The Residential Segregation of the Middle Eastern and North African and South Asian Populations from the White Population in U.S. Metropolitan Areas, 2012–2016

Abstract

We examine the residential segregation of the Middle Eastern and North African (MENA) population, South and East Asian people, and Black from white people in the U.S. Using data from the 2012–2016 American Community Survey (ACS) and the 2012–2016 Integrated Public Use Microdata Sample (IPUMS) at the metropolitan level, descriptive analyses of segregation reveal that Black–white segregation is significantly greater than the segregation of MENA and East Asian people from white people. South Asian–white segregation is higher than Black–white segregation, but the difference is not statistically significant. Multivariate analyses of average dissimilarity indices show that relative to Black–white segregation, MENA–white, South Asian–white, and East Asian–white segregation are not significantly different after controlling for relevant variables. The results for the isolation index also follow a similar pattern. While MENA and both Asian ancestry groups are significantly less isolated than Black people in the unadjusted results, the differences in average isolation indices between Black people and these groups disappear after controlling for relevant characteristics. The results suggest evidence that supports these hypotheses in terms of spatial assimilation.

1. Introduction

Research on residential segregation in the United States has focused on traditional racialized groups, such as the non-Hispanic Black (hereafter “Black”), Hispanic, and Asian people (Massey and Denton 1993; Kim and White 2010; Iceland et al. 2014). Studies in this tradition collectively indicate that Black people are the most segregated from non-Hispanic white people (hereafter “white”), and Hispanic and Asian people are moderately segregated from whites (Massey and Denton 1993).

The residential segregation of the populations identifying as Middle Eastern and North African (hereafter “MENA”)1 and South Asian2 has largely been ignored. The focus of Bozorgmehr et al. (1996) on Middle Eastern people (Iranian, Armenian, Israeli, and Arab people) and Holsinger (2009) on Arab residential segregation are partial exceptions. These studies are limited to examining the Arab MENA population or geographically by only focusing on New York, Los Angeles, Chicago, and Detroit. These studies are restricted in their use of older data from 2000 or earlier. Only a small amount of research explores the residential segregation of South Asian people, and it is limited to specific places and relies on older data (Iceland et al. 2014). The existing research finds that these groups have moderate levels of segregation from white people, thereby necessitating more research on this topic (Holsinger 2009; Iceland et al. 2014).

Between 2000 and 2019, the MENA and South Asian populations within the U.S. doubled in size, from about 600,000 to 1.2 million and from 1,900,000 to 4,606,000, respectively, and are now fairly dispersed throughout the country (Harjanto and Batalova 2022; Pew Research Center 2011). Given this tremendous growth in these populations, we re-examine their segregation with more recent data and at a national level. We choose to focus particularly on MENA and South Asian people because of the housing discrimination they experience and the widespread stereotypes that exist about these two groups, both of which could affect their residential segregation from white people. Several studies have identified bias and discrimination in the housing market against Arab, Muslim (Carpusor and Loges 2006; Ahmed et al. 2010; Hogan and Berry 2011; Gaddis and Ghoshal 2015; Friedman et al. 2019), and Asian people (Turner and Ross 2003; Wong et al. 2011). South Asian people have been mistaken as MENA and/or Muslim because of their phenotypical similarities and common religious indicators (e.g., turban for Sikhs and hijab for Muslims), and as a result of this mistaken identity, they have been subjected to racial discrimination (Maira 2004; Love 2017; Lee and Kye 2016).

The purpose of this paper is to build upon the very few studies that have examined the residential segregation of the MENA and South Asian populations from white people in the United States and compare their segregation from white people to that of East Asian and Black people. Our analysis is guided by the following research two questions: (1) What is the level of segregation of the MENA population and South Asian people from white people compared to that of East Asian and Black people across metropolitan areas in the U.S., and how does segregation vary by metropolitan areas? (2) Does controlling for group differences in immigrant-related, socioeconomic, and demographic characteristics attenuate group differences in residential segregation of these minority groups from white people? Our metropolitan-level analyses rely on Summary File (SF) data from the 2012–2016 American Community Survey (ACS). We also utilize data from the 2012–2016 Integrated Public Use Microdata Series (IPUMS) to create several group-level predictors of segregation that are absent from the ACS data.

2. Theories on Residential Segregation

Sociologists have analyzed residential segregation by relying on two primary theoretical approaches: the spatial assimilation and place stratification models (Massey and Denton 1993; Charles 2003). Spatial assimilation refers to the process of residential attainment by groups to live near the majority members of the society (Massey 1985). The spatial assimilation model contends that groups’ residential patterns are shaped by their degree of acculturation and socioeconomic advancement (Massey 1985). The residential isolation of immigrants is thought to be a temporary phenomenon; residential segregation from the mainstream population will gradually decline as groups attain increased cultural familiarity and socioeconomic status.

There is a large body of scholarship on the residential segregation of groups from white people that supports the general tenets of the spatial assimilation model—that advancement in socioeconomic status (SES) results in spatial assimilation (Timberlake and Iceland 2007; Fong and Wilkes 2003). As immigrants accumulate cultural, human, and economic capital, they achieve upward residential mobility by investing in higher-quality housing and neighborhoods with more resources—neighborhoods where middle-class, non-Hispanic white people often reside (Logan et al. 2002). Such transitions have been observed for European (Fong and Wilkes 2003), Asian (Iceland et al. 2014; Timberlake and Iceland 2007), Arab (Holsinger 2009), and Latin American people (Iceland et al. 2014; Massey and Denton 1993).

Relying on spatial assimilation theory, we expect that when members of minority groups acculturate and advance their socioeconomic attainment, they will convert these gains into locations in whiter neighborhoods. It is expected that in descriptive analyses, the MENA population and South and East Asian people may be more segregated from the White population than Black population because greater shares of the former groups are immigrants; on the other hand, because these immigrant groups’ educational attainment and income may be greater than that of Black people, the residential segregation from white people of the MENA population and South and East Asian people could be somewhat lower than Black–white segregation (Bozorgmehr et al. 1996; Schachter 2014). According to the spatial assimilation model, however, after controlling for variation in acculturation and socioeconomic characteristics across these groups, differences in segregation from white people should diminish or disappear among the MENA population and South and East Asian and Black populations.

Nevertheless, such patterns are not generalizable to every non-White group in the United States. The existence of racism and discrimination have been used as an apparatus by dominant groups, hindering the spatial assimilation of minority groups, thereby giving rise to a second theoretical perspective, the place stratification model (Charles 2003; Massey and Denton 1993). The place stratification model suggests that enduring prejudice and discrimination by powerful groups and structural factors that distribute housing opportunities unequally across race, ethnicity and nativity prevent minority access to residential proximity with white people, the majority group (Charles 2003; Massey and Denton 1993). Discrimination practices include realtors steering minority populations to specific neighborhoods where their race and ethnicity predominate and housing providers constraining home-seekers’ access to housing on the basis of their race and ethnicity (Besbris 2020; Korver-Glenn 2021; Iceland and Nelson 2010; Massey and Denton 1993; Turner and Ross 2003).

