New Destinations and the Early Childhood Education of Mexican-Origin Children

Abstract

This study examined differences in exposure to early childhood education among Mexican-origin children across Latino/a destinations. Early childhood educational enrollment patterns—which are highly sensitive to community resources and foundational components of long-term educational inequalities—can offer a valuable window into how destinations may be shaping incorporation among Mexican-origin families. Integrating data from the Early Childhood Longitudinal Study-Birth Cohort with county-level data from the decennial census, multilevel logistic regression models revealed that Mexican-origin, Black, and White children had lower odds of enrollment in early childhood education programs if they lived in new Latino/a destinations versus established destinations. The negative association between new destinations and early childhood education enrollment persisted despite controls for household selectivity, state and local early childhood education contexts, Latino/a educational attainment, Latino-White residential segregation, and immigration enforcement agreements. Within the Mexican-origin subgroup, the enrollment gap between new and established destinations was widest among the least acculturated families, as measured by parental nativity, duration of residence, citizenship status, and English proficiency. These findings highlight how both place and acculturation stratify outcomes within the large and growing Mexican-origin subset of the Latino/a population.

Introduction

The dramatic dispersion of Latin American immigrants and U.S.-born Latinos/as across a wider range of states and communities in the U.S. is a major demographic shift during the last three decades. The rise of new destinations is especially important to consider given evidence that, relative to established gateways, these locations may lack the co-ethnic ties, strong intergroup relations, and institutional supports that promote social mobility and general wellbeing of immigrants and their descendants (Lichter 2013; Waters and Jiménez 2005). In other words, place may be an increasingly salient source of inequality for Latinos/as in general but particularly for the Mexican-origin population—the largest Latino/a subgroup, which faces high levels of socioeconomic disadvantage and discrimination.

This study focuses on Mexican-origin children’s enrollment in early childhood education across Latino/a destinations to provide insight into how place may be stratifying outcomes within the Mexican-origin population. Children’s exposure to early childhood education programs can promote school readiness, which in turn lays the foundation for long-term educational mobility and socioeconomic attainment (Duncan and Magnuson 2013; Heckman 2011). Mexican-origin children have lower rates of enrollment in early childhood education programs than both African American and non-Latino/a white children (referred to in this text as Black and White children), which puts them at a disadvantage in ways that could fuel inequality later in the life course (Brandon 2004; Bridges et al. 2004; Crosnoe 2007; Gormley 2008; Magnuson, Lahaie, and Waldfogel 2006; Turney and Kao 2009; Votruba-Drzal et al. 2015). Do the places where Mexican-origin families live reduce or exacerbate this disadvantage in early childhood?

To explore this important question, this study compared rates of enrollment in early childhood education programs among young children of Mexican origin across different destinations including new Latino/a destinations and established Latino/a destinations with data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B) integrated with county-level data from the 2000 decennial census. Multilevel modeling tested the hypothesis that early childhood education enrollment in this population would be lowest in new Latino/a destinations and highest in established ones. We also examined moderation of the association between destinations and early childhood education by race/ethnicity (Mexican-origin versus non-Latino Black and White) and by levels of acculturation within the Mexican-origin group. The hypothesis was that barriers to educational opportunities in new destinations would be higher for Mexican-origin children than Whites, and, within the Mexican-origin population, for the least acculturated families. Finally, we assessed whether associations between destinations and early childhood education could be explained by differences in household selectivity, early childhood education institutional context, Latino/a group status, and/or intergroup relations.

This research can bridge demographic and educational literatures focused on racial/ethnic and immigration-related inequalities in the U.S. by using early childhood education as a prism through which to understand contexts of immigrant incorporation and also by using destinations as a context for understanding racial/ethnic stratification in the critical early childhood period (Lichter 2013; Turney and Kao 2009). By highlighting the places where Mexican-origin children are the most at risk of missing educational opportunities, and by identifying the specific types of Mexican-origin families who are the most vulnerable in these contexts, this study can inform the development of policies to boost Mexican-origin early childhood education enrollment.

Early Childhood Education and Destinations in the Mexican-Origin Population

This study connects two major fields of research and theory concerning the Mexican-origin population, which makes up the largest share of the U.S. immigrant population, the undocumented population, and the Latino/a population. One field focuses on educational settings and processes and the other on immigrant destinations and incorporation. Both areas of inquiry are important in their own right, but they can also inform one another. Early childhood racial/ethnic disparities can be a mechanism for assessing the implications of destinations for incorporation, and destinations can be potential determinants of early childhood disparities.

First, researchers from multiple disciplines have documented disparities in early childhood education between children from Mexican-origin families and their peers (Crosnoe 2007; Crosnoe, Bonazzo, and Wu 2015; Schneider, Martinez, and Owens 2006). For a number of reasons, including socioeconomic disadvantage, language barriers, lack of opportunities, and many parents’ unfamiliarity with the U.S. educational system, Mexican-origin children are less likely to be enrolled in early childhood education programs in the years before entry into formal schooling than non-Latino White and African-American children (Brandon 2004; Crosnoe 2007). Although this gap has been narrowing in recent years, it is a concern given the links between early childhood education and school readiness—academic and behavioral skills at the start of the K-12 system (Duncan and Magnuson 2013; Magnuson and Waldfogel 2005). In fact, children from Mexican-origin, Latino/a, and immigrant families receive comparable or greater returns to exposure to early childhood education with respect to their school readiness than other groups (Crosnoe 2007; Gormley 2008; Turney and Kao 2009; Votruba-Drzal et al. 2015). For this reason, under-enrollment in early childhood education is cited as one mechanism for disparities between Mexican-origin children (and, more broadly, Latino/a children) and many of their peers in elementary school achievement based on test scores (Crosnoe 2007; Reardon and Galindo 2009).

Second, demographers have highlighted the recent dispersion of the Latino/a population to a broader swath of the country and have elucidated the implications of this increasing geographic diversity in settlement patterns for Latino/a incorporation (Lichter 2013; Massey 2008; Singer 2004, 2004; Waters and Jiménez 2005; Zúñiga and Hernández-León 2005). Prior to the 1990s, the vast majority of Latino/a families resided in a few settlement areas in a handful of states (e.g., California and Texas). By 2000, approximately one-quarter of all Latinos/as were living outside of established Latino/a settlement areas (Lichter and Johnson 2009), in “new destinations” such as North Carolina and Nebraska (Massey and Capoferro 2008).

Areas of Latino/a settlement in the U.S. are now characterized by the size of the Latino/a population base around the year 1990 and Latino/a population growth thereafter (Lichter and Johnson 2009; Singer and Suro 2002). Based on these two dimensions, destinations can be divided into four groups: new Latino/a destinations, established Latino/a destinations, fast-growing Latino/a hubs, and minor Latino/a destinations. Conceptually, new destinations are places that had a negligible presence of Latinos/as prior to the 1990s and experienced rapid growth in the Latino/a population thereafter, whereas established destinations and fast-growing hubs had a noticeable Latino/a base in 1990 (Massey and Capoferro 2008; Zúñiga and Hernández-León 2005). Established destinations experienced stable Latino/a growth rates in the 1990s, whereas “fast-growing Latino hubs” had large Latino/a bases and experienced high rates of Latino/a growth thereafter (Singer and Suro 2002). Finally, some places are “minor” or “other” gateways that had neither a large Latino/a base nor high rates of Latino/a growth (Hall 2013; Lichter and Johnson 2009; Lichter, Parisi, and Taquino 2015).

Amidst the diversification of Latino/a settlement, two major questions are how patterns of incorporation among Latinos/as differ across destinations and which mechanisms account for these differences (Lichter 2013). Theoretically, variations in the historical presence or absence of Latinos/as and the scale of recent Latino/a population growth should be associated with determinants of incorporation such as group size, intergroup relations and the potential for intergroup contact, and institutional resources (Waters and Jiménez 2005). The empirical research comparing Latinos/as’ outcomes across destinations, however, has largely neglected the early childhood period.

