## Abstract

A rich literature has documented the negative association between dark skin tone and many dimensions of U.S.-born Americans’ life chances. Despite the importance of both skin tone and immigration in the American experience, few studies have explored the effect of skin tone on immigrant assimilation longitudinally. I analyze data from the New Immigrant Survey (NIS) 2003 to examine how skin tone is associated with occupational achievement at three time points: the last job held abroad, the first job held in the United States, and the current job. Dark-skinned immigrants experience steeper downward mobility at arrival in the United States and slower subsequent upward mobility relative to light-skinned immigrants, net of human and social capital, race/ethnicity, country of origin, visa type, and demographics. These findings shed light on multiple current literatures, including segmented assimilation theory, the multidimensionality of race, and the U.S. racial hierarchy.

## Introduction

Skin tone is at the center of recent scholarship on the process of changing racial hierarchies and racial inequality. The greater influx of immigrants of color, some scholars have predicted, will change the current Black-White racial divide in the future. Some expect the American stratification structure to become more like that of Latin America, where inequality is shaped along color scales rather than racial categories (Bailey et al. 2016; Bonilla-Silva and Dietrich 2009; Telles 2014).1 Others expect that the racial hierarchy in the United States will change from Whites/non-Whites to non-Blacks/Blacks as the majority group embraces lighter-skinned racial groups whose socioeconomic status becomes comparable to Whites, including Asians and light-skinned Latinxs—a process called “whitening” (Gans 2012; Lee and Bean 2007). Another prediction of the future racial hierarchy in the United States is a move “toward the eventual elimination of distinct racial and ethnic groups in favor of a skin-color hierarchy” given that the skin color hierarchy has changed little over time but the meaning of race and ethnicity has changed substantially (Hochschild 2005:81).

The diverse racial and ethnic composition of contemporary immigrants also suggests that skin tone plays a key role for immigrants in navigating their positions in the U.S. racial hierarchy and their relation to native-born Americans at the individual level. When immigrants encounter the United States’ prevailing system of colorism and racism, they experience a different hierarchy of skin tone status than in their country of origin (Foner 2000; Roth 2010; Waters 1999). Skin tone can be expected to influence the immigrant assimilation process and ethnoracial identity formation, even among the second generation, because skin tone is an important phenotype by which native people perceive immigrants’ race (Alba and Nee 1997; Gans 1992; Portes and Rumbaut 2006; Portes and Zhou 1993; Zhou 1997).

Nonetheless, empirical studies examining immigrant skin tone discrimination are scarce, in large part because representative survey data to measure immigrants’ skin tone in addition to race are also rare.2 Longitudinal data that include skin tone are even scarcer.3 Existing empirical, survey-based studies of immigrants find a negative effect of dark skin tone on labor market outcomes (e.g., Frank et al. 2010; Hersch 2008; Mason 2004; Rosenblum et al. 2016). A drawback of these studies is that they examined the effect of skin tone at only one time point, leaving unexamined the question of how skin tone influences the immigration and labor market assimilation process over time. Nonetheless, these existing cross-sectional studies suggest that reception in the United States is deflected downward for darker-skinned immigrants.

On the other hand, those studies that examined immigrant assimilation processes over time tend to neglect skin tone effects (e.g., Akresh 2008; Chiswick 1978; Chiswick and Miller 2009). They explain downward mobility experienced by immigrants when they cross the border as being due to imperfect skill transferability and show that immigrants later experience upward mobility as they accumulate destination country-specific human/social capital, resulting in U-shape labor market trajectories. Despite their focus on how country of origin, visa type, education level, and social ties influence the depth of the U-shape, none of these studies examined immigrants’ skin tone or race. That only country of origin is the main variable of interests in those studies indicates an assumption that immigrants from one country are racially homogenous while ignoring the possibility that skill transferability is influenced by racial or skin tone discrimination.

In this study, using the New Immigrant Survey (NIS) 2003 data, I analyze how skin tone influences immigration and assimilation processes in occupational trajectories, including the baseline pre-immigration period. Using a representative survey of immigrants who earned their legal permanent residency (LPR), this study is the first to analyze the mechanisms of dark skin tone penalties for immigrants in the U.S. labor market that focuses on temporal incorporation, controlling for pre-immigration selection across the full range of immigrant origins. This study also contributes to the immigrant assimilation literature with its focus on the effect of skin tone and racial group membership on intragenerational mobility for the first generation, which has been less developed in the current literature. In addition, this study sheds light on the role of skin tone in the current literature on skill transferability by examining the extent to which skin tone influences the process.

## Colorism and Racial Inequality in the United States

A rich literature has documented the negative association between dark skin tone and various social stratification outcomes in the United States, such as occupational status, income, educational attainment, and mate selection (Hamilton et al. 2009; Hughes and Hertel 1990; Hunter 1998; Keith and Herring 1991; Monk 2014). Psychological domains, such as self-esteem, perceptions of attractiveness, racial identity, and Whites’ affective reactions to minorities have been explored as well (Bond and Cash 1992; Hagiwara et al. 2012).

Stratification by skin tone is the consequence of colorism, defined as “the process of discrimination that privileges light-skinned people of color over their dark-skinned counterparts” (Hunter 2007:234). Colorism is conventionally understood as a within-ethnoracial process that operates in relation with racism but in many ways is distinctive from it. Hunter (2007:238), for example, theorizes that “racism is a larger, systemic social process and colorism is one manifestation of it,” characterizing the degree of racial discrimination as moderated by skin tone, with lighter-skinned individuals facing less racial discrimination. Reflecting this perspective, most empirical studies on the negative effects of dark skin tone among Americans explore these effects for only one specific racial/ethnic group, often Blacks (Hughes and Hertel 1990; Hunter 1998; Keith and Herring 1991) or Hispanics (Espino and Franz 2002; Mason 2004).