The subjectivity in the perception of MENA and South Asian groups by others, particularly society’s majority-group members, leads to negative interactional experiences, and these groups may face greater residential segregation than other groups (e.g., East Asian people), especially considering this longstanding and negative sociopolitical climate that configures them as inherently criminal and potential enemies of the state (e.g., President Trump’s “Muslim ban”) (Love 2017). Both Arab and South Asian people report being marginalized by policies that single them out as a group that is dangerous and potentially subversive and express concerns about their civil rights and safety (Cainkar and Maira 2005). The report published by the Council on American-Islamic Relations (CAIR) indicates a dramatic increase in civil rights and hate crime complaints, from 48% to 62% filed by Muslims from 2007 to 2016, respectively (CAIR 2009, 2017). Among Asian people, Asian Indian people are the most likely to report being a victim of a hate crime in the U.S. (Wong et al. 2011).

The MENA and South Asian populations have been subject to various forms of discrimination in the housing market. The National Fair Housing Alliance and other studies have documented bias, discrimination, harassment, and hate-motivated practices toward Muslim and Arab people (Carpusor and Loges 2006; Tehranian 2009; Ahmed et al. 2010; Hogan and Berry 2011; Wong et al. 2011; Gaddis and Ghoshal 2015; Garner and Selod 2015; Friedman et al. 2019). Black Muslims have experienced residential disadvantages, even after accounting for differences in socioeconomic status (Friedman et al. 2019). When selecting potential roommates, those with Arab-sounding names are only 57 percent as likely as those with white-sounding names to receive a response to a roommate request (Gaddis and Ghoshal 2015), and Arab names receive significantly fewer positive responses when attempting to rent in corporate and privately owned apartment complexes (Carpusor and Loges 2006). Similarly, Asian Indian people experienced the highest racial and immigrant-based discrimination in the United States compared to other Asian groups (e.g., Chinese, Filipino, Japanese, Korean, and Vietnamese people) in the workplace, employment, neighborhood and communities (Wong et al. 2011; Jeung and Lim 2012; Schachter 2014).

It is unclear how discrimination against the MENA and South Asian populations will translate into differences in their residential segregation from white people, especially as it compares to Black–white segregation. The place stratification model offers three possible scenarios: (a) the residential segregation of the MENA and South Asian populations from white people will be greater than Black segregation from white people even after taking into account acculturation, socioeconomic, and demographic characteristics; (b) taking into consideration the longstanding housing discrimination that Black people have faced in the United States, it is expected that the residential segregation of the MENA and South Asian population from white people will be similar to Black segregation from white people, controlling for relevant factors; or (c) given the history of residential segregation that has disadvantaged the Black population for decades relative to other non-Black groups, it is expected that the residential segregation from white people of the MENA and South Asian populations will be lower than Black people, controlling for other factors.

3. Data, Measures, and Methods

3.1. Data

We use two data sources for this research provided by the University of Minnesota: (1) the 2012–2016 American Community Survey (ACS) Summary File (via the National Historical Geographic Information System—NHGIS) and (2) the 2012–2016 Integrated Public Use Microdata Sample (IPUMS)—5% sample (IPUMS USA) (Ruggles et al. 2022). Data from the ACS are extracted at two levels of analysis—the census-tract level and the core-based statistical area (CBSA) level. We restrict our analysis to only metropolitan CBSAs because the accuracy of our calculation of segregation measures relies on larger areas that exceed the population size of micropolitan areas. ACS tract-level data are used to calculate the dependent variables—dissimilarity and isolation indices. The ACS CBSA-level data are used to construct some of the control variables for our analysis.

We use IPUMS data to calculate percentages and average characteristics at the CBSA level for each group—MENA, South Asian, East Asian, and Black—including English language proficiency, educational attainment, income, nativity status, housing tenure, residence in the suburbs, marital status, presence of children, and age. These variables are our focal independent variables and control variables used in our multivariate analyses (discussed in more detail below). The use of these data is critical because the aggregated 2012–2016 ACS data in the Summary File do not contain summary tables for these characteristics by our specific groups (e.g., educational attainment of the MENA population). The IPUMS data maintained by the University of Minnesota provides a CBSA code that we use to aggregate these characteristics for each group by each metropolitan area (Ruggles et al. 2022).

We limit the analytical sample to CBSAs with at least 40003 members of each group; this yields a total sample of 192 metropolitan CBSAs from our 2012–2016 ACS data. We do not include metropolitan areas with less than 4000 for each group because, in such metropolitan areas, the calculation of the segregation indices is less accurate (Iceland et al. 2002). The units upon which our measures of residential segregation are constructed are census tracts, which is consistent with the approach of previous research (Iceland et al. 2002; Massey and Denton 1993).

The ACS data are the best available for studying MENA individuals in the United States (Holsinger 2009). Despite there being no consensus on who comprises the MENA population among scholars or well-known research centers (e.g., Pew Research Center, United Nations, U.S. Census Bureau, and the World Bank), the literature converges on the fact that the MENA category is comprised of a multiethnic, multicultural, multilingual and multinational population—which usually includes Egyptians, Iraqis, Jordanians, Saudi Arabians, Lebanese, Moroccans, Palestinians, Syrians, Somalis, Sudanese, Armenians, Israelis, Cypriots, Iranians, Afghans, and Turks (Haddad 2009; Marvasti and McKinney 2004; Awad 2010; Bozorgmehr et al. 1996). Using the ancestry question4, we create the category of MENA from individuals who identified as Egyptian, Iraqi, Jordanian, Lebanese, Moroccan, Palestinian, Syrian, Turkish, Iranian, Afghan, Arab, and Other Arab5. That is, we include explicit responses to ancestry or origins from seven of these nationalities. Yemeni, Kuwaiti, and Saudi Arabian are not choices for the ancestry question. We suspect that people who identify as having ancestry or origins from those countries may identify as Arab or Other Arab, which are categories that we include in our definition of MENA6.