Connecting Destinations to Mexican-Origin Early Childhood Education Enrollment

The accommodations framework offers a theoretically grounded way to understand how the two issues of early childhood disparities and destinations among Mexican-origin children might be connected. This framework conceptualizes children’s entry into early child care and education settings as a process by which parents balance preferences concerning child care, demand for child care, and the local supply of child care to actively or passively decide who will care for young children (Meyers and Jordan 2006). This conceptualization emphasizes early child care and education enrollment as arising from the interplay of families’ contexts and their individual circumstances (Crosnoe et al. 2016; Levine et al. 2014). Drawing from the accommodations framework as well as the demographic and theoretical literature on new destinations (Lichter 2013; Waters and Jiménez 2005), this study tests the premise that Mexican-origin children in new Latino/a destinations will have lower enrollment in early childhood education programs than their peers in established destinations. There are three reasons why enrollment in early childhood education programs are predicted to be lower among Mexican-origin new Latino/a destinations than in established places.

First, Mexican-origin families with young children in new destinations may be less selective than Mexican-origin families in established destinations on common predictors of early childhood education enrollment. Children of all racial/ethnic backgrounds are not distributed randomly across destinations, and background factors related to both early childhood education enrollment and selection into destinations may explain a geographic gradient in early childhood education enrollment. For example, families in new destinations may have fewer socioeconomic resources including parental education and income, which are key predictors of early childhood education enrollment (Augustine, Cavanagh, and Crosnoe 2009; Crosnoe et al. 2016). Even though some research indicates that Latino/a immigrants in new Latino/a destinations have higher educational attainment than those in established places (Stamps and Bohon 2006), other research finds that Latino/a infants and children are more likely to live in poverty in new destinations than in established gateways (Crowley, Lichter, and Turner 2015; Lichter, Sanders, and Johnson 2015).

Some background factors related to selection into destinations and early childhood education enrollment may also be specific to the processes of immigration and incorporation. New Latino/a destinations have tended to attract foreign-born workers and their families (Massey and Capoferro 2008), which means Mexican-origin families in these areas may be more likely to be headed by an immigrant parent or parents, who are less likely to enroll children in early childhood education (Brandon 2004; Crosnoe 2007; Karoly and Gonzalez 2011; Turney and Kao 2009). Given the strong linkage between immigrant status and unauthorized status in the Mexican-origin population (Krogstad, Passel, and Cohn 2017), families in new destinations may be more likely to have unauthorized or mixed-status parents, which imposes further barriers to participation in voluntary schooling.

Second, new destinations may hinder early childhood education enrollment among children of all racial/ethnic backgrounds. Previous research has not investigated racial/ethnic differences in early childhood education enrollment across Latino/a destinations, but Black and White children are less likely to enroll in early childhood education in some new destinations than in established ones. For example, the difference in 3–5 year-olds enrolled in nursery school, preschool, or kindergarten between North Carolina (a new destination) and California (an established destination) is around 7 percentage points for both Black and White children (Kids Count Data Center 2017). As proposed by the accommodations framework, factors related to the preschool context, such as state and local policy support for early childhood education and the supply of programs, influence parental decision-making about early care and education. Prior research has not determined how these institutional factors vary across Latino/a destinations. To the extent that they lack adequate policy support and infrastructure for preschool programs, however, new destinations will pose challenges for early childhood education enrollment for children of all racial/ethnic backgrounds.

Third, new destinations may hinder early childhood education enrollment specifically for Mexican-origin children because of contextual factors that are salient for immigrant incorporation; namely, targeted institutional resources, co-ethnic support, and intergroup relations (Portes and Rumbaut 2001; Waters and Jiménez 2005). Areas with a longer history of Latino/a settlement are characterized by embedded migration streams and rooted immigrant enclaves (Portes and Rumbaut 2006). As such, established gateways (and fast-growing hubs) should theoretically have developed institutions to promote early childhood education enrollment among children in immigrant families. In addition, formal and informal co-ethnic institutions often prevail in established immigrant enclaves, helping to connect immigrant-origin families with educational resources (Bankston and Zhou 2002; Zhou and Kim 2006). New Latino/a destinations and minor Latino/a gateways, in contrast, are only beginning to develop such targeted institutional resources and co-ethnic support mechanisms (Hernández-León and Zúñiga 2005) or may lack these types of support altogether.

Intergroup relations between Latinos/as and non-Latinos/as may also be more negative in new destinations. Latino/a-white residential segregation is higher in new destinations than in established gateways (Hall 2013; Lichter et al. 2010), and non-Latino whites have out-migrated from areas experiencing rapid growth of immigrants (Crowder, Hall, and Tolnay 2011; Hall and Crowder 2014). The “receptivity climate”—attitudes of U.S. citizens towards immigrants—is also more unfavorable, on average, in new versus established destinations (De Jong et al. 2017).

Regardless of the size of the Latino/a base, places that have experienced more rapid Latino/a growth may struggle to meet the educational needs of Mexican-origin children to a greater extent than those that have experienced low or stable growth. In new destinations, for example, early childhood education centers may not be able to quickly develop resources to support a fast-growing Latino/a population, such as dual language services and/or culturally-relevant curriculum, and there may be greater competition for resources that do exist (see, for example, Gouveia, Carranza, and Cogua 2005; Kandel and Parrado 2006). In minor Latino/a gateways, Mexican-origin families may face less discrimination from a lack of visibility and less in-group competition for existing programs than in new destinations; they may also be “innovators” who make their own way amidst contextual constraints (Stamps and Bohon 2006). In fast-growing Latino/a hubs, targeted programs similar to those found in established destinations may exist but could be oversubscribed due to higher demand from increased Latino/a growth. Established destinations with stable growth, therefore, should offer Mexican-origin families the most opportunities to access early childhood education.

The Roles of Racial/Ethnic Background and Acculturation

The accommodations framework specifies that families will be responsive to context when making decisions about early care and education, but it does not address racial/ethnic differences in how contexts shape early childhood education enrollment. Racial/ethnic groups may face a similar set of opportunities and constraints for enrolling children in early care and education, but their responses to similar conditions may differ because of variation in factors such as household resources and overall familiarity with early care and education opportunities. Additionally, Mexican-origin families may face group-specific opportunities and constraints on early care and education enrollment decisions, such as co-ethnic support and/or discrimination. Black children may also encounter group-specific constraints, but they have a longer history in the early childhood education sector, especially Head Start. For these reasons, this study anticipates that the negative association between living in a new destination and early childhood education enrollment will be strongest for Mexican-origin children, followed by Black children, and will be weakest for White children—the most racially/ethnically advantaged group.

The association between destinations and early childhood education may also vary within the Mexican-origin population according to levels of acculturation, or adaptation to U.S. society based on time-since-migration and degree of exposure to U.S. society and its institutions (Rumbaut 2004). Some Mexican-origin families are immigrant newcomers, others have resided in the U.S. for generations, and still others have been here for years but not their whole lives (Alba, Jiménez, and Marrow 2013). All Mexican-origin parents may face more barriers to accessing opportunities for themselves and their children in new destinations than in established ones, but these barriers could be even greater for less-acculturated parents in such destinations, especially within an early childhood system that is often private and optional (Duncan and Magnuson 2013; Clarke-Stewart and Allhusen 2005). To be sure, factors associated with lower acculturation, such as language barriers and a lack of familiarity with U.S. institutions, may exacerbate difficulties for families facing contextual obstacles to enrolling children in early childhood education. In fact, studies looking at the association between new destinations and school dropout show that it is strongest among Mexican immigrants, and, within this group, among newcomer migrants (Ackert 2017; Fischer 2010).

Previous Research on Educational Differences across Destinations

Prior studies have not compared early childhood educational outcomes among Mexican-origin children across Latino/a destinations. Several studies, however, suggest that these communities may be worse contexts of educational incorporation and opportunity than established gateways. Mexican-origin and immigrant youth in new destinations are less likely to be enrolled in school than those in established destinations (Ackert 2017; Fischer 2010). Latino/a high school students in new destinations also have larger gaps with Whites in college-preparatory coursework enrollment than those in new destinations (Dondero and Muller 2012). Educational institutions in some new destinations also lack the resources to meet the needs of immigrant-origin children (Gouveia et al. 2005; Kandel and Parrado 2006).

