Abstract
A contested racial identity refers to incongruence between personal racial identification and external racial categorization. For example, an individual may self-identify as White, but be perceived by most others as non-White. Documenting racial contestation is important because racialized experiences are shaped not only by the racial classification that individuals claim for themselves but also the external racial attributions placed on them by others. Focusing solely on monoracial identifying adults, this study answers three key questions about racial contestation: (a) How common is it? (b) Who is most likely to report experiencing it? and (c) How is it related to aspects of racial identity such as racial awareness, racial group closeness, and racial identity salience? Employing the 2006 Portraits of American Life Study, results suggest that reports of racial contestation among monoracial identifying adults are more common than some studies suggest (6% to 14%)—particularly among the fastest growing racial groups in the United States, including Latina/os and Asians—and that experiences of racial contestation are often associated with immigrant generation, ancestry, and phenotypical characteristics. Ordinal logistic regression analyses indicate that individuals who report experiencing racial contestation are no more aware of race in everyday life than other U.S. adults, but they feel less close to other members of the self-identified racial group and report lower levels of racial identity salience than their noncontested counterparts. These results point to a thinning of racial identity among the racially contested.
Race is multidimensional; experienced not only in accord with how one self-identifies but also in relation to how one is perceived by others. In contrast to this well-established understanding, the vast majority of survey-based sociological research employs only unidimensional measures of racial classification, often self-reported by the respondent. Consequently, some sociologists have called for a more dynamic approach that better measures the multidimensionality of racialized experiences (Flores-González, Aranda, & Vaquera, 2014; Irizarry, 2015; Roth, 2010; Zuberi, 2001; Zuberi & Bonilla-Silva, 2008).
In an attempt to answer these calls, a recent and rapidly growing set of survey-based sociological studies employ multidimensional measures of racial classification and illustrate that individuals’ racial self-classification does not always match how they are perceived racially by others (Bailey, Loveman, & Muniz, 2013; Campbell & Troyer, 2007, 2011; Garcia, Sanchez, Vargas, Ybarra, & Sanchez-Youngman, in press; Penner & Saperstein, 2008; Rastogi, Liebler, & Noon, 2016; Saperstein, 2012; Saperstein & Penner, 2010, 2012, 2014; Stepanikova, 2010; Telles, 2002; N. Vargas, 2014, 2015; E. D. Vargas, Sanchez, & Kinlock, 2014; Veenstra, 2011)—a phenomenon that has been referred to as “racial contestation,” “racial misclassification,” and “racial mismatch.” On the whole, this literature is still emerging and many key questions remain unanswered. Perhaps one of the most basic questions is how common is racial contestation? Because survey research has been hampered by data limitations, it may have underestimated how commonly people report experiencing it. Most studies have examined rates of racial contestation only in limited geographic regions (Veenstra, 2011), only for specific national origin or racial/ethnic groups (i.e., Native Americans, Whites, or Blacks; Campbell & Troyer, 2007; Saperstein, 2006; Saperstein & Penner, 2010, 2012), and tend to rely heavily on an operationalization that could be influenced by survey bias (see Campbell & Troyer, 2011).
Substantively, survey research on racial contestation has also been limited. To date, it has focused primarily on issues related to stress and mental health (Campbell & Troyer, 2007, 2011; Stepanikova, 2010; Veenstra, 2011), and has yet to consider how perceived experiences of racial contestation might influence key elements of racial identity (but see Saperstein & Penner, 2014). As variants of identity theory propose, it is plausible that as a response to experiencing identity contestation, individuals become less committed to their identity (Burke & Stets, 2009). This could have important implications for the U.S. racial landscape. If racial contestation is associated with maintaining thinner (or less personally meaningful) racial/ethnic identities (Cornell & Hartmann, 2007), and if it becomes more common over time with an increase in multiracial families and ensuing phenotypical ambiguity (Gans, 2012), this could pose challenges for movements aimed at alleviating racial and ethnic inequalities. Such movements often coalesce around thick racial identities and ethnoracial unity to air grievances and accomplish goals (Nagel, 1995; Simon et al., 1998). Yet most survey-based studies have not considered how racial contestation could potentially have implications for these elements of racial identity.
In this study, we utilize a nationally representative data set in an attempt to fill some of these significant gaps in the literature. We first document the percentage of monoracial identifying White, Black, Latina/o, and Asian adults who report experiencing racial contestation, and analyze the various phenotypical, ancestral, and social factors associated with those experiences. 1 We then build on scholarship on the malleability of racial identity to examine if individuals who experience racial contestation value and perceive elements of their racial identities in distinct ways. Specifically, we examine if the experience of racial contestation is significantly associated with racial identity awareness, perceived racial group closeness, and racial identity salience.
