## Abstract

The academic performance of foreign-born youth in the United States is well studied, yet little is known about whether and how foreign-born students influence their classmates. In this article, I develop a set of expectations regarding the potential consequences of immigrant integration across schools, with a distinction between the effects of sharing schools with immigrants who are designated as English language learners (ELL) and those who are not. I then use administrative data on multiple cohorts of Florida public high school students to estimate the effect of immigrant shares on immigrant and native-born students’ academic performance. The identification strategy pays careful attention to the selection problem by estimating the effect of foreign-born peers from deviations in the share foreign-born across cohorts of students attending the same school in different years. The assumption underlying this approach is that students choose schools based on the composition of the entire school, not on the composition of each entering cohort. The results of the analysis, which hold under several robustness checks, indicate that foreign-born peers (both those who are ELL and those who are non-ELL) have no effect on their high school classmates’ academic performance.

## Introduction

The question of how well foreign-born youth1 adapt to their communities and schools in the United States is an important and popular topic in demographic research and related disciplines (e.g., Glick and White 2003; Kao and Tienda 1995; Perreira et al. 2006; Portes and Rumbaut 2001; Thomas 2009). Yet, very little is known about how the presence of immigrant youth in the nation’s schools influences the human capital development of their schoolmates. One assumption is that immigrants will reduce the learning climate of the school because of their English language needs and their disproportionate representation among low social class and racial minority groups. Assumptions such as these typically lead to anti-immigrant sentiment; in 2010 alone, more than 1,400 immigration-related bills and resolutions were introduced by state legislatures, with the majority of them aimed at restricting the rights of immigrants (National Conference of State Legislatures 2011). Many of these initiatives originate in states and localities that have no tradition of welcoming new immigrants. Indeed, the recent demographic phenomenon of U.S. newcomers settling in regions outside major immigrant gateways has generated concerns about the potential harms of these new settlement patterns to both immigrant and native-born individuals (e.g., Hall 2013; Singer 2004). At the same time, research on immigrant children’s school performance has found that these children often outperform what would be expected for students of their social class and racial category. The educational achievement found among some first-generation youth has been attributed to their positive attitudes, such as high educational aspirations, strong attachments to school, and respect for elders, teachers, and other community members (Suárez-Orozco and Suárez-Orozco 2001). Findings such as these suggest that students might benefit from sharing schools with immigrants who have some proficiency in English. In short, conventional wisdom and prior research yield ambiguous predictions about the consequences of integrating foreign-born and native-born youth across U.S. schools.

The goal of this article is to offer an empirical test of foreign-born peer effects. Toward that goal, I first describe a set of mechanisms to explain how students might be influenced by the foreign-born composition of their classmates. Importantly, this conceptual framework distinguishes between the effects of immigrant peers who are not yet proficient in English and those who are at least minimally proficient. I then use administrative data on multiple cohorts of public high school students in Florida and three measures of high school performance to answer the following two questions: (1) do immigrant students who are English Language Learners (ELL) and immigrants who are not ELL impact the academic outcomes of their schoolmates?; and (2) do these immigrant peer effects matter differently for youth who are themselves immigrant ELL, immigrant non-ELL, and native-born? The challenges to causal identification of peer effects are well known and difficult to overcome given that parents and students in the United States typically self-select into their neighborhoods and schools. In the absence of random assignment to schools, I use a design that identifies the effect of peer demographics from plausibly random deviations in peer group characteristics across cohorts over time within the same school. The key identifying assumption to this across-cohort/within-school design is that families and students choose schools based on the average composition of the school, not on the composition of the entering cohort, such that small deviations in immigrant shares across cohorts are uncorrelated with unobserved differences in the characteristics of the students who choose the school in different years. To provide justification for this approach, I show that the changes in immigrant shares across cohorts within schools are uncorrelated with key observable characteristics of the students and their families, such as their gender, race, age, and poverty status.

The results of this inquiry, which are robust to several sensitivity checks, reveal no changes in the academic performance of students as the share of immigrants who are ELL or the share of immigrants who are not ELL across school cohorts increases. The absence of a negative immigrant peer effect is consistent with most of the immigrant student achievement literature, which points to equal (and often higher) achievement for immigrants with some proficiency in English relative to native-born students after adjusting for race and class. Yet, the results here also suggest no harm to sharing schools with immigrants who are not yet minimally proficient in English, a result that holds even when native-born ELL students are included in the peer ELL share. Although the data permit analysis of native-born ELL peer effects, the analysis does not examine the effect of sharing schools with non-ELL native-born children of immigrants. In practice, this means that this article provides an analysis of the short-run consequences of immigration on schools, where second-generation peer effects might shed light on the longer-run consequences.

## Academic Performance of Immigrant Students in the United States

Research on immigrant students in the United States tells a complex story. On the one hand, most foreign-born youth face considerable obstacles that might be expected to interfere with their schooling. In addition to being less familiar with U.S. norms and institutions, they often lack citizenship or documentation (and by extension, access to public benefits) and a fluent command of the English language. They are also more likely to be racial minorities, to come from poor families, and to live in large urban areas where school systems are underresourced and achievement falls below national norms (Hernandez and Charney 1998; Van Hook and Fix 2000; Van Hook et al. 2004). Naturally, then, when newly arrived immigrants are compared with white, nonpoor, fully English-proficient, native-born youth who attend endowed schools, most attain lower test scores, lower grades, and fewer years of schooling (e.g., Kao 1999; Kao and Tienda 1995; Perreira et al. 2006).

On the other hand, much of the empirical research on immigrant students has found that they fare relatively well when compared with native-born students who have similar racial and socioeconomic profiles, a finding that has been termed the “immigrant paradox” because it defies what would be predicted given the difficulties faced by immigrant children. With the exception of some Central American immigrants (e.g., Suárez-Orozco 1989), much of this research points to a general advantage of being an immigrant, despite the challenges of learning a foreign language and new customs (Caplan et al. 1989; Conger et al. 2007; Fuligni 1997; Glick and White 2003; Kao 1999; Kao and Tienda 1995; Perreira et al. 2006; Portes and Rumbaut 2001; Schwartz and Stiefel 2006; Tillman et al. 2006; Zhou and Bankston 1998). Immigration scholars often attribute these paradoxical findings to selective migration on characteristics that are difficult to observe. That is, although many immigrants arrive with observably low levels of human capital, they possess a unique mindset (such as high aspirations regarding education, work, family, and community) or innate ability that puts them ahead of native-born individuals with similar human capital inputs (Caplan et al. 1989; Gibson 1988; Kao and Tienda 1995; Portes and Rumbaut 2001; Raleigh and Kao 2010; Suárez-Orozco and Suárez-Orozco 2001; Zhou and Bankston 1998). Another possibility is that foreign-born children experience and react to U.S. schools and customs in a way that provides advantages. Some research has found, for instance, that immigrants who belong to racial or ethnic minority groups process racial discrimination differently than native-born minorities. Although native-born minorities harbor negative feelings about their position with respect to white Americans, foreign-born minorities compare their conditions with their home country and, consequently, perceive a relatively positive experience (Ogbu 1992; Waters 1999). These positive attitudes may also stem from being treated differently by native-born teachers and employers (Waters 1999).