Other studies, however, provide a more positive or neutral view of educational opportunities and outcomes for children in new versus established destinations. Community members and non-profit agencies have been quick to respond to the educational needs of immigrant-origin families in some new destinations (Hernández-León and Zúñiga 2005). Public schools in new destinations also have schooling characteristics that are more amenable to educational success than those in established gateways, such as lower pupil-to-teacher ratios and lower aggregate poverty levels (Fry 2011). Two studies also reported more advantageous educational outcomes in new versus established destinations, such as higher Latino/a adult educational attainment (Stamps and Bohon 2006) and higher academic achievement (Potochnick 2014). These place-based differences, however, are largely explained by factors related to selection, such as duration of residence and socioeconomic resources.

Aims and Hypotheses

Following the accommodations framework, this study connects the demographic literature on new destinations to the educational literature on early childhood disparities by focusing on a context x family interplay in predicting early childhood education enrollment. In line with prior research (Hall 2013; Lichter and Johnson 2009; Singer and Suro 2002; Stamps and Bohon 2006), we categorized destinations (counties) based on the size of the Latino/a base in 1990 and Latino/a growth from 1990 to 2000. The first aim was to test the hypothesis that all children (Mexican-origin, Black, and White) would have lower enrollment in early childhood education in new Latino/a destinations (small Latino/a base, high Latino/a growth), followed by those in minor Latino/a destinations (small base, low/medium growth), followed by fast-growing Latino/a hubs (large base, high growth), and then finally established Latino/a destinations (large base, low/medium growth). The second aim was to test the hypothesis that these associations would be strongest among Mexican-origin children, followed by Black children, followed by White children. The third aim was to test the hypothesis that the association between new destinations and enrollment would be the most pronounced among the least acculturated Mexican-origin families. The final aim was to determine if observed associations between destinations and early childhood education enrollment reflected differences in household selectivity, institutional resources (preschool context), group status (Latino/a educational attainment), and/or intergroup relations (Latino/a-White residential segregation and immigration enforcement).

Data

ECLS-B is a nationally representative dataset that includes information on children and families from birth through kindergarten entry. Children in the ECLS-B were born in the U.S. in 2001 and studied at age 9 months, 2 years, 4 years, and at kindergarten entry. The age 4 wave is referred to as the preschool wave. ECLS-B is somewhat dated for understanding the most recent patterns of early childhood education enrollment, but holds several advantages over other datasets, such as following children and families prospectively as they make decisions about early care and education and including identifiers for Mexican-origin status and geocodes.

We supplemented ECLS-B with county-level data on community contexts from the 1990 and 2000 decennial censuses. Other county-level data were also obtained from the Census County Business Patterns (to measure county child care center supply), the National Center for Children and Families (to measure state preschool contexts), and the Migration Policy Institute (to measure county 287g agreements). All county-level data were merged into the restricted-use ECLS-B based on zip codes where children were living at age 2. Zip codes were assigned to counties via a crosswalk for the year 2000. In cases where zip codes spanned two counties, the zip code was assigned to the county where the majority of the zip code population resided.

The analytical sample was initially restricted to children who participated in all three waves of ECLS-B through the preschool wave. It was further restricted to cases with a valid county identifier and non-missing number of child care centers in the county (based on Census County Business Patterns). A small number of cases in the ECLS-B (less than 50) were excluded because their zip codes were located outside of the U.S. or were reported erroneously. Finally, the sample was restricted to children with Mexican-origin Latino/a parents, non-Latino Black parents, and non-Latino White parents. The ECLS-B contains information on parental Hispanic identification and racial/ethnic identification from the child’s birth certificate. A child was considered to be of Mexican origin if one or both parents was identified as Mexican-origin Latino/a of any race/ethnicity on the birth certificate. A child was considered Black if both parents were non-Latino Black or if the mother was Black and the father’s information was missing. A child was considered White if both parents were non-Latino White or if the mother was White and the father’s information was missing. The final analytic sample of 6,000 children consisted of 1,100 Mexican-origin children, 1,300 Black children, and 3,600 White children (sample sizes rounded to nearest 50 to comply with NCES restricted-use data regulations).

Measures

Early childhood education enrollment

The binary dependent variable indicated whether the child attended preschool, Head Start, or formal center-based care at age 4. The reference category includes children with no exposure to any form of early childhood education (who were in parental, relative, and/or non-relative care in informal settings). Even though program quality varies within the formal early care and education sector (Duncan and Magnuson 2013; Magnuson and Waldfogel 2005), children in this sector are much more likely to have exposure to activities that can promote school readiness, such as daily scheduled reading and math activities, than children in the informal sector (Bassok et al. 2016).

Destinations

Destination county was measured at age 2, prior to children’s early care and education status at age 4. Children who moved between counties from ages 2 to 4 were flagged in multivariate analyses. Following the approach taken by Hall (2013), destinations were defined with a group-specific typology measuring the presence and growth of the co-ethnic Latino/a population. The main criteria used to distinguish new from established destinations was the percent of Latinos/as in the county in 1990 and percent Latino/a growth from 1990 to 2000 (see Ackert 2017; Fischer 2010; Hall 2013; Lichter, Parisi, and Taquino 2015; Singer and Suro 2002; Stamps and Bohon 2006). Counties were divided into four destinations: 1) new Latino/a destinations; 2) established Latino/a destinations; 3) fast-growing Latino/a hubs, and; 4) minor Latino/a destinations.

Counties were first grouped into two categories—high versus low Latino/a base in 1990—defined by a cut-off level of 10% Latino/a in the county in 1990. Median county Latino/a growth rates were then calculated separately for high and low Latino/a base groups among the Mexican-origin sample. The median Latino/a growth rate for Mexican-origin cases in counties with a high Latino/a base in 1990 was 55%, and was 160% for cases in counties with a low Latino/a base. New Latino/a destinations were defined as counties with a low Latino/a base in 1990 with Latino/a growth rates that were 160% or higher from 1990 to 2000. Established Latino/a destinations had a high Latino/a base in 1990 and less than 55% Latino/a growth from 1990 to 2000. Fast-growing Latino/a hubs had a high Latino/a base in 1990 and 55% or higher Latino/a growth from 1990 to 2000. Minor Latino/a destinations had a low Latino/a base in 1990 and less than 160% Latino/a growth from 1990 to 2000. These cut-offs were similar to prior studies (Lichter and Johnson 2009; Singer and Suro 2002; Stamps and Bohon 2006), and this typology ensured adequate sample sizes of Mexican-origin cases within destinations.

Acculturation

To measure acculturation, we created an index of four variables: 1) immigrant household (dummy variable for 1 or 2 foreign-born parents); 2) parents’ (usually mothers’) age of arrival in the U.S. (non-immigrant, arrived before age 5, arrived between ages 5 and 12, or arrived after age 12); 3) household non-citizen status (dummy variable for 1 or 2 non-citizen parents); and 4) mother’s English proficiency (scale from 1–4 based on ability to speak, read, write, and understand English, with higher values indicating lower proficiency). Because these measures were on different scales, they were rescaled to be bounded between 0 and 1, and then summed and divided by 4 (the total number of items). Cases that were missing values on any of the items (<1% of the analytic sample) received a missing value on the index which was imputed using multiple imputation. The Cronbach’s alpha for this index was .82, and the index was reverse coded so that higher values represented households that were more acculturated. Child and household selectivity

Multivariate models adjusted for child and household characteristics that are predictive of early childhood education enrollment and may also influence selection into destinations (Brandon 2004; Coley et al. 2014; Crosnoe et al. 2016; Yesil-Dagli 2011). These variables were typically measured at child age 2, so that they occurred prior to the age 4 decision on early childhood enrollment status. They included child gender (female), a low birth weight indicator, mother’s age at child’s birth (mean-centered), whether the mother was working (child age 2), non-resident father (child age 2), total number of siblings (child age 2), the presence of three or more adults in the household (child age 2), household socioeconomic status index (child age 2, combining parental educational attainment, parental occupation/s, and household income), and mobility within or between counties and states (between child ages 2 and 4).

