Experiencing Discrimination in Los Angeles: Latinos at the Intersection of Legal Status and Socioeconomic Status

Abstract

Despite its recent slowdown, immigration from Latin America continues to be a controversial issue. Some scholars argue that the social climate is increasingly inhospitable to Latinos, potentially fueling discriminatory attitudes and behaviors. However, little research has examined Latinos' experiences with discrimination, especially variation by nativity and legal status. We address this issue with research on perceived discrimination among Mexican and Central American residents of Los Angeles County, a major destination for Latin American immigrants. Using data from the Los Angeles Family and Neighborhood Survey and the American Community Survey, the analyses consider immigrants' legal status, intersectionality, and competing perspectives on assimilation. The results show that undocumented immigrants do not report especially high levels of discrimination. Instead, young U.S.-born Latinos are the most likely to report mistreatment in interpersonal and institutional domains. Neighborhood ethnoracial and income diversity also have implications for perceived exposure to different types of discrimination.

1. Introduction

The ongoing “diversity explosion” in the United States has profoundly transformed racial and ethnic relations. In the mid-twentieth century, African Americans were the largest minority group and a major focus of research on intergroup relations, including studies of actual and perceived discrimination (Allport 1954; Anderson and Massey 2004). Since that time, sustained immigration from Asia and Latin America has contributed to an increasingly multi-ethnic society (Frey 2014; Lee, Iceland and Sharp 2012). It is only recently, however, that studies of discrimination have shifted their focus to reflect this trend. Attention is turning to questions about the experience of being a “new minority” and an immigrant in an increasingly diverse society.

In recognition of their growing prominence and unique constellation of structural disadvantages, Latinos occupy an important place in this nascent literature (Arellano-Morales et al. 2015; Becerra et al. 2013; Flippen and Parrado 2015; AUTHOR, DATE). ¹ But because fully half of Latino adults are foreign born, continued progress in understanding their experience with discrimination requires expansion of theoretical and analytic frameworks (Krogstad and Lopez 2014). Dimensions of inequality that are unique to immigrants must be considered, including those that are grounded in distinct legal statuses. Such factors intersect with traditional axes of inequality in ways that may affect Latinos' actual and perceived encounters with negative treatment. Nonetheless, we know little about how interlocking social statuses are related to experiences with discrimination among Latinos.

Intersectionality provides a useful sensitizing framework for understanding how multiple forms of structural disadvantage jointly influence experiences and perceptions of discrimination. Ethnoracial group membership can be viewed as a potential master status that intersects with other statuses such as gender, age, socioeconomic position, and authorization to live in the United States (Arellano-Morales et al. 2015). Because statuses that shape social interactions and access to opportunities overlap, discrimination based on one form of disadvantage may increase in the presence of another form of disadvantage. Individuals who are structurally positioned to experience multiple forms of disadvantage are likely to have the highest levels of perceived discrimination (Grollman 2012).

Using data from the Los Angeles Family and Neighborhood Survey (L.A. FANS) and the 2005-2009 American Community Survey (ACS), this research focuses on perceived discrimination among Mexican and Central American residents of Los Angeles County.^2,3 We consider the experiences of both native-born and foreign-born Latinos as well as non-Latino whites. Importantly, these data provide a rare opportunity to examine the role of legal status in perceived discrimination. This is important because increased immigration enforcement and restrictions on access to social benefits may have heightened distinctions based on legal status (Waters and Pineau 2015). As a result, legal status may increasingly intersect with other forms of disadvantage in ways that have implications for actual and perceived discrimination. However, even though it is critical to processes of incorporation and assimilation, legal status is an unmeasured source of heterogeneity in most empirical studies of immigrant groups (Massey and Bartley 2005).

Guided by an emphasis on intersectionality and competing perspectives on immigrant assimilation, our analysis first delineates the multiple dimensions of disadvantage experienced by Latinos of Mexican and Central American descent as well as non-Latino whites in Los Angeles County. This is accomplished with latent class analysis (LCA), a methodological approach that is ideally suited for identifying a set of underlying statuses that reflect the multidimensional nature of social disadvantage (or advantage). We then examine the relationships between latent class membership and two distinct aspects of perceived discrimination: everyday encounters with interpersonal discrimination and major experiences with discrimination in institutional settings. These analyses extend prior research by explicitly considering intersectionality using a novel method for delineating subgroups and by including legal status as a major axis of inequality among Mexican and Central American immigrants. Additionally, multilevel modeling is used for insights into the implications of neighborhood ethnoracial diversity and income diversity for perceived discrimination. Neighborhood characteristics are potentially important because they may affect interactions with other residents or representatives of institutions (e.g., police, employers, merchants, etc.).

2. Background

Immigration continues to be a contentious issue in the United States due to the long-term growth of the foreign-born population and large numbers of unauthorized immigrants in some national-origin groups. Since 1970, the immigrant population has quadrupled in size; roughly 13 percent of U.S. residents are now foreign born. During the same period, there has been a twentyfold increase in the immigrant population from Mexico and Central America. About 60 percent of these immigrants are unauthorized (Baker and Rytina 2013; Brick, Challinor, and Rosenblum 2011). Thus, the potential impact of undocumented status on immigrants and the larger social consequences of their presence are at the forefront of national debates about the future of the United States. Because immigration from Mexico and Central America is a “flashpoint” in public debates, there are heightened opportunities for conflict around national origins (Menjívar and Kanstroom 2013; Motomura 2014).

Unauthorized immigrants are highly vulnerable due to their position at the bottom of multiple status hierarchies. First, they are at the bottom of an institutionally-derived legal status hierarchy that places naturalized citizens at the top and lawful permanent residents (LPRs) in an intermediate position among the foreign born. After meeting a strict set of requirements, naturalized citizens have rights that are identical to those of native-born citizens. LPRs occupy a lower position in the hierarchy because they have fewer rights overall. Although authorized to live and work in the United States, LPRs are ineligible to vote in elections that have candidates for federal offices on the ballot and they cannot receive most public benefits prior to completion of a probationary period. LPRs are issued green cards verifying their legal status that they must carry at all times. At the bottom of the hierarchy are undocumented residents without a green card or a visa permitting temporary residence (Romero 2009).