Measuring Racial Contestation
The first aim of this study is to provide descriptive data on racial contestation. Recent studies suggest that it may be fairly rare in the United States. Using the General Social Survey, an early study compared interviewer observed race with respondents’ self-reports of racial classification and found that racial incongruence occurs for approximately 1% of self-identified Whites and 3% of self-identified Blacks (Saperstein, 2006). Studies of racial contestation have relied heavily on this interviewer–respondent measure of incongruence. However, it is plausible that interviewers serve as nonrepresentative proxies for an external review of the respondent (Hitlin, Brown, & Elder, 2007). Interviewers are often attentive and trained to conduct social scientific research. Over the course of an interview, they are privy to extensive background information about respondents. For example, in multiple studies, Saperstein and Penner (2010, 2012) found that interviewers’ classifications of respondents’ race varied systematically by other measures of social status learned during the interview. Utilizing the National Longitudinal Survey of Youth (NLSY79), they found that regardless of how respondents were coded by interviewers in prior waves of the study, interviewers were more likely to code respondents as Black after they learned about experiences of incarceration (Saperstein & Penner, 2010). Other aspects of social status learned throughout the course of the interview, such as bouts of unemployment or poverty, also tend to be associated with how interviewers racially code respondents (Penner & Saperstein, 2008). Yet in most daily interactions wherein racial attributions take place, people do not have access to, and therefore do not rely on the same types of knowledge as survey interviewers. Perhaps most problematic, for some studies of racial contestation, interviewers only code their assessment of the respondent’s race after they hear how respondents racially classify themselves (Campbell & Troyer, 2007). As Campbell and Troyer (2007) note, interviewers may be “biased toward agreement with the self-identification of the respondent” and “without the observer hearing the self-identifications of the respondents, we would expect higher levels of misclassification” (p. 754).
These methodological issues pose a dilemma for documenting racial contestation with national survey data. Aside from an imagined nationally representative survey that would ask a racially diverse set of strangers, acquaintances, friends, and family of each survey respondent to racially classify the respondent (a very costly and complicated endeavor), we cannot examine with certainty the extent to which people self-classify in ways that are inconsistent with others’ perceptions. However, one nationally representative survey, the PALS, provides data on an important and related notion—whether respondents believe that other Americans generally agree with his or her own self-classification. The key strength to the measure employed in this article, then, is that it does not refer to how one particular person classifies the respondent after hearing a great deal of background information, but rather, how the respondent feels that they are generally perceived in everyday life by a wide swath of other Americans. Campbell and Troyer (2011) argue that respondent reports of external racial classification are superior indicators because racial contestation should have the greatest influence on peoples’ lives when respondents are aware that it occurs. This should be especially true of studies that examine the relationship between racial contestation and stress or mental health, or in this case, studies on the relationship between racial contestation and racial identity.
Drawing from PALS data, this study is the first to document how many monoracial identifying Blacks, Latina/os, Asians, and Whites report experiencing racial contestation in the United States, and to then examine some of the most common characteristics associated with those experiences for each racial and ethnic group. For example, research indicates that racial attributions are heavily informed by phenotypical features (Brown, Dane, & Durham, 1998; Feliciano, 2016). Therefore, we examine how phenotypical characteristics such as skin tone, hair color, and eye color are associated with reports of racial contestation across the four racial groups. We also examine how sociocultural characteristics that may influence external racial attributions, such as immigrant generation, socioeconomic status, age, gender, and political orientation, are associated with reports of racial contestation. Many of these indicators have been associated with incongruent interviewer–respondent classification in previous research (Saperstein, 2006), but have not been analyzed for Latina/os and Asians, or with this newer measure of racial contestation.
Contested Racial Identity
Documenting perceived racial contestation provides an opportunity to not only better identify the proportion of U.S. adults at the margins of racial categorization but also to examine questions about the malleability of racial identity. For as long as sociologists have studied the “self,” much emphasis has been paid to the impact that others’ expectations have on personal identities. Cooley’s (1902) looking glass-self, for example, posits that individuals shape their self in accordance with others’ perceptions. In an effort to seek internal and external consistency, individuals are believed to consider the perceptions of others in the process of creating and maintaining their own personal identities (P. J. Burke, 1991). Others’ perceptions are of particular importance for understanding the implications of socially constructed categories, such as race, whereby there is often no consensus, or official comprehensive criteria for defining group membership.
As Jenkins (1994) argues, racial group membership is a product of both internal definitions and external definitions. Internal definitions (or self-categorization) provide information about how individuals perceive themselves vis-a-vis others. Yet it is generally argued that internal definitions are constrained by external racial attributions. Put differently, it is difficult to maintain an identity if it is regularly contested by others. This multidimensionality of racial categorization is aptly elucidated by recent studies on multiracial classification (R. Burke & Kao, 2013; Doyle & Kao, 2007). Studies show high levels of variation in how multiracials classify themselves (Rockquemore, 1998), but such choices are often constrained by a number of key factors, including reflected appraisals (individuals’ perceptions of how others view them racially; Brunsma, 2005; Khanna, 2004; Khanna & Johnson, 2010). For example, Khanna (2004) explains that Asian/White biracials have several racial identification options available to them (Asian, White, multiracial), but when asked to choose one over the other, they are significantly more likely to self-identify as the race that others perceive them as.