State and county-level explanatory mechanisms

Multivariate models also adjusted for state- and county-level characteristics that could potentially explain associations between destinations and early childhood enrollment. These measures were intended to capture differences in state and local preschool contexts, local Latino/a group status, and local intergroup relations. Like the destination measures, these measures were all assigned to states and counties where children were living at age 2. State legislatures set the policy context for preschool program enrollment via decisions regarding preschool investment, child care tax provisions and subsidies, and child care regulations. Using the Policy Choices dataset from the National Center for Children and Families (Rigby and Brooks-Gunn 2005), models incorporated three measures from 2002 (the last year available): state preschool investment score, child care subsidy generosity score, and child care regulation stringency scores. All of these scores were standardized composite scores based on a battery of items such as maximum income eligibility for child care subsidies and teacher-to-child ratios in child care centers (see Rigby and Brooks-Gunn 2005 for details).

The county supply of child care centers was measured by obtaining the number of child care centers in the Census County Business Patterns for the year 2003. Child care centers were identified using the 6-digit North American Industry Classification System code, “624410- Child Day Care Services.” This code identifies establishments that primarily provide day care for infants or children, including nursery schools, preschools, day care centers, home-based day care centers registered as businesses, and Head Start programs separate from schools (U.S. Census Bureau 2017). The total number of child care centers in the county was divided by the number of children aged 0 to 5 in the county from the 2000 decennial census and multiplied by 1,000. The final measure represented the county number of child care centers per 1,000 young children. This supply measure underestimates the total supply of early childhood education options, because it did not capture school-based Head Start or publicly-funded school-based preschool programs.

County Latino/a educational attainment assessed Latino/a group status within destinations. Using data from the 2000 census, a measure of the percentage of Latinos/as in the county with a B.A. or higher was constructed. Measures of residential segregation and immigration enforcement captured intergroup relations between Latinos/as and non-Latinos/as within counties. The index of dissimilarity is a standard measure of residential segregation (Massey and Denton 1988) that gauges how evenly two groups are spread across sub-units (tracts, in this case) within a larger unit (counties, in this case). Latino-White index of dissimilarity values were calculated for each county using counts of Latinos/as and non-Latino Whites within tracts from the 2000 census. Following Rugh and Hall (2016), 287g agreements within counties were used to measure local immigration enforcement—a marker of hostilities towards immigrant-origin populations. A dummy variable indicated whether a county law enforcement agency or any agency within the county had signed an active 287g memoranda of agreement between 2005 and 2010, which deputized local police officers to serve as immigration agents who could facilitate detention and deportation of suspected unauthorized immigrants. Even though the majority of 287g agreements were signed after 2005, when early childhood education enrollment was measured, the 287g dummy variable was intended to reflect latent hostilities towards immigrant-origin groups.

Models also controlled for potential contextual confounds. A dummy variable indicated whether children were living in an urban versus rural area (at age 2), based on ECLS-B. County-level variables from the 2000 census that could have been confounded with destination characteristics included per capita income (in $1,000s, mean-centered) and total population (logged, mean-centered), and state-level variables from the 2000 census that may be confounded with preschool contexts included per capita income (in $1,000s, mean-centered) and total population (logged, mean-centered). Models with the Latino-White residential segregation variable also controlled for the total number of census tracts in the county. Table 1 displays weighted descriptive characteristics for all variables used in the analysis.

Table 1.

Weighted sample means for Mexican-origin, Black, and White children, ECLS-B

	Mexican Origin		Black		White
Variable	Mean	S.E.	Mean	S.E.	Mean	S.E.
Enrolled in pre-kindergarten (age 4)	0.524	0.022	0.725	0.020	0.721	0.012
New Latino/a destination (age 2)	0.116	0.034	0.262	0.050	0.256	0.035
Established Latino/a destination (age 2)	0.387	0.053	0.212	0.038	0.083	0.009
Fast-growing Latino/a hub (age 2)	0.411	0.044	0.095	0.017	0.099	0.020
Minor Latino/a destination (age 2)	0.087	0.018	0.431	0.051	0.562	0.035
Immigrant household	0.628	0.034	0.110	0.020	0.066	0.005
Acculturation index	0.468	0.027	0.928	0.013	0.962	0.003
Acculturation index missing	0.003	0.002	0.002	0.002	0.000	0.000
Gender (female)	0.467	0.018	0.501	0.014	0.491	0.012
Mother’s age at child’s birth	26.0	0.2	25.3	0.3	28.2	0.2
Low birth weight	0.065	0.005	0.129	0.006	0.071	0.003
Household SES index (age 2)	−0.601	0.031	−0.511	0.033	0.169	0.027
Mother working (age 2)	0.425	0.019	0.574	0.017	0.573	0.013
Mother working (age 2) missing	0.009	0.004	0.008	0.003	0.008	0.002
Non-resident father (age 2)	0.148	0.014	0.587	0.019	0.126	0.007
Number of siblings (age 2)	1.3	0.0	1.3	0.1	1.1	0.0
Three or more adults in household (age 2)	0.287	0.018	0.172	0.014	0.105	0.007
Moved within county (ages 2 to 4)	0.428	0.014	0.453	0.020	0.284	0.010
Moved counties within state (ages 2 to 4)	0.051	0.009	0.061	0.008	0.086	0.006
Moved counties between states (ages 2 to 4)	0.030	0.006	0.053	0.008	0.070	0.005
Moved to unknown destination (age 2 to 4)	0.006	0.003	0.007	0.003	0.002	0.001
Urban (age 2)	0.960	0.009	0.915	0.019	0.778	0.014
County number of child care centers per 1,000 children
ages 5 and under, 2003 (age 2)	2.8	0.1	4.0	0.1	4.1	0.1
State preschool investment score, 2002 (age 2)	1.2	0.1	1.6	0.3	1.0	0.2
State child care subsidy generosity score, 2002 (age 2)	0.8	0.1	0.1	0.1	0.3	0.1
State child care regulation stringency, 2002 (age 2)	0.5	0.0	0.4	0.1	0.5	0.1
County Latino/a with BA or higher, 2000 (%, age 2)	8.8	0.4	14.6	0.7	15.0	0.6
County 287g agreement, 2005–2010 (age 2)	0.432	0.046	0.185	0.026	0.111	0.016
County Latino-white index of dissimilarity, 2000 (age 2)	0.480	0.008	0.440	0.012	0.376	0.007
County per capita income, 1999 (in $1,000s, age 2)	20.6	0.4	21.8	0.5	21.5	0.4
County population total, 2000 (in thousands, age 2)	2,574.7	167.7	1074.8	63.4	656.5	40.3
County census tracts, 2000 (number, age 2)	554	37	250	15	149	10
State per capita income, 2000 (in $1,000s, age 2)	30.9	0.1	30.0	0.3	29.8	0.2
State population total, 2000 (in millions, age 2)	20.9	1.2	10.2	0.6	9.7	0.6
Number of observations (rounded to nearest 50)	1,100		1,300		3,600
Number of counties (rounded to nearest 50)	200		250		700
Number of states	35		38		48

Analytical Strategy

A series of 3-level multilevel logistic regression models, with children nested in counties nested in states, predicted the log-odds of enrollment in early childhood education programs. These models had varying intercepts that allowed the mean log-odds of enrollment to vary by county and state. The first set of models were estimated on the pooled sample of Mexican-origin, Black, and White children, and the second set of models were estimated on the Mexican-origin only sample to test the hypothesis regarding acculturation. The second set of models were restricted to the Mexican-origin sample because the acculturation measure did not vary significantly within the Black or White subgroups.

Pooled sample models were estimated sequentially, first examining baseline associations between destinations and early childhood education enrollment, then adding household selectivity variables, then urbanicity and county confounders, then early childhood education context variables and state confounders, then Latino/a educational attainment, and finally the county Latino-White index of dissimilarity (and county number of tracts) and 287g agreement indicator. Interactions between race/ethnicity and destinations were then added to the models for the pooled sample. The full model was then estimated for the Mexican-origin subsample, and interactions between the destination and acculturation index variables were then added to this model. All 3-level models were estimated using Stata 14.0 software, and descriptive statistics incorporated ECLS-B panel weights. The small number of missing values (see Table 1) were imputed using chained equations in Stata, with 10 multiply imputed datasets.