Mexicans and Central Americans experience this hierarchy in tandem with other disadvantaged statuses. Relatively low education, employment in unskilled jobs, and a high rate of poverty place these groups on the bottom rungs of society in terms of socioeconomic status (SES; Zong and Batalova 2015, 2016). While low SES characterizes both the native born and the foreign born, it is most striking among those at the bottom of the immigration status hierarchy. Other sources of vulnerability for these immigrants are low English proficiency and limited experience with U.S. institutions and culture.

Positions in hierarchies based on socioeconomic, demographic and immigration statuses potentially influence the occurrence and perception of discrimination (Pager and Shepherd 2008). Specifically, both native- and foreign-born Latinos may be more likely than non-Latino whites to be exposed to negative treatment and to attribute such treatment to discrimination. However, those positioned at the bottom of multiple status hierarchies and subject to public animosity that is rooted in legal status distinctions may be the most affected. Undocumented Latinos would be expected to experience more unfair treatment than their relatively advantaged documented or U.S.-born co-ethnic counterparts (Hall, Greenman and Farkas 2010; Massey and Gentsch 2014).

In contrast, there are grounds for expecting minimal differences by legal status. The current anti-immigrant climate may have spillover effects that increase the likelihood of experiencing discrimination among all Latinos, in part because legal status is not directly observable or easily inferred by potential discriminators. In other words, animosity toward the undocumented could spread to those who are living in the United States legally as documented immigrants, naturalized citizens, or even native-born citizens. This would lead to weaker distinctions in perceived discrimination by legal status (Quiroga, Medina and Glick 2014).

2.1. Intersectionality, immigrant assimilation, and perceived discrimination

The intersectionality framework emphasizes interlocking forms of disadvantage. Understanding the effects of one type of disadvantage requires consideration of how it intersects with and mutually reinforces other forms of disadvantage. Many of those at the bottom of a given dimension of inequality are simultaneously located at the bottom of other dimensions of inequality. A key theoretical proposition of intersectionality is that overlapping positions in multiple status hierarchies result in complex social inequalities that shape individuals' experiences, including those with discrimination. In addition to ethnicity and race, one must consider demographic, socioeconomic, and immigration-related statuses that may engender discrimination (Arellano-Morales et al. 2015).

As alluded to above, intersectionality implies that individuals with multiple disadvantages are especially likely to have experiences that reflect and are recognized as discrimination. For example, non-English-speaking Latinos who live in poverty may contend with more types of discrimination or greater frequency of discriminatory acts than individuals who have only one disadvantaged status (Grollman 2013). However, care must be taken in applying the intersectionality framework to immigrants (Viruell-Fuentes, Miranda and Abdulrahim 2012). Immigration-related variables may signal disadvantage, but their meaning and their relationships to assimilation and acculturation must be taken into account.

This was pointed out over thirty years ago by Portes and Bach (1985), who highlighted predictions of two perspectives on variation in perceived discrimination among immigrants and their descendants. The classic assimilation perspective implies that as immigrants spend time in United States, they learn English and become familiar with the larger culture. Through this process, the native-born majority becomes more open to them and they become increasingly integrated and accepted into the mainstream. With the passage of time, immigrants and their descents ascend various status ladders. This ascension presumably lowers the likelihood of experiencing negative treatment. This perspective is consistent with intersectionality theory in expecting the newest immigrants and those lowest on the immigration status hierarchy to be the most disadvantaged on other axes of inequality. Consequently, they would also be the most likely to experience and perceive discrimination.

In contrast, the ethnic resilience perspective argues that perceptions reflect the accumulation of experiences through exposure to American society. As immigrants spend time in the United States, they become more aware of their place in the larger system of stratification (Flippen and Parrado 2015; Portes and Bach 1985; Portes and Rumbaut 2006). Although their position may improve over time, the formation of ethnic consciousness provides the foundation for interpreting negative experiences in terms of discrimination (AUTHOR, DATE). This may also be the case for the native born when long-term exposure to U.S. society is coupled with limited opportunities for upward mobility.

Nativity and immigration status may also combine with other forms of disadvantage in ways that have complex implications for perceived discrimination. Some evidence suggests that advantaged positions in status hierarchies do not necessarily lower perceived discrimination (see Arellano-Morales et al. 2015; Flippen and Parrado 2015; AUTHOR, DATE; Pérez, Fortuna and Alegría 2008). Likewise, the structural disadvantages of undocumented immigrants may be coupled with lower perceived discrimination because the undocumented are less likely to feel that they have received unfair treatment and less likely to interpret negative treatment as due to their ethnicity or race. Still, to our knowledge only one recent study of immigrants in North Carolina has considered the role of legal status along with other socioeconomic disadvantages in perceived discrimination among Latinos (Flippen and Parrado 2015).

Finally, gender occupies an ambiguous place in perspectives on assimilation and perceived discrimination even though intersectionality highlights gender as an additional source of inequity. The conventional assumption is that women are more likely than men to experience negative treatment. Women may also be especially disadvantaged due to overlapping statuses, at least according to a “double jeopardy hypothesis” that draws attention to the combined negative effects of two stigmatized statuses (Levin et al. 2002). Alternatively, in line with the ethnic resilience perspective, ethnicity may override gender in perceptions of discrimination. For immigrant and ethnic minority women, ethnicity may be more salient when assessing experiences with discrimination. Prior studies do not provide clear guidance on this unresolved issue (Levin et al. 2002).

2.2. Context of reception: The role of neighborhood characteristics

To fully investigate experiences with discrimination, it is necessary to go beyond individual characteristics to contextualize lives in terms of the places in which people live. The neighborhood context is especially important in this regard because it circumscribes many daily activities and social interactions. Because increasing diversity at both the national and local levels is one of the most far reaching demographic changes that can be traced (in part) to immigration during the past half century, the first consideration is the role of neighborhood ethnoracial diversity in perceived discrimination.

Although this topic has suffered from inattention in studies of Latinos, insights are available from studies of African Americans that suggest that perceived discrimination is positively associated with residence in a racially diverse neighborhood (Dailey et al. 2010; Hunt et al. 2007; Stewart et al. 2009). This typically is explained as a byproduct of greater intergroup contact in neighborhoods that are racially heterogeneous. Contact may heighten competition over limited resources, concerns about safety and the value of property, and overt acts of discrimination (Blau 1977; Hunt et al. 2007; Logan and Molotch 1987; Putnam 2007). Intergroup contact also rises with neighborhood ethnoracial diversity for Latinos, but the extent to which such contact has implications for their perceptions of discrimination has received scant attention (but see Shell, Peek and Eschbach 2013).