In accordance with the purpose of this study, we use the following inclusion and exclusion criteria in generating the MENA category: (1) self-identification, (2) common categorization by former scholars, and (3) Muslim or Arab country of origin (or both). We argue that it is important to consider self-categorization, as it conveys personal experiences in the wider society. For instance, although previous scholars and many research institutions include Armenian, Israeli and Cypriot people under the MENA moniker, the overwhelming majority of these people identify themselves as white7 rather than MENA (Mathews et al. 2017). Among MENA groups who come from Muslim countries, only Somalis, Sudanese, Afghans and Turks do not identify with the MENA category. The majority of the former two groups self-identified as Black, while the latter groups identified with some other race or as white (Mathews et al. 2017). Afghan and Turkish populations are included in this study because they have been categorized as MENA populations by previous scholars (Bozorgmehr et al. 1996; Bakalian and Bozorgmehr 2011; Marvasti and McKinney 2004) and come from predominantly Muslim countries (Pew Research Center 2015). That is, we added ancestry or origins from the countries Morocco, Palestine, Turkey, and Afghanistan to our definition of MENA countries because other research has also included these countries (Bozorgmehr et al. 1996; Marvasti and McKinney 2004; Bakalian and Bozorgmehr 2011), and these countries’ populations are at least 98% Muslim (Pew Research Center 2015). One potential limitation of our approach is that we do not know if people who claim ancestry or origins from these countries are practicing Muslims; we only know that they identify with a country where the religion is largely Muslim.

We also created a South Asian category from respondents who reported their ancestry or origins as Asian Indian or Sri Lankan8. Although Bangladeshis and Pakistanis are part of the South Asian population, we exclude them from the South Asian category as they come from majority-Muslim countries (91 percent and 96 percent, respectively), and we do not want to conflate geographic region with religion in our MENA and South Asian categories. We compared the percentage of respondents identifying their ancestry or origins from India or Sri Lanka to those that identify their race as Indian or Sri Lankan in order to determine how well the ancestry answers correspond to the race-based answers, the latter of which are more commonly used; 1.12% of respondents answered those categories in the ancestry question, and 1.29% answered them for the race question, thereby revealing a fairly strong concordance between the responses to these questions. We create an East Asian category from individuals identifying as Chinese, Korean, Taiwanese, and Japanese. We found that 2.52% of respondents answered these categories in responding to the ancestry question, and 2.49% of respondents answered them in responding to the race question, revealing a high level of concordance in the responses to these questions. Finally, we create the white and Black groups using the race and ethnicity question on the ACS and counting those that identified as “non-Hispanic White” and “non-Hispanic Black,” respectively, in those categories. In order to keep the white and MENA categories distinct, we subtract the MENA tract count from the non-Hispanic White count9.

3.2. Measures

Dependent variables. To measure the residential segregation of the MENA, South and East Asian populations, and Black population from White populations, we use two commonly used segregation measures: the dissimilarity index and the isolation index (see Logan and Stults 2011; Massey and Denton 1988; Massey and Tannen 2015). The dissimilarity index is a measure of the evenness of two populations and ranges from 0 (complete integration) to 100 (complete segregation). Conceptually, it describes the percentage of any members of either of the two groups that would have to change their residence to achieve an even racial and ethnic distribution in the broader metropolitan area (Iceland 2009).

The isolation index is a measure of residential isolation, which refers to the level of potential interaction or contact between minority or majority group members and members of their own group. In Massey and Denton’s words, the isolation index will “capture the experience of segregation from the viewpoint of the average” minority or majority member (Massey and Denton 1993, p. 65). It also ranges from 0 (lowest level of isolation) to 100 (highest level of isolation). Unlike the dissimilarity index, the isolation measure is sensitive to the number of the group in the CBSA.

Both measures are important to include in any analysis of residential segregation because they gauge different dimensions of segregation—the evenness of the group relative to white people versus the isolation of the non-White group. Although MENA and South Asian people compose small shares of the population in any given metropolitan area, their isolation can still be relatively high. For example, Logan and Stults (2011) found that although Asian people compose over 6% of the population in the U.S., the average isolation index across metropolitan areas is 25%, which is notably high given how small their size is in the population. Given the importance of the use of the dissimilarity and isolation indices in previous research (Massey and Denton 1993; Iceland et al. 2002; Logan and Stults 2011), we employ these measures in this analysis.

Focal independent variables. Our analyses use two sets of focal independent variables. The first set is the dummy variables that allow us to compare the MENA–white, South Asian–white, and East Asian–white segregation scores to the Black–white segregation scores, controlling for other relevant factors. The second set of focal independent variables focuses on the acculturation characteristics and socioeconomic status of groups. We measure the acculturation characteristics of groups10 within each CBSA using: (1) the percentage of each group that reported speaking English only or very well and (2) the percentage that are born in the U.S. The socioeconomic characteristics within each CBSA are measured using the: (1) average household income (in 2016 USD) for each group; (2) the percent education level with BA or more; and (3) the percentage of each group that own their homes.

Control variables. For each CBSA, we also create the following control variables for each group: (1) the percentage living in suburban areas; (2) the percentage of households headed by a married couple; (3) the percentage with children; and (4) the average age. Metropolitan-area level characteristics are also associated with residential segregation (Logan et al. 2004; Wilkes and Iceland 2004). To account for variation in these characteristics, we draw on data from the ACS SF file for each CBSA and include the following variables: (1) the log of the CBSA population size; (2) the number of each group in the CBSA; (3) the percentage employed in: (a) manufacturing and (b) the government; (4) the percentage of the population in the armed forces; (5) the percentage of the population age 3+ enrolled in college, graduate school, or professional school; (6) the percentage of the population living in housing built since 2000; (7) the percentage of non-white population; and (8) region—Northeast, Midwest, South or West with Northeast as reference category.

3.3. Methods

Analytic approach. To measure the degree to which residential segregation is explained by acculturation, SES, and other independent variables, we conducted multivariate analyses. We used OLS, and we adjusted the standard errors because there are multiple observations within each metropolitan area.

Formally—and generally—the multivariate OLS models can be written as shown in the equation below:

$$Y_{gj}={\beta }_0+{{\displaystyle \sum }}_{g=1}^3{\alpha }_g\left(X_{gj}\right)+{{\displaystyle \sum }}_{k=1}^K{\beta }_k\left(S_{kgj}\right)+{{\displaystyle \sum }}_{l=1}^L{\gamma }_l\left(C_{lj}\right)+{\epsilon }_{gj}$$

where $Y_{gj}$ is the score on the dependent variable for group g in metropolitan area j, ${\beta }_0$ is the constant, representing the predicted value on the dependent variable for the omitted group (Black people) when all other variables in the model are scored 0, the α_g are the increments or decrements to the constant for the three included groups (MENA, South Asian, and East Asian), the ${\beta }_k$ are average effects across groups of the K acculturation and SES measures (such as income, which varies by group), the γ_l are average effects across CBSAs of the L CBSA-level measures (such as region, which does not vary by group), and ${\epsilon }_{gj}$ is an error term, representing unexplained variation in the dependent variable11. The primary coefficient of interest is the α_g, which indicates the extent to which MENA, South Asian, and East Asian segregation—compared to white people—is greater or less than the level of segregation experienced by Black people compared to white people.

