Vous êtes sur la page 1sur 22

Ethnic and Racial Studies

ISSN: 0141-9870 (Print) 1466-4356 (Online) Journal homepage: http://www.tandfonline.com/loi/rers20

How family, immigrant group, and school contexts


shape ethnic educational disparities

Cynthia Feliciano

To cite this article: Cynthia Feliciano (2017): How family, immigrant group, and
school contexts shape ethnic educational disparities, Ethnic and Racial Studies, DOI:
10.1080/01419870.2017.1355974

To link to this article: http://dx.doi.org/10.1080/01419870.2017.1355974

View supplementary material

Published online: 09 Aug 2017.

Submit your article to this journal

Article views: 53

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rers20

Download by: [Therapeutic Goods Administration] Date: 07 September 2017, At: 06:56
ETHNIC AND RACIAL STUDIES, 2017
https://doi.org/10.1080/01419870.2017.1355974

How family, immigrant group, and school contexts


shape ethnic educational disparities
Cynthia Feliciano
Department of Sociology, University of California, Irvine, USA
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

ABSTRACT
Scholars have long questioned why average educational attainments among
children of immigrants vary greatly by country of origin. Immigrants’ children
from the same country share similar contexts of exit and reception and often
similar school and family contexts. What is the relative importance of these
factors in explaining ethnic differences in educational attainment? Using
cross-classified multi-level models, this study shows that family contexts and
immigrant group educational selectivity, but not school contexts, help explain
ethnic differences. Immigrant selectivity is more decisive in shaping the
second-generation’s educational attainment than other group characteristics
related to immigrants’ contexts of exit and reception. While school
socioeconomic status (SES) only influences the attainment of immigrants’
children from high-SES families, immigrant group selectivity matters
regardless of the SES of the family or school, thus shedding light on why
members of some national-origin groups tend to complete more education
than others despite similar family and school contexts.

ARTICLE HISTORY Received 17 June 2016; Accepted 27 June 2017

KEYWORDS Second generation; children of immigrants; education; ethnicity; family; immigrant


selectivity

Research on the adaptation of children of contemporary immigrants has con-


sistently shown that educational attainment varies significantly by origin
country (Kasinitz et al. 2008; Portes and Rumbaut 2001). Immigrants’ children
from certain (mostly Asian) countries tend to achieve more educational
success than those from other (mostly Latin American) countries (Feliciano
2005a; Portes and Rumbaut 2001). Even while tremendous diversity exists
within immigrant groups, immigrant group membership appears to power-
fully shape the lives of the second generation1 in the United States (Kasinitz
et al. 2008).
Why country of origin is associated with individuals’ educational out-
comes is debated. Which factors best explain variation between ethnic

CONTACT Cynthia Feliciano felician@uci.edu


Supplemental data for this article can be accessed https://doi.org//10.1080/01419870.2017.1355974
© 2017 Informa UK Limited, trading as Taylor & Francis Group
2 C. FELICIANO

groups in educational attainment? In this article, I consider the role of three


contexts that may shape the lives of immigrants and their children: (1)
families, (2) national-origin groups, and (3) schools. First, family contexts
likely shape ethnic-group differences because children of immigrants from
certain countries are more likely to grow up in poor families than those
from others. Second, immigrants from different countries face different con-
texts of reception – the characteristics and responses of the society receiv-
ing newcomers – as well as different contexts of exit – the circumstances
within origin countries shaping migration patterns (Portes and Rumbaut
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