The second aspect of neighborhood context is economic diversity, a topic of national importance that few local-level studies of discrimination broach. While not focusing on diversity within neighborhoods per se, Dailey et al. (2010) found that African American women living in socioeconomically disadvantaged neighborhoods reported significantly lower levels of perceived discrimination than African American women living in relatively advantaged neighborhoods. Their unadjusted comparisons also showed the lowest perceived discrimination in the most racially segregated neighborhoods. However, these relationships did not persist in multivariate models that included individual characteristics and both of these neighborhood characteristics. The authors concluded that to assess the role of neighborhood socioeconomic make-up in perceived discrimination, researchers must also consider neighborhood racial segregation.

This conclusion connects with a broader debate about the consequences of growing racial diversity in American communities for other types of diversity. Some scholars argue that the trend toward ethnoracial diversification is accompanied by rising demographic and economic diversity across multiple geographic scales (Tittle and Rotolo 2010). This implies that racially diverse communities are also likely to be economically diverse. Despite the potential for such a pattern to exacerbate discrimination, the opposite is also possible if the salience of social distinctions declines through intergroup contact (Allport 1954). Intergroup contact may increase familiarity and promote acceptance. Still, the persistent positive relationship between racial diversity and perceived discrimination suggests that this is unlikely (Frey 2014).

An alternative view is that the relationship between these two dimensions of diversity is more complex (Fischer 2003). Economic diversity is declining or remaining the same as ethnoracial diversity rises (Fischer et al. 2004; Tach and Lee 2016). In fact, Tach and Lee (2016) found no substantively significant associations between ethnoracial and economic diversity within “places” (communities within counties). This finding held for both levels and changes in diversity. The implications of a lack of correspondence between ethnoracial and economic diversity for perceived discrimination are ambiguous, again suggesting that it may be important to consider both to understand the role of neighborhood characteristics.

2.3. Context of reception: Los Angeles County

In some respects, Los Angeles County is typical of the largest counties in the United States. It is a majority-minority county where more than half of the population does not self-identify solely as non-Latino white. Nationwide, 11% of all counties and 76% of the largest counties are majority-minority (Krogstad 2015). At the same time, Los Angeles stands out as the most populous county, both in terms of the number of residents (9.9 million) and the number of Latino residents (4.8 million). The latter figure eclipses that for the county with the next highest number of Latinos (Harris County, Texas) by almost a 3:1 ratio (Brown and Lopez 2013). Close to half (48% in 2014) of the population of Los Angeles County is Latino and 27% is non-Latino white (U.S. Census Bureau 2015).

These figures show that Los Angeles County has a Latino plurality that is on the verge of becoming a Latino majority. One reason for this is that it is the destination of more immigrants from Mexico and Central America than any other U.S. county (Zong and Batalova 2015 Zong and Batalova 2016). In short, Los Angeles County is in some respects typical of the most established Latino immigrant gateway. But it is also distinct in terms of the scale of the Latino population, especially immigrants. We return to this issue in the interpretation of our findings.

2.4. The present study

This investigation advances the literature on perceived discrimination by illuminating the experiences of Mexicans and Central Americans. Given that half of adults in these groups are immigrants, we frame our analysis in terms of both intersectionality and assimilation. A key analytic aim is to investigate how intersecting positions in status hierarchies—including the legal status hierarchy—influence perceived discrimination. To address this aim, we use latent class analysis (LCA), an innovative methodology for delineating groups based on multiple dimensions of disadvantage. Our use of LCA is consistent with the intersectionality framework's emphasis on how individuals simultaneously experience a set of characteristics or statuses (Collins and Lanza 2010). ⁴ As such, it is a particularly useful analytic method for identifying classes or groups based on the alignment of multiple characteristics. This advantage sets it apart from traditional analytic methods (e.g., multiple regression and structural equation modeling) that identify variable-centered patterns of association among separate characteristics (Collins and Lanza 2010). We combine this methodology with multilevel modeling to determine how latent class membership and neighborhood context jointly matter. In this way, we provide new insight into perceived experiences with discrimination among Latinos.

Because the first goal of our analysis is to delineate the latent classes that best represent the interconnected social statuses of Latino and non-Latino white adults in Los Angeles County, we cannot present specific hypotheses a priori regarding differences in perceived discrimination among the classes that will emerge. Nonetheless, several broad expectations are consistent with the aforementioned frameworks. First, the intersectionality framework reveals the potential for individuals at the bottom of multiple status hierarchies to be at very high risk of experiencing and perceiving discrimination. But this is not necessarily the case because some social statuses may modify the risks posed by interrelated social statuses. Second, the classic assimilation and ethnic resilience perspectives draw attention to specific intra-ethnic and inter-ethnic comparisons. The classic assimilation perspective implies that among Latinos, those at the top of the legal status hierarchy will be the most socially and economically integrated into mainstream society. They will therefore be the least likely to encounter or perceive discrimination. Conversely, Latinos at the bottom of the legal status hierarchy will be the most likely to experience discrimination because of their multiple disadvantaged statuses and marginalization. The classic assimilation perspective also predicts declining differences in perceived discrimination between Latinos and non-Latino whites as Latinos ascend the legal status hierarchy. In contrast, the ethnic resilience perspective hypothesizes that as Latinos ascend the socioeconomic and legal status hierarchies they will have heightened awareness of barriers to advancement and increasingly attribute such barriers to ethnoracial discrimination. Native-born Mexicans and Central Americans will have higher levels of perceived discrimination than their co-ethnic counterparts who are documented or undocumented immigrants. They will also be the most distinct from non-Latino whites.

3. Data, measures, and analytic strategy

3.1. Data

The individual-level data for this study come from adult respondents in Wave 2 of the Los Angeles Family and Neighborhood Survey (L.A. FANS). Conducted in 2000-2002, Wave 1 included a stratified probability sample of individuals in 3,085 households across 65 neighborhoods throughout Los Angeles County. Within each household, one adult and one child (if present) were selected at random to participate. The child's primary caregiver also completed the adult survey if she had not already done so as the randomly selected adult. Conducted in 2006-2008, the second wave included re-interviews with the randomly selected adults and children from the first wave who were living in Los Angeles County. Attempts were also made to collect information from those in the original sample who moved elsewhere and to “refresh” the sample by adding households that moved to each neighborhood between waves. As before, primary caregivers also completed the adult survey. Some caregivers participated in both waves, while others were new respondents.