However, research on the multidimensionality of race (Roth 2010; Saperstein 2006) suggests that colorism should also be conceptualized beyond a within-race discrimination process because skin tone, along with other phenotypes, influences the boundaries of ethnoracial group membership itself. Ethnoracial boundaries are not naturally given or fixed but are instead created and changing as a consequence of constant negotiations between actors with different strategies in defining memberships of ethnoracial groups (Wimmer 2008), through interactions across individual (micro), institutional (meso), and cultural (macro) levels (Saperstein et al. 2013). In this process, skin tone plays a key role as a signal of race, particularly when the racial classification process is an interactional accomplishment. Research has noted that along multiple dimensions of ethnoracial classification, perception of race by others is more critical than is self-identified race in predicting discrimination (Golash-Boza 2006; Mason 2004; Monk 2015; Roth 2010; Saperstein 2006). A mismatch between ethnoracial self-identity and ethnoracial classification by others based on phenotype is often observed in daily life interactions. For example, light-skinned Hispanic high school students are often perceived as White European descendants despite their self-identity as Latinxs (Fergus 2009). Because of such a mismatch, Latinxs experience a dark skin tone penalty in particular because immigrants from the Latin Americas tend to identify themselves as Whites despite their dark skin, by which potential employers perceive them as Blacks (Rosenblum et al. 2016). Conversely, the finding by Goldsmith et al. (2006) that earnings of Blacks with light skin tone are statistically indistinguishable from earnings of comparable Whites suggests that light-skinned self-identified Blacks are treated in the same manner as Whites in the labor market.

These findings altogether suggest that categorical racial lines are fluid contingent on skin tone as well as on other social contexts, particularly at the phenotypic borders. Reflecting such a discrepancy between self-identified and perceived race, empirical studies show that the combination of skin tone and self-identified race serves as a better predictor of inequality in the United States and in the majority of Latin American countries than either race or skin tone alone does (Bailey et al. 2016). This scholarship thus suggests that skin tone stratification must be understood beyond one ethnoracial group because skin tone and self-identity are two different dimensions of race (Roth 2010).

## Downward Mobility of Immigrants With Dark Skin Tone

### Dark Skin Tone Penalties

Empirical studies of variation in skin tone show a dark skin tone penalty in the U.S. labor market among immigrants as well, net of race and other individual demographics. Hersch (2008), using the NIS data, finds that immigrants with the lightest skin color earn, on average, 17% more than comparable immigrants with the darkest skin tone, net of race. Such a dark skin tone penalty is also found within immigrant groups. Frank et al. (2010), for example, using the same data, find dark skin tone to be associated with wage loss among Latinx immigrants: darker-toned Latinx immigrants earned, on average, \$2,500 less per year than their lighter-skinned counterparts.

The existence of dark skin tone penalties indicates that skin tone is a critical factor in immigrants’ assimilation to the U.S. mainstream. Assimilation theory predicts that immigrants will be unilaterally assimilated into the U.S. mainstream, although it may take time (Alba and Nee 1997, 2003). Scholars of assimilation theory argue that dark skin tone may slow the pace of acculturation because of the resulting racial discrimination but that dark skin tone is not an absolute obstacle. Gans (1992) raised the possibility that immigrants with dark skin color or from a low socioeconomic class in the country of origin, in particular, may be trapped at the bottom of stratification in the United States. Because skin tone separates those deemed phenotypically Black from Whites, immigrants with darker skin tone—like those from the Caribbean—will likely have more difficulty assimilating to the United States than immigrants with lighter skin tone (Alba and Nee 1997). However, assimilation theorists further argue that skin tone is not an all-encompassing obstacle given that there are some immigrant groups—such as South Asians—whose skin tone is relatively dark but who have successfully achieved higher socioeconomic status. Instead, in their new assimilation theory, Alba and Nee (2003) contend that the types of capital (i.e., human and cultural capital) immigrants bring with them are stronger predictors of immigrants’ assimilation than are skin tone or race.

In contrast, segmented assimilation scholars suggest that race is a singularly critical determinant of immigrants’ assimilation paths, especially among immigrant children. Segmented assimilation theory posits that immigrants’ assimilation into U.S. society is not a singular path but rather that the context of reception in the United States determines the direction of assimilation (upward, lateral, and downward) (Portes and Rumbaut 2006). Skin tone is one of the factors that set the context of reception because some of the new immigrants experienced culturally different racialization hierarchies in their sending societies. Relative to light-skinned European descendants whose assimilation to the American mainstream was less influenced by phenotypical discrimination, new immigrants and their children often encounter racial barriers to upward mobility (Portes and Zhou 1993; Zhou 1997).

As such, both assimilation theory and segmented assimilation theory contend that the dark skin tone of immigrants is an obstacle to assimilation and upward mobility, largely because a dark skin tone is associated with Black Americans, who are often stuck at the bottom of the racial hierarchy in the United States. Alba and Nee (1997:846), for example, argue that “not dark skin color per se, but the appearance of connection to the African American group raises the most impassable racist barriers in the United States.” In other words, a dark skin tone matters only as long as immigrants’ skin tone is dark enough to be perceived as African American.

However, a dark skin tone penalty is found both within and across ethnoracial groups, beyond whether dark skin tone is categorically connected to an African American phenotype. Even within African American racial groups, dark-skinned members face stronger discrimination in the labor market than lighter-skinned members (Kreisman and Rangel 2015; Monk 2014), and similar results are found among Hispanics (Frank et al. 2010; Mason 2004). Golash-Boza (2006:35) insists that the extent to which Latinxs “fit the Hispanic somatic norm image” of the Indian/mestizo phenotype—a stereotype widely shared among Americans as being hardworking, undocumented, low wage earners—rather than their association with a Black racial phenotype, determines whether they will face racial discrimination instead of ethnic discrimination. Kreisman and Rangel (2015) found a larger wage gap between light-skinned and dark-skinned African Americans than that between Whites and Blacks, suggesting that a dark skin tone penalty results from more complex mechanisms beyond self-identified membership in the Black racial group.

### Downward Mobility of Immigrants With Dark Skin Tone

While both assimilation and segmented assimilation theories put more focus on the intergenerational mobility of subsequent generations rather than intragenerational mobility of first-generation immigrants themselves, segmented assimilation studies suggest that downward mobility at arrival is influenced by skin tone as well. The implicit assumption of the argument that race is one of the main factors determining the context of immigrants’ assimilation process is the premise that immigrants have experienced prejudice and discrimination based on their phenotypes in their home countries in different ways than in the United States. Scholars have argued that immigrants of dark skin tone in particular have to redefine their phenotypic attributes as obstacles to upward mobility in the United States after immigration (Portes and Zhou 1993; Zhou 1997). As a consequence, immigrants with dark skin tone often stress their ethnic identities in order to avoid the subordinate status attached to American Blacks (Bonilla-Silva 1997). For example, Foner (2000:260) noted that “dark-skinned (West) Indian immigrants, whose skin color might put them at risk at being confused with African Americans, emphasize their ethnic identity and distinctive history, customs, and culture as a way to avoid such mistakes.”

