Pathways to Skin Color Stratification: The Role of Inherited (Dis)Advantage and Skin Color Discrimination in Labor Markets

Abstract

Research has uncovered associations between skin color and myriad outcomes. What drives these associations? We develop a theoretical framework that synthesizes the multiple pathways linking skin color with life chances. Skin color stratification should be conceptualized in historical, structural terms: as the result of unequal treatment and inherited (dis)advantage, that is, unequal resources transmitted by families with different skin tones. We assess the role of two pathways— discrimination and inherited (dis)advantage—for Blacks’ and Latinos’ employment, earnings, and occupational prestige. We use the National Longitudinal Study of Youth 1997, which includes a visual skin color measure; multiple indicators of family background; and a sibling subsample that allows us, using fixed-effects models, to recover the effect of skin color net of family background. First, we find that darker skin tone is associated with worse labor market outcomes. Indicators of family background account for 29 to 44 percent of skin color’s associations with employment, earnings, and occupational prestige. Second, using sibling fixed-effects models, we find that darker skin tone is associated with worse labor market outcomes, but these associations are not statistically significant. In sum, our findings suggest that we pay attention to the multiple pathways linking skin color with life chances.

Full text

Citation:
Abascal, Maria, and
Denia Garcia. 2022. “Pathways
to Skin Color Stratification: The
Role of Inherited (Dis)Advantage
and Skin Color Discrimination in
Labor Markets.” Sociological Sci-
ence 9: 346-373.
Received: May 31, 2022
Accepted: July 13, 2022
Published: August 29, 2022
Editor(s): Ari Adut, Filiz Garip
DOI: 10.15195/v9.a14
Copyright: c

2022 The Au-
thor(s). This open-access article
has been published under a Cre-
ative Commons Attribution Li-
cense, which allows unrestricted
use, distribution and reproduc-
tion, in any form, as long as the
original author and source have
been credited. c b
Pathways to Skin Color Stratification: The Role
of Inherited (Dis)Advantage and Skin Color
Discrimination in Labor Markets
Maria Abascal,aDenia Garciab
a) New York University; b) University of Wisconsin–Madison
Abstract:
Research has uncovered associations between skin color and myriad outcomes. What drives
these associations? We develop a theoretical framework that synthesizes the multiple pathways
linking skin color with life chances. Skin color stratification should be conceptualized in historical,
structural terms: as the result of unequal treatment and inherited (dis)advantage, that is, unequal
resources transmitted by families with different skin tones. We assess the role of two pathways—
discrimination and inherited (dis)advantage—for Blacks’ and Latinos’ employment, earnings, and
occupational prestige. We use the National Longitudinal Study of Youth 1997, which includes a visual
skin color measure; multiple indicators of family background; and a sibling subsample that allows
us, using fixed-effects models, to recover the effect of skin color net of family background. First,
we find that darker skin tone is associated with worse labor market outcomes. Indicators of family
background account for 29 to 44 percent of skin color’s associations with employment, earnings, and
occupational prestige. Second, using sibling fixed-effects models, we find that darker skin tone is
associated with worse labor market outcomes, but these associations are not statistically significant.
In sum, our findings suggest that we pay attention to the multiple pathways linking skin color with
life chances.
Keywords: colorism; skin color; race; inequality; labor markets
A
growing body of work explores the association between skin color and life
chances. What drives observed associations with skin color? By many ac-
counts, these associations are due primarily to contemporary skin tone discrimi-
nation: darker-skinned Blacks, Latinos, and others are treated relatively worse in
schools and in the labor market, and their adverse treatment drives lower educa-
tional attainment, employment, earnings, and occupational prestige (Branigan et al.
2013; Hill 2000; Hunter 2002; Keith and Herring 1991; Kreisman and Rangel 2015;
Monk 2014, 2015, 2019; Telles and Murguia 1990).
Other factors, in addition to contemporary skin tone discrimination, might
drive associations between skin color and socioeconomic outcomes. Families with
different skin tones possess and transmit different resources as a result of skin tone
discrimination in past generations, resources inherited from White or European
ancestors, or both. Stated differently, lighter-skinned Black and Latino families
transmit a different set of resources by virtue of their ancestors’ position in both
categorical (e.g., racial) and continuous (e.g., phenotypic) hierarchies. We refer to
this pathway as “inherited (dis)advantage.”1
The contributions of this article are twofold. On a theoretical front, we develop
a framework that synthesizes the multiple pathways through which skin color is
346

Abascal and Garcia Pathways to Skin Color Stratification
linked to life chances. Our second contribution is empirical: we assess the role of
inherited (dis)advantage and contemporary skin tone discrimination in explaining
the labor market outcomes of Blacks and Latinos. Blacks and Latinos have long
been at the center of the skin color stratification literature; they are also the largest
non-White groups in the United States.
Our empirical analyses unfold in two stages. In the first stage, we model Blacks’
and Latinos’ labor market outcomes as a function of their skin tone, individual
sociodemographic characteristics, and family background. We control for a wider
set of family background indicators than prior work, including parents’ wealth,
which reflects the effects of historically accumulated inequalities (Conley 2009;
Oliver and Shapiro 2006). We also characterize the portion of the association
between skin color and labor market outcomes that is accounted for by indicators
of family background. In the second stage, we use sibling fixed-effects models to
recover the statistical effect of skin color on employment, earnings, and occupational
prestige, net of the advantages or disadvantages inherited by Blacks and Latinos
from their families. Both analyses use the 2008-to-2013 waves of the National
Longitudinal Study of Youth 1997 (NLSY97).
To anticipate the findings: dark-skinned Blacks and Latinos experience worse
outcomes in employment, earnings, and occupational prestige, and indicators of
family background account for 29 to 44 percent of skin color’s associations with
employment, earnings, and occupational prestige. Across our sibling fixed-effects
models, darker skin tone is associated with worse labor market outcomes, but these
associations are not statistically significant.
In short, we find evidence that inherited (dis)advantage is one source of skin
color stratification. Our results should not be interpreted as proof that associations
between skin color and labor market outcomes are due primarily to inherited
(dis)advantage and that contemporary discrimination based on skin tone plays a
smaller role or no role. We do, however, recommend that scholars pay attention to
the multiple pathways through which skin color and life chances are linked. We also
underscore the importance of conceptualizing colorism, like racism, in structural
and historical terms: as both the result of unequal treatment in the present day and
an accretion over generations of inequalities. Finally, our work carries implications
for public policy and inequality. The findings suggest that even if we succeeded in
eradicating skin tone discrimination today, skin color stratification would persist
through the inertia of inherited (dis)advantage.
The Historical Roots of Skin Color Stratification
“Colorism” is a form of stratification that is manifest even within racial categories:
people who identify (or are identified) with the same racial or ethnic group often
exhibit social and economic differences by phenotype. Although stratification by
skin color has been repeatedly studied, conclusions differ based on the outcome
under scrutiny, respondents’ race/ethnicity or gender, and the skin color measure
used. For example, the darkest Black Americans attain, on average, six months less
schooling than the lightest Black Americans (Monk 2014). The darkest Mexican
American men (with indigenous phenotypes) attain, on average, 1.5 years less
sociological science | www.sociologicalscience.com 347 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
schooling than the lightest Mexican American men (with European phenotypes)
(Murguia and Telles 1996). In general, darker-skinned Blacks and Latinos also
earn less than their lighter-skinned coethnics, with substantial variation by gender
(Goldsmith, Hamilton, and Darity 2006, 2007; Hersch 2008; Hunter 2002; Keith and
Herring 1991; Kreisman and Rangel 2015; Telles and Murguia 1990). In the criminal
justice system, dark-skinned Blacks receive harsher sentences than light-skinned
Blacks (Eberhardt et al. 2006; King and Johnson 2016; Viglione, Hannon, and DeFina
2011), but evidence is mixed regarding arrests (Branigan et al. 2017; Kizer 2017a;
Monk 2019).
For Black Americans, colorism originated in slavery when White men impreg-
nated enslaved women by force, conceiving mixed-race children, or “mulattoes.”
Whites granted greater opportunities to light-skinned individuals based on their
complexions or White ancestry (Hill 2000; Jablonski 2012; Keith and Herring 1991).
“Mulattoes” translated these advantages into higher social and economic standing in
the Black community and reproduced their status by marrying other light-skinned
individuals. For example, census data from 1860 to 1880 reveal that households
with two “mulatto” partners had “between 30 to 90 percent more wealth than
households with at least one black spouse” (Bodenhorn 2006:259). Although the
codification of the one-drop rule eventually threatened the status of light-skinned
Blacks, Reece (2019) argues that their economic head start allowed light-skinned
families to retain their position in the Black community and, ultimately, to avail
themselves of opportunities following the Civil Rights Movement. Other studies
show that the importance of White ancestry among Blacks declined in the post–Civil
Rights era (Gullickson 2005). Nevertheless, there is continued evidence of skin color
stratification among Blacks in the twenty-first century (Monk 2014).
For Latinos, colorism has origins in colonial stratification systems in the Ameri-
cas that subordinated people of indigenous and African origin (Hunter 2007; Telles
and Murguia 1990). Following independence, many Latin American states pro-
moted nation-building narratives that hailed mixed-race people as prototypical
citizens (Holt 2003; Telles and Garcia 2013; Wade 1993). Nevertheless, pigmentocra-
cies, that is, skin tone hierarchies favoring light-skinned individuals, endured in the
region (Telles and the Project on Ethnicity and Race in Latin America [PERLA] 2014).
As the diverse people we now consider Latinos entered the United States—through
migration and territorial annexation—they supplemented the autochthonous sys-
tem of skin color stratification with their own understandings of the same (Roth
2012). In recent years, a debate has emerged regarding Latinos’ place in the U.S.
racial hierarchy (Abascal 2015; Bonilla-Silva 2004; Frank, Akresh, and Lu 2010; Lee
and Bean 2004). There is a consensus that Whites still occupy the top rung of the
racial hierarchy and are likely to remain there and that the experiences of Latinos
with different phenotypes will likely diverge (Frank et al. 2010; Hersch 2018; Painter,
Holmes, and Bateman 2016).