When properly weighted, the Wave 2 cross-sectional sample of adults (18+) can be considered representative of the adult population in Los Angeles County. Prior to weighting, the subsample that we examine consists of 1,789 Mexican/Central American and non-Latino whites. Of the 1,275 Mexican and Central American respondents, 406 were U.S.-born, 514 were naturalized citizens or documented immigrants, and 355 were undocumented immigrants. The analysis also utilizes five-year estimates from the American Community Survey (2005-2009) to describe neighborhood contexts from information on census tracts. The contextual measures (described below) were linked to the individual observations with geocodes that are only available for restricted L.A. FANS data.

3.2. Measures

3.2.1. Perceived discrimination

Discrimination refers to differential treatment of individuals based on group membership (Pager and Shephard 2008; Reskin 2012). In interpersonal domains, discrimination entails negative treatment that is inconsistent with social equality in interactions. In institutional domains, it includes negative treatment that is inconsistent with equity; that is, decisions by institutional agents that violate the principle of fairness by restricting access to opportunities.

This study examines perceived discrimination in both domains (following Williams et al. 1997; Williams 2012). The everyday discrimination index relies on eight items that report the frequency of negative interpersonal experiences in daily life. Respondents reported how often in their “day-to-day” life: (a) they were treated with less courtesy than other people; (b) they were treated with less respect; (c) they received poorer service at restaurants or stores; (d) people acted as if they were not smart; (e) people acted afraid of them; (f) people acted as if they were dishonest; (g) they were called names or insulted; and (h) they were threatened or harassed. After reverse coding, the response categories ranged from 1 (never) to 6 (almost every day) for each item. The index is the sum of all values, logged to correct for skewness.

The index of major experiences of discrimination in institutional domains summarizes reports of unfair treatment by the agents of organizations such as government and businesses.⁵ Respondents indicated whether they experienced any of the following during the previous five years for “unfair reasons” (1=yes; 0=no): (a) fired from a job/denied a promotion; (b) not hired for a job; (c) stopped, searched, questioned, physically threatened or verbally abused by the police; and (d) refusal of an offer to buy or rent a dwelling by a landlord, home owner, or realtor. Values on these items were summed to create the index.

We also conducted supplementary analyses of the individual items that comprise both indexes. The results of those analyses (reviewed in detail below) directed our attention to one particular aspect of perceived discrimination in institutional domains—treatment by the police. Thus, perceived discrimination in interactions with the police is analyzed both as a part of the index of major experiences of discrimination and a separate item.

3.2.2. Variables included in the latent class analysis

Latent class analysis provides the analytic tool for addressing the intersection of key demographic, socioeconomic, and immigration-related sources of disadvantage. Demographic characteristics include ethnicity and age. Ethnicity contrasts adults who self-identified as Mexican or Central American with those who self-identified as non-Latino white (hereafter “white”).⁶ Age is a dummy variable differentiating those who were under age 30 from those who were 30 or older. Socioeconomic indicators are education (less than high school vs. high school or more), lack of English proficiency (does not speak vs. speaks English), employment status (employed vs. not employed), and poverty status (below vs. above the 2007 poverty line).

Gender is another potentially important consideration. However, LCA models that included gender along with the other variables could not be satisfactorily identified (Collins and Lanza 2010). Therefore, gender is omitted at this stage of the analysis but remains as a separate covariate in the regressions that follow.

An additional variable combines information on nativity and legal status to distinguish those who were U.S.-born, documented immigrants, and undocumented immigrants. Adult respondents were first asked whether they were born in the United States. Foreign-born respondents were then asked a series of questions to identify: (a) naturalized citizens; (b) legal permanent residents (with a green card); (c) refugees and asylees (including Temporary Protected Status); and (d) holders of a valid visa. Those who fell into one of those four categories are classified as “documented.” Using the standard residual approach, those who did not fall into any of those groups are treated as “undocumented” (Bachmeier, Van Hook and Bean 2014).

3.2.3. Variables included in the multilevel models

In addition to the latent classes identified below and gender, the multilevel models include two key indicators of neighborhood diversity measured at the census tract level: ethnoracial and household income diversity. Ethnoracial diversity is measured with a tract-level nominal entropy score based on Latinos, whites, blacks (African Americans), and Asians:

$$E_n=\sum _{r=1}^R({\Pi }_r)\text{ln}\left[\frac{1}{{\Pi }_r}\right]$$

Where:

E_n = the nominal entropy score

Π_r is an ethnoracial group's (r) proportion of the population within a given tract.

Because its maximum value is a function of the number of groups, we rescaled the entropy index accordingly. After rescaling, the measure ranges from 0 to 1, with the maximum reached when all ethnoracial groups have equal representation in the census tract (Iceland 2004).⁷

Household income diversity is assessed with an ordinal entropy score for each census tract. The income distribution for each tract was divided into five categories based on approximate cut-points defined relative to the median household income for Los Angeles County in 2007: less than 20%, 20-40%, 40-60%, 60-80%, and higher than 80% of the median.⁸ The ordinal entropy score for each census tract is:

$$E_o=\frac{-{\sum }_{r=1}^R[c{\pi }_r\ast {log}_2(c{\pi }_r)+(1-c{\pi }_r)\ast {log}_2(1-c{\pi }_r)]}{R-1}$$

Where:

E_o = the ordinal entropy score

r = 1, 2, …, R subgroups

cπ_r = the cumulative proportion of individuals in groups 1 through r in the tract.

Household income diversity ranges from 0 to 1, and assumes its maximum value of 1 when the two groups at the lowest and highest extremes each constitute 50% of the geographic area. The ordinal entropy score differs from the nominal entropy score in that it recognizes the inherent order of categories (Reardon et al. 2006; Tach and Lee 2016).

3.3. Analytic strategy

The first step in our analysis was to use LCA to estimate a set of latent classes that capture the multidimensional nature of disadvantage among Mexican/Central American and non-Latino white adults. LCA resembles cluster analysis conceptually, but it is based on a measurement model that is akin to factor analysis. This person-centered approach postulates that the true, underlying, classes (or subgroups) of individuals are not directly observable. Instead, they can be inferred from the associations among a set of observed characteristics. The statistical model uses information on these associations to identify the optimal number of latent classes required to represent variation across respondents in their values on the observed variables, as well as the size and characteristics of each latent class (Scharoun-Lee et al. 2011). A fundamental assumption of LCA is local independence—that is, conditional on latent class, the observed variables are independent (Collins and Lanza, 2010).