Our analysis proceeded as follows. First, we presented weighted (by the total population size within CBSAs) mean indices of dissimilarity and isolation by ancestry and racial groups, and we conducted differences in means tests for those comparisons. Next, we report descriptive statistics for the focal independent and control variables by ancestry and racial groups. Then, we conducted our multivariate analyses for each of our two dependent variables—the index of dissimilarity and isolation index. Following the methods of previous research (Wilkes and Iceland 2004; Iceland and Wilkes 2006; Iceland and Nelson 2010; Friedman et al. 2022), each of the two models contained the three dummy variables (i.e., measuring MENA–white, South Asian–white, and East Asian–white, relative to our reference group Black–white) as well as the other focal independent and control variables. In each of these two models, the dummy variables indicate the nature of the differences in segregation scores between Black–white segregation and each of the other ancestry–white groups’ segregation, as well as whether those differences are statistically significant, controlling for other focal independent and control variables.

4. Results

Our descriptive results begin in

Table 1. Residential Segregation of the MENA, South Asian, East Asian, and Black Populations from Non-Hispanic Whites in CBSAs, 2012–2016.

	Index of Dissimilarity					Isolation Index
	Mean	Sig. a,b	St. Dev.	Min.	Max.	Mean	Sig. a,b	St. Dev.	Min.	Max.
Group	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
MENA	51.2	**	6.0	40.3	63.9	6.3	***	7.1	1.9	28.8
South Asian	59.9		5.0	48.1	72.6	11.1	***	4.6	1.9	18.9
East Asian	51.9	**	6.1	40.9	65.0	15.3	***	8.5	2.0	24.7
Black	59.8		10.3	33.2	78.8	47.5		15.4	3.7	68.5
N	192

^a *** p < 0.001; ** p < 0.01, indicating significance relative to Black-white segregation. ^b Shading of cells in columns 1, 6 indicates significant at p < 0.001, p < 0.01, or p < 0.05 from MENA-white segregation. Note: We use bivariate regressions to indicate whether differences in residential segregation are significant, relative to Black-white and MENA-white segregation, as indicated by footnotes ‘^a’ and ‘^b’, respectively.

. We report the mean index of dissimilarity scores and isolation index scores for our key ancestry and racial groups across CBSAs weighted by the size of the minority group. We begin by discussing the index of dissimilarity and the differences between the groups relative to the average Black–white score, which is shown by the significance in column 2. Column 1 shows that the average MENA–white index of dissimilarity across CBSAs is 51.17, meaning that just over 51% of the MENA or white population would have to move for these groups to achieve an even residential distribution. This average score falls in the “moderate” range of segregation scores, although across CBSAs, there is a maximum value of 63.89, which falls into the “high” range (Massey and Denton 1988). Appendix A,

Table A1. Metropolitan Areas with Highest and Lowest Residential Segregation of the MENA, South Asian, East Asian, and Black Populations, 2012–2016.

MENA
Highest Dissimilarity		Highest Isolation
Nashville-Davidson--Murfreesboro--Franklin, TN	63.9	Detroit-Warren-Dearborn, MI	28.8
Birmingham-Hoover, AL	62.8	Nashville-Davidson--Murfreesboro--Franklin, TN	6.61
New Orleans-Metairie, LA	62.7	Chicago-Naperville-Elgin, IL-IN-WI	5.95
Milwaukee-Waukesha, WI	62.2	San Diego-Chula Vista-Carlsbad, CA	5.90
Indianapolis-Carmel-Anderson, IN	62.1	Washington-Arlington-Alexandria, DC-VA-MD-WV	5.79
Lowest Dissimilarity		Lowest Isolation
San Jose-Sunnyvale-Santa Clara, CA	40.3	Virginia Beach-Norfolk-Newport News, VA-NC	1.93
San Francisco-Oakland-Berkeley, CA	40.6	Oklahoma City, OK	2.08
Providence-Warwick, RI-MA	41.2	Kansas City, MO-KS	2.08
Boston-Cambridge-Newton, MA-NH	42.8	Hartford-East Hartford-Middletown, CT	2.08
Washington-Arlington-Alexandria, DC-VA-MD-WV	44.3	Minneapolis-St. Paul-Bloomington, MN-WI	2.09
SOUTH ASIAN
Highest Dissimilarity		Highest Isolation
Birmingham-Hoover, AL	72.6	San Jose-Sunnyvale-Santa Clara, CA	18.9
San Antonio-New Braunfels, TX	72.1	San Francisco-Oakland-Berkeley, CA	16.9
Buffalo-Cheektowaga, NY	71.4	New York-Newark-Jersey City, NY-NJ-PA	15.4
Kansas City, MO-KS	71.4	Raleigh-Cary, NC	14.0
St. Louis, MO-IL	71.4	Dallas-Fort Worth-Arlington, TX	14.0
Lowest Dissimilarity		Lowest Isolation
Orlando-Kissimmee-Sanford, FL	48.1	Virginia Beach-Norfolk-Newport News, VA-NC	1.90
Washington-Arlington-Alexandria, DC-VA-MD-WV	51.0	Las Vegas-Henderson-Paradise, NV	2.19
San Jose-Sunnyvale-Santa Clara, CA	52.8	New Orleans-Metairie, LA	2.29
Boston-Cambridge-Newton, MA-NH	54.8	Miami-Fort Lauderdale-Pompano Beach, FL	2.61
Austin-Round Rock-Georgetown, TX	57.2	Salt Lake City, UT	2.65
EAST ASIAN
Highest Dissimilarity		Highest Isolation
Pittsburgh, PA	65.0	New York-Newark-Jersey City, NY-NJ-PA	24.7
Birmingham-Hoover, AL	62.2	San Francisco-Oakland-Berkeley, CA	24.1
Buffalo-Cheektowaga, NY	61.7	San Jose-Sunnyvale-Santa Clara, CA	20.5
Cleveland-Elyria, OH	59.9	Chicago-Naperville-Elgin, IL-IN-WI	13.0
St. Louis, MO-IL	59.7	Boston-Cambridge-Newton, MA-NH	12.4
Lowest Dissimilarity		Lowest Isolation
Denver-Aurora-Lakewood, CO	40.9	San Antonio-New Braunfels, TX	2.02
San Jose-Sunnyvale-Santa Clara, CA	42.2	Tampa-St. Petersburg-Clearwater, FL	2.04
Seattle-Tacoma-Bellevue, WA	42.6	Milwaukee-Waukesha, WI	2.20
Portland-Vancouver-Hillsboro, OR-WA	42.6	Jacksonville, FL	2.21
Salt Lake City, UT	43.0	Birmingham-Hoover, AL	2.24
BLACK
Highest Dissimilarity		Highest Isolation
Milwaukee-Waukesha, WI	78.8	Memphis, TN-MS-AR	68.5
Detroit-Warren-Dearborn, MI	73.5	Detroit-Warren-Dearborn, MI	67.6
Chicago-Naperville-Elgin, IL-IN-WI	72.6	New Orleans-Metairie, LA	64.0
Cleveland-Elyria, OH	71.5	Milwaukee-Waukesha, WI	63.8
Newark-Jersey City, NY-NJ-PA	71.3	Chicago-Naperville-Elgin, IL-IN-WI	63.0
Lowest Dissimilarity		Lowest Isolation
Las Vegas-Henderson-Paradise, NV	33.2	Salt Lake City, UT	3.72
Riverside-San Bernardino-Ontario, CA	37.4	San Jose-Sunnyvale-Santa Clara, CA	4.50
Raleigh-Cary, NC	39.7	Portland-Vancouver-Hillsboro, OR-WA	8.04
San Jose-Sunnyvale-Santa Clara, CA	39.8	Phoenix-Mesa-Chandler, AZ	10.2
Phoenix-Mesa-Chandler, AZ	40.2	Fresno, CA	10.7