Although not limited to immigrants, Kreisman and Rangel’s finding (2015) that the wage gaps between light-skinned Blacks and dark-skinned Blacks increase over time in the National Longitudinal Survey of Youth 1997 data suggests that similar processes may also apply to immigrants. The authors speculate that the cumulative disadvantage for darker-skinned Blacks results from mismatches and job instability due to labor market discrimination. Furthermore, the discrimination is more likely preference-based discrimination against darker-skinned Blacks rather than statistical discrimination for the whole Black racial group because the negative effect of their dark skin tone on wage has not been ameliorated despite their accumulation of experiences over their working careers.

Many immigrants are known to find their first job in coethnic niches, but dark-skinned immigrants are likely less able to enter into the better-paying general labor market. Morales (2008), for example, finds that dark-skinned Latinxs are more likely than light-skinned Latinxs to find employment in coethnic niches. Applying queuing theory, Morales (2008) explained that based on employers’ preference, workers are sorted by skin tone: lighter-skinned workers are preferred in the general labor market, leaving dark-skinned immigrants with fewer chances to be hired in the general labor market (regardless of earnings) and resulting in limitations on upward mobility. In a similar way, residential immobility of immigrants of dark skin tone (e.g., South et al. 2005) may create a job mismatch as well by prohibiting them from finding housing close to better jobs.

From the preceding discussion, I hypothesize the following:

Hypothesis 1: Immigrants with darker skin tone will experience steeper downward mobility at arrival to the United States net of race.

Hypothesis 2: Immigrants with darker skin tone will experience less steep upward trajectories post-immigration net of race.

On the other hand, a dark skin tone penalty at arrival in the United States may not emerge if dark-skinned immigrants already experienced similar penalties in their country of origin. Preferential treatment toward people with lighter skin tones is also found in many countries around the world, including some Asian countries (Glen 2009), Mexico (Campos-Vazquez and Medina-Cortina 2019; Villarreal 2010), and Brazil (Telles 1992). Stratification by skin color—pigmentocracy—is prevalent across many Latin American countries (Bailey et al. 2016; Telles 2014). However, it is difficult to test this hypothesis unless the immigrants under study can be compared with the population of their sending countries and with the U.S. population in order to measure the relative penalty of dark-skinned immigrants across their sending countries and the United States. Thus, in this study, comparing the skin tone effects net of race between the pre-immigration baseline period and post-immigration periods will test the level of dark skin penalty among immigrants.

## Analytic Strategy

I analyze data from the New Immigrant Survey (NIS),4 which surveyed immigrants who obtained LPR in 2003. Jasso et al. (2000) stressed that the NIS was designed to overcome three deficiencies in previous immigrant-related surveys: (1) cross-sectionality, with a lack of pertinent information on individual immigrants’ dynamics; (2) small sample sizes, which limited the number of immigrant groups that could be analyzed; and (3) missing data on crucial variables, such as specific visa categories, in earlier surveys. The NIS data include information on pre-immigration history and are designed as panel data. Such longitudinal information enables researchers to study dynamic aspects of immigration.

Importantly, the NIS 2003 data has an unusually precise measure of skin tone, ranging from 0 to 10, with 0 being lightest and 10 darkest. The Massey and Martin skin color scale was printed in the field interviewer manual, and interviewers were asked to measure respondents’ skin tone after the survey regardless of race (respondents could not see the scale) (Massey and Martin 2003). Skin tone is reported for the 4,652 face-to-face survey respondents (of 8,573 respondents total); phone interview respondents are necessarily excluded.5 The skin tone measure in the NIS has been tested for precision and judged to be both valid and reliable regardless of interviewer’s race or other identities (Hersch 2008: appendix A). Following the previous studies using the NIS data, skin tone in this study is treated as an interval variable.

The main research aim of this study is to evaluate the effects of skin color on immigrants’ occupational trajectories over the immigration process. The dependent variable is immigrants’ occupational status and its trajectory over time. The survey asks respondents’ occupation at three time points: last job held before immigration (Time 1 (T1)),6 first job in the United States (Time 2 (T2)), and current job at interview date (Time 3 (T3)).7 Occupational status is coded with the International Socio-Economic Index (ISEI) of 2008 (Ganzeboom 2010a). The ISEI is a standardized scale of occupations that represents the “weighted sum of mean education and mean income” of incumbents of each occupation, which maximizes indirect effect of education but minimizes its direct effect on earnings (Ganzeboom et al. 1992:12). This index, constructed using data from pooled International Social Survey Programme 2002–2007 waves (200,000 men and women in 42 countries, including the United States) (for more details and a complete list of ISEI, see Ganzeboom 2010a, b), is validated in its high correlation with job skills and occupational mean earnings across countries (Le Grand and Tåhlin 2013). The census 2003 occupation codes in the NIS 2003 are recoded into ISEI 2008.

I analyze ISEI instead of wage/earnings for two reasons. First, some of the countries of origin are grouped into several regions for confidentiality purposes in the NIS 2003 data, which makes it impossible to adjust wages from jobs abroad into comparable U.S. wages based on international currency rates. Second, ISEI has strengths over the wage/earnings variable in that ISEI is based not only on earnings for each occupation but also on education level for each occupation so as to capture one’s relatively stable socioeconomic status rather than potentially transitory income status. Hence, ISEI is more stable and comparable across time and space, and thus it is a better measurement for international comparison (Hout and DiPrete 2006; Treiman 1977). For these reasons, ISEI was used in many previous studies to examine pre- and post-immigration mobility (e.g., Akresh 2008).8

Because not all respondents whose skin tone is reported are employed at all three time points, the analysis is limited to those whose skin tone is reported and who were employed at each period. This selection rule makes the current study comparable with previous research that used the same data set and limited the sample to respondents who were employed (Akresh 2008; Frank et al. 2010; Hersch 2008). Although excluding respondents who were not employed in paid work may be a source of selection bias, I show below that such biases are quite small.

Self-identified race is controlled and is interacted with time in separate models in order to examine the relative role of skin color and race/ethnicity. The NIS asks whether respondents are Hispanic or not regardless of race. A relatively large proportion of respondents self-identify as White (53%); this will be discussed further. Asians make up 26% of the sample, compared with 11% for Blacks and 4% for Native Americans and Pacific Islanders. To avoid bias from missing on the race variable, missing data on race/ethnicity is also controlled for. Hispanics constitute 38% of the sample.