Pathways to Skin Color Stratification
Skin color stratification has roots in histories of slavery and colonization, but how
is it reproduced today? We present a theoretical model (Figure 1) that identifies
sociological science | www.sociologicalscience.com 348 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Low SES
Family Skin tone
High SES
Family
(A) +
(A) −
(B) +
(B) −
(C) −
(C) +
Explanations of skin color stratification:
(A) Contemporary discrimination
(B) Inherited advantage
(C) Endogeneity/“money whitening”
Ancestor
Ancestor Parent
Parent
Outcomes of past
generations
Outcomes of the
current generation
Figure 1:
Theoretical pathways linking skin color and (labor market) outcomes. Note: SES stands for socioeco-
nomic status.
multiple, nonexclusive pathways through which skin tone could be linked to life
chances. Our model conceptualizes skin color stratification in structural, historical
terms, extending scholarship on racism to colorism (Bonilla-Silva 1997; powell
2008; Reece 2019). A structural approach focuses on the institutions and processes
that reproduce hierarchies over generations through cumulative causation (powell
2008). The three pathways identified in Figure 1are all manifestations of this
system. The first pathway is contemporary skin tone discrimination. Scholars
sometimes use “colorism” to mean skin color stratification (as we do) and sometimes
to mean differential treatment based on skin color, that is, one pathway through
which such stratification may emerge. A second pathway, which we call “inherited
(dis)advantage,” highlights the unequal resources inherited by people with different
skin tones as the result of racial and skin tone hierarchies faced by their ancestors.
The third pathway relates to the endogeneity of perceived skin color; it builds on
the insight that perceptions of another person’s skin color are themselves affected
by that person’s social and economic standing. The inherited (dis)advantage and
endogeneity pathways have received considerably less attention than the skin tone
discrimination pathway. Nevertheless, inherited (dis)advantage and endogeneity
directly implicate scholars’ ability to establish the causes of contemporary skin color
stratification.
By developing a unified model of skin color stratification, we clarify the rela-
tionship between historical and contemporary colorism. On the one hand are rich
historical accounts of skin color stratification and its colonial legacy. On the other
hand are accounts of contemporary unequal treatment based on skin color. The
result is an incomplete model of how colorism in the past shapes colorism in the
sociological science | www.sociologicalscience.com 349 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
present. In our model, by contrast, the tendency to discriminate by phenotype may
persist, but past unequal treatment is also theorized to affect the resources that
people possess to begin with.
Contemporary Skin Tone Discrimination
The first explanation of skin color stratification is skin tone discrimination in the
present day. People, even those who identify as members of the same racial or ethnic
group, are likely treated differently by others based on their phenotype (Hunter
2007). In other words, just as people can experience discrimination based on their
position in categorical (e.g., racial) hierarchies, they can experience discrimination
based on their positions in continuous (e.g., phenotypic) hierarchies. In Figure 1, the
lines labeled (A) illustrate how direct discrimination drives associations between
skin color and outcomes. This pathway captures instances of interpersonal discrim-
ination, in which individuals treat others less favorably based on their skin color.
If darker-skinned people are treated relatively worse than their lighter-skinned
counterparts in labor markets, educational institutions, or other domains, this will
produce differential attainment by skin color.
Skin tone discrimination is theorized to stem from a preference for whiteness,
according to which people who approximate the characteristics of Whites—for
example, because of their skin color—gain higher status and preferential treatment
(Goldsmith et al. 2007). Put differently, skin color, along with other physical
characteristics, serves as a form of capital that bestows privileges upon lighter
people who are seen as more beautiful and competent (Hunter 2002; Monk 2015;
Monk, Esposito, and Lee 2021). These preferences are rooted in ideologies created
by Whites to maintain their supremacy (Hannon, DeFina, and Bruch 2013; Hunter
2002). Both Whites and non-Whites in turn internalize and reproduce a preference
for whiteness.
The contemporary discrimination account is useful for understanding outcomes
produced by encounters with gatekeepers or otherwise powerful individuals in-
cluding, for example, representatives of the criminal justice system. However, the
discrimination account has been criticized for its near-exclusive focus on individual
prejudices and behaviors. As Reece argues, “inequality is the result of more than
the collective pathology of individuals” (2019:4). Skin color stratification must also
be understood in historical terms, as an accumulation of inequalities.
Inherited (Dis)advantage
The second pathway to skin color stratification is inherited (dis)advantage. By this
account, skin color stratification in the present also stems from disadvantages that
flow from racial and skin tone hierarchies in the past. On average, lighter-skinned
Blacks and Latinos might be born into more affluent families, and the resources their
families pass on enable them to attain better outcomes. These resources include
economic capital, as well as cultural and social capital (Bourdieu 1986; Lamont and
Lareau 1988). We refer to this mechanism as “inherited (dis)advantage,” recognizing
the role of the family in transmitting advantages and disadvantages, as documented
by a large empirical literature (Blau and Duncan 1967; Jencks et al. 1979; Lareau
sociological science | www.sociologicalscience.com 350 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
2011; McLanahan and Percheski 2008). Our view of inherited (dis)advantage is
consistent with multigenerational effects and cumulative causation, which link
resources in previous generations (including those preceding parents) to outcomes
in the current generation (Mare 2011). Inherited (dis)advantage is illustrated by the
arrows labeled (B) in Figure 1. The left-hand side of the figure shows unobserved
ancestry leading to more advantaged light-skinned households in the present day.
Why would lighter-skinned Blacks and Latinos be born into families with more
resources? The first possibility is related to skin tone hierarchies in the past. On
average, lighter-skinned Blacks and Latinos likely have lighter-skinned Black and
Latino ancestors, who were subjected to less skin tone discrimination (e.g., Hill 2000;
Keith and Herring 1991; Telles and PERLA 2014). These ancestors accumulated
greater social and material resources than their darker-skinned coethnic contem-
poraries. This is the mechanism that scholars most commonly invoke when they
acknowledge the role of inherited (dis)advantage. Keith and Herring, for example,
refer to “disadvantages derived from parents because of past discrimination against
darker blacks” (1991:775).
Categorical, race-based hierarchies could also account for the association be-
tween a person’s skin color and the (dis)advantages they inherit from their family.
Blacks and Latinos with lighter skin likely have more White or European
2
ancestors
in their family trees (Parra, Kittles, and Shriver 2004). White ancestors were exempt
from race-based subordination—ranging from discrimination to slavery. As a result,
these ancestors accumulated greater social and material resources. This account
does not assume that all mixed-race children, including the children of forced sexual
unions, inherited resources from their White parent. It is enough that some inherited
resources from their White parent and that these resources were greater than those
they would have inherited from two non-White parents.
In the following analyses, we are not able to disentangle the roles of color-
and race-based hierarchies, but they are worth distinguishing analytically. One
implicates differential opportunities and outcomes within racial categories, the
other across them. More broadly, discrimination and inherited (dis)advantage
could operate in tandem: inequalities based on how one’s ancestors were treated
might be reinforced by how people are treated today. These pathways are also
worth distinguishing because they carry different implications for intervention. If
skin color stratification today is primarily the result of “historically accumulated
racial privileges and disadvantages,” then “inequality would likely remain simply
because class inequalities by race would persist across generations” (Telles, Flores,
and Urrea-Giraldo 2015:53–54).
3
Thus, eliminating skin tone discrimination in the
present day might not erase inequalities by skin tone.
Endogeneity of Perceived Skin Color
Finally, the causal arrow may also run in reverse, from socioeconomic outcomes to
skin color, as illustrated by the arrows labeled (C) in Figure 1. Scholars typically
assume that associations with skin color are due to the effects of skin color on
outcomes. However, outcomes might shape how we see and rate another person’s
skin color. For example, according to the notion of “money whitening,” dark-
sociological science | www.sociologicalscience.com 351 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
skinned individuals are seen and treated as whiter once they gain upward mobility
(Freeman et al. 2011; Harris 1956; Ianni 1960).
Money whitening emphasizes how socioeconomic status shapes perceptions
of skin color; however, subtle racial cues, like names, attire, or accent, could also
play a role. Garcia and Abascal (2016) experimentally demonstrate that observers
rate the same face darker when it is associated with a distinctively Latino name.
Theirs is a conservative test of the reverse causality hypothesis. First, Garcia and
Abascal rely on a visual color palette that is, according to Flores and Telles (2012),
less sensitive to money whitening than are verbal scales (we return to this issue).
Second, their results are based on a single, subtle signal: a name. In the real world,
people take in more information about those with whom they interact, including
how they talk, dress, and carry themselves. Survey interviewers receive exceptional
access to interviewees’ personal information by the time they record their skin color,
typically at the end of an interview (for exceptions, see Telles and PERLA [2014] and
Villarreal [2010]). All this information is likely to magnify the observed inequality
by skin color if disadvantaged or racialized people are seen as darker. As the arrows
labeled (C) in Figure 1show, people with low earnings could be perceived as having
darker skin based on their socioeconomic status. Endogeneity makes it even more
difficult to interpret skin color stratification as prima facie evidence of skin tone
discrimination (arrows (A) in Figure 1).
Assessing Pathways to Skin Color Stratification
We focus on the two primary pathways to skin color stratification—contemporary
skin color discrimination and inherited (dis)advantage—and review the methods
and evidence used to study them in this section. Addressing the third pathway—
reverse causality—is challenging in the absence of experimental data, and it is
beyond the empirical scope of this study.
A common approach in the study of contemporary skin tone discrimination
relies on multiple regression models. Scholars compare the outcomes of people
with different skin tones, controlling for their sociodemographic characteristics,
including (when available) measures of their parents’ education and/or occupa-
tion. Indicators of parents’ socioeconomic status serve “as present-day proxies for
the transmission of advantages and disadvantages from the past” (Keith and Her-
ring 1991:775). To capture parents’ socioeconomic status, scholars commonly use
parental education and/or occupation (Bailey, Fialho, and Penner 2016; Branigan
et al. 2013; Flores and Telles 2012; Goldsmith et al. 2006, 2007; Gullickson 2005;
Keith and Herring 1991; Hill 2000; Hunter 2002; Monk 2014; Murguia and Telles
1996). Only a few studies control for other indicators of family background, like
parental income (Frank et al. 2010; Hersch 2008, 2018; Painter et al. 2016) or housing
conditions (Goldsmith et al. 2006; Hill 2000).