shows that several CBSAs have high MENA–white segregation scores, like Nashville-Davidson--Murfreesboro--Franklin, TN; Birmingham-Hoover, AL; New Orleans-Metairie, LA Milwaukee-Waukesha, WI and Indianapolis-Carmel-Anderson, IN. Three of the five CBSAs in the list of CBSAs with the highest dissimilarity scores lie outside of the five states where more than 50% of the MENA population lives (i.e., CA, NY, TX, NJ, MI—see Harjanto and Batalova 2022). Thus, our analysis shows that there is significant variation in the segregation between the MENA population and white people that previous research has not uncovered by solely being focused on specific geographic areas where the majority of the MENA population lives (Bozorgmehr et al. 1996; Holsinger 2009).

The MENA–white average value is significantly lower than the average Black–white index of dissimilarity of 59.77 (see column 2 of Table 1). The standard deviation of the Black–white dissimilarity scores is the highest among all the scores at a value of 10.29, and the maximum value is 78.80, consistent with other research (Massey and Tannen 2015), CBSAs with the highest Black–white dissimilarity scores include Milwaukee-Waukesha, WI; Detroit, MI; and Chicago, IL (see Appendix A, Table A1).12

The average South Asian–white index of dissimilarity across CBSAs is 59.92, which falls under the high threshold of segregation scores. Notably, the maximum value of the South Asian–white segregation is 72.56, which is in Birmingham-Hoover, AL (see Appendix A, Table A1). Other CBSAs with very high South Asian–white dissimilarity scores include San Antonio-New Braunfels, TX; Buffalo-Cheektowaga, NY; Kansas City, MO-KS; and St. Louis, MO-IL. The difference between the average South Asian index of dissimilarity score and the Black–white average score is not statistically significant (see column 2 of Table 1). Comparing the unweighted D-scores in Appendix A,

Table A2. Residential Segregation of MENA, South Asian, East Asian, and Black Populations from non-Hispanic Whites in CBSAs, 2012–2016 (Unweighted).

	Index of Dissimilarity					Isolation Index
	Mean	Sig. a,b	St. Dev.	Min.	Max.	Mean	Sig. a,b	St. Dev.	Min.	Max.
Group	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
MENA	53.0		6.0	40.3	63.9	3.9	***	3.9	1.9	28.8
South Asian	63.6	***	5.9	48.1	72.6	7.0	***	4.0	1.9	18.9
East Asian	52.0		5.9	40.9	65.0	6.3	***	5.2	2.0	24.7
Black	55.4		10.7	33.2	78.8	36.6		19.6	3.7	68.5
N	192

^a *** p < 0.001, indicating significance relative to Black-white segregation. ^b Shading of cells in columns 1, 6 indicates significant at p < 0.001, p < 0.01, or p < 0.05 from MENA-white segregation. Note: We use bivariate regressions to indicate whether differences in resdiential segregation are significant, relative to Black-white and MENA-white segregation, as indicated by footnotes ‘^a’ and ‘^b’, respectively.

, however, reveals that South Asian people are significantly more segregated from whites than Black people (see column 1). The average East Asian–white index of dissimilarity is 51.90, which falls into the moderate range of segregation scores and is significantly less than the mean Black–white index of dissimilarity. One of the highest dissimilarity scores for East Asian–white segregation is found in Pittsburgh, PA (see Appendix A, Table A1).

We also formally compare these average scores to the average MENA–white dissimilarity score (except for the difference with the Black–white score, which was already compared). The statistical significance of this comparison is noted by a shading of the cell in column 1 of Table 1. This analysis shows that the average South Asian–white and Black–white index of dissimilarity is significantly higher than the average MENA–white score. The former finding is also present in the unweighted analysis (see Appendix A, Table A2).

Turning to the results for the isolation index scores in column 6, it is notable that the average isolation score for Black people is the highest at 47.48, indicating that, on average, Black people live in neighborhoods where 47.82 percent of the population is Black. As was the case for the index of dissimilarity, the standard deviation of the Black–white isolation score is the largest at 15.44, reaching a maximum value of 68.52, which is for Memphis, TN-MS-AR (according to Appendix A, Table A1). The average isolation index for the MENA population is 6.32, which is significantly lower than the average isolation index for Black people, as denoted in column 7. However, the maximum value is 28.84, which is found in Detroit-Warren-Dearborn, MI (see Appendix A, Table A1). The average isolation indices of South Asian and East Asian people are 11.15 and 15.28, respectively, which are both significantly lower than the average isolation score for Black people and significantly higher than the average isolation score of the MENA population, as denoted by the shading of these two cells (in column 6). The maximum values for the isolation indices reach 18.88 and 24.74, respectively, for South and East Asian people, and are located in San Jose-Sunnyvale-Santa Clara, CA and the New York-Newark-Jersey City, NY-NJ-PA, respectively (see Appendix A, Table A1).

The differences in residential segregation from white people between ancestry groups and Black people are likely related to variation in acculturation-related, socioeconomic, and demographic variables.

Table 2. Demographic, Socioeconomic, and Metropolitan Characteristics of the MENA, South Asian, East Asian, and Black Populations in CBSAs, 2012–2016.