2001, 2006). Third, because children of immigrants from the same countries
often attend the same schools, disparities may be driven by school quality
differences.
Drawing on survey data of immigrants’ children in San Diego and Miami,
this study makes several contributions. First, while studies have consistently
shown that family socioeconomic background and structure influence the
education of children of immigrants (Portes and Hao 2004), scholars have
not examined the relative importance of family contexts as compared with
immigrant group and school contexts in influencing ethnic educational dis-
parities. Second, while scholars have argued that the characteristics of pre-
existing co-ethnic communities shape immigrants’ children’s lives through
social networks, community institutions, and group resources (Zhou and
Bankston 1998; Zhou and Kim 2006), existing studies have not quantitatively
assessed which particular characteristics associated with different national-
origin groups’ contexts of reception and contexts of exit most strongly influ-
ence the educational attainment of the second generation. In particular, I con-
sider the importance of immigrant group selectivity – a factor related to
contexts of exit not measured directly in most studies – in relation to other
group characteristics associated with immigrants’ contexts of exit (such as
home country income inequality) and contexts of reception (such as group
size). Finally, I extend previous research by assessing whether and how
family, immigrant group, and school contexts interact to influence edu-
cational attainment. Which contexts most strongly influence attainment and
under what conditions?

Explanations for national-origin differences


The three key contexts potentially shaping the education of the second gen-
eration reflect different levels of analysis. As shown in Figure 1, individuals are
nested within families, the most immediate context shaping children of immi-
grants’ lives. At a broader level, individuals and families are members of
national-origin groups, and individual children of immigrants attend different
schools. Factors associated with each context likely influence eventual edu-
cational attainment and ethnic disparities.
ETHNIC AND RACIAL STUDIES 3

National-Origin Group School

Family

Individual

Figure 1. Three contexts shaping second-generation educational attainment.


Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Family contexts
At a micro-level, the childhood family is a key context shaping educational
attainment. Parents’ socioeconomic status (SES) is strongly and positively
associated with the educational attainment of immigrants’ children
(Rumbaut 2005b). Controlling for family SES background substantially
reduces national-origin educational differences among the second gener-
ation, suggesting that these compositional differences partly explain
ethnic educational disparities (Kasinitz et al. 2008, 144–145; Rumbaut
2005b, 1075). Studies also show that children from two-parent households
tend to perform better in school (Portes and Hao 2004). However, the sig-
nificant “effect” of national-origin group membership remains even after
controlling for family characteristics (Haller, Portes, and Lynch 2011;
Levels and Dronkers 2008). For example, net of family SES and structure,
second-generation Chinese and Vietnamese complete more years of
schooling than their Mexican counterparts (Haller, Portes, and Lynch
2011, 747; Rumbaut 2005b, 1075).
Furthermore, the relationship between parents’ SES and their children’s
educational attainment is weaker in immigrant vs. native-born families
(Luthra and Soehl 2015), perhaps because immigrants’ children are more influ-
enced by non-familial co-ethnics (Feliciano 2005a). While family context is cer-
tainly important, existing research has not adequately addressed the question
of how much variation in second-generation educational attainment is
explained by family characteristics relative to immigrant group or school con-
textual characteristics.

Immigrant group contexts


The role of immigrant group contextual factors in shaping the second-gener-
ation’s education has rarely been empirically tested. However, theories
suggest that contexts immigrant groups face after migrating, as well as pre-
migration factors, are important.
4 C. FELICIANO

Contexts of reception
Segmented assimilation theory contends that contextual factors in the receiv-
ing society such as societal prejudice, co-ethnic community resources, and
government reception, shape ethnic differences among immigrants’ children
(Portes and Rumbaut 2001; Portes and Zhou 1993). For example, Portes and
Rumbaut (2006) argue that Vietnamese immigrants’ children have greater
educational opportunities than many others because Vietnamese immigrants
can more easily attain citizenship. However, quantitative research from a seg-
mented assimilation approach has used country of origin itself as a proxy
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