Results

Mexican-Origin, Black, and White Children across Latino/a Destinations

Table 1 displays weighted means for Mexican-origin children in the sample, as well as means for Black and White children. Table 2 displays weighted means for the Mexican-origin group across Latino/a destinations. Approximately half of Mexican-origin children were enrolled in an early childhood education program such as Head Start, preschool, or center care. Consistent with prior research (Brandon 2004; Crosnoe 2007), this enrollment level was substantially lower than for Black and White children, whose levels were closer to 72%. Figure 1 displays racial/ethnic differences in early childhood education enrollment across Latino/a destinations. Appendix Figure A1 also displays early childhood education enrollment by program type (e.g., center care versus Head Start) across destinations for Mexican-origin children.

Table 2.

Latino/a destinations and early childhood education enrollment among Mexican-origin, Black, and White children at age 4

	New Latino/a Destination		Established Latino/a Destination		Fast-Growing Latino/a Hub		Minor Latino/a Destination
Variable	Mean	S.E.	Mean	S.E.	Mean	S.E.	Mean	S.E.
Enrolled in pre-kindergarten (age 4)	0.401	0.044	0.596	0.032	0.488	0.030	0.537	0.066
County Latino/a 1990 (%, age 2)	2.4	0.3	36.8	4.5	20.8	1.1	4.8	0.6
County Latino/a change 1990–2000 (%, age 2)	430.9	79.9	35.6	1.3	94.5	9.0	83.2	10.3
Immigrant household	0.822	0.040	0.633	0.051	0.642	0.033	0.513	0.072
Acculturation index	0.336	0.036	0.485	0.045	0.464	0.029	0.582	0.062
Acculturation index missing	0.012	0.010	0.003	0.002	na	na	na	na
Gender (female)	0.474	0.044	0.438	0.026	0.487	0.029	0.496	0.058
Mother’s age at child’s birth	25.2	0.5	26.5	0.3	25.9	0.3	25.4	0.5
Low birth weight	0.035	0.009	0.079	0.006	0.055	0.008	0.088	0.025
Household SES index (age 2)	−0.686	0.055	−0.582	0.048	−0.624	0.046	−0.462	0.094
Mother working (age 2)	0.413	0.053	0.433	0.035	0.415	0.033	0.450	0.059
Mother working (age 2) missing	0.014	0.014	0.008	0.007	0.008	0.005	0.009	0.007
Non-resident father (age 2)	0.056	0.015	0.167	0.021	0.157	0.022	0.147	0.044
Number of siblings (age 2)	1.3	0.1	1.2	0.1	1.4	0.1	1.2	0.1
Three or more adults in household (age 2)	0.265	0.046	0.317	0.029	0.275	0.027	0.234	0.044
Moved within county (ages 2 to 4)	0.488	0.047	0.412	0.026	0.419	0.027	0.459	0.042
Moved counties within state (ages 2 to 4)	0.057	0.023	0.042	0.014	0.055	0.0144	0.066	0.020
Moved counties between states (ages 2 to 4)	0.027	0.018	0.038	0.011	0.0123	0.006	0.085	0.029
Moved to unknown destination (age 2 to 4)	0.000	0.000	0.006	0.004	0.010	0.007	0.000	0.000
Urban (age 2)	0.883	0.041	0.982	0.011	0.971	0.011	0.917	0.029
County number of child care centers per 1,000 children
ages 5 and under, 2003 (age 2)	3.3	0.5	2.8	0.1	2.5	0.2	3.8	0.3
State preschool investment score, 2002 (age 2)	1.6	0.7	1.4	0.3	1.1	0.1	0.3	0.2
State child care subsidy generosity score, 2002 (age 2)	−0.1	0.1	1.1	0.3	1.0	0.2	0.3	0.2
State child care regulation stringency, 2002 (age 2)	0.4	0.1	0.6	0.0	0.5	0.1	0.3	0.2
County Latino/a with BA or higher, 2000 (%, age 2)	10.3	2.0	8.7	0.3	7.5	0.6	12.8	1.2
County 287g agreement, 2005–2010 (age 2)	0.341	0.139	0.378	0.066	0.592	0.078	0.033	0.023
County Latino-white index of dissimilarity, 2000 (age 2)	0.410	0.021	0.545	0.016	0.461	0.018	0.369	0.022
County census tracts, 2000 (number, age 2)	68.2	15.9	1033.4	150.5	329.6	34.6	121.7	22.3
County per capita income, 1999 (in $1,000s, age 2)	21.1	1.1	20.0	0.9	20.8	0.5	21.1	0.7
County population total, 2000 (in thousands, age 2)	304.3	68.1	4,717.2	671.8	1,634.2	172.4	505.4	90.1
State per capita income, 2000 (in $1,000s, age 2)	29.1	0.5	31.5	0.2	31.3	0.1	28.7	0.5
State population total, 2000 (in millions, age 2)	6.9	1.0	26.3	1.3	22.9	2.0	6.5	0.9
Observations (rounded to nearest 50)	100		400		450		100
Counties (rounded to nearest 50)	50		50		50		50
States	19		8		10		30

Figure 1.

Mean early childhood education enrollment (weighted) among Mexican-origin, Black, and White children by Latino/a destination type, with 95% confidence intervals

Mexican-origin children were most likely to be enrolled in early childhood education when they lived in established Latino/a destinations, and they were least likely to be enrolled in new destinations. Black and White children followed a somewhat similar pattern. Their enrollment was highest in established destinations and was significantly lower in the other three types of destinations. Notably, large enrollment gaps between Mexican-origin and their Black and White peers prevailed across all four types of destinations.

Table 2 shows that both factors related to selectivity and contextual characteristics varied among Mexican-origin children across destinations in ways that could be meaningful for observed differences in early childhood education enrollment. On the one hand, relative to their counterparts in established destinations, for instance, Mexican-origin children in new destinations had lower levels of acculturation, were more mobile, were more likely to live in rural areas, and lived in states that had less generous child care subsidies. On the other hand, on many dimensions, Mexican-origin children in new destinations were either statistically indistinguishable from their peers in established destinations or had household characteristics (e.g., lower incidence of low birth weight, less likely to live in an extended household with three or more adults) or contextual characteristics (e.g., lived in counties with lower Latino-White segregation levels) that might be more conducive to early childhood education enrollment. Destinations as Contexts of Early Childhood Education

Multilevel logistic regression models first examined differences in the log-odds of early childhood education enrollment among the pooled Mexican-origin, Black, and White sample. Table 3 displays some results for these models, with full results in Appendix Table A1. Model 1 in Table 3 indicates that, net of racial/ethnic background, young children in established destinations were the most likely to be enrolled in early childhood education. Gaps in the log-odds of enrollment were widest for children in new versus established destinations; they were smaller but still significant for children in fast-growing hubs and minor destinations relative to those in established destinations.

Table 3.

Log-odds of early childhood education enrollment by Latino/a destination

Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
New Latino/a destination (Ref. Established destination)	−0.689*** (0.188)	−0.586*** (0.174)	−0.832*** (0.183)	−0.812*** (0.174)	−0.806*** (0.175)	−0.750*** (0.182)	−0.909*** (0.234)
Fast-growing Latino/a hub (Ref. Established destination)	−0.610** (0.197)	−0.570** (0.185)	−0.604*** (0.174)	−0.541** (0.169)	−0.543** (0.169)	−0.465* (0.184)	−0.488 (0.256)
Minor Latino/a destination (Ref. Established destination)	−0.484** (0.170)	−0.397* (0.159)	−0.631*** (0.166)	−0.554*** (0.163)	−0.544** (0.166)	−0.502** (0.171)	−0.631** (0.222)
Mexican origin (Ref. White)	−0.870*** (0.093)	−0.379** (0.116)	−0.356** (0.116)	−0.352** (0.116)	−0.353** (0.116)	−0.351** (0.116)	−0.505* (0.224)
Black (Ref. White)	−0.050 (0.081)	0.308*** (0.093)	0.339*** (0.094)	0.332*** (0.094)	0.333*** (0.094)	0.334*** (0.095)	0.113 (0.262)
New destination x Mexican origin							0.059 (0.306)
Fast-growing hub x Mexican origin							0.003 (0.272)
Minor destination x Mexican origin							0.510 (0.306)
New destination x Black							0.418 (0.299)
Fast-growing hub x Black							0.004 (0.374)
Minor destination x Black							0.176 (0.283)
Controls: Acculturation, gender, mother’s age at child’s birth, low birth weight, household SES (age 2), mother working (age 2), non-resident father (age 2), number of siblings (age 2), three or more adults in the household (age 2), mobility between ages 2 and 4 (within county, counties within state, counties between states, to unknown destination)		X	X	X	X	X	X
Urban, county per capita income, county log population total			X	X	X	X	X
County number child care centers per 1,000 young children, state preschool investment scores, state child care subsidy generosity score, state child care regulation stringency, state per capita income, state log population total				X	X	X	X
County Latino/a with BA or higher (%)					X	X	X
County Latino-White index of dissimilarity, county 287g agreement, county number of census tracts						X	X
Random Effects (State)	0.082 (0.124)	0.077 (0.122)	0.084 (0.110)	0.000 (1.845)	0.000 (41.218)	0.000 (8.115)	0.000 (613.655)
Random Effects (County)	0.483 (0.062)	0.414 (0.062)	0.367 (0.064)	0.355 (0.063)	0.355 (0.063)	0.354 (0.063)	0.357 (0.063)
Observations	6,000	6,000	6,000	6,000	6,000	6,000	6,000
Counties	850	850	850	850	850	850	850
States	48	48	48	48	48	48	48