	MENA	South Asian	East Asian	Black
Variables	(1)	(2)	(3)	(4)
Spatial assimilation variables
% speaking English well or only	69.2	79.6	52.6	95.6
% native-born	26.3	7.6	20.2	83.4
Percent education with BA or more	51.8	76.8	57.2	22.3
Average household income (in $000s)	11,155	15,714	12,248	6923
% homeowners	64.4	65.6	71.3	54.3
Other demographics
% with children present	53.7	61.0	48.3	40.0
% married	69.9	82.2	70.5	37.8
Average age (in years)	48.0	44.6	49.5	51.0
% in suburbs	60.5	57.5	56.4	62.5
Metropolitan area characteristics
Log of CBSA population (average)	15.4	15.5	15.6	15.3
Number of the group in CBSA (average)	105,367	222,041	405,780	1,209,927
Percent:
Employed in:
Manufacturing	9.1	8.9	8.5	8.6
Government	4.7	4.5	4.5	4.9
In armed forces	0.47	0.34	0.42	0.52
Aged 65 and over	13.4	13.0	13.2	13.2
With a college degree or more	7.7	7.8	7.9	7.6
In housing built since 2000	16.0	16.0	14.6	17.8
Non-white population	45.7	48.6	49.6	46.3
Midwest	0.21	0.16	0.09	0.21
South	0.31	0.32	0.21	0.52
West	0.23	0.24	0.38	0.08
Northeast	0.25	0.29	0.32	0.19
No. of CBSAs	48	48	48	48

reports descriptive statistics on these variables plus metropolitan-level characteristics for the MENA population, South and East Asian people, and Black people. The results for the acculturation-related variables of English proficiency and nativity status suggest that the ancestry groups should be disadvantaged, relative to Black people, in terms of their segregation from white people. Still, the results for segregation in Table 1 do not reveal such disadvantages. For example, the percentage of the MENA population that speaks only English or English well is almost 70%, but for Black people, it is nearly 96%; however, segregation of the MENA population from white people is significantly lower than the segregation of Black people from white people. The English proficiency level for South and East Asian people is also less than that of Black people, but their segregation scores are similar to or significantly less, respectively, than that of Black people. A similar pattern of results is found for the percentages of the ancestry groups that are native-born; they are significantly lower than that for Black people.

The results for the socioeconomic characteristics, however, likely reveal why the segregation of ancestry groups from white people is lower as Black–white segregation. The average percentage with at least a BA degree of the MENA population and South and East Asian people is at least two times higher than that for Black people. Similarly, the average household income of the MENA and South and East Asian people is at least 1.6 times higher than that of Black people. The percentage of the ancestry groups that are homeowners is also significantly higher than that of Black people, while just over 54% of Black people own their homes, 64.37%, 65.64%, and 71.32% of MENA and South and East Asian people own their homes, respectively.

The results for some of the demographic characteristics indicate that these variables may also be important in predicting the differences in segregation between white people between the ancestry groups and Black people. With respect to the differences in the percentage of households with children present, larger shares of the MENA population (54%) and South (61%) and East Asian people (48%) report having children as compared to Black people (40%). Thus, their generally lower segregation from white people, relative to that of Black people, could reflect their desires to locate in school districts of higher quality that tend to be in whiter communities (Reardon et al. 2015). Similarly, the percentages of ancestry groups that are married are more than 1.5 times higher than that of Black people, which could also be a demographic characteristic that reduces their segregation from white people. The results for the groups’ average age and location in the suburbs, however, reveal no consistent pattern that might be associated with the differences in segregation observed in Table 1.

For the most part, the variations in metropolitan characteristics between the ancestry groups and Black people reveal no clear-cut patterns that might relate to the differences in residential segregation observed in Table 1. For example, the average percentage employed in manufacturing across CBSAs is 9.12% among the MENA population, 8.91% among South Asian people, 8.48% among East Asian people, and 8.62% among Black people. Similar equivalent percentages are observed across these groups for the percentages employed in the government and in the armed forces and in other characteristics, including the percentages aged 65 and over, with a college degree or more, in housing built since 2000, and non-white population. Given that these are average percentages across CBSAs, it is not surprising that little variation exists between ancestry groups and Black people.

The low average values of isolation scores observed in Table 1 for the MENA population and South and East Asian people are not surprising in light of the results in Table 2 for the average number in each group in the CBSA. For the MENA population, the average number in the group across CBSAs is about 105,366, which is relatively small. While small numbers of the MENA population in each CBSA make it difficult for values of the isolation index to be very large, it is notable that the maximum value is 28.84, which is in the Detroit CBSA and is consistent with other research that has found that many MENA immigrants have located among their co-ethnics (Bakalian and Bozorgmehr 2013; Holsinger 2009; Lin 2009). The South and East Asian average population sizes across CBSAs are 222,040 and 405,779, respectively, which are also lower than for Black people (i.e., 1,209,927) and are consistent with the fact that the average isolation scores for South and East Asian people are 11.15 and 15.28, respectively.

Turning to the last set of results in Table 2 for regional location, there are somewhat different geographic patterns of location of the groups. The MENA population is more evenly distributed across the Midwest, South, West, and Northeast than the other groups. Among South and East Asian people, the lowest percentages—16% and 9%—live in the Midwest, respectively. South Asian people tend to be most likely to live in the South and Northeast, whereas greater shares of East Asian people are located in the West and Northeast. With respect to Black people, 52% live in the South, which is consistent with research that has revealed a return migration of Black people to the South (Falk et al. 2004).

We now turn to our multivariate analyses in

Table 3. Coefficient and Standard Error Estimates from OLS Regression Models of Dissimilarity and Isolation Indexes by Group, with All Control Variables, 2012–2016.

	Dissimilarity			Isolation
Variables	Coeff.	SE	Sig.	Coeff.	SE	Sig.
Group (ref. Black-white)
MENA−white	1.32	9.51		5.65	11.92
South Asian−white	15.92	10.38		23.72	13.88
East Asian−white	−0.68	13.37		15.64	17.03
Spatial assimilation variables
% speaking English well or only	−0.19	0.15		−0.61	0.26	*
% native−born	−0.18	0.07	**	0.41	0.14	**
Percent education with BA or more	0.92	0.18	***	0.89	0.30	**
Average household income (in $000s)	0.00	0.00	***	0.00	0.00
% homeowners	0.23	0.18		0.80	0.33	*
Other demographics
% with children present	0.92	0.29	**	3.17	0.43	***
% married	−1.62	0.30	***	−3.53	0.53	***
Average age (in years)	0.84	0.64		1.00	1.06
% in suburbs	−0.02	0.05		0.07	0.10
Metropolitan area characteristics
Log of CBSA population	0.50	0.95		−2.14	1.78
Percent:
Employed in manufacturing	0.11	0.16		−0.25	0.36
Employed in government	0.75	0.26	**	−0.20	0.50
In armed forces	0.00	0.45		2.67	0.72	**
Aged 65 and over	−0.88	0.27	**	−0.33	0.65
With a college degree or more	−2.64	0.83		−3.56	1.28	**
In housing built since 2000	−0.63	0.12	***	−0.53	0.22	*
Non−white population	−0.04	0.05		0.12	0.13
Region (ref. Northeast)
Midwest	0.23	1.77		−0.63	3.99
South	−2.03	2.38		0.92	5.89
West	−3.48	1.54	*	−12.03	3.77	**
Constant	99.19	43.09	*	26.99	65.30
Adjusted R-squared	0.82			0.88
No. of observations			192