Variables that may influence occupational status and immigrants’ assimilation are additionally controlled: demographics of gender and age, human/social/cultural capital, visa type, country of origin,9 U.S. experiences, and regions of U.S. residence. Definitions and measurements are provided in Table A1 in the online appendix.

Table 1 summarizes the descriptive statistics of the variables. I construct the data as person-time longitudinal data. The total number is 8,159 person-time observations for the sample whose skin tone is reported and who were employed at each time point, excluding those whose age is missing and those not working in the United States.10 The mean ISEI across the three time points is 39.10 (e.g., machinery mechanics and repairers). The mean transition between T1 and T2 is –8.17 (e.g., stonemason), and that between T2 and T3 is 2.48 (e.g., building trade workers). These means clearly show that immigrants experience downward mobility with immigration and then recover their occupational status over time. The average trajectory follows a U-shape, as documented in the literature.

## Analysis

Figure 1 describes the mean ISEI scores by skin tone at three time points and clearly shows that occupational status is stratified by skin tone: immigrants with light skin have higher occupational status than those with medium and dark skin tone during and after immigration, all follow U-shape trajectories. However, the depth of the U-shape varies by skin tone. Immigrants with dark skin tone experience steeper downward mobility and less steep upward mobility after immigration than immigrants with lighter skin tone.

To examine the net effect of skin tone on occupational status at three time points, I use a generalized least square (GLS) random-effects model, which adjusts for correlations among observations and heteroscedasticity.11 Fixed-effects models are often applied to panel data in order to capture the net effects of time-varying variables on outcome variables while controlling for both observed and unobserved heterogeneities across entities. I apply random-effects models here because the main research interest is the effect of skin tone, a time-constant variable, as in Eq. (1):
$yit=μt+αi+βSkinTonei+γTime2i+δTime3i+ηSkinTone×Time2i+θSkinTone×Time3i+λXi+εit,$
1
where y = ISEI score, i = individual, t = time point, Xi = a set of time-constant control variables, μt = an intercept that may be different for each period, and εit = individual and time-specific error term.12 In random-effects models, αi is assumed to be a set of random variables that are normally distributed, have constant variance, and are independent of all other independent variables. Whereas αi is controlled for in fixed-effects models, it creates a random intercept combined with μt but not controlled for in Eq. (1).13 In addition to the coefficient β of the main independent variable (skin tone), time dummy variables for T2 and T3 are estimated to measure the mean of the time-specific effects across individuals relative to those at T1. Interaction terms of skin tone with T2 and T3 are estimated to capture how the effect of skin tone varies across time.

Results show that immigrants with dark skin tone are likely to have lower occupational status at all three time points. Table 2 summarizes the coefficients of the GLS random-effects model predicting occupational status. Model 1 shows that skin tone that is one unit darker is associated with 0.88 lower occupational status, on average, across the three time points. Immigrants experience a steep downward mobility at T2, the first job in the United States: occupational status is 8.77 points less at T2 than T1, the last job abroad. Then, immigrants catch up in occupational status by 2.43 points at T3.

Time and skin tone interaction terms are added in Model 2, which shows that the dark skin tone penalty is stronger at T2 and T3, after immigration to United States, relative to before immigration. Having skin tone that is one unit darker additionally decreases occupational status by 0.44 points at T2. That is, on average, immigrants experience downward mobility after immigration to the United States at T2 by 6.99 points, and immigrants with one scale darker skin tone experience 0.44 points additional downward mobility. Immigrants with the darkest skin tone experience downward mobility by 0.44 × 10 = 4.4 more points at T2 than those with the lightest skin tone. For example, in service and sales occupations, an immigrant who worked as a transport conductor (ISEI = 40) in the country of origin is likely to have a first job in the United States as a cleaning and housekeeping supervisor in offices/hotels (ISEI = 33) but is likely to have a lower level occupation, such as waiter (ISEI = 28) if he or she has the darkest skin tone.

The dark skin tone penalty in the United States diminishes slightly but persists at T3. The interaction effect of dark skin tone at T3 is –0.34, which means that immigrants with skin tone that is one scale unit darker have a 0.34 point lower occupational status at T3 in addition to the average downward mobility experienced by immigrants relative to T1. This coefficient is slightly less than –0.44 at T2 but still larger than before immigration (at T1) and is statistically significant. Controlling for additional covariates in Model 3 yields the similar dark skin tone penalty at T2 and T3, and the results are robust to controlling for race in Model 4.

Employment status is also a critical measure of labor market outcomes because unemployment can be an extreme example of downward mobility. However, the NIS 2003 data contain respondents’ employment status in detail only at T3: employed, unemployed and looking for work, temporarily laid off/on sick or other leave, disabled, retired, or a homemaker.14 At T1 and T2, the survey asked about a respondent’s job only if they ever worked for pay. Thus, by necessity, I exclude nonworking individuals from this analysis. As a sensitivity test for resulting bias, I examine whether dark skin tone is associated with nonworking status, including but not limited to unemployment among respondents who have valid skin color information and were in the labor force. Similar to Monk (2014), I find no evidence for association between dark skin tone and nonworking status at any of the three time points.15 Nor do I find an association of skin tone with unemployment when I limit the analysis to T3, for which detailed nonworking status is specified, and exclude those aged 65 and older, most of whom are likely retired.

Next, to further examine the relative role of skin tone and race, I model race and interacted it with time. The results are provided in Models 5–7 in Table 2. Hispanics, on average, are likely to have 11.12 points lower occupational status than non-Hispanics across three time points (Model 5). In addition, compared with Whites, Asians have 3.54 higher occupational status, whereas Blacks have 4.99 lower status across all time points. When race is interacted with T2 and T3 in Model 6, Hispanics have 1.23 points and 3.13 points higher occupational status, respectively, relative to non-Hispanic immigrants, but the interaction is statistically significant only at T3. Despite the higher occupational status at T3 than at T1, Hispanics have an occupational status that is 12.5 points lower than non-Hispanics at T1. Similarly, Asians have 4.27 and 3.28 points higher occupational status at T2 and T3 relative to White immigrants. Considering the higher occupational status of Asians relative to White immigrants at T1, Asians continue to maintain higher status. On the other hand, Blacks have an occupational status that is 4.39 and 4.02 points lower at T2 and T3, respectively. Results suggest that Hispanics and Asians experience upward mobility over time at T2 and T3 relative to their reference groups, whereas Blacks continue to remain in lower occupational status over time. A similar pattern is found when additional covariates are controlled for in Model 7.16

Finally in Model 8 (Table 2), both race and skin tone are interacted with T2 and T3. The interaction effects of race with time are similar to those in Model 7, where skin tone and time interactions are not included. The skin tone and time interaction effects decrease to –0.11 and –1.10 at T2 and T3, respectively, but become statistically nonsignificant in Model 8. However, this does not mean that there is no additional skin tone effect on immigrants’ occupational mobility once race interaction effects are also controlled for. Considering the stronger dark skin tone penalty at T2 and T3 (interaction effects) net of race (not interacted with time) in Model 4, one plausible interpretation is that skin tone is a strong indicator of race and that the inclusion of a race interaction absorbs the variance of occupational status associated with skin tone. A supplementary analysis of ordinary least squares (OLS) regression at each time point separately (not shown here) showing a dark skin tone penalty at T2 and T3 net of race also supports this interpretation.