Using Mplus statistical software, LCA was used to determine the probability of membership in latent classes formed by eight binary indicators for all respondents. An important advantage of this technique is that each class reflects higher-order interactions among all indicators considered. Thus, this technique reduces complex information to a parsimonious set of latent classes reflecting dominant profiles of disadvantage.

Two sets of parameters are estimated in LCA models. The first set contains the latent class membership probabilities, reflecting the relative proportion of individuals in each latent class. These parameters represent a vector of probabilities that sum to 1. The second set contains item-response probabilities that indicate the likelihood of a particular value for each indicator given latent class membership. In expressing the relationship between observed and latent variables, these item-response probabilities provide a basis for interpreting the latent classes.

The model selection process involves both statistical and substantive criteria. Well-identified latent class models with different numbers of classes can be compared using information criteria such as AIC and BIC.⁹ In addition, the substantive interpretation of latent classes can play an important role in model selection. We used both criteria to select the final model after eliminating those that were under-identified.

In subsequent analyses, latent class membership serves as a predictor of perceived discrimination. Based on the LCA parameter estimates and individuals' observed responses, the posterior probability of membership in each latent class was calculated for each individual. Respondents were then classified into the latent class that best represents their characteristics according to their individual maximum posterior probability. This class assignment is treated as a fully-observed dependent variable in the regression analyses. Collins and Lanza (2010) provide details about the LCA mathematical model, posterior probabilities, model identification, and model selection.

We used a hybrid approach to handling missing data. Mplus was used to perform the latent class analysis with full-information maximum-likelihood estimation, thereby handling incomplete information on the eight indicators of latent disadvantage status (Lanza et al., 2011). After individuals were assigned to a latent class, multiple imputation was used to handle missing values on other items (see Rubin, 1987; Schafer, 1997). Twenty-five data sets were imputed using Stata 13 to take the uncertainty of imputed values into account. Class assignment was included in the imputation model.¹⁰ Models were fit with each imputed data set, and the results were combined across imputations to generate the parameter estimates and standard errors.

Hierarchical linear modeling was used to estimate our multivariate regression models. The L.A. FANS has a complex sample design and a nested structure. Hierarchical modeling allowed us to take the complex sample design and nested structure of the L.A. FANS into account. It also allowed for rigorous tests for cross-level interactions between latent disadvantage status and the indicators of neighborhood ethnoracial diversity and neighborhood income diversity.

Our multivariate analysis proceeded in several steps. Null models without explanatory variables were tested first to assess whether HLM is appropriate. After checking that the variance of the intercept was statistically significant in our null models, individual- and neighborhood-level variables were then included. Because individuals are nested within neighborhoods, parameter estimates and standard errors were generated with a two-level hierarchical linear model with random intercepts. The estimates also reflect contextual-level weights that were created and rescaled to remove the unequal probabilities of selection for the neighborhoods (West, Welch and Galecki, 2014).

4. Results

4.1. Descriptive Statistics

Table 1 provides descriptive statistics for all respondents. Focusing first on the dependent variables, the mean for the everyday discrimination index is 12.4 (2.5 logged). Additional analyses (not shown) revealed that although 74% reported at least one type of discriminatory treatment in everyday interactions, various “slights” occurred relatively infrequently for most. On the majority of items included in this index, about 11-16% of the sample reported negative treatment at least a few times a month. The behaviors most commonly reported are being treated with less courtesy than others, being treated with less respect, and people acting as if the respondent was not smart. As for the index of major experiences of discrimination, the value of .3 indicates that most respondents did not experience discrimination in any of the included institutional domains during the five years before the survey. The most common events reported were being unfairly fired or denied a promotion (9.5%) or being unfairly stopped, searched, questioned, physically threatened or verbally abused by the police (14%).¹¹

Table 1. Descriptive Statistics for Dependent and Independent Variables, White, Mexican and Central American Adults.

	Total
	Mean/%
Everyday discrimination (log transformed)	2.52
Major experiences of discrimination	0.27
Individual characteristics
Ethnicity
Non-Latino White	41.4%
Mexican/Central American	58.6%
Immigration status
U.S. born	58.5%
Documented	28.0%
Undocumented	13.5%
Other individual characteristics
Female	63.2%
Age < 30	26.6%
Less than high school	30.1%
No English	29.9%
Employed	72.5%
Poor	19.1%
Neighborhood characteristics
Racial and ethnic entropy	0.70
Income diversity	0.53
Unweighted N = 1,789

Data Source: Los Angeles Family and Neighborhood Survey; ACS 2005-2009; Analysis based on weighted data

This table also shows that 41% of the Los Angeles County adults who met the criteria for inclusion in our study were white and 59% were Mexican or Central American. About 58% were U.S.-born citizens, 28% were documented immigrants (i.e. naturalized citizens or permanent residents), and 14% were undocumented immigrants. Slightly more than one-fourth were under age 30, while 30% had not completed high school and 30% did not speak English. Nearly three-fourths were employed and one-fifth lived below the poverty line. With respect to neighborhood diversity, the means for the entropy scores are .7 for racial and ethnic diversity and .5 for income diversity.

4.2. Latent Class Analysis

Table 2 presents parameter estimates for the latent class analysis. Before reviewing the results, it should be noted that they are the culmination of an extensive preliminary investigation that compared alternative models with one through ten latent classes. This investigation revealed that models with more than five classes were under-identified (see note 8). Models with fewer classes were evaluated using the AIC, BIC, and substantive interpretability as the criteria for selection. A five-class solution was optimal.

Table 2. Latent Class Profiles from 5 Class Model, White, Mexican and Central American Adults.

