was expected that men’s reactions in the DAS sample would be more robust than in the nationally representative NSAL sample, and this was supported by results of these analyses.

The results of this paper extend the literature from the early NSBA sample (Thompson & Keith, 2001) and an earlier analysis of this DAS sample (Brown, 1998) which both found that interviewer-rated skin tone was not associated with African American men’s self-esteem. Those analyses, however, did not consider the possibility that income and discrepancies in skin tone ratings may be connected to self-esteem. Results in this current paper suggest that future studies should explore the possibility that skin tone bias may operate in a more complex manner for African American men than previously thought, challenging past studies that concluded that men’s skin tone was not associated with psychological outcomes (e.g. Keith & Herring, 1991).

It is important to consider the various mechanisms that could underlie these associations. Among African American men of lower social economic status groups, Pleck’s (1995) discrepancy-strain framework may account for some of the relationships of skin tone with self-esteem. This framework predicts that perceiving that one has failed to live up to an ideal “image” results in lowered self-esteem in men. Although these analyses are correlational, it is plausible that skin tone discrepancies may manifest dissatisfaction with one’s complexion (i.e. skin tone “discrepancy strain”) or stress related to not feeling that one “fits in” or “belongs” (Baumeister & Sommer, 1997) to their racial in-group; these outcomes could ultimately threaten self-esteem.

Future studies could also explore factors related to why low SES African American men may be particularly influenced by skin tone, as illustrated by larger standardized discrepancies. Possible factors could include findings that low SES African American men may identify more strongly with their ethnic groups (Abreu et al., 2000), endorse more traditional masculinity ideologies (Levant, Majors, & Kelley, 1998), and may be less forgiving of racial discrimination (Hammond, Banks, & Mattis, 2006). If they are more ethnically identified and collectivistic than higher SES men, when they perceive intra-racial skin tone discrimination, it is plausible that they could be especially vulnerable to lowered self-esteem and feelings of shame (a typical collectivistic reaction) to evaluate their skin tone in a manner biased towards cultural ideals.

This study has a number of strengths, including the use of both self- and interviewer-rated skin tone and intra- and inter-group appraisals of biased treatment. The use of interviewer-reported skin tone measures may be an important predictor of how other in-group members regard the respondents in daily interactions, which may not be properly captured in either self-ratings alone or reflectance meter recordings. The availability of two forms of skin tone reports was beneficial in examining less direct associations (i.e. self-esteem and discrepant ratings).

Limitations and Concluding Thoughts

However, this investigation of the DAS and NSAL has limitations. The most important may be the low internal consistency estimates for self-esteem in the DAS; thus, the results in the third hypothesis may not be valid reflections of DAS men’s experiences. Additionally, although the interviewers were trained, it is likely that there was human bias in their skin tone ratings (Hill, 2002a). However, this analysis controlled for important interviewer demographic characteristics in an attempt to account for some of this potential bias. Finally, the DAS and NSAL are cross-sectional, precluding any causal interpretations. Results from these two samples suggest the need for future research that reliably measures self-esteem and directly measures endorsement of traditional masculinity ideology. Neither the DAS nor NSAL measured masculinity endorsement; therefore, more research is needed to explore the possibility that masculinity independently (or interactively) influenced reactions to skin tone discrimination.

Since colorism is deeply rooted in African American culture, the differences found in many of the associations in the regional, Detroit sample were similar and generally stronger than those in the more recent nationally representative NSAL sample. Findings in the NSAL may be weaker because skin tone may be becoming less important in life outcomes among younger cohorts of African Americans (Gullickson, 2005). Alternatively, findings in the Detroit sample may be more robust because of Detroit’s unique history of race and segregation (Eisinger, 2003).

Very few social scientists collect data on skin tone and even fewer have attempted to examine its role in masculinity construction. Since skin tone bias and masculinity are both socially constructed, we can expect ideals of both constructs to change over time. Future research should investigate whether African American men’s reactions to skin tone discrimination may be connected to deeper influences of the meanings of manhood. It is also important for researchers to track how these norms evolve as society progresses, and develop more thorough models to understand the heterogeneity of influences on African American masculinity, particularly as they relate to mental health among members of this understudied population.

Table 1.