rather than disentangled the particular aspects of immigrants’ contexts of


reception that produce national-origin differences (Haller, Portes, and Lynch
2011; Portes and Rumbaut 2001).
Other quantitative research has found the average SES of the immigrant
generation positively influences SES outcomes among immigrants’ children
(Borjas 1992; Levels, Dronkers, and Kraaykamp 2008). Borjas (1992) suggests
that post-migration SES of the immigrant group is important because it
exposes the second generation to “social, cultural, and economic factors
that increase their productivity … ” (126).
Levels, Dronkers, and Kraaykamp (2008) consider immigrant group size,
finding that it positively influenced the math achievement of immigrants’ chil-
dren in Europe, consistent with the theory that larger immigrant groups
provide supportive social networks (Zhou and Bankston 1998). While the
“ethnic threat” hypothesis predicts larger immigrant groups experience
more discrimination (Quillian 1995), studies suggest that only perceptions
of group size relate to anti-immigrant attitudes; larger groups engage in
more inter-group contact, which reduces prejudice, ultimately producing a
more favourable context of reception (Schneider 2007). While these studies
suggest that group size is positively associated with the education of immi-
grants’ children, this question has not yet been examined in the United States.

Contexts of exit
While theories focus much more on contexts of reception, scholars also
acknowledge that contexts of reception are influenced by contexts of exit,
such as the political climate and economic conditions in home countries
that ultimately shape who migrates (Feliciano 2005b; Portes and Rumbaut
2006). A few studies have examined these factors’ influence on children of
immigrants’ education.
Levels, Dronkers, and Kraaykamp (2008) find that immigrants’ children
from more politically stable countries have higher math achievement,
perhaps because migrants from highly volatile political climates experience
trauma, negatively impacting their adaptation. Levels, Dronkers, and Kraay-
kamp (2008) also find that immigrants’ children from less developed countries
ETHNIC AND RACIAL STUDIES 5

perform better. This paradoxical finding that may be attributed to the high
selectivity of immigrants from less developed countries, a factor Levels, Dron-
kers, and Kraaykamp (2008) did not measure directly, discussed further below.
van Tubergen, Maas, and Flap (2004) find more favourable labour market
outcomes among adult immigrants from more geographically distant
countries with less income inequality, which bodes well for their children’s
attainment. To explain, the authors draw from Borjas’s (1987) theory that
immigrants from more unequal societies and geographically closer countries
are less positively selected, although they do not include a direct measure of
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

immigrant selectivity.
Immigrant group selectivity is speculated to be an important influence on
children’s education because highly educationally select groups may have
ethnic capital in the form of collective cultural frameworks and community
resources supporting educational achievement (Lee and Zhou 2015). Such
ethnic capital stems from class-specific resources related to the relatively advan-
taged pre-migration class statuses of some immigrant groups (Feliciano 2005a;
Lee and Zhou 2015). The few studies quantitatively linking immigrant group
selectivity to educational outcomes show that the children of more highly edu-
cationally select immigrant groups have higher educational expectations and
college enrolment (Feliciano 2005a, 2006). However, this work does not
examine overall educational attainment or employ multi-level models to
assess how much variation between national-origin groups immigrant selectiv-
ity explains, how other aspects of groups’ contexts of exit and reception matter,
and how immigrant selectivity may interact with school and family contexts.

School contexts
School contexts have generally not been considered an important part of the
“contexts of reception” framework (Portes and Rumbaut 2001). This is surpris-
ing since schools have historically served as incorporation agents (Gonzalez
2013). Further, school racial segregation is more pronounced than ever,
suggesting that immigrants’ children from the same country are often clus-
tered within the same schools (Fiel 2013).
Portes and Zhou (1993, 81) suggest that exposure to native-born minorities
has negative consequences for immigrants’ children. High-minority schools
are often disorganized and have lax academic standards (Valencia 2000),
and research on immigrants’ children shows that attending a predominantly
minority school negatively influences educational attainment (Haller, Portes,
and Lynch 2011).
Children of immigrants attending schools with higher proportions of socio-
economically advantaged peers complete more schooling (Haller, Portes, and
Lynch 2011; Portes and Hao 2004). However, Raudenbush and Bryk (1986) show
that family SES is more consequential for school achievement in high SES as
6 C. FELICIANO

opposed to low-SES schools. Research on immigrants’ children similarly shows


family SES predicting grades in high SES, but not low SES, schools (Portes and
Hao 2004). However, we do not know whether the effect of schools’ socioeco-
nomic composition differs by family SES for eventual educational attainment.
School size and location may also matter. Attendance in larger high schools
may exacerbate achievement inequalities (Lee and Bryk 1989). Mexican immi-
grant children are overrepresented in larger high schools and tend to do
worse in these schools than other ethnic groups (Crosnoe 2005). In general,
immigrants’ children tend to have worse educational outcomes in inner city
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

schools (Rumbaut 2005a).