	Non-Latino Whites	Latinos
	U.S. born & advantaged	U.S. born & advantaged	U.S. born, young & jobless	Documented & Older	Undocumented & disadvantaged
	(1)	(2)	(3)	(4)	(5)
Percent in latent class	0.43	0.17	0.10	0.24	0.07
Ethnicity
Non-Latino White	0.99	0.01	0.15	0.00	0.00
Latino	0.01	0.99	0.85	1.00	1.00
Immigration Status
U.S. born	0.87	0.74	0.85	0.00	0.02
Documented	0.13	0.24	0.08	0.73	0.03
Undocumented	0.00	0.02	0.06	0.27	0.96
Other Characteristics
Age < 30	0.15	0.53	0.96	0.00	0.55
Less than high school	0.07	0.11	0.34	0.78	0.76
No English	0.00	0.07	0.04	0.92	0.99
Employed	0.74	0.93	0.30	0.71	0.58
Poor	0.22	0.14	0.36	0.21	0.71
Unweighted N = 1,789

Data Source: Los Angeles Family and Neighborhood Survey

This table is organized as follows. Row labels identify the eight variables that were entered into the LCA. Column labels identify each class that emerged from the analysis. The top value in each column shows the overall proportion of respondents in each class, which sum to 1.0 across columns. The remaining values within each column are item-response probabilities, which indicate the likelihood of having a particular characteristic conditional on latent class membership. These values were used to create substantively meaningful labels for each class.

The results strongly indicate that ethnicity and immigration status are major axes of differentiation, as the item-response probabilities for these variables are high within each latent class. As shown in the first column, the largest latent class includes 43% of all cases. Almost all of the individuals in this class are U.S.-born and advantaged whites, with item-response probabilities for ethnicity and nativity of .99 and .87 respectively. Their advantages are evident from a high probability of employment and access to various forms of capital. None are unable to speak English and few have less than a high school education or are impoverished.

The remaining four latent classes are either overwhelmingly or exclusively Latino. Nearly a fifth (17%) of cases fall into the U.S.-born and advantaged Latino class. These individuals are very unlikely to have less than a high school education, to lack English proficiency, and to live in poverty. In contrast, one-tenth of cases form a separate class labeled U.S.-born, young and jobless Latinos. This class is almost exclusively under age 30 and the least likely to be employed, despite being able to speak English proficiently. These young native-born Latinos are also more likely than the other native-born groups to have low education and to live below the poverty threshold, but neither of these characteristics defines the class per se.

The last two latent classes are comprised exclusively of foreign-born Latinos. Specifically, one fourth of the study population falls into the documented and older Latino class that is incorporated into the labor market. Without anyone under the age of 30, this group is defined by a high likelihood of employment and a low likelihood of poverty, despite low levels of formal education and English proficiency. In contrast, the undocumented and disadvantaged Latino class (7%) is comprised of individuals who are disadvantaged on nearly every indicator. Lacking cultural (English proficiency) and human (formal education) capital, this group has the second lowest likelihood of employment and the highest likelihood of falling below the poverty line. This class is formed by those who are the most disadvantaged along multiple axes of inequality.

Overall, the analysis shows substantial clustering of potential disadvantages. Undocumented Latinos (column 5) are positioned most precariously with respect to their endowments of personal resources and rights. Advantaged U.S.-born whites (column 1) and Latinos (column 2) are at the opposite end of the spectrum from undocumented Latinos. If the intersection of multiple disadvantages is associated with high perceived discrimination, the undocumented group would be expected to report the highest levels of everyday discrimination and major events signaling institutional discrimination. Advantaged native-born whites and Latinos should report the lowest levels of negative treatment. These expectations are also consistent with the assimilation perspective. On the other hand, the ethnic resilience perspective posits that for ethnic groups, perceived discrimination may be heightened by U.S. birth and higher social status. This suggests that Latinos in the two latent classes dominated by the native born may have the highest levels of perceived discrimination.

The multilevel models that follow address how latent class membership and characteristics of the neighborhood context are associated with perceived discrimination in interpersonal and institutional domains. As a preliminary step, Table 3 shows how the five latent classes vary with respect to the dependent variables and neighborhood diversity. The relationships between the dependent variables and latent class membership are explored in detail in subsequent models, but a comparison of the means by latent class shows that the two U.S.-born Latino groups have the highest means on each of the measures of perceived discrimination. With respect to the neighborhood characteristics, U.S.-born whites tend to live in more ethnically diverse neighborhoods than Latinos. Their mean racial/ethnic entropy score is .78, while those for the Latino groups range from .58-.67. As for income, whites tend to live in the least diverse areas with a value of .46. This is lower than the values for the four Latino groups, who live in neighborhoods with similar income diversity scores (.58-.61). Such results suggest that mean levels of neighborhood ethnoracial diversity vary more than mean levels of income diversity, with little variation in the latter across latent classes for Latinos.

Table 3. Dependent Variables and Neighborhood Characteristics by Latent Class, non-Latino White, Mexican and Central American Adults.

	Non-Latino Whites	Latinos
	U.S. born & advantaged	U.S. born & advantaged	U.S. born, young & jobless	Documented & Older	Undocumented & disadvantaged
	(1)	(2)	(3)	(4)	(5)
Percent in latent class	0.43	0.17	0.10	0.24	0.07
Dependent variables
Everyday discrimination (log transformed)	2.55	2.64	2.69	2.35	2.35
Major experiences of discrimination	0.26	0.43	0.44	0.20	0.30
Discrimination experience with police	0.10	0.23	0.23	0.05	0.05
Neighborhood characteristics
Racial and ethnic entropy	0.78	0.67	0.62	0.58	0.60
Income diversity	0.46	0.58	0.59	0.61	0.61
Unweighted N = 1,789

Data Source: Los Angeles Family and Neighborhood Survey; ACS 2005-2009

4.3. Multilevel models of discrimination

4.3.1. Everyday discrimination

Results for the multilevel analysis are shown in Table 4. Focusing on everyday discrimination, the first two columns present estimates generated from a bivariate model and a multivariate model that includes all individual and neighborhood characteristics (Model 1).¹² The reference category for these comparisons is U.S.-born, young and jobless Latinos. Although insights from models using alternative reference groups are provided below, these results show that U.S.-born, young and jobless Latinos report the highest level of everyday discrimination regardless of the model under consideration. Put differently, each other latent class reports less experience with day-to-day interpersonal slights than the reference group. In short, neither the most disadvantaged (undocumented and disadvantaged Latinos) nor the most advantaged Latino group (U.S.-born and advantaged Latinos) is particularly likely to report experiencing interpersonal discrimination in everyday life. This would not be predicted by hypotheses that emphasize the intersection of multiple disadvantages among the former group or the potential for ethnic resilience among the latter. On the contrary, these results suggest that ethnic consciousness and awareness of discriminatory behaviors may be highest among young native-born members of minority groups, who often struggle with their identity and finding a foothold in adult roles. Young native-born Mexicans and Central Americans undoubtedly compare themselves to their native-born age mates in other ethnoracial groups, yet they often experience limited opportunities due to low family SES or spillover effects related to rising anti-immigrant sentiment. Additionally, Model 1 challenges arguments about discrimination based on gender. Women report lower levels of everyday discrimination than men do.