Note:

^*p < .10

^**p < .05

^***p < .001.

Standard errors in parentheses. Sample sizes have been rounded to the nearest 50 to comply with NCES data regulations.

These patterns persisted after controlling for differences in household selectivity (Model 2), household urbanicity and county confounds (Model 3), state and county preschool contextual variables (Model 4), Latino/a B.A. attainment levels (Model 5), and the county Latino-White dissimilarity index and 287g indicators (Model 6). Several individual and household characteristics positively predicted early childhood education enrollment among the pooled sample, including mother’s age at child’s birth, household SES, mother working outside the household, non-resident father, and others negatively predicted enrollment, including total number of siblings and mobility across counties within states (see Appendix Table A1). State child care regulation stringency (positive predictor), county per capita income (positive predictor), and county population size (negative predictor) were the only significant contextual predictors of early childhood education enrollment. Controlling for race/ethnicity, selectivity characteristics, and contextual explanatory factors did not eliminate significant differences in children’s enrollment across destinations. In the full model (Model 6), children in new destinations, fast-growing hubs, and minor destinations were all less likely to be enrolled in early childhood education than those in established destinations. In supplementary models (available upon request), the destination reference groups in Model 6 were rotated. These models showed that the log-odds of enrollment, net of race/ethnicity and all covariates, were significantly higher in minor destinations versus new destinations but were not statistically different in fast-growing hubs versus new destinations.

Figure 2 displays the predicted probability of early childhood education enrollment across the four destination groups, based on the log-odds coefficients in Model 6 of Table 3 and holding all covariates at their mean values. The predicted probability of enrollment in early childhood education was highest in established destinations (80%), followed by fast-growing hubs and minor destinations (71.8% and 70.7%, respectively), and was lowest in new destinations (65.4%).

Figure 2.

Predicted probability of early childhood education enrollment across destinations

Were Mexican-origin children even less likely to be enrolled in early childhood education in new destinations relative to established destinations than their White peers? To answer this question, destinations were interacted with race/ethnicity in a model controlling for all covariates. The negative association between new destinations (versus established) and early childhood education enrollment that was observed for Whites did not vary significantly for either Mexican-origin or Black children (Model 7). Additionally, the associations of fast-growing hubs and minor destinations (versus established destinations) with enrollment did not vary by race/ethnicity. In supplementary models with rotated destination reference groups (available upon request), none of the interactions between destinations and race/ethnicity were significant. Moderation by Levels of Acculturation among Mexican-Origin Families

Multivariate models next examined whether the negative association between new destinations and early childhood education enrollment was strongest for Mexican-origin children in the least acculturated families. Model 1 in Table 4 re-estimated the full model from Table 3 (Model 6) for only the Mexican-origin subsample (with full model results available in Appendix Table A2). In this more restricted sample, only children in new Latino/a destinations had a significantly lower likelihood of enrollment in early childhood education versus children in established destinations, net of all covariates.

Table 4.

Log-odds of early childhood education enrollment by Latino/a destination and acculturation among Mexican-origin children

Variable:	Model 1	Model 2
New Latino/a destination (Ref. Established destination)	−0.954** (0.342)	−1.388** (0.436)
Fast-growing Latino/a hub (Ref. Established destination)	−0.280 (0.234)	0.057 (0.312)
Minor Latino/a destination (Ref. Established destination)	−0.276 (0.327)	−0.565 (0.498)
Acculturation index	0.186 (0.194)	0.277 (0.281)
New destination x Acculturation		1.352* (0.638)
Fast-growing hub x Acculturation		−0.587 (0.359)
Minor destination x Acculturation		0.521 (0.577)
Random Effects (State)	0.000 (4.020)	0.000 (6.742)
Random Effects (County)	0.000 (1.067)	0.000 (4.543)
Observations	1,100	1100
Counties	200	200
States	35	35

Note:

^*p < .10

^**p < .05

^***p < .001.

Standard errors in parentheses. Models control for full set of covariates (see Table 3, Models 6 and 7). Sample sizes have been rounded to the nearest 50 to comply with NCES data regulations.

Model 2 added cross-level interactions between the acculturation index and destination categories. Higher levels of acculturation attenuated the negative association between new Latino/a destinations (versus established destinations) and early childhood education enrollment. The log-odds of enrollment for those in the least acculturated households (i.e., acculturation index value of zero) living in new destinations (versus established destinations) were −1.39, but they were nearly zero (−1.39 + 1.35) for those in the most acculturated households (i.e, acculturation index value of 1.0) living in new destinations. In supplementary models (available upon request), the destination reference group was rotated in order to test for significant differences among Mexican-origin children in other types of destinations and to test interactions between destinations and acculturation. These models revealed that children in minor Latino/a destinations and in fast-growing hubs both had significantly higher log-odds of enrollment in early childhood education than children in new destinations. This association did not vary significantly by acculturation among Mexican-origin children in minor Latino/a destinations, but it did vary for children in fast-growing hubs, with acculturation attenuating differences in enrollment between children in these locations and new destinations.

To illustrate these points further, the log-odds coefficients from the full interaction model in Table 4 (Model 2) were used to calculate predicted probabilities of early childhood education enrollment for Mexican-origin children with low versus high levels of acculturation by destination (see Figure 3). In the calculation of these predicted probabilities, the acculturation index was set at 0.1 for low levels of acculturation (one standard deviation below the sample mean) and at 0.9 for high levels of acculturation (one standard deviation above the sample mean), with all other covariates held at their means for the Mexican-origin sample. They show that Mexican-origin children whose families had low levels of acculturation living in new destinations had the lowest predicted probability of early childhood education enrollment of any acculturation by destination group (26.8%). Gaps in the predicted probability of early childhood education enrollment were widest between Mexican-origin children with low levels of acculturation living in new versus established destinations and in new destinations versus fast-growing hubs (nearly 30 percentage points in both cases). Within destinations, the disparity in the predicted probability of early childhood education enrollment between children in low versus high acculturation households was largest in new destinations (23.2 percentage points) and was smallest in established destinations (5.4 percentage points).

Figure 3.

Predicted probability of early childhood education enrollment across destinations among Mexican-origin children by levels of family acculturation

Discussion and Conclusion

Examining differences in early childhood education enrollment among Mexican-origin children across destinations provides a lens from which to understand how place may be stratifying outcomes within the Mexican-origin population. The accommodations framework and its applications highlight how macro-level context and household-level factors shape family decision-making about early care and education (Coley et al. 2014; Crosnoe et al. 2016; Meyers and Jordan 2006). The demographic and theoretical literature on immigrant destinations suggests that Latino/a historical presence and growth could be important contextual determinants of early care and education patterns among immigrant-origin populations because of the ways in which they influence institutional arrangements, group status, and intergroup relations (Lichter 2013; Waters and Jiménez 2005). This study examined Latino/a destinations as contextual predictors of early childhood education enrollment among Mexican-origin children, compared the linkages across destinations and early childhood education enrollment between this group and non-Latino Whites and Blacks and within this group by family levels of acculturation, and determined whether early childhood education institutional contexts, Latino/a group status, and intergroup characteristics account for these associations.