Though the constraining effect of reflected appraisals on identification choices is well-studied for mixed-race persons, we know much less about the role of reflected appraisals for monoracial identifying persons. For example, we do not know if elements of racial identity such as racial group closeness, racial identity salience, or racial awareness might be affected by experiences of racial contestation. Variants of identity theory suggest that individuals modify behavior, attitudes, and sometimes even elements of their identity to relieve the stresses of identity incongruence (P. J. Burke, 1991). If true in the case of race, what aspects of racial identity would be associated with racial contestation?
Along these lines, we first ask if people who report experiencing racial contestation are any more or less aware of their race in daily life. This question serves as a robustness check for our operationalization of self-reported racial contestation. If respondents who report experiencing racial contestation are those who express disproportionately high levels of racial awareness, this could threaten construct validity. For example, it could be that those who are highly aware of race are actually no more likely to experience racial contestation, but are simply more likely to identify it. If true, our measure of racial contestation would do a poor job of reflecting the lived components of race that we aim to examine. However, we hypothesize that individuals who report experiencing racial contestation are no more or less aware of race than their noncontested counterparts.
More substantively, we also ask if the racially contested feel less of an affective connection to other members of their self-identified racial group, an indicator of ethnoracial unity, and if race is as salient a component of their identity as it is for the noncontested. Such questions serve to highlight the social and interactive processes by which elements of racial identity are navigated, even for individuals who identify as monoracial. To date, these questions have gone mostly unasked, but their answers will help provide a better understanding of how malleable these components of racial identity are in U.S. society, where racial boundaries are sometimes ambiguous and in flux.
Drawing from previous research, there are important reasons to suggest that individuals who experience racial contestation perceive their racial identities in distinct ways. Prior studies indicate that racial contestation can induce stress among those who experience it (Campbell & Troyer, 2011; Stepanikova, 2010). Discordant and unstable racial attributions are associated with elevated risks of hypertension, poorer mental health, and suicidal thoughts (Campbell & Troyer, 2007; Laster Pirtle & Brown, 2015; Veenstra, 2011). Results of the aforementioned studies parallel arguments of identity theorists who assert that we experience stress when we do not receive external confirmations of our personal identities (Burke & Stets, 2009). It is plausible, then, that some individuals may seek to offset the stress of racial contestation by affectively distancing themselves from coracial others and decreasing the extent to which they believe their racial identity is an important or valuable aspect of their lives.
Specifically, we hypothesize that the racially contested report lower levels of racial group closeness and lower levels of racial identity salience than individuals who do not experience racial contestation. For our hypothesis about contested racial identity to hold, these findings should persist net of characteristics typically associated with racial identity outcomes and potentially associated with racial contestation, including immigrant generation, gender, age, political orientation, and socioeconomic status. For example, it is possible that immigrants are more likely to experience racial contestation and have lower levels of racial identity salience given that many are adapting to a new racial label. If immigrant generation explains the relationship between racial contestation and racial identity outcomes, then it is not the experience of racial contestation that perpetuates thinner identities, but the experience of immigration. We hypothesize, however, that the relationship between contestation and identity outcomes will hold net of immigrant generation and these other factors.
Data and Method
Data for this study come from the first wave of the (2006) PALS, a nationally representative survey of 2,610 noninstitutionalized, English- or Spanish-speaking civilian households in the contiguous United States. Interviews were primarily conducted face-to-face but also included audio computer-assisted self-interviewing for more sensitive questions about deviant and private behaviors. One of the primary goals of the study was to obtain a racially diverse sample. Thus, RTI International conducted the survey by using a four-stage sampling design to oversample for Blacks, Asians, and Latina/os. The survey yielded an 83% contact rate, an 86% screening rate, and an 82% cooperation rate, for an overall response rate of 58% (.83 × .86 × .82). More information about the PALS sampling design and recruitment are available at http://www.thearda.com/pals/researchers/ (see also Emerson, Sikkink, & James, 2010). These data are particularly suited for these analyses because PALS is the only known nationally representative study that asks respondents directly about racial self-classification and perceived racial contestation.
Key Independent Variable
The racial contestation measure is generated from two items about racial identification. The first item asks, “What race or ethnic group do you consider yourself? That is, are you White, Black, Hispanic, Asian American, Pacific Islander, American Indian, or of mixed race?” Later in the interview, respondents were asked, “Earlier you told us that you were (x racial identity—i.e., White, Black, Asian, etc.). Do you think other Americans would say that you are (x racial identity), or something else?” Response categories include “x racial identity,” “something else,” “varies,” and “doesn’t matter.” This second measure was recoded into a dichotomous variable to represent racial contestation, whereby 1 = something else and 0 = all other responses. An open-ended up follow-up question was asked of respondents ensuring that by “something else,” respondents did not simply mean an alternative label for their self-identified racial group (e.g., “Anglo” for White or “Chicana/o” for Latina/o).