However, the extent to which skin tone serves as a signal for race differs across ethnoracial groups. Table 3, which summarizes coefficients of skin tone and interaction terms with time for each subsample of ethnoracial group, shows that within-group dark skin tone penalties appear for Hispanics when covariates are controlled as well as for Whites when covariates are not controlled.17 Because Hispanics can be any race, negative coefficients of skin tone in both Models 1 and 2 suggest that skin tone is an indicator of race among Hispanics, although there is no time interaction effect. Interestingly, an even stronger negative effect (–2.11) of darker skin tone appears among Whites in Model 3 than among Hispanics (–0.83) in Model 1, which suggests that discrepancy between self-identified White race and their perceived dark skin tone is larger among Whites than among Hispanics (and Blacks and Asians).

It is worth noting that interaction effects of skin tone with time remain negative for Blacks and Asians (and White at T2), although they are not statistically significant. This loss of significance results, at least in part, from the reduction in statistical power when the sample is stratified by ethnoracial groups. Inclusion of interaction terms between race and skin tone (not shown here) in Models 3 and 4 in Table 2 (pooled sample) does not change the interaction effects of skin tone with T2 and T3, suggesting that the within-race dark skin tone penalty likely exists, although the statistical power is decreased in subsamples. In sum, results show that skin tone not only serves as an indicator of perceived race but also creates inequality within self-identified race.

The dark skin tone penalty in the immigration process discussed so far is summarized in Fig. 2, which shows the means of predicted ISEI for the sample by skin tone and time after each individual is fitted to OLS regressions at each time point with all covariates controlled. Even before immigration, immigrants with darker skin tone are predicted to have lower occupational status. The association, however, is not linear: immigrants with darker skin tone had higher occupational status than those with medium skin tone at T1. Indian immigrants and highly selective African immigrants with very dark skin color belong to this group. After immigration at T2 and at T3, immigrants of all skin tone scales have lower occupational status than in their home country. At this time, however, the relationship between skin tone and occupational status is linear: immigrants with darker skin tone have lower occupational status than those with lighter skin tone. Thus, immigrants with the darkest skin tone are expected to experience the most downward mobility and to have a slower assimilation process after immigration.18

## Discussion and Conclusion

Because of the lack of available data, previous empirical research using large-scale survey data has examined mainly the dark skin tone penalty for immigrants cross-sectionally in the United States, failing to examine the influence of skin tone during the immigration process and the post-immigration assimilation process. In this study, using the NIS 2003 data, which measured both immigrants’ occupational history including pre-immigration jobs and their skin tone, I examine the effects of skin tone and race on immigrants’ occupational trajectories, including the transition from their home country to the U.S. labor market.

Consistent with Hypothesis 1, I find that immigrants whose skin tone is darker are more penalized in the process of migration to the United States by experiencing steeper downward occupational mobility relative to those whose skin tone is lighter. Although some scholars find a U-shape pattern of immigrants’ occupational mobility trajectory (Akresh 2008; Chiswick 1978; Chiswick and Miller 2009), they focus mainly on human capital aspects without incorporating discrimination factors caused by phenotypic attributes, such as skin tone. However, the current study suggests that skin tone and race influence the skill-transferability processes of immigrants. Steeper downward mobility of darker-skinned immigrants may imply that immigrants begin to face discrimination based on their skin tone upon arriving to the United States but they had not experienced skin tone–based discrimination, or experienced it to a lesser degree, in their home country. Hispanics and Asians are likely to experience upward mobility after immigration, whereas Blacks continue to remain at a lower occupational status than White immigrants. These findings support previous assimilation and segmented assimilation studies suggesting that phenotypic attributes, such as skin tone and race, set the context of reception for immigrants in the United States and thereby compel immigrants to redefine the meaning of their phenotypic attributes in a new cultural stratification system (Alba and Nee 1997; Gans 1992; Portes and Rumbaut 2006; Portes and Zhou 1993; Zhou 1997).

Furthermore, my results are consistent with Hypotheses 2, which predicts that immigrants with darker skin tone will experience less rapid upward trajectories over post-immigration time: the dark skin tone penalties in the U.S. labor market do not diminish over time among immigrants even as they develop skills and accumulate work experiences in the United States, resulting in a lopsided U-shape pattern. This finding challenges assimilation theory’s prediction that phenotypic attributes are not impassible obstacles for immigrants in the long run even if they do slow the pace of assimilation (Alba and Nee 1997, 2003). Instead, this finding is consistent with Hersch’s study (2011) (and with segmented assimilation theory), which found that the dark skin tone penalty persists over time among immigrant spouses of the respondents in the NIS 2003 data, whose duration of residence in the United States is more heterogeneous than that of the primary respondents. This conclusion may be premature because of the short duration of observation in the NIS 2003 sample. A longer period of observation may answer the question more clearly. However, because dark skin penalties extend even to intergenerational mobility (Campos-Vazquez and Medina-Cortina 2019; Chetty et al. 2018), we may expect the initial penalties for dark-skinned immigrants at arrival to the United States to continue for a longer period.

Assimilation theory predicts a declining impact of skin tone in that even immigrants with dark skin tone, such as South Asians, overcome the obstacles they encounter. However, the opposite may also be true because there is no reason to expect employers’ skin tone preference to change with immigrants’ length of time in the United States, especially if such a preference is based on their biases (Kreisman and Rangel 2015). The optimistic prediction of assimilation theory stems from an emphasis on the behavior of immigrants rather than that of employers.

