Abstract
Diverse Latino families continue to settle in suburbs, hunting for better neighborhoods and educational opportunities. But do they discover more integrated schools relative to segregated city schools? We find that Latino children attending suburban elementary schools were exposed to a greater share of White peers nationwide between 2000 and 2015 than were Latinos attending urban schools. But exposure to White peers in suburbs declined on average during the period. Demographic forces within suburban districts, especially rising family poverty, contribute to worsening segregation of Latino children, as do institutional features. Districts enrolling fewer children and increasing spending per pupil remained more integrated during the period, as identified by two-level fixed-effect (Mundlak) estimation. Many heavily White districts served growing shares of Latino children without losing White families.
Latino families in the United States have fanned out to the suburbs over the past century in search of jobs, better neighborhoods, and better schools.1 A majority came to live in suburban parts of metropolitan areas more than a decade ago (Lichter et al. 2010). Recent immigrants have moved directly to suburbs as well rather than settling only in urban gateways. This wide dispersion of Latino families adds color to the demographic rainbow that already marks America’s suburbs.
But do Latino families find more integrated schools in diversifying suburbs than the segregated urban schools experienced historically? What demographic or institutional features of school districts contribute to variability in the levels of racial or economic segregation Latino children feel in contemporary suburbs? Theorists of social integration differ on their expectations for diversifying suburbs. Theory also remains underdeveloped in regard to the role of school authorities in advancing integration independent of local demographics, housing patterns, and job structures.2 Little is known regarding how institutional attributes and actions of school districts may discretely contribute to segregation in suburbs.
Notions of modernity and opportunity, along with faith in schooling, promise that the school institution will incorporate minoritized and foreign-born groups into a mainstream culture and labor force (Flippen and Farrell-Bryan 2021). Yet race or class-reproduction theorists counter that schools often act to reinforce and legitimate inequalities, hardening racialized separation (Bourdieu and Passeron 1990; Alba 2020). Middle-class Latino families complicate extant theory as they radiate out to suburbs, perhaps making racial integration palpable for White families of similar class origins. We find, for example, significant counts of suburban districts that host rising numbers of Latino children without losing White families; elsewhere, contemporary White flight unfolds with a vengeance (for more detail, see Frankenberg et al. 2023, this issue).
The school institution endeavors to widen opportunity by integrating Latino children with White or middle-class peers under the same school roof. Many Latino parents migrate to suburban areas with this expectation in mind. We know that poor children benefit when learning within integrated classrooms (for cognitive facets of achievement), mostly based on the African American experience (Cook 1984; Johnson 2019). Disparities in Latino children’s oral language and preliteracy skills emerge early in life, yet these disparities can be narrowed by quality pre-K and elementary schools (Fuller et al. 2015; Reardon and Galindo 2009).
Still, these pivotal dimensions of child development, along with social-emotional growth, are hampered by isolating poor or racialized minorities from their middle-class peers (Owens 2017; Reardon, Kalogrides, and Shores 2019). Little is known about how the integration of Latino children has unfolded in suburbs and whether the trajectory of racial segregation maps closely to the historical experience of cities. How to maintain integrated schools as suburbs diversify, and the strength of local actors relative to demographic and economic forces, remain key questions in civic and scholarly circles.
This article advances understanding of suburban schooling—as one mechanism of racial integration or inequality—by tracing trends in segregation levels and comparing urban and suburban trajectories between 2000 and 2015, then asking whether the organizational structuring of districts contributes to variation in segregation levels beyond the force of local demographics and economics. Dominant accounts of social integration have yet to consider the contemporary diffusion of Latino families and how this dynamic interacts with efforts by local educators to isolate or embrace diverse students. In addition, we test the sensitivity of our findings to how one defines suburban school districts, as Ann Owens and Peter Rich (2023) explore in this issue.
First we clarify differing expectations that scholars of social integration have put forward, recognizing that theory has yet to recognize demographic or institutional dynamics that may differ between suburbs and urban centers for diverse Latino families. We then review trends in Latino segregation found in urban and suburban schools, emphasizing how the organizational features of districts may shape whether integrated schools in suburbs persist or diminish over time. Against this backdrop, we trace change in school segregation levels between 2000 and 2015 in suburban districts, estimating how demographic and institutional factors account for variation in Latino segregation among the nation’s school districts.
LATINO FAMILIES SETTLE IN SUBURBS
Three analytic frameworks offer differing expectations for what families discover as they seek wider opportunity in suburbs, including variably integrated schools. These classic theories focus more on the racialized character of parents and children than on variation in the neighborhoods and school districts. The role of pivotal institutions in fostering acculturation remain underspecified relative to the porous character of labor markets, especially how integrating opportunities may vary among the nation’s diverse suburbs. How might trends and institutional mechanisms tied to Latino integration differ from the historical experience of Black families (Massey and Tannen 2018)?
Optimistic proponents of spatial assimilation argue that minority children will benefit from rising educational attainment and job mobility, bolstered in part by integrated schools of sufficient quality, eventually acculturating into a middle-class mainstream (Alba and Nee 2009). But race theorists (and the place-stratification frame) remind us that many Latinos or immigrant communities remain stigmatized, isolated by housing arrangements and unequal institutional openings in many local communities (Lichter et al. 2010).
A nuanced version of race theory, segmented assimilation, postulates that racialized groups may assimilate into limited segments of the economy or institutions but not others, and Latinos’ progress can stall after the second generation (Portes and Zhou 1993). The interplay between attributes of subgroups with varying conditions found in local communities is what shapes the likelihood of finding integrated schools. Yet we know little about how the attributes and behavior of local educators or civic activists mitigate suburban segregation.