Much of the research on immigrant student achievement explores whether children who are more acculturated (socially, linguistically, and psychologically) to American values and customs experience lower levels of school attachment and achievement. As a proxy for acculturation, this work often distinguishes between the achievements of first-generation, second-generation, and later-generation immigrants, with the assumption that first-generation youth are less acculturated than second-generation youth, and so on. Comparisons between immigrant generations also offer a partial test of whether the relative performance of foreign-born is driven primarily by immigrant parents or by the combination of immigrant parentage and immigrant status. The results on the achievements of second-generation immigrants are mixed, with some early studies showing higher achievement among both first- and second-generation students than later-generation students (e.g., Kao and Tienda 1995) and more recent studies finding no evidence of a second-generation advantage over first or later generations (e.g., Pong and Zeiser 2012; Pong et al. 2005). Thus, there does not appear to be irrefutable evidence that foreign-born student achievement is driven solely by immigrant parents, nor that first- and second-generation immigrant children have the same experiences in school.

There is evidence, however, that immigrants who are designated as ELL possess the same beneficial attitudes as non-ELL immigrants, but they are less likely to overcome the barriers associated with ELL status. Indeed, the academic achievement gap between ELL and non-ELL students is one of the largest and most persistent gaps in the nation’s schools (August and Hakuta 1997). All schools identify their ELL students and develop strategies for increasing their English language skills. Thus, there are good reasons to expect the effect of immigrant peers to matter differently depending on whether those peers are also ELL—a set of expectations that I expand upon in the next section.

## Why Immigrant Peers Might Matter to Student Performance

Immigrant peers might affect their fellow classmates in a number of ways. Most notable are peer effects that operate through academic aspirations and abilities. From the tracking and peer achievement literatures, we know that peer group knowledge, motivation, and behavior can both directly and indirectly affect the performance of an individual student (Gamoran and Mare 1989; Hallinan 1994; Hanushek et al. 2003; Hoxby 2000; Pallas et al. 1994; Vigdor and Nechyba 2007). Students’ decisions may be influenced by the behavior of other students, including how much they study, how excited they are about learning, and how well they listen. Spillover effects of peers might also operate indirectly through the teachers: peers that demand extra attention from teachers because of learning needs (such as language assistance or emotional issues) could detract from the learning of the other students (Fletcher 2010). Higher-performing peers, in contrast, might ask more advanced questions of the teacher, which increases the quantity and rigor of material provided to the class. High-achieving peers could also attract better educators to the school or classroom; teachers often report that they prefer teaching to higher-ability students and find combination classes and classes for lower-ability students challenging and unrewarding (Burns and Mason 1998; Finley 1984). The direction of the academic component of the immigrant peer effect will, therefore, depend on the academic abilities of the immigrant students in the school. If the immigrant students have limited English skills and require a significant amount of educator and other resources, then students who do not benefit from these resources might receive a lower-quality education. Conversely, if the immigrant students are English proficient and bring a relatively high work ethic and strong attachment to schooling, these attitudes could positively influence nearby students.

Immigrant students could also affect their classmates through social pathways, and these effects are likely to matter differently for students who are immigrant ELL, immigrant non-ELL, and native-born. Such mechanisms are well known to those who study racial and socioeconomic segregation in kindergarten–12 schools. Research on racially segregated schools has suggested that nonwhite and low-income students in isolated schools are excluded from the informal networks and social norms that facilitate access to college and the labor market. These networks could include classmates’ parents, school personnel, or other acquaintances that can provide influential referrals for colleges and jobs. Other benefits from exposure to highly resourced peers include learning how to write a successful essay on a college application or how to present well in a job interview (Wells and Crain 1994). For foreign-born youth, the social consequences of having many foreign-born peers (and few native-born peers) could be even greater. Many newcomers are introduced to American culture through formal educational settings (Kasinitz 2001; Portes and Rumbaut 2001; Suárez-Orozco and Suárez-Orozco 2001), and several studies have documented evidence of cultural and social benefits to Latino students from having friends who belong to the dominant cultural group (Ream 2005; Stanton-Salazar and Dornbusch 1995). Absent frequent contact with native-born students, immigrants in isolated schools (particularly those who are ELL) may have difficulty learning the language, customs, and norms of U.S. society, which could in turn affect their ability to achieve.

Social psychologists have also uncovered psychological effects of school peers’ socioeconomic and demographic makeup. Crosnoe (2009), for example, found better psychological outcomes (namely, self-perceptions and feelings of belongingness) for low-income students in schools with more low-income students (as opposed to fewer). He attributed these findings to the fact that peers of similar class eliminate the comparative disadvantage and emotional strain that some students face in schools where they are in the extreme minority. Related work on immigrants has found that newly arrived immigrant youth with limited English proficiency in high schools often report being teased or singled out because of their mannerisms, accents, or clothing (Matute-Bianchi 1986; Olsen 1997). Another recent study found that the effect of ESL placement on academic outcomes is far more negative for immigrant students in schools with fewer other immigrant students, again pointing to an academic or social marginalization that may occur when immigrant students are in the minority (Callahan et al. 2009). Evidence also suggests that native-born students sometimes resent immigrant students who require extra attention (e.g., language assistance) or who excel in school and earn educational opportunities, such as slots in advanced courses (Olsen 1997). Correspondingly, one recent study showed academic benefits to Latino students from having coethnic friends who provide support for their cultural identity and general encouragements (Riegle-Crumb and Callahan 2009). Thus, having more immigrant peers might prevent immigrant children from feeling isolated or different, which may in turn improve their academics (and, for opposite reasons, reduce the achievement of the native-born students in the school).

To summarize, immigrant peers could affect individual students through academic or more social and psychological mechanisms. The literature on immigrant performance suggests that immigrant students who are not fully proficient in English are likely to have negative effects on the learning climate of the school, whereas those who are English proficient might generate positive externalities. In addition, the effect of more immigrant classmates should matter differently depending on whether the student is immigrant or native-born. Immigrant youth may be more likely to develop friendships with other immigrants, and thus be more influenced by their presence. If there are academic and social impacts of this higher level of exposure, then the anticipated effects of immigrant peers (either positive or negative) will be largest for students who are also immigrants. Given the administrative nature of the data (namely, no measures of social or psychological outcomes), this article is unable to document the precise pathways through which immigrant peers operate. Nevertheless, the reduced-form estimates shed light on immigrant integration and inform policy debates about the potential benefits or harms of such integration.

## Prior Relevant Studies on Peer Effects

Launched by the widely cited Coleman Report (Coleman et al. 1966), there is now a long line of studies on the effect of peer group race/ethnicity and/or socioeconomic status on student achievement. Recent examples include Angrist and Lang (2004), Bifulco et al. (2011), Crosnoe (2009), Hanushek et al. (2009), Hoxby (2000), Rivkin (2000), Rumberger and Palardy (2005), and Ryabov and Van Hook (2007). In contrast, a relatively small body of work (five studies, to my knowledge) has examined the effect of immigrant-origin concentrations in school on the academic achievement of the students.