Demographic Characteristics
To gather a baseline understanding of who experiences racial contestation, we compare contested and noncontested members of the same self-identified racial group (e.g., contested Whites vs. noncontested Whites; contested Blacks vs. noncontested Blacks, etc.) via means tests across a variety of demographic and phenotypical characteristics. Specifically, we compare groups across age, gender, generational status, ancestry, parental language use, political party, education, household income, eye color, hair color, and skin tone, as all of these factors could plausibly be associated with how one self-identifies or is perceived racially.
Age is a continuous variable ranging from 18 to 80 years and above. Gender is a dichotomous measure with 1 indicating “male,” and 0 indicating “female.” To account for generational status, we utilize a measure of whether or not the respondent was born in the United States (1 = yes; 0 = no). Ancestry is measured by a series of dichotomous variables representing the most common origins for each racialized group. For example, self-identified Latina/os were asked to identify with one of the following groups: Mexican, Cuban, Puerto Rican, Dominican, Colombian, Central American, “other” or “no preference.” Language use is measured by whether or not either of the respondents’ parents are bilingual (1 = yes; 0 = no). Political party is a seven-category response variable ranging from “strong democrat” to “strong republican.” Education is a 1 to 5 ordinal measure ranging from “less than high school” to “graduate degree.” Income is a 19-category response variable ranging from less than $5,000 to $200,000 and above. We transformed the income variable by assigning respondents their midpoint value for the dollar range identified and then divided the value by 1,000. Eye color is a dichotomous variable whereby 1 indicates brown or black, and 0 represents all other eye colors. On completion of the in person surveys, interviewers were asked about the hair color and skin tone of each respondent. Hair color is scored from 1 to 6 with higher scores representing darker shades. Skin tone is scored from 1 to 7 ranging from light to dark.
Dependent Variables
To examine if racial contestation is associated with elements of racial identity, we employ measures of racial awareness, racial group closeness, and racial identity salience as dependent variables in three sets of multivariate ordinal logistic regressions. Racial awareness is measured by a survey item that asks, “How often, if at all, do you think about or are you aware of what race you are?” Response categories vary from 1 to 6 and range from “everyday” to “never.” Racial group closeness is measured by a survey item which asks, “How closely connected do you feel to your racial group?” Response categories vary from 1 to 4 and range from “extremely close” to “not close at all.” Racial identity salience is operationalized by a survey item that asks, “When you think about yourself, how important to you is being [self-identified race] to your sense of who you are?” Response categories range from 1 to 4, from “very important” to “not important at all.” We reverse coded all three measures so that higher scores represent greater awareness, higher levels of closeness, and greater importance to self, respectively. Correlations between the three dependent variables range from .25 to .42, indicating that they measure related but distinct elements of racial identity. See Table 1 for a distribution of responses across the four racial groups analyzed.
Dependent Variables by Self-Identified Race.
N = 2,435. bN = 2,419. cN = 2,442.
Results
Table 2 illustrates that approximately 6% of all monoracial identifying White, Black, Latina/o, and Asian American respondents report experiencing racial contestation. We also see in Table 2 that racial contestation is much more common for self-identified Latina/o (13%), Asian (8%), and Black (7%) respondents than it is for Whites (3%). This suggests that the experience of racial contestation, at least when measured from the perspective of the respondent, is more common than previous studies indicate and that it is experienced at different rates across racial group classifications. These descriptive findings indicate that unidimensional measures of self-classification are not reflecting how a small but important segment of U.S. adults live and experience race in daily life.
Frequency of Racial Contestation.
Comparing the Contested and Noncontested
In Tables 3 and 4, we compare demographic characteristics of contested and noncontested members of the same self-identified racial group via means comparisons tests. These analyses provide additional descriptive information about those who report experiencing racial contestation. The first two columns of Table 3 illustrate how contested Whites compare with noncontested Whites. Contested Whites are significantly younger (39 years old vs. 47 years old), and report a more recent immigration history than their noncontested White counterparts. The majority of contested Whites were born in the United States (75%), are at least third generation (60%), and have monolingual parents (62%) but they still vary significantly from noncontested Whites on all of these indicators. Skin tone, hair color, and eye color make up some of the most substantial differences between contested and noncontested Whites. Contested Whites, on average, have a darker complexion, darker hair, and darker eyes than noncontested Whites. There is also significant variation by ancestry. The majority of contested Whites report European lineage (59%), but a higher proportion of noncontested Whites reports the same (79%). Moreover, those with Middle Eastern descent make up 5% of contested Whites and only 1% of noncontested Whites. Similarly, self-identified Whites with Hispanic ancestry make up approximately 5% of contested Whites, and only 1% of noncontested Whites. In sum, it appears that contested and noncontested Whites vary in accordance with age, immigration history, phenotype, and ancestry, but not on political party, or socioeconomic indicators such as education and household income.