Although immigrants’ cultural or behavioral dimensions are not incorporated in the current study, considering that immigrants are more likely able and motivated workers (Chiswick 1978), the observed dark skin tone penalty in this study may be underestimated. Immigrants may try to overcompensate for their minority status but inevitably face some degree of dark skin penalties from employers. If immigrants have levels of human capital and motivation that are comparable to those of the American population generally, they may have experienced harsher dark skin penalties in the U.S. labor market than observed in this study. Thus, it will be worth examining further how employers’ conscious and unconscious biases work toward immigrants’ skin tone over the employment period.

The findings from this study will expand the discussion of the role of skin tone in the racial identification process in the future. The dark skin tone penalty findings imply that self-identified race alone may not be a precise proxy for immigrant racial group membership. In an additional analysis for each subsample of ethnoracial groups in Table 3, I find a dark skin tone penalty for Whites and Hispanics but not for Asians, and I even find a positive effect of dark skin tone among Blacks. These results may be due to the possible discrepancy between how immigrants’ race is perceived and categorized in the United States depending on their skin tone and how immigrants identify their own racial category. Frank et al. (2010), using the same data but limiting their sample to Latinxs, found that Latinxs tend to identify themselves as White rather than non-White or “other.” Darity et al. (2005) also pointed out that Latinxs, even those with very dark skin tone, disproportionately prefer to identify themselves as White. As a consequence, although most Latinxs identify as White, dark-skinned Latinx immigrants encounter a wage penalty in the labor market (Frank et al. 2010; Rosenblum et al. 2016).

Thus, using racial self-identity only as a proxy for how others may treat individuals based on their race is problematic in survey data. Using skin tone data (as identified by the interviewer) in conjunction with self-identified race may be a way to better calibrate how racialized outcomes are measured (Bailey et al. 2016). Roth (2010) conceptualizes multiple dimensions of racial identity, emphasizing that how others perceive one’s race—rather than one’s self-perception—is central to discrimination. In this process, the subcategory of skin tone plays a more critical role in understanding and constructing interactions than one’s racial identity given that discrimination varies according to the extent to which “individuals are perceived to fit a particular category” (Monk 2015:406). Similarly, Kreisman and Rangel (2015) suspect that the perceived differences by skin tone in interactions, rather than the categorical classification of race, create the earnings gap among African Americans. This is not limited to immigrants with dark skin tone. Maghbouleh (2017), for example, documents how even groups categorically defined as a White racial group—specifically, Iranian Americans—face discrimination in daily life interactions. On the other hand, observers can “whiten” immigrants’ race relative to their self-identified race (Saperstein 2006). Thus, future studies should examine how dark skin phenotype interacts with other dimensions of race in differing social contexts to create different meanings of race in American society.19

## Acknowledgments

The author thanks Jennifer Lundquist, Donald Tomaskovic-Devey, and David Cort for their support and generous comments on this article. Thanks also to the anonymous reviewers and the editors for thoughtful comments and suggestions.

## Notes

1

On the other hand, some scholars point out that the United States and Latin America are not much different in terms of the construction and understanding of race because complexion, rather than lineage, centers in racial identification processes in both regions (Goldsmith et al. 2006).

2

The National Survey of Black Americans 1979–1980, the 1979 Chicano National Survey, the 1990 Latino National Political Survey, and the National Survey of American Life 2001–2003 are the major national-level surveys that measure respondents’ skin color and race. The Multi-City Study of Urban Inequality 1992 and the Detroit Area Study 1995 data are the commonly used regional studies. Data on skin tone are also collected in health-focused surveys, such as the National Heart, Lung, and Blood Institute’s Coronary Artery Risk Development in Young Adults.

3

The National Longitudinal Survey of Youth 1997 and the Add Health survey are among the few longitudinal data sets that measure respondents’ skin tone. However, they contain limited immigration-related information.

4

The NIS second wave data were publicly released but are not included in this analysis. Because of high attrition rates of the sample in the second wave, inclusion of the second wave data in the analysis reduces the sample size to approximately one-half of the original sample. A thorough analysis of sample selection in the second wave is underway.

5

Because dropping samples whose skin tone is not reported and those without a job at each time point complicates applying sampling weights, the analyses here are unweighted, following previous research using the same data set (e.g., Frank et al. 2010; Rosenblum et al. 2016).

6

The missing observations in the last job held before immigration were imputed using their first job held before immigration.

7

One limitation of this analytic frame is that time spans between T1, T2, and T3 are inconsistent across the sample. However, controlling for age, U.S. labor market experience, and whether they achieved LPR while residing in the United States mitigates this problem.

8

Despite the strength of the ISEI, using an occupational index as a proxy measure of labor market status in the United States may also have limitations: both within-occupation and between-occupation wage inequality constitute a considerable portion of total wage variance (Avent-Holt and Tomaskovic-Devey 2014; Kim and Sakamoto 2008), and within-occupation bias in job sorting can result.

9

Although some may suggest a subgroup analysis by sending regions, such an analysis is beyond the focus of the current study on the destination reception. Rosenblum et al. (2016) have discussed this issue, although cross-sectionally.

10

For the sample construction, see Table A2, online appendix.

11

As a robustness check of the GLS random-effects model, I also analyze the effects of skin tone at each time point separately using OLS regression and find that it yields results similar to the current analysis.

12

Here I assume that skin tone is time-constant. However, it also should be acknowledged that skin tone may change. For example, construction workers who tend to work outdoors may have darker skin tone than their original tone (Hersch 2008), and some people intentionally bleach their skin (Glen 2009).

13

A trade-off exists between random-effects models and fixed-effects models: random-effects models risk omitted variable bias by assuming that unobserved attributes are independent of observed variables, but they have a higher efficiency than fixed-effect models. In addition, fixed effects include only estimates for measures that vary over time, excluding time-invariant cases and variables (Allison 2009). Although the result of the Hausman test indicates that the coefficients in the two models are different at p < .05, results from a fixed-effects specification show patterns that are quite similar to the current random-effects model. The fixed-effect result is available upon request.

14

Treating all nonworking states as unemployment can cause bias. Of the full sample of 8,573 observations (including those without a skin tone measure), 58.3% are employed, and 16.4% are unemployed, constituting 39.3% of total nonworking individuals at T3. The majority (60.7%) of the nonworking sample are retirees, homemakers, disabled, other, on leave, or temporally laid off.