Using Israeli administrative data, Gould et al. (2009) found that foreign-born concentrations in elementary school have modest negative effects on the probability that Israeli native-born will pass their matriculation exams (a prerequisite for college enrollment), with no observed impacts on the quality of the high school attended or the likelihood of high school dropout. In a similar analysis, Schwartz and Stiefel (2011) used administrative data on New York City primary school students and found that attending a school with more foreign-born had no impact on students’ reading and math scores. Using data from the Children of Immigrants Longitudinal Study, a sample of first- and second-generation immigrant youth from San Diego and Miami, Cortes (2006) also found no effect of immigrant-origin concentrations in middle school on other immigrant-origin students’ test scores. Finally, despite the perception of stronger peer effects among teens than among younger children, in two separate analyses that used the National Longitudinal Study of Adolescent Health (Add Health) data (a representative sample of U.S. high school–age youth), Crosnoe and Lopez-Gonzalez (2005) and Riegle-Crumb and Callahan (2009) revealed no relationship between the share of first- and second-generation immigrants in the school and Latino students’ grades.

A summary of this recent research suggests that in U.S. schools, immigrant students appear to have no influence on their classmates. Even the studies that have expanded the peer share to include second-generation immigrants have found no negative externalities from sharing schools with native-born children of immigrants. Although these prior studies are informative, they are few and can be extended in several ways. Most notably, this article distinguishes between the effects of immigrant classmates who are ELL versus those who have a stronger command over English, with ample reasons to expect differences in these effects. This study also provides estimated effects of immigrant peers in secondary school on all immigrant and native-born students (with the prior studies of secondary school youth focusing exclusively on Latino students). Finally, the data come from one of the largest immigrant-receiving states in the United States (Florida), and consequently applies to a relatively larger share of U.S. immigrants than data sources from smaller geographic areas.

## Data Sources, Measures, and Samples

Data for the study were obtained from the Florida Department of Education (FLDOE), which maintains an integrated statewide longitudinal database that tracks the demographic characteristics and educational experiences of students in Florida’s public elementary, secondary, and postsecondary schools. From this data source, I obtained longitudinal records on four cohorts of 9th grade students, with the first cohort entering the 9th grade in 2000–2001 and the last cohort entering in 2003–2004. The cohorts are “progressive,” meaning that any student who enters the Florida system after the 9th grade and who would be in the same graduating class is included in the cohort. For instance, a student who entered in the 10th grade in 2001–2002 is appended to the 9th grade 2000–2001 cohort, and so on.

Student data include race/ethnicity,2 gender, disability status, grade level, birth date, birthplace, language most frequently spoken at home, and whether the student is classified as ELL. Using the birthplace variable, I identify immigrants as students not born on U.S. soil, which means that students born in Puerto Rico, Guam, and the other U.S. territories are treated as native-born. I further classify immigrant students into those who are ELL and those who are not, where ELL students are those who fall below predetermined scores on tests that measure students’ proficiency in listening, speaking, reading, and writing English.3 As is characteristic of public school system records, the available measure of income is whether the student is eligible for the subsidized meal program; students eligible for free lunch live in families with incomes at or below 130 % of the federal poverty level, and reduced-price lunch eligible students live in families with incomes between 130 % and 185 % of the poverty level. Students are also linked to the schools attended in each year, which permits aggregation from the student to the grade and school level and, correspondingly, measures of the characteristics of students in the school, grade, and year.4

Three measures of academic performance are used for the analyses. The first two measures are students’ 10th grade scores on the reading and math portions of the Florida Comprehensive Assessment Test (FCAT). The FCAT, which was first administered in 1998 to public school students in grades 4, 5, 8, and 10, is a criterion-referenced test aligned with Florida’s Next Generation Sunshine State Standards. FCAT scores are standardized within subject, grade, and year to a mean of 0 and a standard deviation of 1 for all analyses. The third measure is an indicator variable set to 1 if a student took the Scholastic Aptitude Test (SAT), the nationally recognized college-entrance exam and a measure of the students’ academic motivation and college aspirations.

The four cohorts of 9th graders include a total of 592,293 students. Approximately 6 % of these 9th grade students were not enrolled in the 10th grade; as a result, they are missing data on all three outcomes.5 From the group of students enrolled in the 10th grade, I delete students with missing sociodemographic data as well as students who attended grades with fewer than 50 students. These omissions result in a sample reduction of roughly 2 % and an analytic sample of 544,290 students. All students in the analytic sample have scores on the reading FCAT and a value for whether they took the SAT exam; approximately 2,000 students in the analytic sample did not have a score for the FCAT math exam, resulting in a sample of 541,895 for this outcome.

Three limitations of the data are worth noting. First, the data are restricted to public school students. Because U.S. immigrant students have a lower rate of private school attendance than native-born students, public school records may capture the bulk of the story (Betts and Fairlie 2001); however, the findings may not apply to the private school system. Second, the data may best generalize to students in Florida, which has a disproportionately large share of Cuban immigrants. Sixty-eight percent of Cuban Americans live in Florida, and they tend to be more educated and have higher incomes than other Hispanics (Pew Hispanic Center 2006). In a robustness check (described later herein), I drop the Cuban-born immigrants from the sample to determine the sensitivity of the results to this unique population. Third, although the data identify native-born youth who are ELL, the data do not permit identification of non-ELL second-generation immigrants; thus, the estimates capture the effect of sharing schools only with first-generation immigrant children.

This data set also has key advantages. Unlike data sets that sample a single cohort of high school students across the nation, the multiple cohorts provided by the statewide administrative data allow for causal identification using the within cohort across school design. Further, the census of schools and students in the data set provides sufficient numbers of immigrant students within schools to reliably estimate the effects of immigrant peers across cohorts over time. The large numbers of students also provide sufficient statistical power for analyses of heterogeneity in the immigrant peer effect by nativity and English language status.6

## Methodology

### Main Empirical Models

One of the main difficulties in identifying peer effects is the endogeneity of choices: the same characteristics that determine where and with whom students go to school may also determine their achievement in school. Moreover, student and peer achievement occur contemporaneously (for instance, when an individual student convinces his peers to cheat on an exam). Studies that rely on random assignment mechanisms to estimate peer or school effects tend to provide the strongest causal evidence (see, e.g., Park et al. 2013).

Of the prior relevant studies, the studies of Israeli and New York City primary school students have provided the strongest identification by exploiting random variation in the number of immigrants across grades within the same school. Gould et al. (2009) noted, for example, that independent of the total number of immigrants in a school, the number of immigrants in a specific grade is due to random factors, such that endogeneity is overcome by examining conditional grade-level effects.7

I take a related approach to reduce bias that uses data on students who attend the same school but who are enrolled in different years. The main specification is as follows:
$yist=α1PIELLst+α2PINonELLst+βXi+δs+γt+εist,$
(1)
where yist is the outcome for student i in school s in year t. The primary variables of interest are PIELLst and PINonELLst, which are calculated as the percentage immigrant ELL and percentage immigrant non-ELL, respectively, in the students’ 10th grade. Note that the immigrant shares are calculated at the grade level and not the school level; however, the effect of immigrant peers in this model is not identified from across-grade differences within school because the outcomes are not measured for each high school grade. The peer composition is calculated based on the fall semester of the 10th grade, whereas all academic performance outcomes occur in the spring of the 10th grade year or later.8 The models are estimated with ordinary least squares (OLS) for the continuous dependent variables and probit for the binary dependent variable (with the average of the marginal effects reported); standard errors are corrected for heteroskedasticity and for clustering at the school level.