We are not aware of any U.S.-based study of colorism that has been able to
control for an important family background characteristic: parental wealth. To our
knowledge, only Telles, Flores, and Urrea-Giraldo (2015) control for parental wealth,
although these were supplementary analyses for a subset of Latin American coun-
tries (personal communication with Edward Telles, November 10, 2019). Wealth
sociological science | www.sociologicalscience.com 352 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
provides individuals with greater advantages than income, has the potential to last
generations, and is a more pronounced indicator of historical racial inequalities
(Conley 2009; Oliver and Shapiro 2006). Furthermore, wealth is more correlated
across generations than other indicators of family background, such as income,
education, and occupation (Mare 2011).
Having controlled for individual human capital and available indicators of
parental background, scholars attribute the residual association with skin tone to
discrimination. Most scholars acknowledge the role for social origins and past dis-
crimination in shaping present-day outcomes (Bailey et al. 2016; Flores and Telles
2012; Reece 2019; Telles et al. 2015). However, some interpret a significant statistical
effect of skin tone as undermining evidence for inherited (dis)advantage altogether.
For example, in their classic study, Keith and Herring (1991) concluded that dis-
crimination against darker-skinned Blacks was a more “powerful determinant” of
skin color stratification than “parental socioeconomic status” (1991:775).
However, interpreting the residual association between skin color and labor
market outcomes as evidence of differential treatment by skin color is challenging.
Even when we control for indicators of family background, unobserved differences
across families might be correlated with both skin color and labor market outcomes,
confounding their relationship. (These unobserved characteristics are illustrated by
the dashed line in Figure 1.)
Dealing with Unobserved Heterogeneity
Scholars can use two strategies to account for unobserved background character-
istics: experiments and sibling fixed-effects analyses. Experiments are the gold
standard for establishing discrimination because random assignment ensures that
any difference observed is the outcome of unequal treatment (or random chance)
(Pager and Shepherd 2008). However, using experiments to study skin tone discrim-
ination in labor markets has proved challenging because of difficulties in signaling
and manipulating skin color, for example, on resumes. In-person audits face other
challenges related to the feasibility of matching testers who differ only in terms
of skin tone. Indeed, only a handful of experiments have examined skin tone dis-
crimination in labor markets around the world (Dias 2020; Saeed, Maqsood, and
Rafique 2019). In the United States, to our knowledge, there is only one laboratory
experiment in which undergraduates evaluated the resume of a Black applicant for
a marketing position (Harrison and Thomas 2009). External validity represents a
limitation of laboratory-based studies, especially those based on student samples.
Sibling fixed-effects models are another strategy for addressing unobserved
background characteristics, in this case, using observational data; this is the strategy
we use in this article. Sibling fixed-effects models provide for a rigorous test of the
causal link between individual characteristics and life chances by netting out the
effects of shared genes, household and neighborhood environment, and inheritance.
Scholars of stratification and intergenerational mobility have long analyzed sibling
data to gain insight into the effect of “global family background” on life chances
(Conley and Glauber 2007:134).
sociological science | www.sociologicalscience.com 353 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Sibling samples with skin color measures are rare. Nevertheless, some sibling
analyses of skin color stratification exist. The first sibling studies of colorism
originated in Brazil, where scholars examined differences between siblings who
identified with different color/racial categories (Francis-Tan 2016; Marteleto and
Dondero 2016; Rangel 2015; Telles 2004). For example, the first study on this
topic by Telles (2004) found that White siblings had better educational outcomes
than their non-White siblings based on 1991 census data. In the United States,
a handful of studies examine the association between outcomes and skin color
differences across siblings. In terms of income and earnings, Kizer (2017b) finds
that, among Blacks, Latinos, and Asian Americans, darker-skinned siblings have
lower household incomes than lighter-skinned siblings. Findings are mixed with
respect to educational attainment. Ryabov (2016) finds that darker-skinned Asian
Americans and Latinos are less likely to complete high school and transition to
college than lighter-skinned siblings. However, Kizer (2017b) finds skin color to be
a significant predictor of college completion among Asian American women but
not among Blacks and Latinos. Finally, looking beyond socioeconomic outcomes,
Kizer (2017a) finds that darker-skinned Black, Latino, and Asian American men
are more likely to experience arrest than their lighter-skinned siblings, and Laidley
et al. (2019) find that dark skin is associated with hypertension among Black and
Latino siblings.
Sources of Sibling Skin Color Data
To our knowledge only two U.S. sibling data sets have a measure of skin color:
the National Longitudinal Study of Adolescent to Adult Health (Add Health) and
NLSY97. Both data sets use interviewer-rated skin tone, which is appropriate
for the study of discrimination because discrimination hinges on how people are
perceived by others (Roth 2016; Telles and Lim 1998).
4
All previous U.S.-based
sibling studies of skin color stratification rely on Add Health, which we do not use.
Add Health interviewers rated respondents’ skin tones at the end of interviews
using a five-point scale anchored to the following verbal categories: “black,” “dark
brown,” “medium brown,” “light brown,” and “white.” They did not receive visual
aids to guide their ratings. We rely instead on NLSY97, which uses a 10-point visual
palette to capture skin color. Like Add Health interviewers, NLSY97 interviewers
rated respondents’ skin tones at the end of interviews.
The NLSY97 measure improves on Add Health’s measure in several ways.
First, measures that rely on visual palettes—as opposed to verbal scales—might be
less sensitive to money whitening, that is, reverse causality from socioeconomic
outcomes to perceived skin color (Flores and Telles 2012). Second, some of the
verbal anchors on Add Health’s scale (e.g., “black” and “white”) correspond to
racial categories, rather than commonly observed skin tones. As a result, Add
Health skin tone ratings likely reflect interviewers’ social understandings of racial
classification (who is, e.g., “Black” or “White”), not just skin color.
Third, NLSY97’s 10-point scale can reveal more variation than Add Health’s five-
point scale, which is especially important for sibling analyses. On this issue, Bucca
(2019) writes, “one fundamental challenge when studying the effect of skin color on
sociological science | www.sociologicalscience.com 354 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
socioeconomic outcomes [is] the limited heterogeneity that remains after accounting
for race” (P. 12). In fact, he finds through a variance decomposition analysis of Add
Health that just 20 percent of skin tone variation is found within racial categories;
even less (less than 11 percent) is found within families. By contrast, more than
three in five Black and Latino siblings in NLSY97 have a different skin tone than one
or more of their siblings, a finding to which we return in the section on Descriptive
Statistics. These advantages notwithstanding, all measures of skin color have
limitations (see Dixon and Telles [2017] for a review). For example, both visual and
verbal scales are sensitive to endogeneity and interviewer effects, and they exhibit
moderate levels of inter-rater reliability (Campbell et al. 2020; Garcia and Abascal
2016; Hannon and DeFina 2020).
Data and Methods
Data
This study contributes to the empirical study of skin color stratification by using
a rich set of family indicators; a fine-grained, visual skin color measure; and sib-
ling fixed-effects models to examine the role of both inherited (dis)advantage and
contemporary color-based discrimination for labor market outcomes. We use data
from NLSY97, a nationally representative cohort study of youth born between 1980
and 1984. It consists of a cross-sectional sample of respondents plus a supplemental
oversample of Blacks and Latinos.
Our analyses rely on two analytic samples: (1) a pooled sample of self-identified
Black and Latino respondents who participated in at least one of the waves when
skin color was recorded (2008, 2009, or 2010)
5
and (2) a subsample of Black and
Latino respondents with siblings who identified with the same race/ethnicity. The
pooled sample (1) consists of 2,155 Black and 1,718 Latino respondents (3,873 total).
The sibling subsample (2) comprises those 819 Black respondents and 745 Latino
respondents (1,564 total) with at least one full or half same-race/ethnicity sibling
in NLSY97.
6
Because of the sampling design of NLSY97, all sibling respondents
were living in the same household during the initial round. These respondents
were distributed across 717 unique households. Of these households, 84.14 percent
contain two sibling respondents. The maximum number of sibling respondents per
household is five; this describes just two households (less than one percent).
Analytic Strategy
The following analyses unfold in two stages. In the first stage, we characterize the
portion of the association between skin color and labor market outcomes that is
accounted for by indicators of family background. We use ordinary least squares
(OLS) regressions to model labor market outcomes first as a function of respondent
skin tone and individual characteristics, then as a function of respondent skin tone,
individual characteristics, and family background characteristics. These models are
estimated for the pooled sample of 3,873 Black and Latino respondents. Standard
errors account for clustering at the household level.7
sociological science | www.sociologicalscience.com 355 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Our goal is to gauge the magnitude reduction in skin color’s statistical effect once
we control for family background characteristics. If an outcome were uncorrelated
with other correlates of skin tone, conditional on all other covariates in the model,
then the statistical effect of skin tone would be an unbiased estimate of its true
causal effect. This assumption is unverifiable and implausible. However, it is worth
stressing that, when compared with prior work, we account for a larger set of family
background indicators, including parental wealth.
In the second stage, we use sibling fixed-effects regressions to model the labor
market outcomes of siblings as a function of differences in their skin tones, individ-
ual characteristics that vary between siblings, and a family fixed effect (intercept)
that captures all family characteristics that are common across siblings. The stan-
dard errors for the fixed-effects regressions are also adjusted for clustering at the
household level.
8
These models are estimated for the subsample of 1,564 Black and
Latino respondents with one or more siblings in NLSY97.
To interpret the statistical effect of skin tone as evidence of its causal effect, we
must assume that unobserved factors that simultaneously affect both skin tone and
our outcomes are invariant across siblings. Put differently, we must assume that the
way in which families distribute resources between their children is uncorrelated
with each child’s skin tone. There is scant evidence on this in the United States, an
issue to which we return in the Conclusion.
Key Independent Variable and Dependent Variables
Key independent variable. Interviewers rated respondents’ skin tone using a 10-
point visual palette where 1 represents the lightest color and 10 represents the
darkest. Skin tone was initially recorded in round 12 (2008). In rounds 13 (2009) and
14 (2010), interviewers rated the skin tone of respondents whose skin tones were not
recorded previously. Interviewers rated skin tone at the end of in-person interviews.
As a guide, interviewers relied on a memorized “color card” that illustrates the
skin tones associated with each number. The questionnaire design and color card
provided conflicting information regarding whether albino respondents were to be
coded as 0 or 1 (personal communication with National Longitudinal Survey User
Services, August 22, 2019). We therefore recoded cases (
n=
15) that were assigned
a color of 0 as 1.9
Dependent variables. We examine three labor market outcomes: employment
status, earnings, and occupational prestige. To mitigate noise associated with one-
off measurements, we examine three-year averages for these variables based on
rounds 14 through 16, which were collected in 2010 (the last year in which skin
color was recorded), 2011, and 2013.