15

The results from the balanced panel (sample members with a job at all three time points only) are broadly same as those from the current sample in the magnitude of the coefficients. The only difference is that skin tone × T3 interaction effects become marginally statistically significant at p < .10, which is likely due to the reduced sample size.

16

R2 in models that fit race is larger than in models fitting skin tone. R2 is .18 in Model 6, where race/ethnicity and its interaction with time are included, compared with .06 in Model 2, where skin tone and its interaction with time are included. This suggests that the explanatory power of race and ethnicity is larger than that of skin tone. However, the difference in R2 may be due to interval versus categorical variable differences in explanatory power. R2 is not different between Model 3, where skin tone is fitted with additional control for covariates (.43), and in Model 7, where race/ethnicity is fitted instead (.43). An additional control for race/ethnicity in Model 4 does not change R2 from Model 3, where only skin tone and its time interaction are fitted. These results suggest that self-identified race and skin tone are two dimensions of race, as discussed earlier.

17

Similarly, the model can be stratified by gender given that skin tone influences may be different for men and women. However, gendered immigration assimilation processes are complex because skin tone effects are compounded with other factors, such as visa type (e.g., a spouse of U.S. citizen visa would be granted more to females), and thus deserve a separate study.

18

Skin tone has a curvilinear effect on occupational mobility: the negative skin tone × Time2 and skin tone × Time3 interaction effects are stronger among darker-skinned immigrants.

19

One limitation of this study is that undocumented immigrants are not included in the analyses. The majority of undocumented immigrants are from Mexico and Latin America, having emerged as a racialized class in the United States (Massey and Pren 2012). I speculate that including them in the analyses would not change the results significantly. They are likely to have held lower occupational status even before immigration because of their relatively low human capital, and the dark skin tone penalty in the United States relative to pre-immigration is less salient. Nevertheless, the skin tone effect for undocumented immigrants’ assimilation process is worth further research.