Table 4. HLM Analyses of Everyday Discrimination and Major Experiences of Discrimination: White, Mexican, and Central American Adults.

	Everyday Discrimination				Major Experiences of Discrimination						Discrimination Experience with Police
											Odds Ratios
	Bivariate		Model 1		Bivariate		Model 1		Model 2		Bivariate		Model 1		Model 2
Latent disadvantage status
U.S. born & advantaged whites	-0.28	***	-0.29	***	-0.37	**	-0.31	*	-0.31	*	0.76	***	0.77	***	0.76	***
U.S. born & advantaged Latinos	-0.15	*	-0.15	*	-0.26	+	-0.23		-0.23		0.85	*	0.85	*	0.85	*
U.S. born, young & jobless Latinos	ref.		ref.		ref.		ref.		ref.		ref.		ref.		ref.
Documented & older Latinos	-0.44	***	-0.43	***	-0.38	***	-0.37	**	-0.34	**	0.74	***	0.75	***	0.74	***
Undocumented & disadvantaged Latinos	-0.36	**	-0.36	**	-0.34	*	-0.33	*	-0.30	+	0.75	***	0.75	***	0.75	***
Other individual characteristics
Female	-0.15	***	-0.14	***	-0.18	**	-0.15	**	0.25		0.86	***	0.86	***	1.07
Neighborhood characteristics
Racial and ethnic diversity	0.14	**	0.10	*	0.02		-0.04		-0.04		1.05		1.02		1.02
Income diversity	0.07		0.08		0.46	**	0.45	*	0.65	***	1.26	***	1.21	*	1.36	***
Cross-level interactions
Income diversity * female									0.03	*					0.66	**
Constant			2.78	***			0.44	**	0.32	*			1.37	***	1.30	***
Log-likelihood			-691				-1657		-1632				-376		-311
ICC			0.57				0.53		0.53				0.52		0.52

Data Sources: Los Angeles Family and Neighborhood Survey; ACS 2005-2009

Analysis based on weighted data

⁺p<.10

^*p<.05

^**p<.01

^***p<.001

As for neighborhood characteristics, the association between ethnoracial diversity and everyday discrimination is positive. This significant association indicates that individuals in racially-diverse neighborhoods report more frequent experiences with negative interpersonal treatment (irrespective of their personal characteristics). In contrast, everyday discrimination is unrelated to the income diversity of neighborhoods.

Separately tests were conducted for cross-level interactions between each measure of neighborhood diversity and latent disadvantage status. Both sets of interaction terms were nonsignificant, as were interactions between the neighborhood diversity measures and respondent gender. Thus, the effects of latent status and gender are not conditioned by either measure of neighborhood context.

To shed further light on the association between latent disadvantage status and perceived everyday discrimination, we ran two sets of supplementary models (not shown) in which we shifted the reference group. These models provide additional information pertinent to evaluating the assimilation perspective. With U.S.-born and advantaged whites treated as the reference group, U.S.-born and advantaged Latinos and U.S.-born, young, and jobless Latinos had significantly higher levels of everyday discrimination, while documented and advantaged Latinos perceived less. This pattern held across all models and the contrast between non-Latino whites and undocumented Latinos was nonsignificant. With undocumented and disadvantaged Latinos treated as the reference group, the two U.S.-born Latino groups again perceived more everyday discrimination. This pattern held across all models and no other contrasts were significant. The implications of these results are discussed in the conclusion.

4.3.2. Major experiences of discrimination

Turning to major experiences with discrimination in institutional domains, the results are largely consistent with those just reviewed. In the bivariate model, U.S.-born, young, and jobless Latinos report a broader range of experiences with unfair treatment in this domain than all other groups. This pattern changes little in Model 1 after gender and neighborhood characteristics are controlled, except for the attenuation of a borderline significant bivariate coefficient for advantaged U.S.-born Latinos.¹³ Furthermore, gender remains negatively related to major experiences of discrimination in institutional domains. Women are less likely to report discriminatory treatment.

Neighborhood context also matters, but the results differ from those for everyday discrimination. Reports of major experiences of discrimination by institutional agents are unrelated to ethnoracial diversity, but positively associated with income diversity. As before, tests for cross-level interactions between each measure of neighborhood diversity and latent disadvantage status were nonsignificant. However, tests for cross-level interactions involving gender revealed that the positive relationship between income diversity and reports of experiences with institutional discrimination holds for men, but not women (see Model 2).

In models in which the reference group was changed to U.S.-born and advantaged whites (not shown) there was only one significant contrast. U.S.-born, young, and jobless Latinos reported more experience than their advantaged white counterparts with unfair treatment in institutional domains. Similarly, when the reference group was shifted to undocumented and disadvantaged Latinos (not shown), the only group that was significantly different was U.S.-born, young, and jobless Latinos. Once again, the multiple disadvantages experienced by undocumented Mexican and Central American immigrants do not together combine to heighten perceived discrimination.

4.3.3. Additional analyses

As a final step, we replicated the analyses for each of the component items included in both indexes. For everyday discrimination, the results were essentially identical across items. For major discriminatory experiences in institutional domains, the results for latent disadvantage status were driven by reports of unfair treatment by the police. In the models for the other types of discriminatory events, there were no significant results for latent status. Consequently, the last three columns of Table 4 report odds ratios from logistic regression models of unfair treatment by the police. Both the bivariate models and Model 1 show that reports of negative encounters with the police are most likely among young, native-born Latinos and males. As was the case for the overall index, neighborhood income diversity is positively associated with reports of unfair treatment by the police (see AUTHOR, DATE for a similar finding). As discussed below, this could reflect a greater police presence, styles of policing, and sensitivities of populations in areas characterized by greater economic diversity. Moreover, the significant cross-level interaction between neighborhood income diversity and gender in Model 2 indicates that the former influences men's—but not women's—perceptions of unfair police practices. Men may be more sensitive to heightened policing because they have a higher level of engagement with the police or are more often targets of negative police behavior.