10
Supplementary analyses confirm that results
are similar using only 2010 values for employment, earnings, and occupational
prestige.
Interviewers contacted respondents every week to obtain their employment
status. Following earlier analyses of employment in the National Longitudinal
Surveys of Youth (NLSY) (e.g., Western and Beckett 1999), employment status is
represented by the proportion of weeks observed in a year when a respondent
sociological science | www.sociologicalscience.com 356 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
worked 20 hours or more. Twenty hours correspond to part-time employment; the
results are substantively similar using 10- or 30-hour cutoffs.
Earnings are represented by the natural logarithm of total employment income
in the past year.
11
This includes income received from “wages, salary, commissions,
or tips from all jobs before deductions.”
12
We focus on employment income because
this source of income is more plausibly affected by skin tone discrimination in the
labor market than are other sources of income, such as investment or family income.
Our final dependent variable is occupational prestige. We assigned status scores
to civilian and military occupations.
13
Scores are drawn from the International
Socio-Economic Index of Occupational Status (ISEI-08). ISEI-08 is one of the most
recent occupational scales, and, in contrast to earlier scales, it takes into account data
from both men and women. Status scores in the data range from 10 for dishwashers
to 89 for physicians and surgeons.
Control Variables: OLS Regression Models
Individual characteristics. All OLS models control for respondent age, marital
status, and cohabitation status in 2010, as well as gender, nativity (U.S.-born citi-
zen or other), and race/ethnicity (Latino or Black). Respondents who identified
as “Hispanic or Latino” of any race are classified as Latino. In addition, analy-
ses based on the pooled sample control for the presence of full or half siblings in
NLSY97. We do not control for respondents’ educational attainment because educa-
tion likely mediates the effect of family background on labor market outcomes. As
a result, it is possible that we are overestimating the statistical effect of skin tone
and underestimating the statistical effect of family background.
Family background characteristics. The second set of OLS models control for family
background characteristics, which were recorded in the first round of the NLSY97
(1997). These characteristics are of two types: characteristics shared across parents in
respondents’ households (hereafter “household characteristics”) and characteristics
of individual parents (hereafter “parent characteristics”).
Household characteristics include household income, government aid receipt,
household net worth, parents’ homeownership, and residence with both biological
parents. Household income represents gross household income in 1997.
14
House-
hold net worth represents the difference between total assets and total debts.
15
In
the regressions, the coefficients for parents’ income and net worth reflect the statisti-
cal effect of a $10,000 difference. Parents’ homeownership takes a value of 1 if the
responding parent or their spouse/partner owns the house or apartment where the
respondent lives. Government aid receipt takes a value of 1 if the responding parent
received government aid (e.g., food stamps) between the age of 18 (or the birth of
the oldest respondent) and the time of the survey. A binary variable represents
whether a respondent was living with both biological parents in 1997. Controlling
for parents’ marital status, instead of whether the respondent was living with both
biological parents, yields substantively similar results.
Parent characteristics include nativity, mother’s education, and father ’s educa-
tion. Parents’ nativity takes a value of 1 if at least one of the respondent’s parents
(biological or residential) was born outside the United States. Mother’s education
sociological science | www.sociologicalscience.com 357 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
and father’s education are captured by two variables, each. For father ’s education,
for example, the first variable reflects the highest grade completed by a respon-
dent’s father.
16
This variable takes a value of 0 if the respondent does not have
educational information for either a biological or residential father. In this case, a
second variable (“Father’s education missing”) takes a value of 1. This strategy,
which treats the absence of a parent’s information as meaningful, is modeled after
other analyses of the NLSY (e.g., Bloome and Western 2011).
Control Variables: Sibling Fixed-Effects Regressions
The sibling fixed-effects regressions control for the following individual charac-
teristics: skin tone, gender, age, cohabitation, and marital status. They do not
control for respondent race/ethnicity (Latino or Black), nativity, or the presence
of full or half siblings in the NLSY97 because these variables are invariant across
siblings; that is, siblings take the same values on these variables. Nor do the sibling
fixed-effects regressions control for family background characteristics. Predictably,
household characteristics are invariant across siblings, and parent characteristics
show extremely limited variation across siblings.
Missing Data and Multiple Imputation
To conserve statistical power and mitigate bias, we multiply imputed missing values
of independent and dependent variables. Dependent variables were used in the
imputation model, and imputed values of these variables were retained.
17
We
implemented multiple imputation by chained equations using the “mice” package
in R. Predictive mean matching ensures that imputed values are plausible, that is,
observed in nonmissing data. Following recommendations by White, Royston, and
Wood (2011), we generated 50 imputed data sets, then used Rubin’s rules to pool
parameter estimates. Results are substantively similar for the subset of respondents
with complete data (Tables A2 and A3 in the online supplement).
Values were imputed at the respondent level, not the household level, meaning
that some siblings may take different imputed values for household variables, like
parents’ income. Disregarding dependencies in the data, in essence treating all
variables as “just another variable,” is a common approach that works well in linear
models (von Hippel 2007; White et al. 2011). The alternative—to enforce dependen-
cies after imputation—can introduce biases stemming from the transformation of
imputed variables.
Results
Descriptive Statistics
Figure 2reports the distribution of skin tones for Black and Latino respondents
in NLSY97. The picture for both Black and Latino respondents, but especially for
Blacks, is one of substantial variation: nontrivial numbers of Black respondents are
observed at every point in the 10-point scale. Variation is also considerable within
sociological science | www.sociologicalscience.com 358 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Figure 2: Skin tone distribution for Black and Latino respondents (nonimputed values).
our subsample of Black and Latino siblings. In this subsample, 63.78 percent of all
respondents have at least one full or half sibling in NLSY97 whose skin tone differs
from theirs. Among these respondents, the average difference between their skin
tone and their sibling’s
18
is 0.93 points on a 10-point scale. A 0.93-point difference in
skin tone would probably be noticeable to many people, including Whites (Garcia
and Abascal 2016; Hannon et al. 2021; Roth 2012; Wilder 2010). Table A1 in the
online supplement reports additional descriptive statistics for Black and Latino
respondents.
The Role of Family Background
Here, we examine the associations between skin tone and labor market outcomes
in order to characterize the portion of these associations accounted for by family
background. We first consider the association with employment. Model P1 (Table 1)
reports the results of a linear regression predicting the proportion of weeks a
respondent was employed as a function of their skin color
19
and demographic
characteristics, including the respondent’s race/ethnicity (Black or Latino) and
whether the respondent has at least one full or half sibling in NLSY97. Skin color
is significantly, negatively associated with employment (
p<
0.10). On average,
respondents who are one point darker on a 10-point scale were employed between
one-half and one week less over the course of one year.
What portion of this association is accounted for by the advantages transmit-
ted by the families of lighter-skinned respondents? Model P2 (Table 1) adjusts
for important indicators of family background: household income, government
aid receipt, household net worth, parents’ homeownership, residence with both
sociological science | www.sociologicalscience.com 359 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Table 1:
Labor market outcomes by skin tone, covariates: Blacks and Latinos with full or half siblings in
NLSY97
Employment Yearly earnings (log) Occupational prestige
Model P1 Model P2 Model E1 Model E2 Model O1 Model O2
Individual characteristics
Skin tone −0.007∗−0.005 −0.064∗−0.046 −0.246∗−0.136
(0.004) (0.004) (0.033) (0.032) (0.127) (0.121)
Latino (vs. Black) 0.038†0.034∗0.336†0.321†1.230∗1.493†
(0.018) (0.018) (0.154) (0.159) (0.643) (0.623)
Female −0.038†−0.032†−0.403†−0.362†3.073†3.247†
(0.012) (0.012) (0.108) (0.106) (0.448) (0.434)
Age 0.008∗0.008∗0.037 0.035 0.446†0.429†
(0.004) (0.004) (0.037) (0.037) (0.154) (0.151)
Cohabiting 0.045†0.054†0.194 0.277†−0.853 −0.322
(0.016) (0.016) (0.140) (0.137) (0.578) (0.559)
Married 0.031†0.024 0.588†0.539†1.682†1.522†
(0.015) (0.015) (0.123) (0.121) (0.544) (0.519)
U.S.-born citizen −0.048†−0.069†−0.320†−0.393∗−1.748†−1.431
(0.016) (0.028) (0.136) (0.214) (0.649) (1.060)
Full sibling(s) in NLSY97 −0.026∗−0.020 −0.212∗−0.136 −1.178†−0.510
(0.013) (0.013) (0.115) (0.113) (0.465) (0.452)
Half sibling(s) in NLSY97 −0.125†−0.087†−0.519 −0.224 −2.705†−1.361
(0.036) (0.037) (0.332) (0.331) (1.244) (1.138)
Family background (1997)
Household income [$10K] 0.002 0.064†0.383†
(0.004) (0.029) (0.133)
Received government aid −0.024 −0.111 −1.085∗
(0.015) (0.128) (0.572)
Household net worth [$10K] 0.000 0.002 0.058
(0.001) (0.008) (0.046)
Parents own home 0.079†0.505†1.212†
(0.016) (0.130) (0.545)
Resided with both biological parents 0.041†0.259†1.346†
(0.015) (0.129) (0.546)
Immigrant parent(s) 0.001 0.215 2.379†
(0.026) (0.198) (0.966)
Mother’s education 0.006†0.050†0.520†
(0.003) (0.023) (0.100)
Mother’s education missing 0.026 0.330 3.666†
(0.039) (0.343) (1.362)
Father’s education 0.005∗0.059†0.382†
(0.003) (0.022) (0.098)
Mother’s education missing 0.024 0.394 3.692†
(0.033) (0.290) (1.176)
Constant 0.500†0.343†8.134†6.436†27.248†14.091†
(0.121) (0.126) (1.044) (1.089) (4.398) (4.597)
N3,873 3,873 3,873 3,873 3,873 3,873
R-squared 0.026 0.064 0.028 0.062 0.045 0.138
Notes: Standard errors adjusted for clustering at the household level. ∗p<0.10; † p<0.05 (two-sided).
sociological science | www.sociologicalscience.com 360 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
biological parents, parents’ nativity, and parents’ education (henceforth, we refer to
these variables as “seven indicators of family background”). Several indicators are
associated with employment in predictable ways: net of other factors, employment
is significantly, positively associated with parents’ homeownership (
p<
0.001),
residing with both biological parents (
p<
0.01), mother’s education (
p<
0.05), and
father’s education ( p<0.10).