Equation (1) also holds constant Xi, a vector of characteristics unique to student i, measured in the students’ 9th grade year, including ELL/immigrant, gender, age, race/ethnicity, eligibility for free or reduced-price lunch (FRPL), disability status, and most frequently spoken home language is other than English. δs are fixed effects for the school that the student attended in the 10th grade,9 which control for unobserved school attributes that correlate with grade-level immigrant shares; and γt are fixed effects for the year the student attended the 10th grade, allowing for different intercepts for each of the four cohorts. As such, the effect of immigrant peers is identified from idiosyncratic shocks to 10th grade immigrant shares within schools over time.

To explore the differential effect of immigrant shares on immigrant ELL, immigrant non-ELL, and native-born students, Eq. (1) is estimated separately as follows:
$yistN=αNPIELLstN+αNPINonELLstN+βNXiN+δsN+γtN+εistN,N=ImmigrantELL,ImmigrantNon‐ELL,Native‐born,$
(2)
where all variables are the same as those in Eq. (1), and the immigrant ELL and non-ELL coefficients in Xi are constrained to 0.

### Sensitivity Analyses

The robustness of the model is tested against alternative samples and specifications. First, to determine the sensitivity of the results to the loss in sample members due to missing demographic data and enrollment in small schools, I add these observations back into the sample and impute the missing values 10 times using multiple imputation by chained equations (Royston 2004; Rubin 1987). Second, to determine the sensitivity of the results to Cuban-born students, I drop Cuban-born students from the analytic sample. Cuban students make up approximately 15 % of the immigrants in the sample and 26 % of the Hispanic immigrants in the sample (see Table 7 in the appendix for the top 10 origin countries of immigrants in the sample overall and by race). The use of this alternative sample sheds light on the generalizability of the results to states and localities outside Florida, where the Cuban-born population represents much smaller shares. Third, to test for the sensitivity of the results that rely on OLS with Huber-White standard errors, I reestimate the equation with a random intercept, which partially accounts for the between-cluster heterogeneity in effects and separates the error term into between- and within-cluster components. This alternative estimator includes the across-cluster (ωst) and within-cluster (εist) error components as follows:
$yist=α1PIELLst+α2PINonELLst+βXi+δs+γt+ωst+εist.$
(3)
The random intercept model also uses the intragroup correlation to produce weighted coefficients; thus, in addition to yielding different standard errors, the inclusion of random effects can yield different estimated coefficients than an OLS model. Fourth, I add school-specific linear trends to control for unobserved factors that change linearly over time within schools and that might confound results obtained from the traditional school fixed-effects specification. For instance, within-school variation in academic outcomes may correlate with school-specific characteristics, such as the quality of the teachers or the other demographic traits of the students. If these unobserved school attributes change over time and correlate with changing immigrant shares, then the school fixed-effects specification will be biased. To address this concern, I estimate the following equation:
$yist=α1PIELLst+α2PINonELLst+βXi+δs+γt+τs⋅t+εist,$
(4)
where τs denotes the time trend for school s and provides a means of controlling for changes in academic outcomes that might be correlated with other school-level trends. In an additional analysis, I add other characteristics of the student’s peer group, including the share of students who are Asian, black, Hispanic, FRPL, in homes where a language other than English is predominantly spoken, and designated as disabled, captured by vector Zst in the following equation:
$yist=α1PIELLst+α2PINonELLst+βXi+ØZst+δs+γt+εist.$
(5)

Finally, as described later herein, a nontrivial number of native-born students are also ELL (they represent only 1 % of all native-born, but 19 % of all ELL). To estimate the complete impact of ELL peers (both native-born and foreign-born), I estimate two additional models. The first model adds to Eq. (1) the share of peers who are native-born ELL and whether the student is native-born ELL. This model permits separate estimation of native-born ELL peer effects as well as performance gaps between native-born ELL and their classmates. The second model replaces the peer share immigrant ELL variable with peer share ELL, which permits estimation of the effect of ELL peers irrespective of their nativity status.

## Results

### Characteristics of the Students and Their Schools

Panel A of Table 1 provides summary statistics on the demographic characteristics and educational needs of the students in Florida secondary schools, where immigrant ELL, immigrant non-ELL, and native-born youth differ in ways that are largely consistent with national trends. Of the approximately 550,000 students, roughly 4 % are immigrant ELL and another 8 % are immigrant non-ELL. Immigrant students in Florida and nationally are disproportionately Asian, Hispanic, FRPL, in homes where a language other than English is the primary language spoken, and ELL (Hernandez and Charney 1998; Van Hook et al. 2004). Between the two immigrant groups, those designated as ELL are also more likely to be Hispanic, FRPL, and in homes with a non-English primary language. Immigrant high schoolers in Florida also have much lower rates of disabilities than native-born children, a pattern that has been documented among students using other national and local data sources (Conger and Grigorenko 2009; Conger et al. 2007). Much like native-born in other immigrant gateways, native-born children in Florida are also touched by immigration, with approximately 13 % of them living in homes where English is not the primary language spoken and 1 % (approximately 5,500 students) designated as ELL despite being born in the United States (not shown in the table).

Also consistent with national trends, immigrant youth who are ELL have raw academic outcomes that are significantly lower than those of English-speaking immigrant and native-born youth: immigrant ELL students score lower on both the math and reading portions of the FCAT and are less likely to take the SAT exams than native-born and immigrant non-ELL students (see Panel B of Table 1). At the same time, immigrant students who are not ELL score higher on both achievement tests (on the order of 0.02 standard deviations) and are slightly more likely to take the SAT exam than their native-born and ELL schoolmates.

The schools attended by the immigrant and native-born students differ as well (see Panel C of Table 1). Although immigrant ELL students make up only about 4 % of the sample, the average immigrant ELL student is enrolled in a school (in the 10th grade, more specifically) that is approximately 27 % immigrant (11 % ELL and 16 % non-ELL). Immigrant students with greater command of the language are not much more integrated with native-born: although they make up approximately 8 % of all students, their schools are roughly 23 % immigrant. Immigrant students also attend schools that are larger and that that have more Hispanic, FRPL, and ELL students than the schools attended by native-born students. The resources of the schools (per pupil expenditures, teacher characteristics) attended by the three groups are very similar, however.

The high degree of isolation among immigrant—specifically, immigrant ELL—students is consistent with national trends (Cosentino de Cohen et al. 2005). The absence of differences between groups on observed school resources is likely due to Florida’s policy of ensuring relatively equal funding across schools; however, there are likely to be other resource differences that are not captured by the administrative data. Other studies that examine more measures of school resources have detected differences in the schools attended by ELL and non-ELL students (for instance, in the principals’ credentials and years of experience), which reinforces the need for identification of peer effects that hold these schoolwide resources constant (Cosentino de Cohen et al. 2005).