Existing research has not determined which of these various school contex-
tual characteristics – racial composition, SES, school size, and location – most
strongly predicts educational attainment for immigrants’ children, nor
assessed the relative importance of school contexts as compared with immi-
grants’ contexts of exit and reception and family contexts, in influencing
ethnic disparities. Moreover, we do not know whether school contextual
effects vary by immigrant group and family contexts.

Research questions
I address several questions. First, which particular characteristics of immigrant
groups and schools are most strongly associated with the educational attain-
ment of children of immigrants? Second, what is the relative importance of
family contexts, immigrant group contexts, and school contexts in explaining
variation by country of origin in educational attainment? Third, do the effects
of immigrant group and school contexts on educational attainment vary by
family contexts, or vice-versa?

Data and methods


Children of Immigrants Longitudinal Study data
Primary data come from the Children of Immigrants Longitudinal Study (CILS),
a three-wave longitudinal study of children with at least one immigrant parent
living in San Diego, California and Miami/Fort Lauderdale, Florida in 1992. The
study’s school-based sampling frame is representative of eighth-grade immi-
grants’ children in these two metropolitan areas in 1992 (Portes and Rumbaut
2001). I use data from the first survey and the third follow-up, conducted in
2001–03, when the respondents were in their mid-twenties. From the CILS
data, I selected the 29 national-origin groups for which the pre- and post-
migration data described below were available to arrive at an analytic
sample of 2,777.2 The largest groups were Cubans, Filipinos, Mexicans, Vietna-
mese, Nicaraguans, Colombians, Jamaicans, and Dominicans.3
ETHNIC AND RACIAL STUDIES 7

The CILS data contain several limitations. Although studies report results
with fewer level-2 groups (Levels, Dronkers, and Kraaykamp 2008; van Tuber-
gen, Maas, and Flap 2004), multi-level results might be more robust if data on
more than 29 immigrant groups were available. Also, respondents were only
in their mid-twenties during the survey’s final wave, so their education may be
in progress. However, most existing work on ethnic differences among immi-
grants’ children relies on the educational outcomes of much younger youth
(Portes and Rumbaut 2001). Finally, because the data are limited to two his-
torically large immigrant destinations, the findings cannot be generalized to
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

the United States as a whole. Despite these limitations, the CILS is the only
available data set to include family background and school context infor-
mation for a large number of U.S. immigrant groups.

Variables
Table 1 shows the means and standard deviations of variables in the analyses.
The dependent variable, years of schooling completed, is measured when
respondents were in their mid-twenties, based on the question: “what is the

Table 1. Means and standard deviations of variables (N = 2777)


Mean SD Minimum Maximum
Dependent variable (mid-twenties)
Ave. years of schooling 14.15 1.94 10.00 19.00
Individual-level variables (eighth grade)
Family context
Family socioeconomic status 4.55 1.88 0.00 10.00
Two-parent household 0.69 0.46 0.00 1.00
Demographics
Age 14.14 0.84 12.00 18.00
Female 0.54 0.50 0.00 1.00
Born in the U.S. 0.55 0.50 0.00 1.00
Macro-level variables
Immigrant group contexts of reception (Miami and San Diego)
Post-migration SES 0.29 0.17 0.00 1.00
Per cent of citizens 39.89 17.43 3.29 100.00
Group size 173.02 167.97 1.72 428.41
Immigrant group contexts of exit
Educational selectivity 0.49 0.16 −0.06 0.88
Distance from U.S.A. (in thousands of miles) 2.58 3.17 0.01 7.98
Income inequality 42.48 10.98 25.81 60.10
GDP per capita in 1000s of dollars 2.46 2.58 0.23 18.40
Political suppression 4.33 1.64 1.00 6.50
School context
Per cent of white students 22.84 19.18 0.10 65.00
Per cent of black students 13.86 17.05 0.00 92.00
Per cent of Hispanic students 48.78 34.08 4.00 99.00
Per cent of Asian students 14.43 17.54 0.00 45.00
Minority school 0.47 0.50 0.00 1.00
Per cent of students eligible for free/reduced lunch 43.45 23.92 0.00 92.30
School population 1,840.15 779.64 707.00 3,568.00
Inner city school 0.31 0.46 0.00 1.00
8 C. FELICIANO