*** p < 0.001; ** p < 0.01; * p < 0.05.

to assess the residential segregation of ancestry groups from white people as it compares to that of Black people from white people once we adjust for the differences in group- and metropolitan-level characteristics discussed in Table 2. As stated above, these analyses are weighted by the total population size in the metropolitan area consistent with the approach taken in previous research (e.g., Lichter et al. 2015).13

In discussing the results of these models, we first focus on our key differences in residential segregation between ancestry groups and white people relative to segregation between Black people and white people. In contrast to the descriptive or unadjusted results in Table 1, we find that MENA–white segregation and South Asian–white segregation, as measured by the index of dissimilarity, are higher than Black–white segregation as denoted by the positive coefficients, controlling for group- and metropolitan-level characteristics, but are not statistically significant. Interestingly, the segregation of South Asian people from white people is about 16 units higher than the segregation of Black people from white people; the magnitude of the difference between MENA–white and Black–white segregation is only 1.3 units. The coefficient for East Asian–white segregation is negative and small (–0.677) but is not significantly different from Black–white segregation, controlling for other factors.

Column 2 of Table 3 reports the results from our model of the variation in isolation indices. Unlike the descriptive or unadjusted results in Table 1, the isolation of indexes of the MENA population and South and East Asian people are higher—5.7, 23.7, and 15.6 units, respectively—than the isolation index of Black. However, after controlling for relevant factors, group- and metropolitan-level characteristics, the differences, relative to Black people, in the isolation indices of the MENA population and South and East Asian people are no longer significant, which is in contrast to the unadjusted results reported in Table 1 where MENA and South and East Asian people had significantly lower levels of isolation than Black people.

What is the association between residential segregation and the acculturation-related, socioeconomic, and demographic variables? Column 1 of Table 3 shows that few characteristics are significantly associated with variation in the index of dissimilarity. With respect to the acculturation-related and socioeconomic variables, the findings are consistent with hypotheses derived under the spatial assimilation model. Controlling for other factors, the average household income of groups and percent native-born are negatively and significantly related to residential segregation from white people, as gauged by the dissimilarity index. The group percentages of speaking English well are negatively related to the dissimilarity index, but the coefficient is not statistically significant. Interestingly and in contrast to the spatial assimilation model, the results show that the percentage group with BA or more education is positively and significantly associated with the dissimilarity index for these groups. With respect to group-level demographic variables, the percentage of children present is positively and significantly related to residential segregation from white people. On the other hand, the percentage of married people is negatively and significantly related to dissimilarity scores.

Column 2 of Table 3 shows the characteristics that are significantly associated with variation in the isolation index. Controlling for other factors, the group percentages speaking English well is negatively and significantly related to residential isolation, which is consistent with hypotheses derived under the spatial assimilation model. However, the native-born group percentages with a BA and more education that are homeowners are all positively and significantly related to residential isolation, which is in contrast to the hypotheses derived under the spatial assimilation model that suggest that more native-born individuals and those of greater socioeconomic status would not necessarily be residentially isolated. On the other hand, these findings do reveal support for the finding that affluent households tend to isolate amongst themselves (Reardon and Bischoff 2011). Household income is the only socioeconomic variable that is not significantly associated with isolation, although it is positively related to isolation. Most of the demographic variables are significantly associated with variation in the isolation index. The percentage of the group with children and the average age are positively associated with isolation indices, and the percentage of the group that is married is negatively associated with isolation, controlling for other relevant factors.

How do the metropolitan-level characteristics relate to segregation? The lower half of Table 3 reports these results. The size of the population in the CBSA is positively related to the dissimilarity index, consistent with other segregation research (e.g., Wilkes and Iceland 2004; Iceland and Nelson 2010). The percentage of the population employed in government is significantly and positively related to segregation from white people, and the percentage of the population that is aged 65 and over is significantly and negatively related to dissimilarity scores, controlling for relevant factors. The percent of the population living in housing built since 2000 is negatively and significantly related to residential segregation from white people, which is consistent with previous research and suggests that areas with housing built after the passage of the Fair Housing Act have lower levels of residential segregation from white people (e.g., Wilkes and Iceland 2004; Iceland and Nelson 2010; Friedman et al. 2022).

The percent employed in the armed forces is significantly and positively related to residential isolation, and the percent of the population aged 65 and over is significantly and negatively related to residential isolation. As with the results for dissimilarity scores, the percentage of the population living in housing built since 2000 is negatively and significantly related to residential isolation. Similarly, the percentage of the population with a college degree or more is significantly and negatively related to isolation scores, controlling for relevant factors.

Metropolitan areas in the West have significantly lower values of the index of dissimilarity than those in the Northeast, controlling for other factors. This finding is also true for the isolation index, controlling for other factors. The adjusted R-squared for the dissimilarity index indicates that approximately 82% of the variation is explained by the independent variables in the model, suggesting that the model fits the data well and that the independent variables provide a strong explanation of the variation in the dissimilarity index. Similarly, the adjusted R-squared value for the isolation index is also very high (88%), indicating that the model provides a very strong fit in explaining the variation in isolation indexes across metropolitan areas based on the given predictors.

5. Discussion

The goals of this study were to examine the residential segregation of the MENA population and South and East Asian people from white people in the United States and to compare their segregation to that of Black people from white people. Little research has examined the residential segregation of the MENA and South Asian populations from white people. Our study builds upon the few studies that exist on this topic using much more recent data from the 2012–2016 Summary File of the ACS and the IPUMS.

The descriptive analyses of segregation as measured by the index of dissimilarity and the isolation index reveal that Black–white segregation, as gauged by both measures, is significantly greater than the other groups’ segregation from white people. The one exception is that there is no statistically significant difference between the average indices of dissimilarity of Black–white segregation and South Asian–white segregation. Taken together, these results could be indicative of support for either the tenets of the spatial assimilation or place stratification models. With respect to the spatial assimilation model, it is likely that the higher levels of Black–white segregation reflect Black people’s poorer socioeconomic status than that of the MENA population and South and East Asian people, consistent with hypotheses derived under this model. Indeed, the results in Table 2 revealed that Black people have much lower levels of income and access to homeownership than the ancestry groups, and Black people’s average years of education were also lower than that of the MENA population and South and East Asian people. In addition, Black people are much less likely than the ancestry groups to be married, which can also put them at an economic disadvantage if there are fewer earners in the household. According to the tenets of the spatial assimilation model, these disadvantages will likely make them more segregated from white people than their ancestry counterparts. The principles of place stratification, however, offer a different explanation. They suggest that higher levels of Black–white segregation, relative to the segregation of other groups from white people, is the result of Black exceptionalism, in that Black people have experienced a legacy of discrimination in the housing market that is above and beyond what other people of color have faced in American history (Logan and Stults 2011; Massey and Denton 1993; Massey and Tannen 2015; Wilkes and Iceland 2004).