### Model Assumptions: Within-School Variation and Balancing Tests

Two assumptions must be met in order to evaluate the appropriateness and validity of the identification approach. First, there must be sufficient within-school variation in immigrant shares over time to generate precise estimates. Second, the characteristics of youth and their families are uncorrelated with the deviations in immigrant shares across cohorts over time.

Regarding the first assumption, Panel A of Table 2 presents the total (across- and within-school) variation in immigrant shares for all four cohorts, and Panel B presents the within-school variation. The distribution shown in Panel B is calculated from the residuals of a regression of peer share immigrant (ELL or non-ELL) on school and cohort fixed effects. As shown, the total variation in immigrant ELL shares averages at approximately 16 % for immigrant ELL, 13% for immigrant non-ELL, and 6 % for native-born, with standard deviations on the order of 7 to 10 percentage points (Panel A). The within-school standard deviation—the variation across cohorts within the same school—shrinks to approximately 1.1 to 1.7 percentage points (Panel B). The minimum to maximum values are quite large, spanning from a roughly 24 percentage point decrease in immigrant ELL shares to an 11 percentage point increase. The within-school deviations in the peer share immigrant non-ELL are slightly smaller, with ranges on the order of a 6 percentage point decrease to a 9 percentage point increase.

The second assumption is that students are essentially randomly assigned to their within-school cohorts. To test this assumption, I estimate separate regressions of student characteristics as a function of peer immigrant share (ELL or non-ELL), cohort fixed effects, school fixed effects, and whether the student is immigrant ELL or immigrant non-ELL. The estimated coefficient, standard error, and t statistics on the peer share immigrant variable from each of these balancing tests are reported in Table 3. The absence of statistically significant estimates suggests that the observable characteristics of students are not highly correlated with the deviations in immigrant shares within schools across cohorts. Although there may be correlation on unobserved attributes of students, the results in Table 3 provide supporting evidence for the internal validity of the identification strategy.

### Regression Results

Table 4 presents results of the regressions of 10th grade reading test scores. Column 1 shows the results on the percentage peer immigrant ELL and non-ELL variables with no controls; the coefficients indicate that a 1 percentage point increase in the immigrant ELL share associates with a statistically significant 0.012 standard deviation decrease in students’ reading scores, and a corresponding increase in the share immigrant non-ELL peers associates with a similarly sized increase in students’ reading scores. Both coefficients are consistent with the academic mechanisms outlined earlier, with non-ELL immigrants having positive spillovers and ELL immigrants having the opposite. Columns 2–4 show the results as control variables are added to the regression. Column 2 reveals no change in the estimated coefficient with cohort fixed effects. Column 3 suggests that controlling for school fixed effects (that is, unobserved across-school differences that correlate with share immigrant within grade and individual test score performance) renders the coefficients on immigrant shares much closer to 0 and statistically insignificant. The dramatic change from column 2 to column 3 reveals that school conditions or differential selection into schools provides an important set of controls; within schools, increased exposure to immigrants seems to have no effect on student reading achievement. Column 4 further shows that adding student-level covariates to the within-cohort/across-school model renders no change to the sign or significance of the estimated effect of immigrant peers.

The resulting estimates from column 4 (the fully estimated Eq. (1)) suggest that adding 10 more immigrant ELL students to a school-grade cohort of 548 students (the average cohort size) associates with a decrease of 0.0004 (1.8 × –0.0002) standard deviations in reading achievement and that adding 10 more immigrant non-ELL students associates with a 0.005 standard deviation increase in reading achievement. Both of these estimated effects are extremely small in magnitude and statistically insignificant. In contrast, the fully adjusted estimates reveal large achievement differences between immigrant and native-born youth, even after adjusting for measured student demographics and educational needs as well as school fixed effects. Immigrant students who are ELL score 0.539 standard deviations lower on the reading FCAT than their native-born peers, and immigrant students who are not ELL score 0.080 standard deviations higher. These two results are consistent with the broader literature on immigrant students described earlier.

Table 5 provides the results of the seven sensitivity analyses. Column 1 presents the findings from the original specification and sample (from column 4 of Table 4) to ease comparisons across models. Column 2 adds back to the sample the students who were omitted due to small cluster size and missing demographics, with the demographic data multiply imputed; column 3 omits Cuban-born students from the analytic sample; column 4 includes a random intercept; column 5 includes a school-specific linear trend; column 6 includes other attributes of the grade-level peers; and columns 7 and 8 add variables that incorporate the effects of native-born ELL. The estimated effect of immigrant ELL peers varies from –0.0004 in column 7 to 0.004 in column 5. Despite the change in the sign of the estimated immigrant ELL peer effect with random effects and school trends, the estimates remains small and statistically insignificant in all models. The results are also robust to expanding the peer share to include ELL students who are native-born. The coefficient on the immigrant non-ELL peer variable is also similar across the alternative models and again is statistically insignificant. The estimated effects of individual students’ ELL and nativity group status are also quantitatively similar across the models, with immigrant non-ELL students earning average reading scores of approximately 0.08 standard deviations higher than native-born (and 0.05 higher than native-born non-ELL), and both immigrant and native-born ELL earning average scores of more than 0.5 standard deviations lower than native-born non-ELL.

Results from estimation of Eq. (1) for all three outcomes are shown in Table 6. The results for the math portion of the FCAT resemble those for reading with a negative coefficient on immigrant ELL peers and a positive coefficient on immigrant non-ELL peers, both of which are precisely estimated zeros. Similarly, the SAT model yields peer effects estimates that are near 0 and insignificant. Immigrant non-ELL students also outperform native-born students on all outcomes after their schools as well as their socioeconomic status and other background characteristics are adjusted for. Taken together, the results from Table 6 indicate that even though nativity gaps in outcomes resemble those found in the larger literature on immigrant student achievement, students who are enrolled in cohorts with more immigrant peers perform no differently than students who enroll in cohorts with fewer immigrant peers. This is true whether those classmates are designated by the school as ELL or more adept at English.

Given the possibility that exposure to more immigrants may affect immigrant and native-born students differently, the final analysis searches for heterogeneity in peer effects. Table 8 in the appendix provides the results from Eq. (2), where separate models are estimated for immigrant ELL (panel A), immigrant non-ELL (panel B), and native-born (panel C) students. The estimated coefficients range in magnitude and direction across the three subgroups. Yet, despite predictions, all estimates are insignificant and near 0. The largest coefficient across all models is 0.004, which indicates that a 1 percentage point increase in immigrant peers would increase student achievement by a modest 0.4 % of a standard deviation. Further, although the small immigrant non-ELL peer coefficient in the reading model from panel C is significant at the 5 % level, the large number of hypotheses being tested, combined with the large sample size, increases the likelihood of finding significant results.10

## Discussion

Immigrant youth with some command of English have often demonstrated an ability to outperform native-born youth from similar race and class backgrounds, and those immigrants who are designated by the school system as ELL tend to underperform. Whether and how foreign-born peers with different language abilities influence their classmates, however, remains an open question. In this article, I estimate the effect of foreign-born peers in school on the academic achievement of individual students using three measures of educational outcomes, data on the census of Florida public high schoolers, and a within–school/across-cohort identification approach.