5. Discussion and conclusion

This research contributes to a growing literature on perceived discrimination among Latinos. Consideration of this topic requires attention to the roles of assimilation and ethnic resilience in assessments of unfair treatment (Portes and Bach 1985). It also requires attention to role of legal status because undocumented individuals are likely to be marginalized during an era marked by the resurgence of anti-immigrant sentiments. Legal status has received little attention in empirical studies of perceived discrimination. Yet legal status is potentially important as a determinant of expectations for equal treatment because it affects the standing of residents in relation to the rights of membership in society. This source of inequality is also intertwined with other sources of inequality that are normally difficult to disentangle.

We addressed this topic using an analytic strategy that highlights the intersectionality of multiple forms of disadvantage. The intersectionality framework proposes that individual experiences must be examined in the context of simultaneous memberships in different status groups. In emphasizing constellations of multiple statuses, this framework requires a person-oriented approach instead of a variables-oriented approach (Bergman and Magnusson 1997). Rather than examining each indicator of inequality separately and with controls for all others, a person-oriented approach delineates major subgroups of individuals based on similar patterns of intersecting characteristics (Collins and Lanza 2010). An important innovation of this study is the use of latent class analysis to identify these major subgroups. While the concept of intersectionality has garnered widespread attention, quantitative methods for delineating key intersections and incorporating them into empirical research have lagged behind (Collins 2015). Latent class analysis provides a promising methodology for understanding how major constellations of statuses that jointly construct inequalities influence outcomes.

Based on self-reports, perceptions of discrimination in interpersonal and institutional domains were examined. Interpersonal discrimination in daily life was assessed by the reported frequency of various forms of negative treatment in interactions that are inconsistent with social equality between individuals. Institutional discrimination was assessed by reported experiences with unfairness and inequity in treatment by agents of institutions. The findings regarding both of these do not show that individuals who experience multiple disadvantaged statuses are the most likely to report discrimination. This is particular the case for those individuals whose disadvantages are anchored in lack of documentation. Undocumented Mexican and Central American residents of Los Angeles who have low education and few financial resources do not report elevated levels of unfair treatment. Consequently, the experience of those at the bottom of multiple status hierarchies is difficult to characterize as “exceptional.”

Such results provide important counterpoints to narratives that reflect the dominant public ideology about the “equalizing” role of assimilation and the structure of opportunity in American society. According to the dominant ideology, forms of discrimination that are disproportionately experienced by immigrants should wither over generational time with the absorption of their native-born children into the social, cultural, and economic mainstream. In contrast to this expectation, experiences with both types of discrimination are the most common among U.S.-born Latinos at the intersection of ethnicity (Mexican/Central American), employment (low), and age (less than 30). Young native-born Latinos are favorably positioned in terms of nativity and legal status, but they are neither the most advantaged nor the least advantaged overall when all factors are considered simultaneously. In short, this segment of the population is not insulated from negative treatment due to their nativity.

Such findings offer greater support for the ethnic resilience perspective, which emphasizes how exposure to the host society may lead to more contact with majority group members, rising ethnic consciousness, and awareness of inequality and negative stereotypes (Flippen and Parrado 2015; AUTHOR, DATE; Portes and Bach 1985). The ethnic resilience perspective also posits that incorporation into the socioeconomic mainstream through educational attainment and English mastery may be associated with greater perceived discrimination, not less. Indeed, our findings that young U.S.-born Latinos and males are especially likely to report experiences with discrimination are consistent with claims that native-born minority men are especially vulnerable to negative experiences in institutional domains (Perez, Fortuna and Alegría 2008). Young native-born Latinos may have higher expectations for equitable treatment. When thwarted, these expectations may become the basis for frustration and heightened sensitivity to discriminatory treatment.¹⁴

An additional important finding is that young native-born Latino males are the most likely to regard their treatment by the police as unfair. Although young men are the demographic group that is most likely to be involved in illegal activities (Tittle, Ward and Grasmick 2003), there is more to the story than their likelihood of committing crimes or the prevalence of neighborhood crime. A study of encounters with police concluded that Latinos (and African Americans) were especially likely to be “over-stopped,” “over-frisked,” and “over-arrested” in Los Angles in 2003-2004 (Ayres and Borowsky 2008). Such results concur with those of studies conducted among other groups in other places. For example, Author (Date) showed that negative contact with the police was particularly likely to be mentioned by young men among Dominican immigrants in a new immigrant destination. Thus, attention to perceived discrimination by agents of institutions should pay particular attention to law enforcement agencies.

Another understudied issue is whether and how neighborhood ethnoracial and income diversity influence perceptions of discrimination. These two dimensions of diversity do not go hand in hand (Tach and Lee 2016); nor are they related to interpersonal discrimination and institutional discrimination in similar ways. Consistent with prior studies, individuals report more routine interpersonal discrimination when they live in neighborhoods that are racially and ethnically diverse (Dailey et al., 2010; Hunt et al., 2007; Stewart et al., 2009). Since neighborhoods may be sites of frequent face-to-face interactions, it is not surprising that ethnoracial diversity would increase intergroup interaction, micro-aggressions, and interpersonal slights.

Neighborhood income diversity is another matter. Unrelated to everyday discrimination, income diversity is positively associated with reports of unfair treatment by agents of institutions. This is especially the case for interactions with the police. This could stem from an increase in crime rates with income diversity at the neighborhood level (Hipp 2007, 2011). Higher crime rates, in turn, might heighten demands by the relatively affluent for a greater police presence and more aggressive policing (Smith 1986; Weitzer 2000). Together, these factors could give rise to police-civilian encounters that residents regard as unfair and unwarranted. It is unsurprising that this relationship holds just for men, given that men are more likely than women to be the targets of forceful police behavior.

In closing, an important aim of our research was to examine the role of legal status in discrimination among Mexicans and Central Americans. This is especially important in Los Angeles County, which receives more undocumented and documented immigrants from Mexico and Central America than any other U.S. county (Zong and Batalova 2015 Zong and Batalova 2016). Our results are unequivocal---undocumented Latinos do not report higher levels of discrimination than their documented or U.S.-born co-ethnics. While one might be tempted to attribute this to the fact that Los Angeles County is a majority-minority county and a major destination for Latino immigrants, Flippen and Parrado (2015) find a similar pattern in Durham, North Carolina, a new destination for Latino immigrants. In short, despite widespread debates about immigration and public animosity toward those who enter without authorization, undocumented Latinos do not appear to be more likely than others are to interpret their experiences as unfair or discriminatory. Nonetheless, our understanding of actual and perceived discrimination in an increasingly diverse America would benefit from future research on the role of legal status in additional communities and regions.