Descriptive Statistics and Means Comparisons.
Asterisks indicatee statistically significant difference from “noncontested” of same racial group.
p < .001. **p < .01. *p < .05.
Descriptive Statistics and Means Comparisons.
Asterisks indicate statistically significant difference from “noncontested” of same racial group.
p < .001. **p < .01. *p < .05.
The third and fourth columns in Table 3 compare demographic characteristics of contested and noncontested Blacks. Contested Blacks differ from noncontested Blacks only on political affiliation, immigration history, skin tone, and ancestry. Contested Blacks are slightly more conservative than noncontested Blacks (2.73 vs. 2.20, respectively). An 80% of contested Blacks are at least third generation, whereas 92% of noncontested Blacks are the same. Interviewers rated contested Blacks as significantly lighter in complexion than noncontested Blacks (4.89 vs. 5.72, respectively). There are, however, no significant differences in hair color and eye color between the two groups. In terms of ancestry, a large majority of contested Blacks report African American descent (68%), but a higher proportion of noncontested Blacks identifies the same way (83%). About 15% of contested Blacks report Caribbean descent, whereas only 4% of noncontested Blacks report the same.
Table 4 compares self-identified Latina/os and Asians across the same characteristics. Though previous work suggests that Latinas may have more flexible racial/ethnic identities than Latinos (Vasquez, 2010), we find no differences by gender. However, there are differences by socioeconomic status, immigrant generation, and phenotype. Contested Latina/os report significantly higher household incomes (8.99 vs. 6.73), and significantly higher educational attainment (2.61 vs. 2.04) than their noncontested Latina/o counterparts. Almost one third of contested Latina/os are at least third generation, whereas 21% of noncontested Latina/os are the same. A greater proportion of contested Latina/os were born in the United States than noncontested Latina/os (62% vs. 40%, respectively). In terms of phenotype, contested Latina/os on average have lighter shades of hair (4.21 vs. 4.64). Additionally, 88% of contested Latina/os report having brown or black eyes, whereas 95% of noncontested Latina/os report the same. Last, a smaller proportion of contested Latina/os report Mexican lineage than noncontested Latina/os (52% vs. 66%). In total, it appears that Latina/os with high incomes, high levels of education, lighter phenotypical characteristics, and those who have long ancestral ties to the United States are more likely to report experiencing racial contestation.
The third and fourth columns compare contested and noncontested Asians. It is important to note that the sample size for Asians is relatively small (N = 173). Therefore, the reported results should be interpreted very cautiously. Though preliminary, there are some patterns that are worthy of reporting. Contested Asians report significantly lower levels of household income than noncontested Asians (8.16 vs. 11.00, respectively). Contested Asians are also significantly more likely to identify with the Democratic Party over the Republican Party (2.07 vs. 3.76, respectively). Perhaps the most substantial difference, however, is related to ancestry. No contested Asians report Chinese descent, whereas 27% of noncontested Asians do. Notably, 67% of contested Asians report Filipino descent, while only 21% of noncontested Asians report the same. This high rate of reported racial contestation among Filipinos coincides with recent qualitative research illustrating that second-generation Filipinos sometimes report being perceived as Latina/o (Ocampo, 2014).
Racial Identity Characteristics-Multivariate Results
In Table 5, we examine how contested and noncontested respondents value and perceive their racial identities across a variety of identity indicators. Specifically, we estimate a series of multivariate ordinal logistic regression analyses to examine variation in reported racial awareness, perceived racial group closeness, and racial identity salience. Column 1, for example, indicates that all three groups of monoracial identifying non-Whites in the sample (Blacks, Latina/os, and Asians) report significantly higher levels of racial identity awareness than Whites. This is not surprising, as Whites need not think about the social costs of race on a regular basis because it does not typically intrude on their everyday experiences (Doane, 1997). Notably, respondents who were born in the United States have significantly lower levels of racial awareness than those who were born elsewhere (odds ratio [OR] = 0.60). This is likely because immigrants are adapting to a new racial structure, and are renegotiating and constructing their identities in a new social context (Massey & Sánchez, 2010). Age and political conservatism are also associated with racial awareness (OR = 0.98 and 0.95, respectively). Older adults, and those who are more politically conservative report thinking about their racial identity less frequently than those who are younger or more politically liberal. Education, on the other hand is positively associated with racial awareness. Each one unit increase in education is significantly associated with an 8% increase in the odds of reporting a higher level of racial awareness. Most central, we find that individuals who report experiencing racial contestation are no more or less aware of their race than other U.S. adults, lending support to the construct validity of the racial contestation measure.