Several theoretical predictions regarding why and how immigrant peers might affect their classmates are offered, with specific attention paid to the effects expected from classmates who are identified as ELL and those who are not. Absent controls for selection into schools, the results are consistent with theoretical predictions that immigrants who are ELL generate negative externalities, and those who are more proficient improve the outcomes of their peers. Yet, after the bias driven by school sorting is accounted for, youth in school cohorts with more immigrant peers who are not identified as ELL have similar academic outcomes to students in cohorts with fewer such peers. No spillover effects from immigrant peers who are ELL are detected, either. In an additional analysis, I expand the ELL peer share to include the 1 % of native-born students who are ELL, and the results suggest that even second-generation immigrants who are not fully proficient in English generate no negative externalities on their classmates. In addition to being statistically insignificant, the magnitude of the estimated immigrant peer effects is small (with the largest estimated coefficient suggesting that a one-point increase in immigrant shares increases student test scores by 0.4 % of a standard deviation). The results are robust to several sensitivity checks, including removing Cuban students from the analysis. This particular robustness check suggests that although the results may be best generalized to high school students in Florida, they are not driven by the high share of Cuban students in the Florida immigrant population, a group known for their relatively high rates of education and income.

The findings counter commonplace beliefs about the deleterious consequences of immigrant arrivals and are consistent with the handful of prior studies of U.S. students that found no association between student performance and the share of immigrants or children of immigrants in the school. A number of implications can be drawn from these results. First, with respect to anti-immigration sentiment—across country, state, or school district boundaries—the absence of a large negative effect of immigrant youth on native-born youth suggests that strong resistance to the presence of immigrant individuals may be unfounded (for instance, anti-immigrant ordinances or other such practices). Immigrant students who are not identified as ELL outperform native-born students with similar characteristics, and little evidence exists to suggest that the immigrants who are ELL detract from the achievement of native-born. In drawing this conclusion, it is important to recognize that although immigrant status itself may not pose an academic risk to other students, immigrant students are disproportionately poor and Hispanic, and these characteristics typically associate with lower academic performance. Thus, it is possible that sharing schools with students who are poor and Hispanic can have harmful effects on individual student achievement. The findings in this article, which mirror most of the research on the academic achievement of individual immigrant students, merely suggest that the potentially negative influence of students with these attributes is not driven by the fact that they are immigrants. Put differently, native-born youth with these attributes are likely to generate more negative externalities than immigrant youth with these attributes.

It is also important to recognize that many native-born youth of color in low-income families are likely children of immigrants; thus, the long-term effect of immigration on student achievement (across generations) may differ from the short-term effect examined in this study. As noted earlier, analyses of generational differences in educational outcomes have suggested that second-generation immigrant students perform differently from their first-generation and later-generation counterparts, with mixed evidence on the direction of these effects. It is therefore possible that second-generation immigrant youth might also impact their classmates, although at least three prior studies indicate no spillover effects of peers who are the children of immigrants (Cortes 2006; Crosnoe and Lopez-Gonzalez 2005; Riegle-Crumb and Callahan 2009).

The strengths of this study lie in the internal validity of the research design, the large number of outcomes examined, and the near-census sample of students. The sheer size of the administrative data, with large numbers of native-born and foreign-born students in the same schools at different points in time, permits reliable estimation of the within-school effects. The study limitations also lie in the research design. First, although immigrant peers are distinguished by their ELL status, the data do not allow for further estimation of the differential effects of being exposed to immigrants from different racial/ethnic or nationality groups. The variation in the share Asian or white immigrant across cohorts within a school, for example, is so small that it produces extremely imprecise estimates on the effect of increases in the share of students from these groups. Similarly, the design does not easily permit estimation of nonlinearities in the immigrant peer effect: that is, whether increases matter differently for schools with few or many immigrants. Second, these findings do not shed light on the potential effect of large increases in immigrant shares on student outcomes or on the school-wide effects of immigrant shares: for instance, whether more immigrants to a school alters the quality or attitudes of the teachers or the school course offerings. Finally, given the reliance on administrative data, this analysis cannot isolate the exact mechanisms through which immigrant peers matter; indeed, several of the mechanisms described herein may cancel one another to contribute to a null effect. The absence of a negative association between immigrant peers and student performance helps to rule out some theoretical links, however. Namely, despite concerns that the isolation of immigrant students from native-born students might harm immigrant students, this study offers no evidence to support this expectation. The nonnegative effects on native-born students also suggests that although there could be tensions between immigrant and native-born youth within schools (e.g., anti-immigrant hostility), these social interactions do not appear to harm native-born students’ academic performance.

Given the shortage of attention to the academic consequences of integration between immigrant and native-born students in the United States and across the world, this topic is ripe for further inquiry. Future research should search for differential effects of sharing schools with immigrants with different educational and demographic traits. Additional inquiry in this area could also explore the potential impacts of immigrant peers on more social and psychological outcomes of students. In the absence of random assignment of students to their schools, these future studies should also continue to search for creative quasi-experimental approaches to identifying the causal effects of immigrant peers.

## Acknowledgments

I thank the Florida Department of Education for maintaining such comprehensive records and for making them available to me for analysis. I am also grateful to the insightful comments provided by Brandon Bartels, Stephanie Riegg Cellini, David Deming, Bruce Fuller, Guanglei Hong, Micere Keels, Paco Martorell, Steven Raudenbush, Amy Ellen Schwartz, Leanna Stiefel, Jacob Vigdor, and seminar participants at the Association for Education Finance and Policy, George Washington University, Public Policy Institute of California, University of California at Berkeley, University of Chicago, and University of Maastricht. Special thanks to Rajeev Darolia and Megan Hatch for exceptional research assistance. Errors or omissions belong to me.

## Notes

1

I use the terms “immigrant,” “foreign-born,” and “first-generation immigrant” interchangeably to refer to children who were born abroad. The terms “children of immigrants” and “immigrant-origin” are used interchangeably to refer to both first-generation and second-generation immigrant youth (born in the United States with one or both parents born abroad).

2

The administrative data do not distinguish between Hispanic ethnicity and race. Instead, students are given the option to identify as Hispanic; black, not of Hispanic origin; white, not of Hispanic origin; Asian; or other.

3

The FLDOE requires all students in homes where a language other than English is primarily spoken to take English language assessments and the scores on these exams are used to determine entry into and exit from ELL status. Unlike most other states with large ELL populations, the FLDOE also relies on parent or teacher referrals and a formal committee review to classify ELL students, which suggests that at least relative to several other states, students in Florida are less likely to be misclassified as either ELL or non-ELL (Ragan and Lesaux 2006).

4

Data on earlier progressive cohorts was used for these aggregations, such that the counts include all students in the grade and year, not just those in the cohort.

5

Consistent with the characteristics of high school dropouts nationally, the students who are not enrolled in the 10th grade are more likely to be male, black, eligible for free or reduced-price lunch, and disabled than students who remain enrolled. The students who exit by the 10th grade are also slightly less likely to be immigrant non-ELL than students who remain and more likely to be native-born, with equal shares of immigrant ELL in both groups.