We do know that school attainment predicts assimilation into mainstream jobs and civic life across generations, at least for Mexican Americans (Telles and Ortiz 2009). Better educated Latino youth move into less-Latino neighborhoods (Goldsmith and Puga 2019). But other Latino families arrive to suburbs only to find racially isolated, low-quality schools (Ee and Gándara 2020; Donato 2008).3 Suburbs may display “segregated diversity” as Angela Simms (2023, this issue) details, marked by the racialized sorting of students across schools within school districts (Pinto-Coelho and Zuberi 2015).4
The steady diversification of suburbs since the 1970s suggests a third analytic frame, replete with less deterministic expectations—the cultural pluralist frame that accents the multiracial character of housing patterns and perhaps schools that has unfolded in many suburbs. Frank Bean and his colleagues (2015) describe how Latino families have spread across a variety of suburbs in southern California. Laura Meckler and Kate Rabinowitz (2019) similarly detail growing Latino populations in suburban schools, advancing integration in previously White and close-in suburbs.
This pluralist frame gains validity when the attribute of race becomes pliable or reconstructed. Assimilation for some groups, for instance, may be less of a “Whitening process” than “mainstream expansion [that will] meld many whites, nonwhites, and Hispanics . . . the prospect of a new kind of societal majority,” Richard Alba argues (2020, 8–10).5 Latinos may widen what mainstream means racially or linguistically in select suburbs. But this multiracial portrayal assumes differing families inhabit the same civic spaces, benefiting from an inclusive opportunity structure, wider pathways enabled by social institutions.
Finding Differing Suburbs
To contextualize integration trends in suburbs, let us first examine how cities and suburbs differ, especially the evolving character and complexion of school districts. The transformation of suburban demographics has been remarkable over the past half-century. The racial composition of suburbs included just 1.6 percent Latino residents in 1970, rising to 14.8 percent in 2010 across the nation’s metropolitan areas (Massey and Tannen 2018). Just over one-fourth of all Latinos lived in suburbs in 1970 (Frey 2018), rising to 57 percent in suburban parts of metro areas by 2020 (Lichter, Thiede, and Brooks 2023).6 Nearly one-third of suburban growth in the 2000s was attributable to the influx of foreign-born residents (of all ethnicities) who occupied a variety of social-class positions (Suro, Wilson, and Singer 2011).
Still, levels of residential segregation for Latinos ticked slightly upward in the nation’s suburbs between 1970 and 2010 (based on the dissimilarity index), and Black segregation eased (Massey and Tannen 2018).7 More than half the nation’s poor (55 percent) lived in suburban areas in 2010 (Kneebone and Berube 2013). Yet we know little about how such trends in residential segregation relate to evolving levels of school segregation in suburbs.8
The rising share of children from Latino backgrounds of course accelerates the diversification of suburban schools (Meckler and Rabinowitz 2019). Yet this likely prompts differing responses by White parents, depending on the racial and class features of the newcomers, along with commitments of educators and civic activists (see in this issue Frankenberg et al. 2023; Simms 2023). We describe how integration ranges from heavily White suburban districts, hosting say 15 percent Latino students, to districts with racially balanced enrollments. So, how we conceive of integration in suburbs is pivotal, informed by the Latino case.
Diverse Latinos Settle in Diverse Suburbs
The interplay of family characteristics and prior neighborhood dynamics likely shapes local levels of school integration. The outward spread of Latino families to the suburbs is certainly not new. In the 1950s, the GI Bill enabled many Latinos to buy suburban homes, especially in expanding metro areas, such as Los Angeles (Ong and González 2019). In contemporary times, differing Latino families have continued to radiate out to a widening diversity of suburbs. Fully one-third of Mexican immigrants to the United States between 1995 and 2000 settled directly in new destinations, not in traditional gateway cities (Lichter et al. 2010; Ludwig-Dehm and Iceland 2017).
We know that residential segregation tends to intensify in metro areas where Latinos are mostly foreign born, displaying lower incomes and school attainment, than neighborhoods dominated by native-born Latinos (Massey and Tannen 2018).9 In addition, low- and higher-income families sort into differing suburbs, a dynamic that worsened the economic isolation of Latino households between 1990 and 2014 (Owens and Rich 2023, this issue).
At the same time, the suburbanization of Latinos co-occurs with rising educational attainment and declining fertility in many regions. In 1970, 69 percent of Latinos in the Northeast had not completed high school, declining to 30 percent by 2015 (Massey and Constant 2017). The number of expected births among Latina mothers fell from 2.8 in 2008 to 2.0 early in the pandemic (Stone 2021). This affects the likelihood that Latino children attend school with other racial groups, odds lent order by the social-class backgrounds of Latino parents.
In sum, the character of Latino families is changing and diversifying over time, as well as the suburbs and districts in which their children attend school. The confluence of these two levels contribute to the variably integrated contexts in which children are raised. Further, demography may not be destiny: educational attainment over time has markedly shaped the fertility, economic mobility, and suburban migration of Latino parents.
Institutional Reception in Suburbs
The extent to which Latino families discover integrated suburban schools depends in part on the actions of local educators and civic activists. Two districts may show differing commitments to integration or varying fiscal capacity to respond in welcoming fashion, even when reflecting identical demographic or residential trends. Our analysis asks whether differing organizational features of districts, easily measured, enhance school integration or worsen segregation among suburbs over time.
Qualitative research vividly reveals the inviting or separatist sentiments that greet Latino families as their children enter school. Angela Valenzuela’s book Subtractive Schooling (1999) detailed how teens arriving from Mexico and into Texas public schools perceived the classrooms and educators they encountered. Valenzuela discovered that large classes and the lack of bilingual specialists to foster English fluency eroded motivation among Latino youth.