The study results provide support for the first hypothesis that, relative to established Latino/a destinations, new destinations pose challenges for early childhood education enrollment for Mexican-origin, Black, and White children. Regardless of race/ethnicity and net of covariates, children in these three groups were the least likely to be enrolled in early childhood education in new destinations, followed by fast-growing Latino/a hubs and minor destinations, and then established destinations. Contrary to the second hypothesis, the association between destinations and early childhood education enrollment was not stronger for Mexican-origin children (or for Black children) than for White children. These results suggest that all three racial/ethnic groups face barriers to early childhood education enrollment in new destinations, not just Mexican-origin children.

Within the Mexican-origin group, acculturation attenuated differences in the likelihood of enrollment between those in new and established destinations, consistent with the third hypothesis. To test the fourth hypothesis, we explored several mechanisms that might explain the association between new destinations and enrollment in early childhood education, including differences in selectivity, state preschool context, county supply of child care establishments, Latino/a B.A. attainment, Latino/a-White residential segregation, and 287g agreements. Even after accounting for these factors, children in new destinations were significantly less likely to be non-enrolled in early childhood education than those in established destinations—in both the pooled sample and the Mexican-origin sample—counter to expectations.

Two main themes can be drawn from these findings. One theme is that living in a new Latino/a destination is associated with increased educational disadvantages among Mexican-origin children. These results are consistent with prior work showing worse educational outcomes among immigrant-origin adolescent youth in new versus established gateways (Ackert 2017; Dondero and Muller 2012; Fischer 2010). Taken together with prior research on residential segregation, poverty, and occupational stratification in new destinations (Crowley et al. 2015; Hall 2013; Lichter 2012; Lichter, Parisi, et al. 2015; Turner 2014), these findings suggest that new destinations are imposing greater barriers to incorporation for the Mexican-origin population than established gateways. Importantly, this study also demonstrates that lower early childhood education enrollment in new versus established destinations was not unique to Mexican-origin children, but was also observed for Black and White children. This finding suggests a generalized pattern of educational disadvantage among children in new destinations.

Given that early childhood education enrollment was measured in 2005 in this analysis, we cannot know whether new destinations have developed supports over time that could ameliorate disparities in early childhood education enrollment with established destinations among children generally and Mexican-origin children specifically. More recent data on early childhood education participation among Latinos/as by state, from 2012, suggests that some new Latino/a destinations such as Kentucky and South Dakota are achieving parity in Latino/a early childhood education enrollment levels with established gateways such as California and Texas, but others, such as Alabama and North Carolina are lagging behind (Table 5).

Table 5.

Latino/a children aged 3 to 5 enrolled in nursery school, preschool, or kindergarten, by Latino/a destination state, 2012

Location	Proportion
States with Largest Latino/a Populations (2011)
California	0.57
Texas	0.53
Florida	0.57
New York	0.63
Illinois	0.56
States with Fastest Latino/a Growth (2000–2011)
Alabama	0.34
South Carolina	0.44
Tennessee	0.47
Kentucky	0.57
South Dakota	0.57
Arkansas	0.53
North Carolina	0.45
Mississippi	0.48
Maryland	0.51
Georgia	0.49

Note: Data on early childhood education enrollments by state in 2012 obtained from the Kids Count Data Center (2017). State categorizations based on Brown and Lopez (2013).

Additionally, future research should further investigate why newer destinations may impose greater barriers to children’s enrollment in early childhood education than more established gateways. This study examined several potential mechanisms that could account for the negative association between new destinations and early childhood education enrollment, including early childhood education institutional supports, Latino/a group status, and intergroup relations. Given that Mexican-origin, Black, and White children all had the lowest levels of early childhood education enrollment in new destinations, investigating resources for early childhood that might affect all children and families, such as the availability of school-based early childhood education programs, could be an important area for exploration. Future work could also look at a broader set of contextual measures associated with immigration and incorporation, such as the proportion of the Latino/a population that is unauthorized and/or deportation levels.

One explanation for these patterns could be that children in new destinations receive lower returns to participation in early childhood education for school readiness. In other words, families in general and Mexican-origin families in particular may not enroll in early childhood education in new destinations because they do not perceive that programs in these places hold value for their children. Supplemental analyses (available upon request) explored this possibility using the ECLS-B Kindergarten cohort, regressing three types of school readiness outcomes at the start of kindergarten (math scores, reading scores, and prosocial skills) on early childhood education attendance, Latino/a destinations, race/ethnicity (Mexican-origin, Black, White), and covariates. This analysis did not find any evidence that school readiness returns to early childhood education participation varied by destination or by group-within-destination.

A second theme is that treating the Mexican-origin population as monolithic can obscure within-group differences among families and communities that are theoretically meaningful and policy relevant. Less-acculturated Mexican-origin families appeared to be the most vulnerable to not enrolling children in early childhood education in new destinations, and more-acculturated families were more resilient to destination effects. This interaction of destinations and household-level factors related to acculturation in shaping individual-level outcomes has been largely overlooked in the previous research (but see Ackert 2017). Future research is needed to determine why some Mexican-origin families are more susceptible to contextual conditions related to destinations as they make educational decisions for their children. For example, future work could examine perceptions of local early childhood education option among less-acculturated Mexican-origin families across destinations.

There are limitations to this study that must be acknowledged. First, as mentioned previously, the ECLS-B data are somewhat dated for understanding the most recent patterns of Mexican-origin early childhood education enrollment. We argue that the benefits of the ECLS-B—such as following families prospectively across the U.S. as they make early care and education decisions and identifying destinations at the county level—outweighed this drawback. Replicating the results of this study using more recent, even cross-sectional data, will be critical in determining whether these early childhood education enrollment patterns across destinations are persisting over time. The results of this analysis are also subject to bias due to non-random selection of families into destinations. Destination-level differences in early childhood education enrollment could be reflective of unmeasured heterogeneity among Mexican-origin families across gateways on factors such as motivation to seek out educational opportunities.

This study contributes to the literatures on immigrant destinations and educational stratification in the critical early childhood period. The history and growth of the Latino/a population in the places where Mexican-origin children live is related to the likelihood that they will attend early childhood education. Thus, destinations are associated with stratified educational outcomes among an already heterogeneous population. Additionally, in new destinations, Mexican-origin children, especially those in less-acculturated families, are missing out on exposure to educational opportunities that could attenuate early educational disadvantages. The ways in which these place-based disparities in early childhood play out over the long term is yet to be determined, but a failure to attend to the educational needs of this large, fast-growing, and disadvantaged group, especially in newer Latino/a destinations, could fuel long-term racial/ethnic socioeconomic inequalities.

Appendix

Fig. A1.

Early care and education type among Mexican-origin children across destinations, ECLS-B

Table A1.