highest grade or year of school you have completed?” I recoded responses to


range from 10 (some high school, grades 9–12) to 19 (master’s or professional
degree). Respondents averaged 14.2 years of schooling.

Family variables
Family socioeconomic context. The CILS data contain a composite parental SES
indicator based on the standardized unit-weighted sum of father’s and
mother’s education, occupational status, and home ownership in 1992. The
measure ranges from 0 to 10, with a sample mean of 4.55.
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Two-parent household was coded one if respondents lived with two parents
in eighth grade. Sixty-nine per cent lived with both parents.
Demographic variables. Age, gender, and nativity are included as controls.
Respondents averaged 14.1 years old in 1992. Fifty-four per cent are female
and 55 per cent U.S. born.

Immigrant group contexts of reception


I used U.S. census data (IPUMS) from 1990 (Ruggles and Sobek 1997), the year
closest to the CILS base year, of immigrants born in the 29 countries living in
the Miami and San Diego metropolitan areas to create context of reception
measures.
Post-migration socioeconomic status is based on the average years of
schooling, Duncan average occupational status score (see Duncan 1961),
and the average income for adult immigrants (18+) from each country. I
standardized and summed these highly correlated measures into a SES
status scale ranging from 0 to 1, averaging 0.29 among the 29 groups
(Table 1).
Percentage of U.S. citizens among immigrants from each country proxies for
political context of reception. This averages 40 per cent across the 29
countries, from a low of only 3 per cent of Nicaraguan immigrants, to a
high of 100 per cent among Puerto Rican migrants (citizens by birth).
Group size. I divided the Census population estimates by 1000 to obtain
group size in the thousands in Miami or San Diego. The smallest group was
from Hong Kong (1,717) and the largest from Cuba (428,410).

Immigrant group contexts of exit


Immigrant group educational selectivity. I used published immigrant groups’
educational selectivity measures, calculated using age-adjusted net difference
indexes based on the percentages of immigrants with the same, more, or less
schooling as nonmigrants in the home country (Feliciano 2005b). The higher
the educational selectivity, the more educated immigrants are relative to the
nonmigrant population in their home country. Educational selectivity
averages 0.49 (Table 1), indicating that immigrants’ educational attainment
ETHNIC AND RACIAL STUDIES 9

exceeds that of nonmigrants from that country approximately 49 per cent


more often than nonmigrants exceed immigrants.
Home country distance is approximated using the Google maps distance
calculator, from the home country to the closest U.S. city considered a
typical port of entry: New York, San Francisco, Los Angeles, or Miami.
Measured in thousands of miles, the mean is 2.58.
Pre-migration income inequality is measured by the World Bank’s Gini coef-
ficient (Milanovic 2005). For each origin country, the Gini score corresponds to
the average year of migration to the United States. The average score among
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

the 29 origin countries is 42.48.


Home country GDP is measured per capita in 2005 U.S. dollars. These data
are drawn from the International Macroeconomic Data Set (United States
Department of Agriculture 2012), for the year closest to the immigrant
group’s average migration year.
Political suppression. I used ratings of political rights and civil liberties drawn
from freedomhouse.org, combined to indicate the level of freedom vs. politi-
cal suppression in each origin country (Freedomhouse.org 2010). Sample
ratings ranged from 1 (Free) to 6.5 (Not Free), averaging 4.3.