African American men’s socio-demographic characteristics—1995 Detroit Area Study

	Total	Dark skin	Medium skin	Light skin	p-value
N (% of sample)	243 (100%)	80 (32.9%)	115 (47.3%)	48 (19.7%)
Age, mean (SD)	41.55 (16.7)	43.57 (15.9)	42.55 (17.6)	35.79 (14.5)	.075
Income scaled score, mean (SD)1	2.98 (1.4)	3.48 (1.4)	2.78 (1.3)	2.95 (1.2)	.001b,c
Income, percentages
<$10,000, %	17.5%	12.5%	22.6%	14.6%
$10,000–$19,999, %	16.2%	13.8%	17.4%	16.7%
$20,000–$39,999, %	29.9%	22.5%	31.3%	39.6%
$40,000–$59,999, %	16.8%	15.0%	17.4%	16.7%
$60,000+, %	19.5%	36.3%	11.3%	12.5%
Education scaled score, mean (SD)2	2.16 (.94)	2.09 (.96)	2.18 (.98)	2.25 (.84)	ns
Education, percentages
Did not complete high school, %	29.1%	32.5%	30.4%	20.4%
High school graduate/GED, %	33.7%	36.3%	31.3%	36.7%
Some college, %	28.8%	22.5%	28.7%	38.8%
College graduate, %	8.4%	8.8%	9.6%	4.1%
In-group appraisal, mean (SD)	3.01 (.55)	3.17 (.62)	2.87 (.48)	3.07 (.53)	.001b,c
% appraising worse in-group treatment	11.5%	17.5%	5.2%	16.7%
Out-group appraisal, mean (SD)	3.03 (.66)	3.30 (.49)	3.06 (.65)	2.51 (.63)	<.000a,b,c
% appraising worse out-group treatment	16.7%	28.8%	15.6%	0.0%
Self-esteem,3 mean (SD)	3.79 (.37)	3.75 (.36)	3.83 (.37)	3.79 (.40)	ns

Note. These measures above are statistically weighted.

¹Income is on a five-point scale with higher scores indicating higher household income

²Education is on a four-point scale with higher scores indicating higher education

³Self-esteem is on a four-point scale with higher scores indicating greater self-esteem

^*Superscripts indicate that there were statistically significant differences between the following pairs: a. light-skinned and dark-skinned; b. light-skinned and medium-skinned; c. medium-skinned and dark-skinned

Table 2.

African American men’s socio-demographic characteristics —2001–2003 National Survey of American Life

	Total	Dark skin	Medium skin	Light skin	p-value
N (% of sample)	944	399 (41.7%)	387 (41.8%)	158 (16.5%)
Age, mean (SE)	41.7 (.72)	43.0 (.88)	41.0 (1.1)	40.2 (1.7)	.090
Income scaled score, mean (SE)1	3.33 (.09)	3.26 (.15)	3.32 (.12)	3.53 (.21)	ns
<$16,000, (%)	22.7%	23.0%	24.1%	18.5%
$16,000–$24,999, (%)	11.9%	11.5%	11.7%	13.4%
$25,000–$34,999, (%)	16.7%	21.0%	15.4%	8.8%
$35,000–$49,999, (%)	22.3%	21.8%	21.0%	27.1%
$50,000–$74,999, (%)	16.1%	11.9%	17.4%	23.2%
$75,000–$99,999, (%)	5.8%	5.2%	6.0%	6.5%
$100,000 or more, (%)	4.6%	5.5%	4.4%	2.7%
Education scaled score, mean (SE)2	2.38 (.05)	2.22 (.07)	2.46 (.06)	2.59 (.10)	.000a
Education
Did not complete high school, %	19.0%	23.1%	17.0%	13.8%
High school graduate/GED, %	44.7%	48.3%	43.0%	39.7%
Some college, %	15.8%	12.5%	17.2%	20.7%
College graduate, %	20.5%	16.1%	22.9%	25.8%
In-group appraisal, mean (SE)	2.19 (.04)	2.30 (.06)	2.09 (.07)	2.20 (.08)	ns
% appraising worse in-group treatment	13.6%	17.0%	10.5%	13.2%
Out-group appraisal, mean (SE)	2.81 (.05)	2.93 (.06)	2.81 (.08)	2.53 (.11)	.012a,b
% appraising worse out-group treatment	29.0%	32.7%	28.4%	21.4%
Self-esteem,3 mean (SE)	3.62 (.02)	3.61 (.03)	3.61 (.02)	3.66 (.04)	ns

Note. These measures above are statistically weighted in the NSAL but the frequencies (N=944) are un-weighted.

¹Income is on a seven-point scale with higher scores indicating higher household income

²Education is on a four-point scale with higher scores indicating higher education

³Self-esteem is on a four-point scale with higher scores indicating greater self-esteem

^*Superscripts indicate that there were statistically significant differences between the following pairs: a. light-skinned and dark-skinned; b. light-skinned and medium-skinned; c. medium-skinned and dark-skinned