6

The three most frequently used national probability samples of high school students do not provide the large numbers of students and multiple cohorts afforded in these data. The National Educational Longitudinal Study (NELS) and its later counterpart, the Education Longitudinal Study (ELS), are both single-cohort designs with small numbers of students in each high school. Add Health provides multiple cohorts, but insufficient numbers of immigrant ELL and non-ELL within each school across cohorts to identify estimates with precision using this design.

7

This technique has also been used to study the effects of the racial composition of peers (see Hanushek et al. 2009; Hoxby 2000).

8

Results from alternative specifications using 9th grade peers and the average of the students’ 9th and 10th grade peers are qualitatively similar to those presented in this article and available upon request.

9

For the students who attended more than one school in the fall of the 10th grade, I chose the school that they attended the majority of the time. If the time was equal, I randomly chose one of the schools.

10

All sensitivity analyses were estimated for the models presented in Table 6 and in Table 8 in the appendix, and the results were robust to these alternative models.

## References

Angrist, J. D., & Lang, K. (
2004
).
Does school integration generate peer effects? Evidence from Boston’s Metco Program
.
American Economic Review
,
94
,
1613
1634
. 10.1257/0002828043052169
D. August, & K. Hakuta (
1997
).
Improving schooling for language-minority children: A research agenda
.
Washington, DC
:
.
Betts, J. B., & Fairlie, R. W. (
2001
).
Explaining ethnic, racial, and immigrant differences in private school attendance
.
Journal of Urban Economics
,
50
,
26
51
. 10.1006/juec.2000.2207
Bifulco, R., Fletcher, J. M., & Ross, S. L. (
2011
).
The effect of classmate characteristics on post-secondary outcomes: Evidence from the Add Health
.
American Economic Journal: Economic Policy
,
3
,
25
53
.
Burns, R. B., & Mason, D. A. (
1998
).
Class formation and composition in elementary schools
.
American Educational Research Journal
,
35
,
739
772
. 10.3102/00028312035004739
Callahan, R., Wilkinson, L., Muller, C., & Frisco, M. (
2009
).
ESL placement and schools: Effects on immigrant achievement
.
Educational Policy
,
23
,
355
384
. 10.1177/0895904807310034
Caplan, N., Whitmore, J. K., & Choy, M. H. (
1989
).
The boat people and achievement in America
.
Ann Arbor, MI
:
University of Michigan Press
.
Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J., Mood, A. M., Weinfeld, F. D., & York, R. L. (
1966
).
Equality of educational opportunity. U.S. Department of Health, Education, and Welfare
.
Washington, DC
:
U.S. Government Printing Office
.
Conger, D., & Grigorenko, E. L. (
2009
).
Special educational needs of children in immigrant families
. In E. L. Grigorenko, & R. Takanishi (Eds.),
Immigration, diversity, and education
(pp.
170
187
).
New York, NY
:
Routledge
.
Conger, D., Schwartz, A. E., & Stiefel, L. (
2007
).
Immigrant and native-born differences in school stability and special education: Evidence from New York City
.
International Migration Review
,
41
,
402
431
. 10.1111/j.1747-7379.2007.00073.x
Cortes, K. (
2006
).
The effects of age at arrival and enclave schools on the academic performance of immigrant children
.
Economics of Education Review
,
25
,
121
132
. 10.1016/j.econedurev.2004.12.001
Cosentino de Cohen, C., Deterding, N., & Clewell, B. C. (
2005
).
Who’s left behind? Immigrant children in high- and low-LEP schools
.
Washington, DC
:
Urban Institute
.
Crosnoe, R. (
2009
).
Low-income students and the socioeconomic composition of public high schools
.
American Sociological Review
,
74
,
709
730
. 10.1177/000312240907400502
Crosnoe, R., & Lopez-Gonzalez, L. (
2005
).
Immigration from Mexico, school composition, and adolescent functioning
.
Sociological Perspectives
,
48
,
1
24
. 10.1525/sop.2005.48.1.1
Finley, M. K. (
1984
).
Teachers and tracking in a comprehensive high school
.
Sociology of Education
,
54
,
233
243
. 10.2307/2112427
Fletcher, J. (
2010
).
Spillover effects of inclusion of classmates with emotional problems on test scores in early elementary school
.
Journal of Policy Analysis and Management
,
29
,
69
83
. 10.1002/pam.20479
Fuligni, A. (
1997
).
The academic achievement of adolescents from immigrant families: The roles of family background, attitudes, and behavior
.
Child Development
,
68
,
351
363
.
Gamoran, A., & Mare, R. D. (
1989
).
Secondary school tracking and educational inequality: Compensation, reinforcement, or neutrality?
.
American Journal of Sociology
,
61
,
61
77
.
Gibson, M. (
1988
).
Accommodation without assimilation: Sikh immigrants in an American high school
.
Ithaca, NY
:
Cornell University Press
.
Glick, J. E., & White, M. J. (
2003
).
The academic trajectories of immigrant youths: Analysis within and across cohorts
.
Demography
,
40
,
759
783
. 10.1353/dem.2003.0034
Gould, E. D., Lavy, V., & Paserman, D. D. (
2009
).
Does immigration affect the long-term educational outcomes of natives? Quasi-experimental evidence
.
The Economic Journal
,
119
,
1243
1269
. 10.1111/j.1468-0297.2009.02271.x
Hall, M. (
2013
).
Residential integration on the new frontier: Immigrant segregation in established and new destinations
.
Demography
,
50
,
1873
1896
. 10.1007/s13524-012-0177-x
Hallinan, M. (
1994
).
Tracking: From theory to practice
.
Sociology of Education
,
67
,
7
91
. 10.2307/2112697
Hanushek, E. A., Kain, J. F., Markman, J. M., & Rivkin, S. G. (
2003
).
Does peer ability affect student achievement?
.
Journal of Applied Econometrics
,
18
,
527
544
. 10.1002/jae.741
Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (
2009
).
New evidence about Brown v. Board of Education: The complex effects of school racial composition on achievement
.
Journal of Labor Economics
,
27
,
349
383
. 10.1086/600386
Hernandez, D. J., & Charney, E. (
1998
).
From generation to generation: The health and well-being of children in immigrant families
.
Washington, DC
:
.
Hoxby, C. M. (
2000
).
Peer effects in the classroom: Learning from gender and race variation
(NBER Working Paper No. 7867).
Cambridge, MA
:
National Bureau of Economic Research
.
Kao, G. (
1999
).
Psychological well-being and educational achievement among immigrant youth
. In D. J. Hernandez (Ed.),
Children of immigrants: Health, adjustment, and public assistance
(pp.
410
475
).
Washington, DC
:
.
Kao, G., & Tienda, M. (
1995
).
Optimism and achievement: The educational performance of immigrant youth
.