School contextual variables


School contextual variables, included in the CILS data set, were drawn directly
from school records in the study’s first wave (eighth grade).
School racial composition. I considered various measures of school racial
composition: percentage white, black, Hispanic, and Asian. I also included a
minority school dummy coded 1 if respondents’ school enrolled 60 per cent
or more black and/or Hispanic students. As Table 1 shows, 47 per cent
attended minority schools.
School socioeconomic composition was based on the per cent of students
eligible for free or reduced lunch, and ranged from 0 to 92 per cent, averaging
43 per cent.
Total school population was the number of enrolled students, and ranged
from 707 to 3,568 students.
Inner city school indicates whether the school was in the inner city. Thirty-
one per cent of respondents attended inner city schools.

Analytic strategy
As Figure 1 shows, children of immigrants are embedded within micro-level
family contexts and macro-level immigrant group and school contexts.
Family contextual measures are considered individual-level determinants,
along with demographic controls (age, sex, nativity). Immigrant group and
school variables are non-hierarchical level-2 macro-level determinants: indi-
viduals are nested within immigrant groups and schools, but schools are
10 C. FELICIANO

not nested within groups, or vice-versa. Cross-classified multi-level models


account for this clustering, providing more accurate estimates of macro-
level effects on individual outcomes than ordinary-least-squares regression
(Snijders and Bosker 1999). Independent variables, all measured when the
respondents were adolescents, logically predict the dependent variable,
measured in young adulthood.
In these data, level-2 sample sizes are relatively small (29 immigrant groups,
42 schools), and many of the macro-level factors are highly correlated (see
Supplemental Table 1). Therefore, to address the first research question, I
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

introduced one macro-level variable at a time into models controlling for


only individual-level factors to assess which immigrant group and school
characteristics relate most closely to second-generation educational attain-
ment. To evaluate how much the most influential immigrant group, school,
and family factors explain ethnic-group variation in educational attainment,
I present and compare various components of a series of multi-level models
that introduce individual-level determinants, immigrant group factors, and
school contextual factors separately. Then, I test for significant interaction
effects between key immigrant group, school, and family variables, to
address whether the effects vary under different conditions. Finally, based
on the preferred model, I calculate predicted years of schooling for individuals
within various immigrant group, family, and school contexts to illustrate the
relative importance of these different contexts for educational attainment.

Results
Family, school, and immigrant group contexts by national origin
Table 2 shows national-origin differences in immigrant group contexts of exit
and reception and school and family contexts, illustrating patterns of disad-
vantage or advantage across several dimensions. For example, Mexicans,
the largest U.S. immigrant group, have the lowest average parental SES
(2.9), low post-migration group SES (0.05), among the lowest citizenship
rates (24 per cent), and relatively low educational selectivity (0.20). They
also often attend inner city, low-SES schools (56 per cent). In contrast, Filipino
immigrant parents have high average SES (5.3), high average group post-
migration SES (0.47), high citizenship rates (59 per cent), relatively high edu-
cational selectivity (0.60); Filipino adolescents often attend suburban, low-
poverty schools. While Mexican immigrants’ children average the lowest edu-
cational attainment (13.1 years), Filipinos complete the third highest (14.3),
following Vietnamese (14.8) and Jamaicans (14.7).
The high educational attainment of Vietnamese, in particular, might appear
surprising given their parents’ SES is lower than average (3.7) and they attend
schools with above average poverty rates (49 per cent free/reduced lunch
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Table 2. Immigrant group, school, and family contexts by select countries of origin.
Dominican
Cuba Philippines Mexico Nicaragua Vietnam Colombia Haiti Jamaica Republic Honduras
Ave. years of schooling 14.29 14.34 13.13 13.99 14.80 14.13 14.23 14.74 13.67 13.86
Family context
Parents’ SES 4.93 5.33 2.87 4.59 3.72 4.59 4.11 5.26 4.39 3.89
Two- parent household 0.62 0.83 0.69 0.70 0.80 0.68 0.55 0.52 0.57 0.53
Immigrant group contexts of reception (Miami and San Diego)
Post-migration SES 0.25 0.47 0.05 0.24 0.37 0.35 0.03 0.50 0.18 0.09
Per cent of citizens 48.62 58.75 23.86 3.29 48.08 29.35 21.01 42.39 30.08 16.88
Group size 428.41 62.18 198.20 67.26 16.48 49.77 37.70 36.28 17.50 13.49
Immigrant group contexts of exit
Educational selectivity 0.41 0.60 0.20 0.67 0.59 0.62 0.71 0.67 0.49 0.43
Distance from USA 0.23 7.30 0.01 1.02 7.66 1.51 0.71 0.52 0.84 0.93
Income inequality 28.00 45.18 50.00 60.10 35.70 54.71 58.50 44.52 45.00 57.06
GDP per capita in 1000s of dollars 1.96 1.05 6.49 1.21 0.23 2.10 0.88 2.56 1.74 1.14
Political suppression 6.50 3.50 2.50 4.00 6.00 3.50 4.50 2.50 2.00 4.00
School context