Moreover, holding these differences constant, skin tone is no longer significantly
associated with the proportion of weeks employed. And, although the point esti-
mate of the coefficient remains negative, it is smaller in magnitude. Specifically, the
seven indicators of family background account for 29.54 percent of the association
between skin tone and employment. To assess whether the reduction in the size of
the coefficient is significant, we follow the procedure proposed by Yan, Aseltine,
and Harel (2013) for comparing coefficients that are common to nested models
based on clustered data. Specifically, for each imputed data set, we calculate a t
statistic for the change in the size of the skin tone coefficient between models P1
and P2 when these models are estimated using generalized estimating equations
(GEEs).
20
The mean tstatistic across imputed data sets is
−
3.33, which corresponds
to a significant reduction in the size of the skin tone coefficient (p<0.001).
Next, we consider the association between skin tone and earnings. Model E1
(Table 1) reports the results of a linear regression predicting respondents’ earnings
(logged) as a function of their skin color and basic demographic characteristics. Skin
color is significantly, negatively associated with earnings (
p<
0.10). On average,
respondents who are one point darker on a 10-point scale earned 6.24 percent less
in one year.
Model E2 (Table 1) additionally adjusts for seven indicators of family back-
ground.
21
Net of other factors, earnings are significantly, positively associated with
household income (
p<
0.05), parents’ homeownership (
p<
0.001), residing with
both biological parents (
p<
0.05), mother’s education (
p<
0.05), and father’s
education (p<0.01).
Holding the seven indicators of family background constant, moreover, skin tone
is not significantly associated with earnings, and the point estimate of the coefficient
is smaller in magnitude. Together, the seven indicators of family background
account for 28.74 percent of the association between skin tone and earnings (on the
log scale). To assess whether this reduction is significant, we follow the procedure
described earlier. The mean tstatistic across imputed data sets is
−
2.94, which
corresponds to a significant reduction in the size of the skin tone coefficient (
p<
0.01).
Finally, we consider the association between skin tone and occupational prestige.
Model O1 (Table 1) reports the results of a linear regression predicting occupational
prestige as a function of respondent skin color and basic demographic characteristics.
Skin color is significantly, negatively associated with occupational prestige (
p<
0.10). On average, respondents who are one point darker on a 10-point scale
are employed in civilian or military occupations that are 0.25 points lower on a
100-point scale.
Model O2 (Table 1) additionally adjusts for seven indicators of family back-
ground. Many of these indicators are significantly associated with occupational
sociological science | www.sociologicalscience.com 361 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
prestige. Occupational prestige is significantly, positively associated with house-
hold income (
p<
0.01), parents’ homeownership (
p<
0.05), residing with both
biological parents (
p<
0.01), immigrant parent(s) (
p<
0.05), mother’s education
(
p<
0.001), and father’s education (
p<
0.001). Occupational prestige is also
negatively associated with government aid receipt (p<0.10).
Holding family background constant, the point estimate on the skin tone coeffi-
cient continues to be negative, but it is not significant. Additionally, the magnitude
of the coefficient has dropped by 44.49 percent. The reduction in the size of the skin
tone is significant (t=−2.52, p<0.05).
Heterogeneity by Race
Are these results different across Black and Latino respondents? To examine this,
first, we reestimate models P1 and P2 separately for Black and Latino respondents
(Table A5 in the online supplement). Controlling for individual characteristics
but not family background characteristics, skin tone is negatively associated with
employment among Blacks, but the association does not reach significance. Among
Latinos, skin tone is significantly, negatively associated with employment (
p<
0.05).
Controlling for both individual and family background characteristics, skin tone
remains a significant predictor of employment among Latinos (
p<
0.10), but its
statistical effect has fallen in terms of both size (by 19.16 percent) and significance.
Next, we reestimate models E1 and E2 predicting earnings separately for Black
and Latino respondents (Table A6 in the online supplement). Controlling for in-
dividual characteristics but not family background characteristics, skin tone is
negatively associated with earnings among Blacks, but the association does not
reach significance. Among Latinos, skin tone is significantly, negatively associated
with employment (
p<
0.10). Controlling for both individual and family back-
ground characteristics, the statistical effect of skin tone is reduced by 26.21 percent
and is no longer statistically significant.
Finally, we reestimate models O1 and O2 predicting occupational prestige sep-
arately for Black and Latino respondents (Table A7 in the online supplement).
Controlling for individual characteristics but not family background characteristics,
skin tone is significantly, negatively associated with occupational prestige among
Blacks (
p<
0.10). Controlling for both individual and family background charac-
teristics, the statistical effect of skin tone is reduced by 43.93 percent, and it is no
longer statistically significant. Among Latinos, skin tone is negatively associated
with occupational prestige, but the association does not reach significance, control-
ling for individual characteristics alone or in combination with family background
characteristics.
In sum, modeling labor market outcomes separately for Black and Latino re-
spondents tells a broadly similar story: where skin tone is significantly associated
with outcomes, controlling for family background reduces the association in both
magnitude and significance.
sociological science | www.sociologicalscience.com 362 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Table 2:
Labor market outcomes by skin tone, covariates, sibling fixed effects: Blacks and Latinos with full or
half siblings in NLSY97
Employment Yearly earnings (log) Occupational prestige
Model P3 Model E3 Model O3
Skin tone −0.006 −0.065 −0.074
(0.009) (0.077) (0.275)
Female −0.048∗−0.450†2.144†
(0.024) (0.201) (0.812)
Age 0.014∗0.119∗0.306
(0.008) (0.065) (0.262)
Cohabiting 0.083†0.374 1.777∗
(0.033) (0.305) (1.027)
Married 0.000 0.350 0.660
(0.032) (0.256) (0.989)
Sibling fixed effects
N1,564 1,564 1,564
R-squared 0.021 0.016 0.018
Notes: Standard errors adjusted for clustering at the household level. ∗p<0.10; † p<0.05 (two-sided).
Results of Sibling Fixed-Effects Models
The findings of the previous analyses suggest that basic indicators of family back-
ground account for a significant portion of the associations between skin tone
and employment, earnings, and occupational prestige. However, the remaining
associations with employment, earnings, and occupational prestige may be due
to other, unobserved family background characteristics rather than interpersonal
contemporary discrimination. To explore this, we model the labor market outcomes
of siblings as a function of differences in their skin tones, other sibling differences,
and a fixed effect (intercept) that captures all family background characteristics that
are common across siblings. These analyses are based on the subsample of Black
and Latino respondents (
N=
1, 564) with coethnic full or half siblings who were
also respondents in NLSY97.
First, we consider the association between skin tone and employment. Model P3
(Table 2) predicts the proportion of weeks sibling respondents were employed as a
function of their skin color, sibling-varying characteristics (gender, age, cohabitation,
and marital status), and household fixed effects. Coefficients should be interpreted
as the difference in the proportion of weeks respondents were employed corre-
sponding to a unit difference between each predictor. If skin color is a significant,
negative predictor of employment in this model, it implies that skin color is causally
related to employment, presumably through interpersonal discrimination. In model
P3 (Table 2), skin tone is negatively associated with proportion of weeks employed,
but the association is not significant (p=0.50).
Next, we consider the association between skin color and earnings. In model
E3 (Table 2), skin tone is negatively associated with earnings,
22
but as before, the
sociological science | www.sociologicalscience.com 363 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
association is not significant (
p=
0.40). Finally, in model O3 (Table 2), skin tone is
negatively associated with occupational prestige, but as before, the association is
not significant (p=0.79).
Heterogeneity by Race
To examine heterogeneity by race/ethnicity, first, we reestimate models P3, E3,
and O3 predicting employment, earnings, and occupational prestige separately for
Black and Latino respondents (Table A9 in the online supplement). With respect to
employment (model P3), earnings (model E3), and occupational prestige (model
O3), results are similar across Black and Latino respondents: skin color does not
significantly predict labor market outcomes, net of sibling-varying characteristics
and household fixed effects.23
Summary of Findings
Skin color and labor market outcomes are linked through inherited (dis)advantage,
not only contemporary discrimination. Indicators of family background account for
29 to 44 percent of the associations between skin tone and employment, earnings,
and occupational prestige. In a pooled sample of Black and Latino respondents, skin
tone is negatively associated with employment, earnings, and occupational prestige,
adjusting for basic sociodemographic characteristics. On average, respondents who
are one point darker on a 10-point scale work about one week less per year, earn 6.24
percent less, and are in occupations that are less prestigious by 0.25 points (on a 100-
point scale). Across our sibling fixed-effects models, darker skin tone is associated
with worse labor market outcomes, although these associations do not achieve
significance. The lack of significance in the sibling fixed-effects models may reflect
insufficient statistical power to detect skin color associations of modest magnitudes
like those observed. It is also possible that unmeasured family characteristics,
24
such as birth order, interact with skin color and affect the way parents treat their
children. If that is the case, sibling fixed-effects models might underestimate the
role of family background.
Conclusion
Studies have repeatedly uncovered an association between skin tone and myriad
outcomes, including labor market outcomes. This article addressed a further ques-
tion: through what pathways are skin color and labor market outcomes linked?
We presented a theoretical framework (Figure 1) that synthesizes three, critical
pathways: inherited (dis)advantages passed down through families, present-day
discrimination based on skin tone, and reverse causality from outcomes to per-
ceived skin tone. Then, we explored the pathways of inherited (dis)advantage
and discrimination using 2008-to-2013 rounds of NLSY97 to examine employment,
earnings, and occupational prestige among Blacks and Latinos. NLSY97 provides a
visual measure of skin tone and an extensive set of family background indicators,
allowing us to build on prior empirical work.
sociological science | www.sociologicalscience.com 364 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
In the first stage, we sought evidence of skin tone differences and inherited
(dis)advantage in labor markets. Darker skin color is associated with worse labor
market outcomes for Black and Latino respondents, adjusting for basic sociodemo-
graphic characteristics. Indicators of family background account for between 29 and
44 percent of skin color’s associations with employment, earnings, and occupational
prestige. Some or all of the residual associations with skin tone could stem from
unobserved confounders related to family background.