Social Science Quarterly
,
76
,
1
19
.
Kasinitz, P. (
2001
).
Fade to black? The children of West Indian immigrants in southern Florida
. In R. G. Rumbaut, & A. Portes (Eds.),
Ethnicities: Children of immigrants in America
(pp.
267
300
).
New York, NY
:
Russell Sage Foundation
.
Matute-Bianchi, M. E. (
1986
).
Ethnic identities and patterns of school success and failure among Mexican-descent and Japanese-American students in a California high school: An ethnographic analysis
.
American Journal of Education
,
95
,
233
255
. 10.1086/444298
National Conference of State Legislatures (NCSL)
(
2011
).
Immigration related laws and resolutions in the States (January 1–December 31, 2010)
Ogbu, J. (
1992
).
Understanding cultural diversity and learning
.
Educational Researcher
,
21
,
5
14
. 10.3102/0013189X021008005
Olsen, L. (
1997
).
Made in America: Immigrant students in our public schools
.
New York, NY
:
The New Press
.
Pallas, A. M., Entwisle, D. R., Alexander, K. L., & Stluka, F. M. (
1994
).
Ability-group effects: Instructional, social, or institutional?
.
Sociology of Education
,
67
,
27
46
. 10.2307/2112748
Park, H., Behrman, J. R., & Choi, J. (
2013
).
Causal effects of single-sex schools on college entrance exams and college attendance: Random assignment in Seoul high schools
.
Demography
,
50
,
447
469
. 10.1007/s13524-012-0157-1
Perreira, K. M., Harris, K. M., & Lee, D. (
2006
).
Making it in America: High school completion by immigrant and native youth
.
Demography
,
43
,
511
536
. 10.1353/dem.2006.0026
Pew Hispanic Center
(
2006
).
Cubans in the United States
(Fact Sheet).
Washington, DC
:
Pew Hispanic Center
Pong, S-L, Hao, L., & Gardner, E. (
2005
).
The roles of parenting styles and social capital in the school performance of immigrant Asian and Hispanic adolescents
.
Social Science Quarterly
,
86
,
928
950
. 10.1111/j.0038-4941.2005.00364.x
Pong, S-L, & Zeiser, K. L. (
2012
).
Student engagement, school climate, and academic achievement of immigrants’ children
. In C. García Coll, & A. K. Marks (Eds.),
The immigrant paradox in children and adolescents: Is becoming American a developmental risk?
(pp.
209
232
).
Washington, DC
:
American Psychological Association
.
Portes, A., & Rumbaut, R. G. (
2001
).
Legacies: The story of the immigrant second generation
.
Los Angeles, CA
:
University of California Press and Russell Sage Foundation
.
Ragan, A., & Lesaux, N. (
2006
).
Federal, state, and district level English language learner program entry and exit requirements: Effects on the education of language minority learners
.
Education Policy Analysis Archives
,
14
,
1
29
. 10.14507/epaa.v14n20.2006
Raleigh, E., & Kao, G. (
2010
).
Do immigrant minority parents have more consistent college aspirations for their children?
.
Social Science Quarterly
,
91
,
1083
1102
. 10.1111/j.1540-6237.2010.00750.x
Ream, R. K. (
2005
).
Toward understanding how social capital mediates the impact of mobility on Mexican American achievement
.
Social Forces
,
84
,
201
224
. 10.1353/sof.2005.0121
Riegle-Crumb, C., & Callahan, R. M. (
2009
).
Exploring the academic benefits of friendship ties for Latino boys and girls
.
Social Science Quarterly
,
90
,
611
631
. 10.1111/j.1540-6237.2009.00634.x
Rivkin, S. G. (
2000
).
School desegregation, academic achievement, and earnings
.
Journal of Human Resources
,
35
,
333
346
. 10.2307/146328
Royston, P. (
2004
).
Multiple imputation of missing values
.
The Stata Journal
,
4
,
227
241
.
Rubin, D. B. (
1987
).
Multiple imputation for nonresponse in surveys
.
New York, NY
:
John Wiley and Sons
.
Rumberger, R. W., & Palardy, G. J. (
2005
).
Does segregation still matter? The impact of social composition on academic achievement in high school
.
Teachers College Record
,
107
,
1999
2045
.
Ryabov, I., & Van Hook, J. (
2007
).
School segregation and academic achievement among Hispanic children
.
Social Science Research
,
36
,
767
788
. 10.1016/j.ssresearch.2006.04.002
Schwartz, A. E., & Stiefel, L. (
2006
).
Is there a nativity gap? Achievement of New York City elementary and middle school immigrant students
.
Education Finance and Policy
,
1
,
17
49
. 10.1162/edfp.2006.1.1.17
Schwartz, A. E., & Stiefel, L. (
2011
).
Immigrants and inequality in public schools. Prepared for the project on Social Inequality and Educational Disadvantage
. In G. J. Duncan, & R. J. Murnane (Eds.),
Wither opportunity? Rising inequality, schools, and children’s life chances
(pp.
419
442
).
New York, NY
:
Russell Sage
.
Singer, A. (
2004
).
The rise of new immigrant gateways
(The Living Cities Census Series).
Washington, DC
:
The Brookings Institution
.
Stanton-Salazar, R. D., & Dornbusch, S. M. (
1995
).
Social capital and the reproduction of inequality: Information networks among Mexican-origin high school students
.
Sociology of Education
,
68
,
116
135
. 10.2307/2112778
Suárez-Orozco, C., & Suárez-Orozco, M. M. (
2001
).
Children of immigration
.
Cambridge, MA
:
Harvard University Press
.
Suárez-Orozco, M. M. (
1989
).
Central American refugees and U.S. high schools: A psychosocial study of motivation and achievement
.
Stanford, CA
:
Stanford University Press
.
Thomas, K. (
2009
).
Parental characteristics and the schooling progress of the children of immigrant and U.S.-born blacks
.
Demography
,
45
,
513
534
. 10.1353/dem.0.0068
Tillman, K. H., Guo, G., & Harris, K. M. (
2006
).
.
Social Science Research
,
35
,
129
156
. 10.1016/j.ssresearch.2004.07.001
Van Hook, J., Brown, S. L., & Kwenda, M. N. (
2004
).
A decomposition of trends in poverty among children of immigrants
.
Demography
,
41
,
649
670
. 10.1353/dem.2004.0038
Van Hook, J., & Fix, M. E. (
2000
).
A profile of the immigrant student population
. In J. Ruiz-De-Velasco, M. E. Fix, & B. Clewell (Eds.),
Overlooked and underserved: Immigrant children in U.S. secondary schools
(pp.
9
33
).
Washington, DC
:
The Urban Institute
.
Vigdor, J., & Nechyba, T. (
2007
).
Peer effects in North Carolina public schools
. In L. Woessmann, & P. E. Peterson (Eds.),
Schools and the equal opportunity problem
(pp.
73
102
).
Cambridge, MA
:
MIT Press
.
Waters, M. C. (
1999
).
Black identities: West Indian immigrant dreams and American realities
.
New York, NY
:
Russell Sage Foundation and Harvard University Press
.
Wells, A. S., & Crain, R. L. (
1994
).
Perpetuation theory and the long-term effects of school desegregation
.
Review of Educational Research
,
64
,
531
555
. 10.3102/00346543064004531
Zhou, M., & Bankston, C. L.III (
1998
).
Growing up American: How Vietnamese children adapt to life in the United States
.
New York, NY
:
Russell Sage Foundation
.