Ordinal Logistic Regression Analyses of Racial Identity Characteristics Among Contested and Noncontested Respondents.
Note. OR = odds ratio; SE = standard error.
p < .001. **p < .01. *p < .05.
Column 2 examines variation in reports of racial group closeness. Results indicate that, on average, self-classified Black and Latina/o respondents report significantly higher levels of closeness with their racial group than do Whites (OR = 3.08 and 2.01, respectively). Notably, age is positively associated with reports of racial group closeness (OR = 1.01), and men report significantly lower levels of closeness than women (OR = 0.61). Socioeconomic indicators and political ideology are not significantly associated with the outcome of interest. However, as hypothesized, perceived racial contestation is negatively associated with racial group closeness net of these factors. On average, and controlling for a host of sociodemographic characteristics, individuals who report experiencing racial contestation feel less of an affective connection to other members of their self-identified group (OR = 0.66).
Column 3 examines variation in how important race is to respondents’ sense of self. As expected, self-classified Black, Asian, and Latina/o respondents report significantly higher levels of racial identity salience than do Whites. Interestingly, gender and income are also associated with racial identity salience. Higher levels of income are negatively associated with the outcome of interest, and men, on average, have significantly lower levels of racial identity salience than women (OR = 0.79). Most important, for this study, we find that racial contestation is significantly and negatively associated with racial identity salience. On average, and controlling for a series of sociodemographic characteristics, individuals who report experiencing racial contestation are substantially less likely to believe that race is an important element of their identity (OR = 0.68).
In supplementary analyses (models not shown), we also analyzed interactions between racial self-identification and contestation to examine if the relationship between contestation and identity outcomes varies by race. We estimated four separate models, each with a statistical interaction between one of the four self-identified racial groups and racial contestation (White × contestation, Black × contestation, Latina/o × contestation, and Asian × contestation) for each of the three racial identity outcomes (Racial Identity Awareness, Racial Group Closeness, and Racial Identity Salience), resulting in a total of 12 supplementary models. Not one of the models indicated a significant difference in the effect of contestation by race. These results indicate that the relationship between perceived racial contestation and the identity outcomes analyzed is relatively uniform across self-identified racial groups. Though Whites express lower levels of racial awareness, identity salience, and group closeness than Blacks, Latina/os, and Asians, the relationship between contestation and these identity outcomes does not vary systematically across the White/non-White boundary, nor any other analyzed boundary (Black/non-Black, Latina/o/non-Latina/o, Asian/non-Asian). We caution, however, that these results should be interpreted as preliminary. Statistical power for such interactions is relatively low, especially for Asians. We encourage future studies with larger samples to continue examining whether the influence of racial contestation may vary by self-identified race.
Discussion
The primary goal of this article was to add basic but important contextual information to a growing literature on racial contestation. To date, studies have been hampered by significant data limitations. Some studies that rely primarily on interviewer–respondent incongruence suggest that racial contestation is a rare occurrence (Smith 1997). The key implication is that unidimensional measures of race (particularly measures of self-identified race), accurately indicate how race is conceptualized and experienced for the vast majority of U.S. adults, and that more complex multidimensional systems of racial measurement—though often called for—may not be necessary). This article suggests that, on the whole, it is true that the majority of monoracial identifying U.S. adults do not report experiencing racial contestation. According to the PALS data, approximately 94% of monoracial identifying Whites, Latina/os, Blacks, and Asians report that their racial self-classification is congruent with how they are perceived racially by others. However, this leaves a little over 6% of the aforementioned groups that are incongruently classified. Moreover, the rates are even higher for self-identified Latina/os, Blacks, and Asians. An additional 8% of PALS respondents report that they are “sometimes” perceived as a race that they do not identify with. This adds to a total of approximately 14% of self-identified monoracial adults in the sample who have experienced racial contestation. These findings contrast with previous studies that suggest racial contestation is a particularly rare occurrence. Moreover, we find that racial contestation is most common among self-identified Latina/os and Asians, the fastest growing racial groups in the United States, indicating that such experiences will likely become more common over time.
One substantial implication is that by employing only unidimensional measures of race, most statistical studies gloss over important systematic variation in how nontrivial subsets of self-identified Latina/os, Asians, Blacks, and to a lesser extent Whites, live and experience race. As previous research illustrates, employing only unidimensional measures can significantly bias the results of survey-based studies that focus primarily on racial issues (Bratter & Gorman, 2011; Saperstein, 2006). If social scientists are to take the social construction of race seriously, we need not only consider how individuals self-identify racially but also whether or not these identities are confirmed or contested by others. This will provide for a much closer approximation of how race is actually experienced in social life. If we do not take up this task, our measures, over time, will become even further divorced from the lived realities of race.