Therefore, in the second stage, we used sibling fixed-effects regressions to
model labor market outcomes for Black and Latino siblings, accounting for family
background. Darker-skinned siblings attain worse labor market outcomes than their
lighter-skinned siblings, but these associations are not statistically significant. To
put these results in context, we know that darker-skinned Blacks and Latinos report
more discrimination than their lighter-skinned coethnics (Monk 2015; Pew Research
Center 2021; Santana 2018). And, in high-stakes situations, such as in encounters
with the criminal justice system, differential treatment based on skin tone could
have life-changing consequences (Eberhardt et al. 2006; King and Johnson 2016;
Kizer 2017a; Monk 2019; Viglione et al. 2011).
That skin tone is nonsignificant in the sibling fixed-effects models does not
suggest that lighter- and darker-skinned Blacks and Latinos attain similar labor
market outcomes. Nor should our findings be interpreted as evidence that con-
temporary skin tone discrimination plays no role or a smaller role than inherited
(dis)advantage in producing skin color stratification. Our results do not speak
to the relative size of the effects played by family background versus skin tone
discrimination in labor market outcomes. More broadly, absence of evidence is not
evidence of absence, especially in the case of modest effect sizes like those observed.
Our study has some limitations. Sibling fixed-effects models assume that siblings
with different skin colors obtain similar resources from their parents. The few
studies to look at this issue in the United States do not reveal a consistent pattern of
preferential treatment by skin tone within families (Drake and Cayton 1945; Landor
et al. 2013; Taylor, Desjardin, and Robles 2016; Tharps 2016). Furthermore, cohort
studies, such as the NLSY97, tell us how a particular cohort (in our case, those
born from 1980 to 1984) is doing at a particular time. In supplementary analyses
predicting labor market outcomes in 2019, instead of between 2010 and 2013, we
did not find that the results changed as respondents aged. Additional research is
needed on these topics.
Short of experimental evidence, however, sibling fixed-effects models represent
our next best hope for uncovering evidence of contemporary color-based discrim-
ination, unbiased by confounders. This is important for theoretical and practical
reasons. Our findings direct attention to inherited (dis)advantage as one pathway
through which skin color is linked to socioeconomic outcomes. Skin color is one
manifestation of the (dis)advantages that Blacks and Latinos inherit from their
families, whether these (dis)advantages are due to color- or race-based hierarchies
faced by their ancestors. Household income, net worth, and parents’ educational
attainment are common proxies of family background, but they are more directly
correlated with material, as opposed to symbolic, resources. Theoretically, the
sociological science | www.sociologicalscience.com 365 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
advantages we inherit from our family also comprise social, cultural, and symbolic
forms of capital that help us secure benefits in the workplace and other institutions.
Our findings underscore the importance of conceptualizing colorism in struc-
tural and historical terms. This means thinking of skin color stratification as the
result not simply of present-day discrimination but also of cumulative, multigen-
erational inequalities. This explanation looks to the past and to a deeper, more
intransigent understanding of how inequalities are reproduced. There is immense
value in eradicating present-day discrimination and ensuring equal access to rights
and opportunities. But, by drawing attention to the structural and historical roots of
skin color stratification, our findings make clear that were skin tone discrimination
to disappear today, we would still see lighter- and darker-skinned Blacks and Lati-
nos achieving different outcomes through the inertia of inherited (dis)advantage.
Indeed, conceptualizing skin color in structural and historical terms carries im-
plications for policy design. Traditionally, when people think of skin tone discrimi-
nation, they think about individual attitudes and behaviors that favor individuals
with lighter skin. Designing interventions based on this understanding of colorism
might lead us to focus on changing individuals and organizations through, for
example, awareness campaigns and anti-bias training. However, conceptualizing
colorism in structural terms requires that we think of other policy interventions.
Improving the socioeconomic outcomes of dark-skinned Blacks and Latinos will
require policies that promote the social and economic well-being of families so that
they can pass on more resources to their children.
We look forward to studies that attempt to disentangle the pathways through
which skin color is linked to well-documented as well as novel outcomes—including
educational achievement, dating, marriage, residential segregation, and criminal
justice contact. Future work would also do well to investigate how skin color
stratification varies by gender and for other ethnic and racial groups who were
beyond the immediate scope of this research. Moving forward, research and policy
should give requisite weight to a structural, historical understanding of phenotypic
stratification. Our findings suggest that skin color stratification is deeply rooted in
inequalities that have been reproduced through centuries of unequal treatment.
Notes
1
Our term is a simplification inspired by Hill’s (2002) term: “ancestrally accumulated
disadvantage.”
2
Recent European ancestry might be more common among Latinos from countries that
encouraged European immigration in the 1800s and 1900s to whiten their populations
(Hernández 2013).
3See also Hill (2000:1444).
4
By contrast, self-rated skin color can be conceptualized as a manifestation of internalized
social status that is better suited to the study of subjective outcomes (Monk 2015).
Measures of skin tone taken with a spectrophotometer, usually from the underarm, may
not be socially meaningful, in part because we perceive facial skin tone (Campbell et al.
2020).
sociological science | www.sociologicalscience.com 366 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
5
We refer to survey rounds based on the first year in which data were collected (e.g., we
refer to round 14, fielded in 2010 and 2011, as the 2010 round).
6
This excludes seven respondents who did not identify with the same race/ethnicity as
any of their siblings. This also excludes 18 respondents who reported another respondent
as their sibling but for whom the other respondent did not mutually report a sibling
relationship.
7Results based on nonadjusted standard errors are substantively similar.
8Results based on nonadjusted standard errors are substantively similar.
9
Interviewer race is associated with ratings of respondent skin tone. Controlling for
interviewer race, however, does not substantively change the results of our analyses.
10
Starting in 2011, NLSY97 moved to biannual surveys; that is, the survey following the
2011 round was fielded starting in 2013.
11
To conserve observations, we assigned dollar amounts to midpoint values based on the
binned earnings variable (T6055600), rather than the exact earnings variable (T6055500).
12
“Questionnaire Public Report. Cohort: National Longitudinal Survey of Youth 1997;
Round: Youth Questionnaire 97 (R14).” National Longitudinal Surveys, accessed August
2, 2022,
https://www.nlsinfo.org/sites/default/files/attachments/121129/
nlsy97r14inc.html.
13
Occupation was recorded for respondents with a valid employer. For respondents
who were not interviewed in a certain round, we do not know who did not have
a valid employer, that is, which respondents would have legitimately skipped the
occupation item. We therefore impute occupation for all respondents in the analytic
sample (
N=
3, 873) with missing values for this variable. For models based on complete
cases, see Tables A2 and A3 in the online supplement.
14
Reported by the responding parent, if the respondent was not independent, or by the
respondent, if the respondent was independent. Just 5.22 percent of the respondents in
the pooled sample were classified as independent in 1997.
15
Reported by the responding parent or by the respondent, if the respondent was inde-
pendent and the responding parent did not report assets and debt.
16
The highest reported grade for respondents with both a biological and a residential
father.
17
Retaining these values conserves statistical power and does not produce substantially
more biased or less efficient estimates if a sufficient number of data sets is generated
(Young and Johnson 2010).
18
For respondents with more than one sibling in NLSY97, this figure corresponds to the
difference between their skin tone and the average skin tone of their siblings.
19
In supplementary analyses, we treat skin color as a five-point categorical variable, rather
than as a continuous variable, to explore whether the statistical effect of skin color is
nonlinear. The results are consistent with a linear association between skin color and
employment and earnings.
20
GEEs can be used to estimate the parameters of a generalized linear model based
on clustered data because estimates are not sensitive to the specification of the error
structure.
21
Earnings might be confounded with employment. In Table A4 of the online supplement,
we reestimate models E1 and E2 controlling for the proportion of weeks a respondent
worked 20 hours or more. Results are substantively similar.
sociological science | www.sociologicalscience.com 367 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
22
Because earnings might be confounded with employment, in Table A8 of the online
supplement, we reestimate model E3 controlling for the proportion of weeks a respondent
worked 20 hours or more. Results are substantively similar.
23
Results from the sibling fixed-effects regressions are similar for the subset of respondents
with at least one full sibling in NLSY97. This is valuable because a family fixed effect is
an imperfect proxy for shared family endowment, especially for siblings with different
biological parents.
24
In supplementary sibling fixed-effects analyses, we did not find evidence that the effect
of skin color was moderated by sibling gender.
References
Abascal, Maria. 2015. “Us and Them: Black–White Relations in the Wake of Hispanic
Population Growth.” American Sociological Review 80(4):789–813.
https://doi.org/10.
1177/0003122415587313.
Bailey, Stanley R., Fabrício M. Fialho, and Andrew M. Penner. 2016. “Interrogating Race:
Color, Racial Categories, and Class across the Americas.” American Behavioral Scientist
60(4):538–55. https://doi.org/10.1177/0002764215613400.
Blau, Peter M., and Otis Dudley Duncan. 1967. The American Occupational Structure. New
York: John Wiley & Sons.
Bloome, Deidre, and Bruce Western. 2011. “Cohort Change and Racial Differences in
Educational and Income Mobility.” Social Forces 90(2):375–95.
https://doi.org/10.
1093/sf/sor002.
Bodenhorn, Howard. 2006. “Colorism, Complexion Homogamy, and Household Wealth:
Some Historical Evidence.” American Economic Review 96(2):256–60.
https://doi.org/
10.1257/000282806777211883.
Bonilla-Silva, Eduardo. 1997. “Rethinking Racism: Toward a Structural Interpretation.”
American Sociological Review 62(3):465–80. https://doi.org/10.2307/2657316.
Bonilla-Silvia, Eduardo. 2004. “From Bi-racial to Tri-racial: Towards a New System of Racial
Stratification in the USA.” Ethnic and Racial Studies 27(6):931–50.
https://doi.org/10.
1080/0141987042000268530.
Bourdieu, Pierre. 1986. “The Forms of Capital.” Pp. 241–58 in Handbook of Theory and Research
for the Sociology of Education, edited by John Richardson. New York: Greenwood Press..
Branigan, Amelia R., Jeremy Freese, Assaf Patir, Thomas W. McDade, Kiang Liu, and Catarina
I. Kiefe. 2013. “Skin Color, Sex, and Educational Attainment in the Post–Civil Rights
Era.” Social Science Research 42(6):1659–74.
https://doi.org/10.1016/j.ssresearch.