The multivariate analyses in Table 3 provide a better way to adjudicate support for the tenets of these two different models because they include controls for socioeconomic, demographic, acculturation-related, and other relevant variables. Our results indicate support for hypotheses derived under the spatial assimilation model. The results regarding the index of dissimilarity reveal that once we account for control variables, the average level of segregation of MENA and South Asian people from white people is not significantly different than Black–white segregation. The results for the isolation index also show a similar pattern of findings. While MENA and both Asian ancestry groups are significantly less isolated than Black people in the unadjusted results (Table 1), after controlling for relevant characteristics, the differences in average isolation indices between Black people and these groups disappear. Thus, making these groups’ acculturation-related, socioeconomic, demographic, and other characteristics reduces the large gap in isolation indices between Black people and MENA and Asian ancestry groups and renders the differences non-significant.

The results for the control variables in the models also suggest support for the tenets of the spatial assimilation model. For example, the percentage of groups speaking English only or well, native-born, married, and the average household income was negatively related to segregation as measured by the dissimilarity index. Thus, consistent with hypotheses under the spatial assimilation model, acculturation and higher levels of socioeconomic status of groups were negatively related to residential segregation. Some results that did not conform with expectations under the model were the findings that the percentages of a group that is native-born, with a college education or more, and homeowners were positively associated with variation in the isolation index. As mentioned above, these alternative findings are consistent with research that shows that affluent households are increasingly isolating from one another (Reardon and Bischoff 2011).

The main contribution of our study is that it expands the existing literature by being the first to document levels of segregation of the MENA population and South Asian people in a much wider range of metropolitan areas, using recent data relative to Black–white segregation. Multivariate analyses that compare the segregation of non-white groups from white people have shown that Black people are significantly more segregated from white people than other non-white groups, controlling for other factors (Wilkes and Iceland 2004; Iceland and Wilkes 2006; Iceland and Nelson 2010). However, research using more recent data has shown that there is no significant difference between Black people and Asian people in their residential segregation from white people after controlling for relevant group- and metropolitan-level characteristics (Rugh and Massey 2014).

Our findings appear to be consistent with this latter study. After controlling for relevant group- and metropolitan-level characteristics, as well as regional differences, we find no significant difference in segregation from whites between Black people and the MENA population and South and East Asian people. While not statistically significant, it is notable that the results in Table 3 show that South Asian residential segregation from white people in terms of the index of dissimilarity and isolation index is higher than Black–white people residential segregation.

Some scholars argue that immigrants with high levels of social capital—such as South Asian people—may prefer to settle in ethnic communities, defined as “ethnic neighborhoods that are selected as living environments by those who have wider options based on their market resources” (Logan et al. 2002, p. 300). Mainly, this approach—the ethnic community model—argues that immigrants form ethnic communities and live among co-ethnics even when living in affluent white neighborhoods is feasible for them to maintain their traditions, cultural practices and lifestyle (Portes and Jensen 1989; Lee and Kye 2016; Li 1998; Logan et al. 2002). Thus, segregation of immigrant groups, particularly those with high socioeconomic capital, could be attributable to in-group, residential preferences.

Several lines of inquiry would be useful for future research to pursue to expand the present analysis. It would be fruitful for future research to use individual-level data to examine the association between individual-level characteristics and locational attainment measured by the percentage of white and non-white in the neighborhood as well as the neighborhood poverty rate and median income, for example. With access to the confidential IPUMS data, the research could examine how acculturation-related, socioeconomic, and other characteristics predict access to neighborhoods inhabited by white people (the majority group) as well as with higher levels of socioeconomic status for specific MENA, South, and East Asian groups. This would be another way to test whether and how the tenets of the spatial assimilation model are supported, but using individual-level data would show how individual-level characteristics are associated with neighborhood characteristics. It would also examine the individual attainment of specific groups underlying the broad categories of MENA, South, and East Asian people. A benefit of focusing on individual-level data is that you would capture much more variation in the experiences of smaller groups like the MENA population. In aggregate analyses of residential segregation, if the metropolitan area is too small, it would not be included in the analysis.

Future research should also explore the mechanisms underlying the segregation of all the groups from white people. Uncovering why minority groups do not differ in their residential segregation from white people is important and can be examined more clearly at the individual level. The results would show whether individual-level socioeconomic and demographic characteristics are associated with location in neighborhoods of better quality. If they are not associated with residential attainment, it could point to the fact that other, more structural factors like housing discrimination could be affecting individuals’ location in neighborhoods of varying quality.

A final direction for future research would be to focus more on variation in segregation of South Asian people who experienced higher levels of residential segregation from white people than Black people in both multivariate models. However, the findings were not statistically significant. Previous research shows that South Asian people have been mistaken as MENA and/or Muslim because of their phenotypical similarities and common religious indicators (e.g., turban for Sikhs and hijab for Muslims), and as a result of this mistaken identity have been subjected to racial discrimination (Maira 2004, p. 219; Lee and Kye 2016; Love 2017). In fact, studies show that compared to Chinese, Filipino, Japanese, Koreans and Vietnamese, Asian Indian people experienced the highest racial and immigrant-based discrimination in the United States (Jeung and Lim 2012; Schachter 2014) in the workplace, employment, neighborhood and community level and have been viewed as “adversarial invaders and troublemakers” (Aptekar 2009, pp. 1512–29; Brettell 2011). These studies use older data, however. Thus, future research should conduct audits or correspondence tests to examine current discrimination and barriers faced by groups on the basis of race/ethnicity and/or religion in the housing market.

By 2030, non-Hispanic whites will constitute about 56% of the overall U.S. population but just under a majority of the population under 35 (i.e., 48%) (U.S. Census Bureau 2017). The Asian population in the U.S. is growing particularly rapidly. It is expected that by 2055, Asian people will surpass Hispanics in being the largest immigrant group in the U.S. (Pew Research Center 2021). Asian Indian people are the second largest subgroup in the overall Asian category and have grown in population by 142% between 2000 and 2019, which is much greater than the 88% growth rate of Chinese, the largest Asian subgroup (Pew Research Center 2021). Given the significant projected growth of South Asian people, it is imperative for future research to monitor their residential segregation relative to the segregation of other Asian and non-white groups, especially given that residence is inextricably linked to other dimensions of stratification and well-being in metropolitan America.