In addition, White teachers tended either to stigmatize what Latino students knew or failed to grasp their aspirations, denying the norms and knowledge necessary for achieving within the majority culture. One student told Valenzuela (1999, 122), “Aquí, nos tienen sentados en las sillas, calladitos, todos humillados” (Here they have us sitting in our chairs, quiet, all of us subdued), he said. Recent work similarly details how minoritized students, even within integrated suburbs, continue to face low achievement expectations from teachers; minority parents remain on the edge of mostly White networks that help identify strong teachers and afterschool academies and sustain influential roles inside schools (Lewis-McCoy 2014).10
Our analysis informs the context-of-reception question by testing for segregating effects that stem from measurable attributes of school districts: organizational features that may operate independent of residential profiles inside suburbs. These institutional features, for instance, include how small elementary school districts may act to fence out Black or Latino families. The spread of private or charter schools may similarly work to segregate poor or minority children within districts as found in North Carolina between 1998 and 2016 (Clotfelter et al. 2021; see also Fiel 2015; Rich, Candipan, and Owens 2021). School funding levels and access to educational resources—pre-K programs, experienced teachers, more demanding curricula—may differ between segregated and integrated schools (Ackert, Crosnoe, and Leventhal 2019; Dondero and Muller 2012; Fuller 2022). We ask whether these institutional features may contribute to steady or declining school integration after accounting for the demographics of suburbs.
TRACKING FACETS OF SCHOOL SEGREGATION
Much is known about overall levels of segregation Latinos have experienced over the past half-century. Little work, however, focuses on integrated school opportunity in suburbs and whether trends over time resemble the worsening segregation observed in cities. Nor do we understand how suburban school districts may differ organizationally in ways that sustain integration, beyond the effects stemming from variation in demographic composition.
Empirical findings rest in part on whether one focuses on segregation between school districts or across schools within districts. Education leaders in the United States tend to focus on the latter, given that they sit in local districts, along with judicial constraints on cross-district integration (for a review, Fuller et al. 2022). Yet to identify the extent to which Latino families settle in particular districts, or Whites leave diversifying areas (Frankenberg et al. 2023, this issue), we must track segregation levels between districts as well.
Racial segregation, when measured by the likelihood that a Latino child attends a school with more or fewer White peers (exposure index) has generally declined over the past half-century. Rising proportions of Latino students, relative to other ethnic groups, necessarily diminish the likelihood of attending school with more White peers. We extend these earlier findings to the 2000 through 2015 period for the exposure index.
Demographers and sociologists, alternatively, may focus on the distribution of racial groups across schools within districts, gauging the evenness with which a group is distributed across campuses (or entropy, per Reardon and Owens 2014). This barometer remained stable or improved slightly from 1993 to 2015 for Black and Latino students in the nation’s school districts, not limited to metropolitan areas or elementary-age children (Richards, Stroub, and Kennedy 2020).
But little is known about whether these trends hold for Latino children attending suburban schools, or how change over time may differ for subgroups. Laura Meckler and Kate Rabinowitz (2019) find a rising count of schools with diversifying enrollments in previously White suburbs and exurbs. “The number of children attending U.S. public schools with students of other races has nearly doubled over the past quarter century,” they report. We detail how many predominantly White districts enrolled increasing, yet modest, counts of Latino students during the 2000 to 2015 period. This trend contributed to the falling exposure index—signaling declining integration—as Latino pupils attend schools with fewer White classmates. From a White student’s vantage point, however, they attend increasingly integrated schools as rising counts of Latinos enroll.
In addition, climbing numbers of Latino students may prompt a response by other racial groups, including the outmigration of White parents, as Matthew Hall and Jacob Hibel (2017) find at modest levels of magnitude between 1980 and 2010. This form of White flight appears more severe when neighboring districts remain mostly White, less so in districts hosting more schools, presumably allowing White parents to find less integrated schools. Yet, little is known about the extent of White flight in districts gaining Latino students, a question we return to in the next section.
Worsening Racial Segregation
After marked progress integrating Black students into predominantly White schools between 1954 and 1980, the isolation of Black and Latino students from White or middle-class peers between school districts has worsened (Orfield 2001; Orfield and Lee 2007). But when examining change in the distribution of Black and White students across schools within districts (entropy), little slippage and some gains have appeared since 1990 (Reardon and Yun 2002; Richards, Stroub, and Kennedy 2020; Stroub and Richards 2013). Similar patterns are discernible for Latino students, although declining likelihoods of Latino children sitting next to White peers are shaped by the proportional growth of Latino enrollments in suburbs (Fuller et al. 2019). The decline in Latino students’ exposure to White peers is consistent with diminishing exposure between Latino and White residents in suburbs between 1990 and 2020 (Lichter, Thiede, and Brooks 2023).
Equally consequential for poor Latino children is the extent to which they attend schools with middle-class peers of any race, defined as economic integration. Much of the impact that race exerts on school achievement operates through class-based isolation of Black or Latino students from middle-class peers (Fahle et al. 2020). This becomes highly relevant in suburbs hosting varying mixes of poor and middling Latino families, where class-based segregation still occurs between schools.11 Little is known about how trends in the economic segregation of Latino students have played out in suburban districts.
RESEARCH QUESTIONS AND METHOD
Settling in suburbs may help Latino families get ahead and enjoy better schools and neighborhoods. These diverse places ideally host integrated schools—a consequential element of the local opportunity structure—serving to advance children’s growth and the household’s class position. But do Latino parents discover integrated schools in suburbs? Has the prevalence of White suburban schools, moving toward diverse enrollments, changed in recent decades? Which Latino families are more likely to find integrated schools in what kinds of suburbs? Our analytic strategy informs these questions by addressing these empirical questions:
Research Question 1. How do suburban school districts differ, relative to districts situated in cities, in terms of demographic and economic characteristics of local residents, student composition of schools, and funding levels and institutional attributes in 2000 and 2015?
Research Question 2. Do segregation levels, isolating Latino from White children, differ between suburban and city districts? Do time trends differ over the 2000 to 2015 period between suburbs and cities?
Research Question 3. Do economic and demographic attributes of Latino subgroups living in districts (such as income, home language, nativity) explain magnitudes of change in the segregation of Latino students in suburbs during the period?
Research Question 4. Do district funding levels or institutional features explain worsening segregation levels for Latino children, after taking into account demographic features of district residents?