2013.07.010.
Branigan, Amelia R., Christopher Wildeman, Jeremy Freese, and Catarina I. Kiefe. 2017.
“Complicating Colorism: Race, Skin Color, and the Likelihood of Arrest.” Socius 3.
https://doi.org/10.1177%2F2378023117725611.
Bucca, Mauricio. 2019. “Does Skin Color Affect Educational and Labor Market Outcomes in
the United States?” Working Paper, European University Institute.
Campbell, Mary E., Verna M. Keith, Vanessa Gonlin, and Adrienne R. Carter-Sowell. 2020.
“Is a Picture Worth a Thousand Words? An Experiment Comparing Observer-Based
Skin Tone Measures.” Race and Social Problems. 12:266–78.
https://doi.org/10.1007/
s12552-020-09294-0.
sociological science | www.sociologicalscience.com 368 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Conley, Dalton. 2009. Being Black, Living in the Red: Race, Wealth, and Social Policy in America.
10th Anniversary Edition. University of California Press.
Conley, Dalton, and Rebecca Glauber. 2007. “Family Background, Race, and Labor Market
Inequality.” ANNALS of the American Academy of Political and Social Science 609(1):134–52.
https://doi.org/10.1177/0002716206296090.
Dias, Felipe A. 2020. “How Skin Color, Class Status, and Gender Intersect in the Labor
Market: Evidence from a Field Experiment.” Research in Social Stratification and Mobility
65(2020):1–15. https://doi.org/10.1016/j.rssm.2020.100477.
Dixon, Angela, and Edward E. Telles. 2017. “Skin Color and Colorism: Global Research,
Concepts, and Measurement.” Annual Review of Sociology 43:405–24.
https://doi.org/
10.1146/annurev-soc-060116-053315.
Drake, St. Clair, and Horace R. Cayton. 1945. Black Metropolis: A Study of Negro Life in a
Northern City. New York: Harcourt, Brace and Company.
https://doi.org/10.2307/
2264023.
Eberhardt, Jennifer L., Paul G. Davies, Valerie J. Purdie-Vaughns, and Sheri Lynn Johnson.
2006. “Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts
Capital-Sentencing Outcomes.” Psychological Science 17(5):383–86.
https://doi.org/10.
1111/j.1467-9280.2006.01716.x.
Flores, René, and Edward Telles. 2012. “Social Stratification in Mexico: Disentangling Color,
Ethnicity, and Class.” American Sociological Review 77(3):486–94.
https://doi.org/10.
1177/0003122412444720.
Francis-Tan, Andrew. 2016. “Light and Shadows: An Analysis of Racial Differences be-
tween Siblings in Brazil.” Social Science Research 58:254–65.
https://doi.org/10.1016/
j.ssresearch.2016.02.003.
Frank, Reanne, Ilana Redstone Akresh, and Bo Lu. 2010. “Latino Immigrants and the U.S.
Racial Order: How and Where Do They Fit In?” American Sociological Review 75(3):378–401.
https://doi.org/10.1177/0003122410372216.
Freeman, Jonathan B., Andrew M. Penner, Aliya Saperstein, Matthias Scheutz, and Nalini
Ambady. 2011. “Looking the Part: Social Status Cues Shape Race Perception.” PLoS One
6(9):e25107. https://doi.org/10.1371/journal.pone.0025107.
Garcia, Denia, and Maria Abascal. 2016. “Colored Perceptions: Racially Distinctive Names
and Assessments of Skin Color.” American Behavioral Scientist 60(4):420–41.
https:
//doi.org/10.1177/0002764215613395.
Goldsmith, Arthur H., Darrick Hamilton, and William Darity, Jr. 2006. “Shades of Dis-
crimination: Skin Tone and Wages.” American Economic Review 96(2):242–45.
https:
//doi.org/10.1257/000282806777212152.
Goldsmith, Arthur H., Darrick Hamilton, and William Darity, Jr. 2007. “From Dark to
Light: Skin Color and Wages among African-Americans.” Journal of Human Resources
XLII(4):701–38. https://doi.org/10.3368/jhr.XLII.4.701.
Gullickson, Aaron. 2005. “The Significance of Color Declines: A Re-analysis of Skin Tone
Differentials in Post–Civil Rights America.” Social Forces 84(1):157–80.
https://doi.org/
10.1353/sof.2005.0099.
Hannon, Lance, and Robert DeFina. 2020. “The Reliability of Same-Race and Cross-Race
Skin Tone Judgments.” Race and Social Problems 12:186–94.
https://doi.org/10.1007/
s12552-020-09282-4.
Hannon, Lance, Robert DeFina, and Sarah Bruch. 2013. “The Relationship between Skin
Tone and School Suspension for African Americans.” Race and Social Problems 5(4):281–95.
https://doi.org/10.1007/s12552-013-9104-z.
sociological science | www.sociologicalscience.com 369 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Hannon, Lance, Verna M. Keith, Robert DeFina, and Mary E. Campbell. 2021. “Do White
People See Variation in Black Skin Tones? Reexamining a Purported Outgroup Homo-
geneity Effect.” Social Psychology Quarterly 84(1):95–106.
https://doi.org/10.1177/
0190272520961408.
Harris, Marvin. 1956. Town and Country in Brazil. New York: Columbia University Press.
Harrison, Matthew S., and Kecia M. Thomas. 2009. “The Hidden Prejudice in Selection: A
Research Investigation on Skin Color Bias.” Journal of Applied Social Psychology 39(1):134–
68. https://doi.org/10.1111/j.1559-1816.2008.00433.x.
Hernández, Tanya Katerí. 2013. Racial Subordination in Latin America: The Role of the State,
Customary Law, and the New Civil Rights Response. New York: Cambridge University Press.
https://doi.org/10.1017/CBO9781139176125.
Hersch, Joni. 2008. “Profiling the New Immigrant Worker: The Effects of Skin Color and
Height.” Journal of Labor Economics 26(2):345–86. https://doi.org/10.1086/587428.
Hersch, Joni. 2018. “Colorism against Legal Immigrants to the United States.” American
Behavioral Scientist 62(14):2117–32. https://doi.org/10.1177/0002764218810758.
Hill, Mark E. 2000. “Color Differences in the Socioeconomic Status of African American
Men: Results of a Longitudinal Study.” Social Forces 78(4):1437–60.
https://doi.org/10.
2307/3006180.
Hill, Mark E. 2002. “Race of the Interviewer and Perception of Skin Color: Evidence from
the Multi-city Study of Urban Inequality.” American Sociological Review 67(1):99–108.
https://doi.org/10.2307/3088935.
Holt, Thomas C. 2003. “Foreword: The First New Nations.” Pp. vii–xvi in Race and Nation in
Modern Latin America, edited by Nancy P. Appelbaum, Anne S. Macpherson, and Karin
Alejandra Rosemblatt. University of North Carolina Press.
Hunter, Margaret. 2002. “‘If You’re Light You’re Alright’: Light Skin Color as Social Cap-
ital for Women of Color.” Gender & Society 16(2):175–93.
https://doi.org/10.1177/
08912430222104895.
Hunter, Margaret. 2007. “The Persistent Problem of Colorism: Skin Tone, Status, and
Inequality.” Sociology Compass 1(1):237–54.
https://doi.org/10.1111/j.1751-9020.
2007.00006.x.
Ianni, Octavio. 1960. “Segunda Parte.” Pp. 153–226 in Côr e Mobilidade Social Em Florianópolis,
edited by Fernando Henrique Cardoso and Octavio Ianni. Sao Paolo, Brazil: Companhia
Editora Nacional.
Jablonski, Nina G. 2012. Living Color: The Biological and Social Meaning of Skin Color. University
of California Press. https://doi.org/10.1525/9780520953772.
Jencks, Christopher, SusanBartlett, Mary Corcoran, James Crouse, David Eaglesfield, Gregory
Jackson, Kent McClelland, Peter Mueser, Michael Olneck, Joseph Schwartz, Sherry Ward,
and Jill Williams. 1979. Who Gets Ahead? The Determinants of Economic Success in America.
New York: Basic Books.
Keith, Verna M., and Cedric Herring. 1991. “Skin Tone and Stratification in the Black
Community.” American Journal of Sociology 97(3):760–78.
https://doi.org/10.1086/
229819.
King, Ryan D., and Brian D. Johnson. 2016. “A Punishing Look: Skin Tone and Afrocentric
Features in the Halls of Justice.” American Journal of Sociology 122(1):90–124.
https:
//doi.org/10.1086/686941.
Kizer, Jessica. 2017a. “Arrested by Skin Color: Evidence from Siblings and a Nationally
Representative Sample.” Socius 3. https://doi.org/10.1177%2F2378023117737922.
sociological science | www.sociologicalscience.com 370 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Kizer, Jessica. 2017b. “Skin Color Stratification and the Family: Sibling Differences and the
Consequences for Inequality.” PhD dissertation, Department of Sociology, UC Irvine,
Irvine, CA. https://escholarship.org/uc/item/29n4r264.
Kreisman, Daniel, and Marcos A. Rangel. 2015. “On the Blurring of the Color Line: Wages
and Employment for Black Males of Different Skin Tones.” Review of Economics and
Statistics 97(1):1–13. https://doi.org/10.1162/REST_a_00464.
Laidley, Thomas, Benjamin Domingue, Piyapat Sinsub, Kathleen Mullan Harris, and Dalton
Conley. 2019. “New Evidence of Skin Color Bias and Health Outcomes Using Sibling
Difference Models: A Research Note.” Demography 56(2):753–62.
https://doi.org/10.
1007/s13524-018-0756-6.
Lamont, Michele, and Annette Lareau. 1988. “Cultural Capital: Allusions, Gaps and
Glissandos in Recent Theoretical Developments.” Sociological Theory 6(2):153–68. https:
//doi.org/10.2307/202113.
Landor, Antoinette M., Leslie Gordon Simons, Ronald L. Simons, Gene H. Brody, Chalandra
M. Bryant, Frederick X. Gibbons, Ellen M. Granberg, and Janet N. Melby. 2013. “Exploring
the Impact of Skin Tone on Family Dynamics and Race-Related Outcomes.” Journal of
Family Psychology 27(5):817–26. https://doi.org/10.1037/a0033883.