## References

Akresh, I. R. (
2008
).
Occupational trajectories of legal US immigrants: Downgrading and recovery
.
Population and Development Review
,
34
,
434
456
.
Alba, R., & Nee, V. (
1997
).
Rethinking assimilation theory for a new era of immigration
.
International Migration Review
,
31
,
826
874
.
Alba, R., & Nee, V. (
2003
).
Remaking the American mainstream: Assimilation and contemporary immigration
.
Cambridge, MA
:
Harvard University Press
.
Allison, P. D. (
2009
).
Fixed effects regression models
.
Los Angeles, CA
:
Sage Publishing
.
Avent-Holt, D., & Tomaskovic-Devey, D. (
2014
).
A relational theory of earnings inequality
.
American Behavioral Scientist
,
58
,
379
399
.
Bailey, S. R., Fialho, F. M., & Penner, A. M. (
2016
).
Interrogating race: Color, racial categories, and class across the Americas
.
American Behavioral Scientist
,
60
,
538
555
.
Bond, S., & Cash, T. F. (
1992
).
Black beauty: Skin color and body images among African American college women
.
Journal of Applied Social Psychology
,
22
,
874
888
.
Bonilla-Silva, E. (
1997
).
Rethinking racism: Toward a structural interpretation
.
American Sociological Review
,
62
,
465
480
.
Bonilla-Silva, E., & Dietrich, D. R. (
2009
).
The Latin Americanization of U.S. race relations: A new pigmentocracy
. In E. N Glenn (Ed.),
Shades of difference: Why skin color matters
(pp.
40
60
).
Stanford, CA
:
Stanford University Press
.
Campos-Vazquez, R. M., & Medina-Cortina, E. M. (
2019
).
Skin color and social mobility: Evidence from Mexico
.
Demography
,
56
,
321
343
.
Chetty, R., Hendren, N., Jones, M. R., & Porter, S. R. (
2018
).
Race and economic opportunity in the United States: An intergenerational perspective
(NBER Working Paper No. 24441).
Washington, DC
:
National Bureau of Economic Research
.
Chiswick, B. R. (
1978
).
The effect of Americanization on the earnings of foreign-born men
.
Journal of Political Economy
,
86
,
897
922
.
Chiswick, B. R., & Miller, P. W. (
2009
).
The international transferability of immigrants’ human capital
.
Economics of Education Review
,
28
,
162
169
.
Darity, W. A.Jr, Dietrich, J., & Hamilton, D. (
2005
).
Bleach in the rainbow: Latin ethnicity and preference for Whiteness
.
Transforming Anthropology
,
13
,
103
109
.
Espino, R., & Franz, M. M. (
2002
).
Latino phenotypic discrimination revisited: The impact of skin color on occupational status
.
Social Science Quarterly
,
83
,
612
623
.
Fergus, E. (
2009
).
Understanding Latino students’ schooling experiences: The relevance of skin color among Mexican and Puerto Rican high school students
.
Teachers College Record
,
111
,
339
375
.
Foner, N. (
2000
).
Beyond the melting pot three decades later: Recent immigrants and New York’s new ethnic mixture
.
International Migration Review
,
34
,
255
262
.
Frank, R., Akresh, I. R., & Lu, B. (
2010
).
Latino immigrants and the U.S. racial order: How and where do they fit in?
.
American Sociological Review
,
75
,
378
401
.
Gans, H. G. (
1992
).
Second-generation decline: Scenarios for the economic and ethnic futures of the post-1965 American immigrants
.
Ethnic and Racial Studies
,
15
,
173
192
.
Gans, H. G. (
2012
).
“Whitening” and the changing American racial hierarchy
.
Du Bois Review
,
9
,
267
279
.
Ganzeboom, H. B. G. (
2010a
).
International standard classification of occupations
Ganzeboom, H. B. G. (
2010b
,
May
).
A new International Socio-Economic Index [ISEI] of occupational status for the international standard classification of occupation 2008 [Isco-08] constructed with data from the ISSP 2002–2007
.
Paper presented at the annual conference of International Social Survey Programme
,
Lisbon, Portugal
.
Ganzeboom, H. B. G., De Graaf, P. M., & Treiman, D. J. (
1992
).
A standard international socio-economic index of occupational status
.
Social Science Research
,
21
,
1
56
.
Glen, E. N. (
2009
).
Shades of difference: Why skin color matters
.
Stanford, CA
:
Stanford University Press
.
Golash-Boza, T. (
2006
).
Dropping the hyphen? Becoming Latino(a)-American through racialized assimilation
.
Social Forces
,
85
,
27
55
.
Goldsmith, A. H., Hamilton, D., & Darity, W.Jr (
2006
).
Shades of discrimination: Skin-tone and wages
.
American Economic Review: Papers & Proceedings
,
96
,
242
245
.
Hagiwara, N., Kashy, D. A., & Cesario, J. (
2012
).
The independent effects of skin tone and facial features on Whites’ affective reactions to Blacks
.
Journal of Experimental Social Psychology
,
48
,
892
898
.
Hamilton, D., Goldsmith, A. H., & Darity, W.Jr (
2009
).
Shedding “light” on marriage: The influence of skin shade on marriage for Black females
.
Journal of Economic Behavior & Organization
,
72
,
30
50
.
Hersch, J. (
2008
).
Profiling the new immigrant worker: The effects of skin color and height
.
Journal of Labor Economics
,
26
,
345
386
.
Hersch, J. (
2011
).
The persistence of skin color discrimination for immigrants
.
Social Science Research
,
40
,
1337
1349
.
Hochschild, J. L. (
2005
).
Looking ahead: Racial trends in the United States
.
Daedalus
,
134
(
1
),
70
81
.
Hout, M., & DiPrete, T. A. (
2006
).
What we have learned: RC 28’s contributions to knowledge about social stratification
.
Research in Social Stratification and Mobility
,
24
,
1
20
.
Hughes, M., & Hertel, B. R. (
1990
).
The significance of color remains: A study of life chances, mate selection, and ethnic consciousness among Black Americans
.
Social Forces
,
68
,
1105
1120
.
Hunter, M. (
2007
).
The persistent problem of colorism: Skin tone, status, and inequality
.
Sociology Compass
,
1
,
237
254
.
Hunter, M. L. (
1998
).
Colorstruck: Skin color stratification in the lives of African American women
.
Sociological Inquiry
,
68
,
517
535
.
Jasso, G., Massey, D. S., Rosenzweig, M. R., & Smith, J. P. (
2000
).
The New Immigrant Survey Pilot (NIS-P): Overview and new findings about U.S. Legal immigrants at admission
.
Demography
,
37
,
127
138
.
Keith, V. M., & Herring, C. (
1991
).
Skin tone and stratification in the Black community
.
American Journal of Sociology
,
97
,
760
778
.
Kim, C., & Sakamoto, A. (
2008
).
The rise of intra-occupational wage inequality in the United States, 1983–2002
.
American Sociological Review
,
73
,
129
157
.
Kreisman, D., & Rangel, M. A. (
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
13
.
Lee, J., & Bean, F. D. (
2007
).
Reinventing the color line: Immigration and America’s new racial/ethnic divide
.
Social Forces
,
86
,
561
586
.
Le Grand, C., & Tåhlin, M. (
2013
).
Class, occupation, wages, and skills: The iron law of labor market inequality
. In G. E. Birkelund (Ed.),
Class and stratification analysis
(Comparative Social Research Vol. 30, pp.
3
46
).
Bingley, UK
:
Emerald Group Publishing
.
Maghbouleh, N. (
2017
).
The limits of Whiteness: Iranian Americans and the everyday politics of race
.
Stanford, CA
:
Stanford University Press
.
Mason, P. L. (
2004
).
Annual income, hourly wages, and identity among Mexican-Americans and other Latinos
.
Industrial Relations
,
43
,
817
834
.
Massey, D. S., & Martin, J. A. (
2003
).
The NIS Skin Color Scale
.
Princeton, NY
:
Office of Population Research, Princeton University
.
Massey, D. S., & Pren, K. A. (
2012
).
Origins of the new Latino underclass
.
Race and Social Problems
,
4
,
5
17
.
Monk, E. P.Jr (
2014
).
Skin tone stratification among Black Americans, 2001–2003
.
Social Forces
,
92
,
1313
1337
.
Monk, E. P.Jr (
2015
).
The cost of color: Skin color, discrimination, and health among African-Americans
.
American Journal of Sociology
,
121
,
396
444
.
Morales, M. C. (
2008
).
The ethnic niche as an economic pathway for the dark skinned labor market incorporation of Latina/o workers
.
Hispanic Journal of Behavioral Sciences
,
20
,
280
298
.
Portes, A., & Rumbaut, R. G. (
2006
).
Immigrant America: A portrait
(3rd ed.).
Berkeley
:
University of California Press
.
Portes, A., & Zhou, M. (
1993
).
The new second generation: Segmented assimilation and its variants
.
Annals of the American Academy of Political and Social Science
,
530
,
74
96
.
Rosenblum, A., Darity, D.Jr, Harris, A. L., & Hamilton, T. G. (
2016
).
Looking through the shades: The effect of skin color on earnings by region of birth and race for immigrants to the United States
.
Sociology of Race and Ethnicity
,
2
,
87
105
.
Roth, W. D. (
2010
).
Racial mismatch: The divergence between form and function in data for monitoring racial discrimination of Hispanics
.
Social Science Quarterly
,
91
,
1288
1311
.
Saperstein, A. (
2006
).
Double-checking the race box: Examining inconsistency between survey measures of observed and self-reported race
.
Social Forces
,
85
,
57
74
.
Saperstein, A., Penner, A. M., & Light, R. (
2013
).
Racial formation in perspective: Connecting individuals, institutions, and power relations
.
Annual Review of Sociology
,
39
,
359
378
.
South, S. J., Crowder, K., & Chavez, E. (
2005
).
Migration and spatial assimilation among U.S. Latinos: Classical versus segmented trajectories
.
Demography
,
42
,
497
521
.
Telles, E. E. (
1992
).
Residential segregation by skin color in Brazil
.
American Sociological Review
,
57
,
186
197
.
Telles, E. E. (
2014
).
Pigmentocracies: Ethnicity, race, and color in Latin America
.
Chapel Hill
:
University of North Carolina Press
.
Treiman, D. J. (
1977
).
Occupational prestige in comparative perspective
.
New York, NY
:
.
Villarreal, A. (
2010
).
Stratification by skin color in contemporary Mexico
.
American Sociological Review
,
75
,
652
678
.
Waters, M. C. (
1999
).
Black identities: West Indian immigrant dreams and American realities
.
New York, NY
:
Russell Sage Foundation
.
Wimmer, A. (
2008
).
The making and unmaking of ethnic boundaries: A multilevel process theory
.
American Journal of Sociology
,
113
,
970
1022
.
Zhou, M. (
1997
).
Growing up American: The challenge confronting immigrant children and children of immigrants
.
Annual Review of Sociology
,
23
,
63
95
.

## Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.