We focus on elementary-age children given the importance of these years in potentially narrowing gaps in early learning (Jenkins et al. 2018; Reardon and Galindo 2009). Elementary schools also serve families in smaller attendance zones, relative to high schools, which may mirror residential segregation.
Data
We draw on school and district information contained in U.S. Common Core of Data (CCD) for the 2000–2001 and 2015–2016 school years, collected by the National Center for Education Statistics (NCES 2017). This includes attributes of students enrolled, such as the racial composition of schools and eligibility for free or reduced-price meals (FRPM), district revenues and spending, and the geographic boundaries (polygons) of each school district.
We first selected districts that host at least one elementary school, including elementary and unified districts, yielding a working file (n = 4,059 districts). Districts operating just one elementary school were then excluded because at least two schools are required to calculate the evenness of enrollments by race between schools (entropy) in each district. Those enrolling fewer than ten Latino elementary-level students were also excluded. This selection procedure yielded 4,293 districts nationwide with complete enrollment data by race for 2000 and 2015. All analyses distinguish between racially White parents of Latino or non-Latino descent.
A variety of demographic and economic indicators for residents within district boundaries were obtained from the 2000 Census and 2015 estimates derived from the American Community Survey (U.S. Census Bureau 2020). We matched these data to the geographic bounds of districts, employing the areal interpolation technique in R Studio to match overlapping polygons. This allowed us to associate each school district’s level of student segregation to the attributes of populations residing inside district boundaries.
MEASURES
Our analysis relies on the NCES (2017) classification of school districts between urban and suburban, derived in collaboration with the Census Bureau and detailed by Education Demographic and Geographic Estimates (EDGE) Program. Suburban areas “are located outside the boundary of a principal city of a metropolitan area,” although “micropolitan areas may contain suburban territory as well” (NCES 2017, 4).12 Schools are then located within suburban areas along with their host school district. We included all such schools and districts, whether tied to central city in a metropolitan area or not. This resulted in the sample of 4,293 districts with at least two elementary schools and ten Latino students, as explained earlier, including 1,851 suburban districts.
Alternatively, Ann Owens and Peter Rich (2023, this issue) draw from the NCES data, yet they start with 398 core-based statistical areas (similar to metropolitan statistical areas). They identify four geographical patterns, most commonly a single central city surrounded by multiple suburban school districts. In some instances, a single district encompasses suburban areas (say, Los Angeles), or an urban county was surrounded by multiple suburban districts. Using this method, the authors identify 6,202 nonurban districts that enrolled at least five fourth-grade students, focusing on elementary schools and host districts as well. We made final estimates for change in Latino segregation levels from 2000 to 2015 using the Owens and Rich set of suburban districts to detect any differences relative to our sample of suburban districts.
We also use and extend two basic measures of racial and economic segregation: the two-way exposure and evenness (entropy) indices. The exposure index reports what the label suggests: for the case of Latino and White students, exposure is the mean percentage of the latter group enrolled in the schools attended by students of the first group. An exposure index score of 0.36, for example, indicates that on average thirty-six of every one hundred peers attending a Latino child’s school is White. A higher score indicates greater integration between Latino and White students.
This measure corresponds with the ethnic composition of a district. When districts host rising shares of Latino students, these children will necessarily be exposed to fewer White peers. Yet the exposure index also traces the extent to which Latino, White, or other groups are sorting into racially isolated districts and remains a relevant gauge of between-district segregation. Scholars, alternatively, have calculated normalized exposure scores for nested units, including children attending schools within districts (Bell 1954). This gauge “measures the gap between the observed exposure level of one group to another,” for example, across schools within a district, “and the exposure level that would occur in a situation of perfect integration,” such as where the Latino and White composition of all schools equals the relative proportions district-wide (Reardon, Yun, and Kurlaender 2006, 56).13
The entropy index measures how close the average school-level ethnic distribution is to the district’s overall racial distribution. A smaller distance yields lower values given that the distribution of Latino children is more even across schools within a district relative to other groups. Entropy scores, as typically calculated, range between 0 and 1.14
Exposure and entropy measures can be applied to the segregation of poor from nonpoor students. We estimated the probability that a FRPM-eligible child interacts with nonpoor peers for all children, regardless of ethnic membership (race-specific data on FRPM eligibility are not available) (Owens, Reardon, and Jencks 2016). We have detailed elsewhere how use of FRPM modestly overstates detected declines in the poor-nonpoor exposure index (Fuller et al. 2022).15
Demographic attributes of residents living within the bounds of school districts likely account for levels of Latino segregation or change over time. We report on economic characteristics of residents, aggregated to the district level for 2000 and 2015: median household income for Black, Latino, and White residents; percentage of population below 150 percent of the poverty line; percentage of households in owner-occupied residences; and median rent paid. We also report educational attainment levels; the share of district residents who are foreign born; and the share of residents, five years old or older, who do not speak English at home.
Beyond the force of demographics, institutional and economic features of school districts may further explain change in segregation levels, covariates available from CCD as well. This includes the percentage of elementary schools that are charter or magnet schools in each district, the geographic size of districts (square miles), whether the district is an elementary-school-only or larger unified district. To examine possible correlates or effects from public funding of school districts, we draw on economic data provided by the Census Bureau’s F-33 file for 2019. We calculated instructional spending per pupil (including teacher salaries) stemming from financing provided by state and federal governments, along with local property taxes. We report fiscal data in current dollars.
Estimation Method
We used a two-step procedure to observe how these covariates might account for cross-sectional variation in racial and economic segregation in 2015, or change in segregation levels during the 2000 to 2015 period. The first step required basic OLS estimates while being attentive to collinearity among predictors. We then used a panel model, reducing the set of independent predictors, estimating how they account for change in segregation between Latino and White children from 2000 to 2015.