Lareau, Annette. 2011. Unequal Childhoods: Class, Race, and Family Life. 2nd ed. University of
California Press. https://doi.org/10.1525/9780520949904.
Lee, Jennifer, and Frank D. Bean. 2004. “America’s Changing Color Lines: Immigration,
Race/Ethnicity, and Multiracial Identification.” Annual Review of Sociology 30:221–42.
https://doi.org/10.1146/annurev.soc.30.012703.110519.
Mare, Robert D. 2011. “A Multigenerational View of Inequality.” Demography 48(1):1–23.
https://doi.org/10.1007/s13524-011-0014-7.
Marteleto, Letícia J., and Molly Dondero. 2016. “Racial Inequality in Education in Brazil:
A Twins Fixed-Effects Approach.” Demography 53(4):1185–205.
https://doi.org/10.
1007/s13524-016-0484-8.
McLanahan, Sara, and Christine Percheski. 2008. “Family Structure and the Reproduction of
Inequalities.” Annual Review of Sociology 34:257–76.
https://doi.org/10.1146/annurev.
soc.34.040507.134549.
Monk, Ellis P. 2014. “Skin Tone Stratification among Black Americans, 2001–2003.” Social
Forces 92(4):1313–37. https://doi.org/10.1093/sf/sou007.
Monk, Ellis P. 2015. “The Cost of Color: Skin Color, Discrimination, and Health among
African-Americans.” American Journal of Sociology 121(2):396–444. https://doi.org/10.
1086/682162.
Monk, Ellis P. 2019. “The Color of Punishment: African Americans, Skin Tone, and the
Criminal Justice System.” Ethnic and Racial Studies 42(10):1593–612.
https://doi.org/
10.1080/01419870.2018.1508736.
Monk, Ellis P., Michael H. Esposito, and Hedwig Lee. 2021. “Beholding Inequality: Race,
Gender, and Returns to Physical Attractiveness in the United States.” American Journal of
Sociology 127(1):194–241. https://doi.org/10.1086/715141.
Murguia, Edward, and Edward Telles. 1996. “Phenotype and Schooling among Mexican
Americans.” Sociology of Education 69(4):276–89. https://doi.org/10.2307/2112715.
Oliver, Melvin, and Thomas Shapiro. 2006. Black Wealth/White Wealth: A New Perspective on
Racial Inequality. 2nd ed. New York: Routledge.
Pager, Devah, and Hana Shepherd. 2008. “The Sociology of Discrimination: Racial Discrim-
ination in Employment, Housing, Credit, and Consumer Markets.” Annual Review of
Sociology 34(1):181–209. https://doi.org/10.1146/annurev.soc.33.040406.131740.
sociological science | www.sociologicalscience.com 371 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Painter, Matthew A., Malcolm D. Holmes, and Jenna Bateman. 2016. “Skin Tone,
Race/Ethnicity, and Wealth Inequality among New Immigrants.” Social Forces 94(3):1153–
85. https://doi.org/10.1093/sf/sov094.
Parra, Esteban J., Rick A. Kittles, and Mark D. Shriver. 2004. “Implications of Correlations
between Skin Color and Genetic Ancestry for Biomedical Research.” Nature Genetics
36(11):S54–60. https://doi.org/10.1038/ng1440.
Pew Research Center. 2021. “Majority of Latinos Say Skin Color Impacts Op-
portunity in America and Shapes Daily Life.” Washington, DC.
https:
//www.pewresearch.org/hispanic/2021/11/04/majority-of-latinos-say-skin-
color-impacts-opportunity-in-america-and- shapes-daily- life/.
powell, john a. 2008. “Structural Racism: Building upon the Insights of John Calmore.” North
Carolina Law Review 86(3):791–816.
Rangel, Marcos A. 2015. “Is Parental Love Colorblind? Human Capital Accumulation within
Mixed Families.” Review of Black Political Economy 42(1):57–86.
https://doi.org/10.
1007/s12114-014-9190-1.
Reece, Robert L. 2019. “Color Crit: Critical Race Theory and the History and Future of
Colorism in the United States.” Journal of Black Studies 50(1):3–25.
https://doi.org/10.
1177/0021934718803735.
Roth, Wendy D. 2012. Race Migrations: Latinos and the Cultural Transformation of Race. Stanford
University Press. https://doi.org/10.1515/9780804782531.
Roth, Wendy D. 2016. “The Multiple Dimensions of Race.” Ethnic and Racial Studies 39(8):1310–
38. https://doi.org/10.1080/01419870.2016.1140793.
Ryabov, Igor. 2016. “Educational Outcomes of Asian and Hispanic Americans: The Sig-
nificance of Skin Color.” Research in Social Stratification and Mobility 44:1–9.
https:
//doi.org/10.1016/j.rssm.2015.11.001.
Saeed, Abubakr, Sidra Maqsood, and Amir Rafique. 2019. “Color Matters: Field Experiment
to Explore the Impact of Facial Complexion in Pakistani Labor Market.” Journal of the
Asia Pacific Economy 24(3):347–63.
https://doi.org/10.1080/13547860.2019.1612974
.
Santana, Emilce. 2018. “Situating Perceived Discrimination: How Do Skin Color and
Acculturation Shape Perceptions of Discrimination among Latinos?” Sociological Quarterly
59(4):655–77. https://doi.org/10.1080/00380253.2018.1506690.
Taylor, John L., Suzanne Desjardin, and Irene Robles. 2016. “From Colored People to Students
of Color: The Complexity of Colorism in Families and Educational Institutions.” Pp.
39–61 in Race and Colorism in Education, edited by Carla R. Monroe. New York: Routledge.
Telles, Edward. 2004. Race in Another America: The Significance of Skin Color in Brazil. Princeton
University Press. https://doi.org/10.1515/9781400837434.
Telles, Edward, René D. Flores, and Fernando Urrea-Giraldo. 2015. “Pigmentocracies:
Educational Inequality, Skin Color and Census Ethnoracial Identification in Eight Latin
American Countries.” Research in Social Stratification and Mobility 40:39–58.
https://doi.
org/10.1016/j.rssm.2015.02.002.
Telles, Edward, and Denia Garcia. 2013. “Mestizaje and Public Opinion in Latin Amer-
ica.” Latin American Research Review 48(3):130–52.
https://doi.org/10.1353/lar.2013.
0045.
Telles, Edward, and Nelson Lim. 1998. “Does It Matter Who Answers the Race Question?
Racial Classification and Income Inequality in Brazil.” Demography 35(4):465–74.
https:
//doi.org/10.2307/3004014.
sociological science | www.sociologicalscience.com 372 August 2022 | Volume 9

Abascal and Garcia Pathways to Skin Color Stratification
Telles, Edward, and Edward Murguia. 1990. “Phenotypic Discrimination and Income
Differences among Mexican Americans.” Social Science Quarterly 71(4):682–96.
Telles, Edward, and the Project on Ethnicity and Race in Latin America (PERLA). 2014.
Pigmentocracies: Ethnicity, Race, and Color in Latin America. University of North Carolina
Press.
Tharps, Lori L. 2016. Same Family, Different Colors: Confronting Colorism in America’s Diverse
Families. Boston: Beacon Press.
Viglione, Jill, Lance Hannon, and Robert DeFina. 2011. “The Impact of Light Skin on
Prison Time for Black Female Offenders.” Social Science Journal 48(1):250–58.
https:
//doi.org/10.1016/j.soscij.2010.08.003.
Villarreal, Andrés. 2010. “Stratification by Skin Color in Contemporary Mexico.” American
Sociological Review 75(5):652–78. https://doi.org/10.1177/0003122410378232.
Von Hippel, Paul T. 2007. “Regression with Missing Ys: An Improved Strategy for Analyzing
Multiply Imputed Data.” Sociological Methodology 37(1):83–117.
https://doi.org/10.
1111/j.1467-9531.2007.00180.x.
Wade, Peter. 1993. Blackness and Race Mixture: The Dynamics of Racial Identity in Colombia.
Baltimore, MD: Johns Hopkins University Press.
Western, Bruce, and Katherine Beckett. 1999. “How Unregulated Is the U.S. Labor Market?
The Penal System as a Labor Market Institution.” American Journal of Sociology 104(4):1030–
60. https://doi.org/10.1086/210135.
White, Ian R., Patrick Royston, and Angela M. Wood. 2011. “Multiple Imputation Using
Chained Equations: Issues and Guidance for Practice.” Statistics in Medicine 30(4):377–99.
https://doi.org/10.1002/sim.4067.
Wilder, JeffriAnne. 2010. “Revisiting ‘Color Names and Color Notions’: A Contemporary
Examination of the Language and Attitudes of Skin Color among Young Black Women.”
Journal of Black Studies 41(1):184–206. https://doi.org/10.1177/0021934709337986.
Yan, Jun, Robert H. Aseltine, and Ofer Harel. 2013. “Comparing Regression Coeffi-
cients between Nested Linear Models for Clustered Data with Generalized Estimat-
ing Equations.” Journal of Educational and Behavioral Statistics 38(2):172–89.
https:
//doi.org/10.3102/1076998611432175.
Young, Rebekah, and David R. Johnson. 2010. “Imputing the Missing Y’s: Implications for
Survey Producers and Survey Users.” Pp. 6242–48 in Papers Presented at the 64th Annual
Conference of the American Association for Public Opinion Research (AAPOR), May 13–16,
2010, Chicago, Illinois.
Acknowledgments:
We thank Siyeona Chang for invaluable research assistance. We are
grateful to Jason Fletcher, Mike Hout, Patricia McManus, Rourke O’Brien, Edward
Telles, and members of the Center for Research on Race and Ethnicity in Society work-
shop at Indiana University, Bloomington for their feedback. Both authors contributed
equally to this work. Direct correspondence to Maria Abascal, 295 Lafayette Street,
New York, NY 10027, m.abascal@nyu.edu, and Denia Garcia, 1225 Observatory Drive,
Madison, WI 53706, dgarcia28@wisc.edu.
Maria Abascal:
Department of Sociology, New York University. E-mail: m.abascal@nyu.edu.
Denia Garcia:
La Follette School of Public Affairs, University of Wisconsin–Madison.
E-mail: dgarcia28@wisc.edu.
sociological science | www.sociologicalscience.com 373 August 2022 | Volume 9