An additional goal of this study was to examine who is most likely to experience racial contestation. Our analyses indicate three key factors of influence. First, racialized phenotypical characteristics such as skin tone, hair color, and eye color tend to covary with experiences of racial contestation. Darker phenotypical features for Whites and lighter phenotypical features for Blacks are associated with having a contested racial identity. Latina/os and Asians, on the other hand, can experience racial contestation for being either lighter or darker in complexion than other members of their ethnoracial groups. In this regard, our analyses are likely to underestimate the influence of phenotypical characteristics for these two groups. Nonetheless, we find that Latina/os with lighter hair colors and lighter eyes are more likely to report experiencing racial contestation. These findings are not particularly surprising given that people often look to physical features of the body in order to make racial attributions (Brown et al., 1998).
A second key factor associated with racial contestation is immigration history. Immigration history is significantly associated with reports of racial contestation across all four self-classified racial groups. However, the direction of the relationship is not uniform. Blacks and Whites who have a more recent immigration history are more likely to report experiencing racial contestation than their counterparts with more established ancestral ties to the United States. For self-identified Latina/os and Asians, however, the opposite is true. It is the later generation Latina/os and Asians who are significantly more likely to report experiencing racial contestation than their earlier generation counterparts. This indicates that in the minds of many, to be White or Black indicates long established ancestral ties to the United States, whereas to be Latina/o or Asian indicates a very recent immigration history. Though approximately two thirds of Latina/os and one third of Asians are U.S. born (Motel & Patten, 2013), this does not coincide with the stereotypical image of “forever foreigner” that many Americans have when making racial attributions (Kibria, 2002).
An additional factor associated with racial contestation across groups is ancestry. When individuals do not report the most common ancestry of their self-identified racial group, they are more likely to experience racial contestation. One potential explanation is that most people in the United States are generally familiar with some of the most common descent groups for each racial and ethnic classification (e.g., European for White, African American for Black, Mexican for Latina/o, or Chinese for Asian), but less familiar with others (e.g., Middle Eastern for White, Caribbean for Black, Puerto Rican for Latina/o, Filipina/o for Asian). In this way, U.S. residents may have more difficulty racially categorizing individuals who do not fit their most common assumptions about race. If true, experiences of racial contestation should vary across regions. For example, Puerto Ricans make up New York City’s largest Latina/o group (8.9% of the total population), and one of Dallas, Texas’ smallest Latina/o groups (0.2% of the total population). If the aforementioned explanation is correct, we would expect that Puerto Ricans in Dallas would be more likely to experience racial contestation than Puerto Ricans in New York City because Dallasites are less familiar with Puerto Ricans as a group, and may therefore have more difficulty racially classifying them as Hispanic/Latina/o. Future research will be necessary to examine how geographical context and population demographics interact with ancestry to potentially influence racial attributions and experiences of racial contestation (e.g., Newby & Dowling, 2007).
Moreover, some descent groups do not always experience race in accord with their official racial designations. Middle Easterners, for example, are classified as White by the Office of Management and Budget, but are not perceived or treated as White in everyday discourse (Jamal & Naber, 2008). Though phenotypically diverse, Latina/os are typically racialized as a distinct group in the United States (Cobas, Duany, & Feagin, 2009; Feagin & Cobas, 2014; Flores-González et al., 2014; Frank, Akresh, & Lu, 2010; Golash-Boza & Darity, 2008; Roth, 2012; N. Vargas, 2015) but are not currently identified as such in the U.S. Census. Despite being officially classified as Asian by the Office of Management and Budget, many second-generation Filipinos are reluctant to self-identify as such, and often report being perceived as Latina/o by others (Ocampo, 2014). Shared cultural legacies of Spanish and U.S. colonialism between Filipinos and Latina/os lead some Filipinos to express high levels of similarity with U.S. Latina/os, and fewer similarities with East Asian groups (Ocampo, 2014). Notably, it appears that the U.S. Census Bureau recognizes some of these discrepancies and has begun testing separate classifications for Latina/os and individuals of Middle Eastern and North African decent along with the categories that are typically understood as racial groups (i.e., White, Black, Asian, Native American) in the 2015 National Content Tests. This could ultimately lead to changes in the system of racial and ethnic classification utilized in the 2020 U.S. Census, providing more options for some racially contested groups to begin identifying in ways that are more congruent with their racialized experiences.