All covariates vary across time i and district j and thus should be treated as time-varying. Because they are time-varying, a random-effects model is unable to separate the within- and between-school-district effect of any explanatory variable (Bell and Jones 2015). To provide reliable estimates, we use the correlated random-coefficients model (Mundlak 1978). This formulation adds one additional term for each time-varying covariate to account for the between-district effect: the district mean of the covariate over the period. The equation can be specified as follows:
where yij represents the average change in the segregation index, x1ij is a (series of) time-varying covariates that are measured at the occasion level with coefficient β1, and x̅jis the district j’s mean across time, which is the time invariant component of those variables (Snijders and Bosker 2011). Here β1 is the estimate of the within effect (or longitudinal effect in this case where we have measures at two time points) and β2 is the estimate of the contextual effect (or the average district effect across time), which models the difference in between and within effects. Note that this β1 resembles the within effect estimated by fixed-effects models when no confounding influence of a level-2 or district-level variable operates.
FINDINGS
Let us situate our sampled school districts within all districts nationwide. The first two rows of table 1 speak to all districts, reference points for interpreting the economic status of residents in our working sample of districts. Median income moved from $47,016 to $54,535 for residents of all districts between 2000 and 2015, an increase of 16 percent in current dollars. In constant 2019 dollars, the Census Bureau reported flat income levels during the same period, moving from $58,609 to $58,476 (Statista 2021); meanwhile, consumer prices rose 37 percent (Federal Reserve Bank of Minneapolis 2021).
Turning to residents of districts that met our selection criteria and reported complete data, income climbed from $49,749 to $57,706 in current dollars. Our enrollment weighting likely emphasizes income trends in larger urban school districts, which moves mean income higher relative to all districts. We see that mean income levels ranged considerably higher in suburban school districts, and rose by 16 percent during the period relative to a 10 percent hike in cities. Black and Latino households earned considerably more within suburban districts relative to their peers residing in cities.16 Poverty rates, falling lower in suburbs, grew significantly during the period in cities and suburbs. The percentage of district residents living below the poverty line climbed in suburban areas from 14 to 20 percent.
The percentage of residents of Latino heritage living in suburban districts rose from 13 to 20 percent between 2000 and 2015, and from 18 to 26 percent in urban districts. Average educational attainment at the district level increased during the period, despite overall economic stagnation (table 2).17 A declining share of residents in suburban districts reported English as their first language, falling from 75 to 70 percent during the period. We see that 739 of the 1,852 suburban districts can be classified as new immigrant destinations as defined by Van Tran and Nichol Valdez (2017).
Table 2 details institutional features of districts included in our sample. Urban school districts enrolled more students (11,077 on average) in 2015 than suburban districts (4,596, contrasts tied to research question 1). The percentage of residents in suburban districts completing some college or more increased from 57 to 63 percent. Urban districts occupy more than twice the land area than their suburban counterparts. This may have implications for the ability of White parents to find less integrated schools inside larger districts.
Suburban districts received higher revenues per pupil ($13,079), totaling all government sources, than city districts ($12,670) in 2015. The gain in revenue per pupil for suburban districts equaled 57 percent between 2000 and 2015, perhaps due to finance reforms yielding additional state revenues exceeding the 37 percent rise in the consumer price index (Johnson 2019). Instructional spending per pupil was 7.3 percent higher in suburban districts for 2015 relative to city districts.
More Integrated Schools in Suburbs?
Patterns of school segregation look different for Latino children attending school in suburban districts relative to their urban counterparts. Levels and time trends also differ for the additional barometers of segregation (research question 2). The Latino-White exposure index declined from 0.50 to 0.38 between 2000 and 2015 (table 3). Exposure levels remained higher in suburban districts relative to cities, yet the decline was sharp—falling from 0.56 to 0.41—during the period.
Figure 1 shows where Latino-White exposure levels have declined since 2000. Districts with at least a 0.05 reduction in their exposure index appear, split between city and suburban districts. Larger dots indicate greater declines in exposure scores. As Latino enrollments grew in California and Texas relative to Whites, exposure levels predictably fell. Latino exposure to White peers worsened throughout the Northeast, Upper Midwest, and parts of the South.
Figure 2 further distinguishes city and suburban trends for the exposure measure of Latino-White segregation. Panel A displays counts of urban districts by their exposure scores for 2000 and 2015; panel B does the same for suburban districts. Many districts in each panel are moving leftward, indicating declining exposure scores for Latino children. The dramatic decline in suburban districts (far right of panel B) shows the impact of rising Latino enrollments in once lily-White districts. We will return to the question of whether this represents eroding integration for Latino children or diversifying desegregation for White students.
At the same time, segregation of Latino children among isolated schools within districts did not worsen overall. Entropy scores remained constant in urban and suburban districts. These scores averaged lower (that is, less isolation of Latinos) in suburban districts (0.11 in 2015), relative to urban districts (0.20, table 3). This steady distribution of Latino children across schools within districts is borne out when estimating normalized exposure scores for urban and suburban school districts, another indication that Latino children experience more equitable exposure to White peers in suburbs, relative to city schools.18
The apparent stability in this even distribution of Latino children is shaped in part by many districts that enrolled low percentages of Latino children, seen on the lower-right corner of figure 3. We display the relationship between entropy and exposure scores for suburban districts. The size of circles indicate district enrollment size. Many small to medium-size districts did experience rising entropy scores, indicating worsening isolation of Latino children in particular schools, after taking into account their district-wide representation.
The lower-right corner of figure 3 reveals a large count of suburban districts with high exposure scores, tied to high shares of White enrollments. As these districts gain Latino students, exposure scores necessarily decline, contributing to national averages. The percentage of Latino students entering these mostly White districts remains modest, however. When we isolate on suburban districts with exposure scores over 0.75 (n = 735 districts), mean enrollments were 5 percent Latino, 85 percent White, and 10 percent other ethnicities in 2015. From a White student’s viewpoint, the school is becoming more integrated; meanwhile, Latino children attend the same school with declining counts of White peers, lowering the exposure score.