Full Model Results from Table 3

Variable:	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
New Latino/a destination (Ref. Established destination)	−0.689*** (0.188)	−0.586*** (0.174)	−0.832*** (0.183)	−0.812*** (0.174)	−0.806*** (0.175)	−0.750*** (0.182)	−0.909*** (0.234)
Fast-growing Latino/a hub (Ref. Established destination)	−0.610** (0.197)	−0.570** (0.185)	−0.604*** (0.174)	−0.541** (0.169)	−0.543** (0.169)	−0.465* (0.184)	−0.488 (0.256)
Minor Latino/a destination (Ref. Established destination)	−0.484** (0.170)	−0.397* (0.159)	−0.631*** (0.166)	−0.554*** (0.163)	−0.544** (0.166)	−0.502** (0.171)	−0.631** (0.222)
Mexican origin (Ref. White)	−0.870*** (0.093)	−0.379** (0.116)	−0.356** (0.116)	−0.352** (0.116)	−0.353** (0.116)	−0.351** (0.116)	−0.505* (0.224)
Black (Ref. White)	−0.050 (0.081)	0.308*** (0.093)	0.339*** (0.094)	0.332*** (0.094)	0.333*** (0.094)	0.334*** (0.095)	0.113 (0.262)
New destination x Mexican origin							0.059 (0.306)
Fast-growing hub x Mexican origin							0.003 (0.272)
Minor destination x Mexican origin							0.510 (0.306)
New destination x Black							0.418 (0.299)
Fast-growing hub x Black							0.004 (0.374)
Minor destination x Black							0.176 (0.283)
Acculturation index		0.021 (0.136)	0.032 (0.137)	0.034 (0.137)	0.035 (0.137)	0.035 (0.137)	0.012 (0.139)
Female		−0.020 (0.061)	−0.022 (0.061)	−0.024 (0.061)	−0.024 (0.061)	−0.024 (0.061)	−0.029 (0.061)
Mother’s age at chid’s birth		0.022*** (0.006)	0.022*** (0.006)	0.022*** (0.006)	0.022*** (0.006)	0.022*** (0.006)	0.022*** (0.006)
Low birth weight		0.127 (0.065)	0.133* (0.065)	0.129* (0.065)	0.129* (0.065)	0.129* (0.065)	0.132* (0.065)
Household SES index		0.529*** (0.051)	0.521*** (0.052)	0.520*** (0.052)	0.520*** (0.052)	0.520*** (0.052)	0.517*** (0.052)
Mother working		0.252*** (0.063)	0.243*** (0.063)	0.247*** (0.063)	0.247*** (0.063)	0.247*** (0.063)	0.252*** (0.063)
Non-resident father		0.271** (0.087)	0.270** (0.087)	0.272** (0.087)	0.272** (0.087)	0.271** (0.087)	0.272** (0.087)
		-	-	-	-	-	-
Number of siblings		0.232*** (0.027)	0.230*** (0.027)	0.230*** (0.027)	0.230*** (0.027)	0.230*** (0.027)	0.229*** (0.027)
Three or more adults in household		−0.105 (0.087)	−0.102 (0.087)	−0.097 (0.087)	−0.097 (0.087)	−0.096 (0.087)	−0.100 (0.087)
Moved within county		−0.009 (0.071)	−0.006 (0.071)	−0.002 (0.071)	−0.002 (0.071)	−0.000 (0.071)	−0.006 (0.071)
Moved counties within state		−0.231* (0.117)	−0.241* (0.117)	−0.240* (0.117)	−0.241* (0.117)	−0.244* (0.117)	−0.248* (0.117)
Moved counties between states		−0.141 (0.132)	−0.140 (0.132)	−0.124 (0.132)	−0.123 (0.132)	−0.123 (0.132)	−0.129 (0.132)
Moved to unknown destination		−0.089 (0.523)	−0.086 (0.522)	−0.075 (0.524)	−0.075 (0.524)	−0.072 (0.524)	−0.062 (0.525)
Urban			0.107 (0.100)	0.125 (0.100)	0.127 (0.100)	0.136 (0.101)	0.133 (0.102)
County number child care centers per 1,000 young children				0.002 (0.022)	0.004 (0.023)	0.003 (0.023)	0.002 (0.023)
State preschool investment				0.041 (0.028)	0.041 (0.028)	0.045 (0.028)	0.044 (0.028)
State child care subsidy generosity				−0.045 (0.046)	−0.046 (0.046)	−0.046 (0.047)	−0.049 (0.047)
State child care regulation stringency				0.123* (0.054)	0.125* (0.054)	0.123* (0.055)	0.128* (0.055)
County Latino/a with BA or higher (%)					−0.002 (0.005)	−0.001 (0.006)	−0.000 (0.006)
County 287g agreement						−0.117 (0.143)	−0.103 (0.144)
County Latino-white index of dissimilarity						0.017 (0.423)	0.050 (0.425)
County per capita income (in $1,000s)			0.035*** (0.010)	0.033** (0.011)	0.034** (0.012)	0.035** (0.012)	0.034** (0.012)
			-	-	-
County population total (log)			0.146*** (0.038)	0.145*** (0.038)	0.144*** (0.038)	-0.159** (0.051)	-0.163** (0.051)
County number of census tracts						0.000 (0.000)	0.000 (0.000)
State per capita income (in $1,000s)				−0.025 (0.025)	−0.025 (0.025)	−0.024 (0.025)	−0.024 (0.025)
State population total (log)				0.067 (0.060)	0.070 (0.061)	0.061 (0.062)	0.065 (0.062)
Constant	1.586*** (0.165)	1.546*** (0.210)	1.579*** (0.223)	1.416*** (0.243)	1.401*** (0.247)	1.326*** (0.265)	1.471*** (0.297)
Random Effects (State)	0.082 (0.124)	0.077 (0.122)	0.084 (0.110)	0.000 (1.845)	0.000 (41.218)	0.000 (8.115)	0.000 (613.655)
Random Effects (County)	0.483 (0.062)	0.414 (0.062)	0.367 (0.064)	0.355 (0.063)	0.355 (0.063)	0.354 (0.063)	0.357 (0.063)
Observations	6000	6000	6000	6000	6000	6000	6000
Counties	850	850	850	850	850	850	850
States	48	48	48	48	48	48	48

Note:

^*p < .10

^**p < .05

^***p < .001.

Standard errors in parentheses. Sample sizes have been rounded to the nearest 50 to comply with NCES data regulations.

Table A2.

Full model results from Table 4.

Variable	Model 1	Model 2
New Latino/a destination (Ref. Established destination)	−0.954** (0.342)	−1.388** (0.436)
Fast-growing Latino/a hub (Ref. Established destination)	−0.276 (0.327)	−0.565 (0.498)
Minor Latino/a destination (Ref. Established destination)	−0.280 (0.234)	0.057 (0.312)
Acculturation index	0.186 (0.194)	0.277 (0.281)
Minor destination x acculturation index		0.521 (0.577)
New destination x acculturation index		1.352* (0.638)
Fast-growing hub x acculturation index		−0.587 (0.359)
Gender (female)	−0.087 (0.131)	−0.094 (0.132)
Mother’s age at child’s birth	0.004 (0.013)	0.003 (0.013)
Low birth weight	0.211 (0.152)	0.227 (0.153)
Household SES index	0.487*** (0.129)	0.475*** (0.130)
Mother working	0.082 (0.142)	0.112 (0.143)
Non-resident father	0.133 (0.190)	0.134 (0.192)
Number of siblings	−0.193*** (0.058)	−0.200*** (0.059)
Three or more adults in household	0.140 (0.149)	0.162 (0.150)
Moved within county	0.035 (0.143)	0.044 (0.144)
Moved counties within state	−0.666* (0.312)	−0.688* (0.315)
Moved counties between states	−0.006 (0.386)	0.020 (0.390)
Moved to unknown destination	−0.369 (0.967)	−0.431 (0.965)
Urban	−0.627 (0.354)	−0.550 (0.359)
County number child care centers per 1,000 young children	0.157* (0.070)	0.161* (0.072)
State preschool investment	0.191* (0.076)	0.225** (0.079)
State child care subsidy generosity	0.167* (0.079)	0.181* (0.080)
State child care regulation stringency	−0.041 (0.162)	−0.067 (0.165)
County Latino/a with BA or higher (%)	0.007 (0.024)	0.003 (0.024)
County 287g agreement	−0.098 (0.225)	−0.141 (0.230)
County Latino-white index of dissimilarity	0.099 (1.133)	−0.163 (1.152)
County per capita income (in $1,000s)	−0.010 (0.024)	−0.011 (0.024)
County population total (log)	−0.089 (0.114)	−0.061 (0.117)
County number of census tracts	0.000 (0.000)	0.000 (0.000)
State per capita income (in $1,000s)	−0.075 (0.064)	−0.076 (0.065)
State population total (log)	−0.052 (0.126)	−0.045 (0.127)
Constant	0.565 (0.533)	0.358 (0.557)
Random Effects (State)	0.000 (4.020)	0.000 (6.742)
Random Effects (County)	0.000 (1.067)	0.000 (4.543)
Observations	1,100	1,100
Counties	200	200
States	35	35

Note:

^*p < .10

^**p < .05

^***p < .001.

Standard errors in parentheses. Sample sizes have been rounded to the nearest 50 to comply with NCES data regulations.