Perhaps the most interesting finding of this study is that racial contestation is significantly associated with key elements of racial identity. Previous research illustrates that the contestation of social identities is associated with higher levels of mental stress and other mental health problems (P. J. Burke, 1991; Stepanikova, 2010). This is particularly true of self-identified Whites and Native Americans who experience racial contestation (Campbell & Troyer, 2007, 2011; Veenstra, 2011). One might hypothesize, then, that individuals may seek to offset the stresses of racial identity incongruence by ignoring, or placing less value on their own racial identity. The results of this study are consistent with these arguments. Multivariate analyses indicate that racial contestation is not significantly associated with how frequently one is aware of their racial identity, but it is significantly associated with perceived racial group closeness and racial identity salience. These findings indicate that external definitions of race likely shape how racially contested individuals value race as a component of identity, and see themselves as an integrated member of a racial group. Consistent with Burke and Stets (2009), we find that racially contested individuals report “thinner,” or less personally meaningful racial identities, and lower levels of ethnoracial unity, perhaps as a subconscious strategy to alleviate the stresses of identity incongruence. Moreover, in supplementary analyses, we found that the relationship between contestation and these racial identity outcomes does not vary systematically by self-identified race. This could come to have important implications. If racial contestation becomes more common and continues to lead to thinner identities, it could result in an increase of racial apathy in the United States—even in the face of sustained and growing racial disparities across many indicators of well-being (Forman & Lewis, 2006). It will be important for future research to continue examining these relationships, especially for the fastest growing racial groups in the United States (Latina/os and Asians) who continue to transform the U.S. racial landscape.
Although this study is especially equipped to examine various multidimensional measures of racial categorization, it is not without limitations. A broad and general measure of how respondents feel they are perceived racially can serve as an improvement over previous measures, but there is no way of verifying the accuracy of these self-reports with PALS data. No known data set is equipped to compare respondents’ reports of how they are perceived with actual racial attributions. Moreover, we do not know if respondents are thinking of racially specific others (i.e., Whites), or just “others” in general when answering the question about how “other Americans” perceive them. Some respondents answered that how they are perceived racially varies, but we are unable to examine the source or extent of that variation. 2 It would be a fruitful extension of this research to systematically examine if and how respondents report being perceived differently by different racial groups. Unfortunately, these questions have not yet been asked in a nationally representative survey.
Related, racial classification is sometimes consciously altered by individuals in different contexts (Doyle & Kao, 2007; Khanna & Johnson, 2010; Rockquemore & Brunsma, 2002; Saperstein & Penner, 2010, 2014). Saperstein and Penner (2014) found that a small subset of individuals who experience racial contestation come to identify as the racial group that they are perceived as. Others change their racial classification in ways that become inconsistent with how they are generally perceived. Though we focus only on monoracial identifying adults in this study, respondents who recognize a mixed-racial background could have still self-identified with only one racial group (Gullickson & Morning, 2011). Unfortunately, we are unable to examine if respondents were born from mixed-racial parentage or if they have modified their racial self-classifications over time with these data. Those who report mixed-race backgrounds and those who change how they self-classify in different contexts may be especially likely to report that they regularly experience racial contestation. This, too, could be a useful avenue for future studies.
Cross-sectional data also limit the ability to determine causal order. We suspect that racial contestation and thin identities may be mutually constitutive. By this, we mean that racial contestation likely leads to thinner identities, and thin identities may also increase the odds of reporting experiences of racial contestation. One potential pathway of influence is that thin identities produce patterns of “racial performance” that defy common stereotypes and racial expectations. If it is true that cultural tastes (i.e., styles of dress, music, speech, etc.) vary systematically by racial group membership and people generally use these cues to help them make racial attributions about others, then any factor that influences things like cultural tastes could also be indirectly associated with racial contestation. Unfortunately, measures of such “racial performances,” are not often included in nationally representative surveys, and we are unable to examine if they covary with racial identity salience or racial group closeness.
Still, with these limitations in mind, we believe that our study and future studies that utilize multidimensional measures of race are better equipped to examine racial identity characteristics and racialized experiences than those that simply use unidimensional measures of interviewer-classified or self-classified race. Quantitative studies may never be able to precisely measure the lived experience of race, but careful analysis of multidimensional measures will bring us closer to accomplishing that goal.
Conclusion
This article adds central knowledge to a growing literature set on racial contestation, and to scholarship on the relationship between reflected appraisals and racial identification. We build on these literatures and extend their insights to include monoracial identifying adults and racial and ethnic groups typically left out of such analyses (i.e., Latina/os and Asians). Employing the only known nationally representative data set to ask respondents directly about perceived experiences of racial contestation, results suggest that it may be more common than previous studies illustrate (6% to 14%, depending on the measure), especially for the fastest growing racial groups in the United States (Latina/os and Asians). Results also show that ancestry, phenotype, and immigration generation are associated with racial contestation. Additionally, findings suggest that perceived racial contestation is associated with less salient racial identities and less of an affective connection with other members of one’s self-identified racial group, indicating a “thinner” racial/ethnic identity and lower levels of ethnoracial unity among the racially contested. We encourage future research to continue considering the implications of racial contestation, and thinning racial/ethnic identities for the small but growing portion of the population (especially contested Latina/os and contested Asians) who may increasingly experience both over time.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.