These trends unfold against the backdrop of rising Latino presence in the nation’s public schools, moving from 22 to 31 percent of all students attending elementary campuses between 2000 and 2015 (table 3). Latino student shares climbed from 19 to 29 percent of enrollments in suburban districts during the period. These enrollment proportions were higher than Latino representation among residents of suburban districts, rising from 13 to 20 percent on average during the period (table 1).
Latino Arrival, White Response
Growing counts of Latino students may prompt responses by White families, including their exit to neighboring districts. We found that one-third of districts experienced rising enrollments of Latino children of at least 250 children between 2000 and 2015 (32 percent of the 4,293 districts). Among suburban districts meeting this growth threshold (n = 656), White enrollments grew in just 131 districts and fell in the remaining 525—evincing a contemporary version of White flight.
Table 4 shows how White enrollments tended to grow in economically stronger suburban districts experiencing Latino enrollment gains. Conversely, the decline of White enrollments was greater in lower-income districts as Latino enrollments rose.19 The overall pattern is consistent with the two metro cases detailed by Erica Frankenberg and her colleagues (2023, this issue) in which White flight was greatest in suburbs most proximal to (low-income) cities. To the extent that middle-class Latinos mark the in-migration to suburbs, they may be seen as more acceptable, relative to the incursion of poor Latinos. In addition, entropy scores fall lower in suburbs with White enrollment gains, reflecting a more even spread of Latino children across schools.
Explaining Change in District Segregation
Do features of district residents or institutional features of districts help explain varying levels of racial segregation or rates of change over time? We turn next to this question, estimating change in exposure and entropy scores between 2000 and 2015, comparing explanatory results for Latino children attending school in urban and suburban districts (research questions 3 and 4). We built estimation models using Owens and Rich’s (2023) sample of suburban districts. Yet no notable differences were observed with regard to the key drivers of change in segregation.20
In estimating exposure and entropy cross-sectionally, we first regressed Latino-White exposure scores among districts in 2015, isolating on distinct sets of covariates (table 5). Because covariates at times are collinear, we enter groups of predictors separately. We see, for instance, that the percentage of district residents in poverty is negatively related to exposure scores, and this related set of covariates explains nearly half the variance in exposure scores (model 2). The ethnic composition of district residents, as expected, is strongly associated with exposure scores (model 4). Districts with higher shares of White residents manifest higher Latino-White exposure scores, whereas the converse is true for districts with high shares of Latino residents.
Institutional features of school districts further contribute to exposure (model 5), accounting for 12 percent of the variance in exposure scores. Districts with larger enrollments have lower exposure scores. Districts that include only elementary schools—often observed in middle-class and affluent communities—show higher exposure scores, as do districts spending more on instruction per pupil.
Table 6 reports cross-sectional results when estimating entropy scores for 2015 among districts. Here too we see that entropy scores are higher, that is, greater segregation of Latino children, in districts with larger shares of residents in poverty (model 2). Less segregation (lower entropy scores) occurs in districts with higher median income among Latino households (model 1). Yet the institutional features of districts explain a larger share of variance, 21 percent, in entropy scores. Entropy scores range higher in larger districts, consistent with lower exposure scores. Similarly, districts spending more per pupil show lower entropy scores, that is, greater integration of Latino children in schools within districts.
To estimate change in exposure between 2000 and 2015, we pull forward the covariates that display significant relationships with exposure or entropy, comparing results for urban and suburban districts. The magnitude of change in Latino-White exposure and entropy scores varies greatly, accenting differing dynamics among suburban districts (figure 4). The vast majority of districts show declining exposure scores over the period, change scores typically falling below zero. Yet, suburban districts generally display small declines, less than 0.15. Change in entropy scores are more muted, consistent with the stability discussed (Richards, Stroub, and Kennedy 2020).
Table 7 summarizes multilevel Mundlak estimates of change in Latino-White exposure between 2000 and 2015. We report fixed-effect (time within districts, level-1) and random-effect coefficients (between districts, level-2). Model 1 excludes ethnic composition of districts; model 2 includes these demographic characteristics.
Changing levels of residential poverty within districts over time predict lower exposure scores, whether we control on ethnic composition or not. This negative effect is apparent between districts as well, including among suburban districts. (To approximate the total effect of each covariate, add fixed and random-effect coefficients.)
Two institutional facets of districts continue to exert a significant influence on change in exposure scores. Larger districts show lower exposure scores (between-district effect), after taking family poverty and ethnic composition into account. Then, districts growing in terms of enrollment size further host Latino students who are becoming more isolated from their White peers.
Significantly, suburban districts that enjoyed rising instructional spending per pupil show declining exposure scores. This may stem from progressive gains in school finance in districts with higher shares of disadvantaged students (Johnson 2019). Overall, these estimation models account for about half the within-district variance in change in exposure scores and four-fifths of between-district variance.
Table 8 reports corresponding results for change in entropy scores. Districts with higher poverty rates show significantly higher entropy scores (greater segregation, adding coefficients), before controlling on district ethnic composition. This relationship blurs after controlling for ethnic composition: districts with growth in White residents, along with higher shares of Whites during the period, show lower entropy scores (greater integration within districts).
Larger suburban districts and those with growing enrollments display higher entropy scores. This finding, taken with the ethnic composition results, suggests that suburban districts that retain larger shares of White households and constrain enrollment growth are able to distribute Latino students more equitably among their constituent schools. We also see that suburban districts that enjoy rising instructional spending show declining entropy scores, perhaps gaining new resources that facilitates a more even spread of Latino children across schools.
DISCUSSION: SUBURBAN DIVERSITY, RISING SEGREGATION
As more Latino families settle in suburbs, they do find neighbors who are better-off economically than urban counterparts. Latino parents discover schools that, on average, host more integrated blends of Latino and White students. Latino children are spread more evenly across schools within suburban districts than in city schools. Similarly, poor children in suburban districts enjoy greater exposure to middle-class peers and less isolation in particular schools, on average, than in urban districts. This manifests richer economic integration for poor Latino children in suburbs than in city schools.
A glimmer of good news also appears in suburban districts that gained Latino students between 2000 and 2015 without losing White families. These districts—hosting significant shares of middle-class Latinos—enjoy schools that provide greater exposure to White peers and fairer distribution of Latino children across schools within districts. At the same time, many districts gained Latino children, lost White families, and became more segregated.
The nation’s already diverse suburbs continue to change. Suburban districts hosted rising shares of Latino children and poor families during the period. Many school districts in America’s suburbs have come to resemble their urban counterparts in terms of modest household incomes, racial diversity, and worsening family poverty. These shifts in residential composition help explain declining integration in many schools. The worsening of school segregation is driven by the racially arranged sorting of poor Latino families into particular school districts. The spread of Latinos across schools within districts has not worsened.
Beyond demographic drivers of school segregation, we find that institutional features of district organizations protect or erode integrated enrollment patterns. Such factors can more readily be addressed by policy remedies, being more malleable than migration patterns or demographic forces. We find greater levels of Latino segregation in larger suburban districts, along with those with growing enrollments. This may not be surprising. After all, we expect greater segregation in large urban districts; big suburban districts now follow suit. This should alert civic leaders that preventive remedies are urgently needed.
Institutional dynamics are observed in smaller districts that host only elementary schools—which may act to buffer the incursion of non-White families, or instead become places of planned integration as surrounding communities diversify. We find that rising school spending is no antidote to declining rates of exposure between Latino and White children. Although suburban districts benefiting from funding increases may be finding ways of more equitably distributing Latino children across schools rather than isolating them on certain campuses.
That district organizations exert some influence on trends in Latino segregation lend support of race-rooted theories of social integration. Racial or class-based markers of populations are not all that matter—institutions and organized opportunity structures matter, too. The qualitative studies that appear in this double issue shine a brighter light on the institutional processes that unfold inside neighborhoods and district offices to shape inclusive or isolating forms of student assignment across schools. The variety of suburban settings in which Latino families settle matches a pluralist conception of integration—that a diversity of families continue to settle in suburbs. Their ability of integrate into an evolving mainstream is conditioned by enveloping demographics, job opportunities, and institutions that potentially integrate groups into shared civic spaces and neighborhoods, especially schools.
Policymakers and committed local educators, for instance, have sustained cross-district integration efforts. Boston’s METCO program is among the best known, operating for a half-century among the city’s schools and surrounding and otherwise White suburban districts, yielding discernible achievement effects for Black participants (Ardon and Hatch 2022). New Jersey continues to operate a cross-district magnet program that yields integrating benefits, along with districts nationwide that have expanded magnet schools and dual-language programs, such as in Los Angeles, which help integrate children across racial or social-class lines (Fuller 2022).
Our findings accent how the Latino experience of school segregation departs from the institutional history of African American families. The expanding Latino presence in public schools necessarily means these children will be schooled alongside declining shares of White classmates. Still, the exposure measure picks up intensifying sorting of Latino, White, and other groups into separate school districts. At the same time, a significant portion of the decline in exposure scores stems from modest gains in Latino enrollments in still mostly White school districts. This represents enriched diversity from one vantage point yet registers as declining exposure by Latino children to White peers as gauged by the exposure index. We reported how this spurs White flight in some (presumably aging) suburbs even as a slice of districts hosts more and more Latino families without losing White households.
Again, we observe varying patterns as Latino families spread to diverse suburban areas. Their diffusion includes middle-class Latino parents whose children may not signal racialized dissonance in the eyes of White suburbanites. Perhaps the social-class attributes of Latino parents matter most, as those with higher incomes find suburbs that remain Whiter and display significantly lower rates of family poverty. This tickles how we theorize about the drivers of school integration, the extent to which racialized or class markers ease settlement patterns of Latino parents, along with responses by suburban White families and local educators.
Our study remains limited by national data available on the organizational and economic features of school districts and schools. We did not detect any effects of charter school strength, for example, in the segregation of Latino children, as suggested by scholars analyzing differing samples of schools or districts (see, for example, Fiel 2015). School-level organizational features—say, ethnic characteristics of teachers or English-language services—may yield more integrated schools than schools with less warm or agile contexts of reception. A spotlight shone on districts that gain Latino children without losing White or middle-class peers could yield illuminating results on how educators may become more inclusive over time.
Overall, policymakers and educators across America’s diversifying suburbs should squarely face the intensifying separation of poor Latino children from White and middle-class peers. Suburbs do offer more inclusive educational opportunities for Latino children, relative to the isolation so often found in cities. But many suburbs already resemble large urban districts, populated by a single racial group or large shares of impoverished families. How to stem this trend and build pluralist suburban areas—preserving racially and economically integrated schools—offers the immediate challenge for civic leaders, reform activists, and local educators.
FOOTNOTES
↵1. We opt to use the term Latino rather than any of the alternatives. A recent national survey found that just 2 percent of all Hispanic respondents identify as Latinx (Bendixen & Amandi 2021).
↵2. We use the term integration to report on the exposure of Latino-heritage children to White or middle-class peers inside schools. This is not to imply cultural assimilation or the erosion of distinct ethnic identities.
↵3. This implies a racially arranged distribution of resources, such as experienced teachers and stronger education funding (Johnson 2019; Rotberg 2020).
↵4. Bruce Fuller (2022) details how funding across schools in Los Angeles tends to reinforce disparities in educational quality and children’s achievement, favoring schools that host mostly White or Asian American students.
↵5. Cross-racial marriage alters the White-minority dichotomy that marks dominant narratives. More than one-fifth of recent marriages in Los Angeles are cross-racial, nearly one-fifth in Chicago and New York (Livingston 2017). Mexico’s earlier project of mestizaje, or the evolving status of mixed-race citizens in modern-day Brazil, offer international cases in point (Fiel 2021).
↵6. Which metro areas are included in studies, along with definitions of suburb, vary over time in this literature.
↵7. The dissimilarity index equals the relative share of minority group members and Whites who would have to exchange neighborhoods to achieve an even residential distribution. About 23 percent of all Latinos live in “highly segregated” metropolitan areas in 2010 (Massey 2020).
↵8. Studying variation among U.S. states, Douglas Massey and Jonathan Tannen (2016) find that neighborhood segregation levels between White and Black residents accounted for three-fifths of the variance in Black-White student segregation among school districts (for an earlier analysis of school segregation in the suburbs, see Reardon and Yun 2001).
↵9. More than one-third (37 percent) of foreign-born Latinos with school children lived in a new destination in 2010 versus 25 percent of native-born parents (defined by Tran and Valdez 2017; Fuller et al. 2019). For earlier demographic shifts in suburbs, see Reardon and Yun 2001.
↵10. Ethnographers detail how White civic leaders in suburbs, often based in schools, may discourage participation by recently arriving South Asian, Latino, or Black parents (Frasure-Yokley 2015; Diamond, Posey-Maddox, and Velázquez 2021).
↵11. We earlier detailed the intensifying isolation of middle-class from poor students between 2000 and 2015 (Fuller et al. 2022). This article also details how liberalized federal eligibility for free or reduced-price meals, the conventional measure of student social-class status, tends to inflate estimated levels of economic segregation.
↵12. NCES further breaks down suburban areas into large suburbs (populations exceeding 250,000 residents), midsize (one hundred thousand to 250,000), and small (less than one hundred thousand). We do not use this differentiation.
↵13. Normalized exposure is an evenness measure that takes into account the relative proportions of two (or more) groups at the next higher geographic level (school district). For this gauge of segregation, higher scores indicate greater isolation of Latinos in particular schools, deviating more from the relative district proportions.
↵14. The barometer of racial (or economic) integration pertains to two groups (such as Latino and White children), calculating the probability that a Latino elementary student interacts with White pupils:
where xi, yi, and Si are counts of Latinos, Whites, and the total enrollment in elementary school i, respectively. X equals elementary school enrollment of Latinos in the host district. Similarly, we calculated the odds that children from low-income (FRPM-eligible) families interact with nonpoor students at school. The evenness or entropy measure first calculates the proportional distribution of at least two groups within each school, then determines the extent to which these levels of evenness approximate or stretch distant from the district-wide distribution of groups. We report entropy values for the spread of Latinos and Whites among elementary schools in districts. It is calculated in two steps (Reardon et al. 2000):
where R is the count of racial groups in a school, and Qr is the proportion of group r. This equation estimates the spread of groups within a school.
In the second-stage equation, H measures the distance of each school’s entropy from the district’s level of entropy (evenness):
where T is total student enrollment in the district. Si is total enrollment in school i. H is then the averaged distance between the district’s entropy score, E, and entropy found in school i.
↵15. Reliance on FRPM eligibility to gauge poverty status modestly distorts estimates of change over time at least for the gauge of economic segregation. This stems from the liberalization of how federal authorities have defined FRPM eligibility over the past quarter century (Domina et al. 2018). We earlier calculated, based on CCD, that the percentage of elementary-age children deemed eligible for FRPM climbed from 28 percent in 1998 to 50 percent in 2015. But the Census Bureau’s supplemental index of child poverty (taking cash transfers to families into account) fell from 21 percent to 17 percent nationwide (for detail, see Fuller et al. 2022). Thus, a portion of any decline in the index stems from the noise emanating from the FRPM measure.
↵16. We know that middle-class Latinos were hit especially hard by the Great Recession in current income and asset valuation (Kochher, Fry, and Taylor 2011).
↵17. The share of suburban residents, foreign born, climbed from 12 percent in 2000 to 16 percent in 2015. This rise is consistent with results from national probability samples of families (Fuller et al. 2019).
↵18. Recall that like entropy, low normalized exposure scores indicate a more even distribution of Latino children across schools within districts. For example, in 2015, Latino youngsters are more evenly spread across schools within suburban districts (0.12) than their Latino peers in urban schools (0.21, seen in table 3).
↵19. In suburban districts gaining Latino and White children, median household income averaged $69,077, compared with $64,054 in districts gaining Latinos yet losing White children in 2015 (table 4). The former districts hosted higher-earning Latino residents, relative to the latter group. In turn, racial exposure scores remained higher in Latino-plus-White gaining suburbs (0.51), compared with Latino-gaining but White-losing counterparts (0.30), driven partly by compositional differences.
↵20. Owens and Rich (2023) identify thirty-eight districts as urban that are defined as suburban by the Census Bureau procedure used by NCES. Results when estimating normalized exposure scores looked nearly identical to entropy results, so we report the latter to allow for comparisons with earlier literature.
- © 2023 Russell Sage Foundation. Bathia, Shruti, Bruce Fuller, Claudia Galindo, Francisco Lagos, and Sophia Rabe-Hesketh. 2023. “Finding Integrated Schools? Latino Families Settle in Diverse Suburbs, 2000–2015.” RSF: The Russell Sage Foundation Journal of the Social Sciences 9(2): 104–31. DOI: 10.7758/RSF.2023.9.2.05. This article stems from the Latino Contexts and Early Development Project, funded by the Spencer Foundation and the Institute of Human Development at Berkeley. Wonderfully critical comments by the editors, three anonymous reviewers, and colleagues attending a Russell Sage Foundation conference greatly improved the article. Special thanks to Ann Owens for advice on measures and methodology. The University of California Data Lab helped compile census data. Drs. Bathia and Rabe-Hesketh, along with Mr. Lagos, led the data analysis. All authors contributed equally. Direct correspondence to: Bruce Fuller, at b_fuller{at}berkeley.edu, University of California, United States.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.