ETHNIC AND RACIAL STUDIES


Per cent of white students 12.45 33.24 28.20 11.37 37.47 18.64 13.60 32.48 12.90 8.86
Per cent of black students 5.57 12.75 15.83 9.80 12.91 8.94 59.22 39.65 25.04 21.20
Per cent of Hispanic students 80.64 15.56 35.35 76.96 19.63 70.20 25.98 24.12 59.33 68.25
Per cent of Asian students 1.23 38.44 20.47 1.80 29.85 2.04 0.98 3.77 2.52 1.54
Minority school 0.90 0.01 0.05 0.91 0.06 0.78 0.65 0.42 0.79 0.86
Per cent of students eligible for free/reduced 38.29 35.91 56.42 45.23 49.10 43.42 60.22 40.85 49.13 51.76
lunch
School population 1,936.49 1,894.62 1,482.96 2,429.70 1,458.71 2,038.38 1,463.62 1,573.76 2,048.26 2,128.50
Inner city school 0.31 0.05 0.56 0.25 0.45 0.27 0.45 0.36 0.54 0.50
N 774 567 396 222 192 136 94 89 61 36

11
12 C. FELICIANO

eligible). However, Vietnamese immigrants’ children are advantaged in other


ways: fairly high citizenship rates (48 per cent) and group educational selec-
tivity (.60). What is the relative importance of these family, immigrant
group, and school contexts in shaping educational attainment? Before addres-
sing this question, I first determine which immigrant group and school factors
are the most important predictors of educational attainment.

Comparing immigrant group and school contextual factors


Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Since analysing the simultaneous effects of highly correlated measures within


schools and immigrant groups is not possible with these data (see Sup-
plemental Table 1), I first must arrive at a preferred model including only
those factors most strongly associated with education. Table 3 compares
various models of macro-level factors predicting years of schooling to
Model 1, with only individual-level family and demographic factors (full
models in Supplemental Table 2). To determine which macro-level factors
fit the data best, I present two fit statistics, AIC and BIC, for which lower
scores indicate a better fit.
Models 2–4 consider three proxies for immigrant groups’ contexts of recep-
tion in San Diego or Miami: average post-migration SES, citizenship rate, and
group size. These models overall fit the data worse that Model 1, with individ-
ual-level determinants alone.
Models 5–9 include measures related to immigrant group contexts of exit
including immigrant group selectivity, home country’s distance from the U.S.,
per capita GDP, political suppression, and income inequality. Only Model 5
with immigrant group educational selectivity fits the data better than Model 1.
I also considered combinations of contexts of exit and reception measures
(not all shown). Model 10 shows that including the best-fitting context of
reception measure, immigrant group post-migration SES, and the best-
fitting context of exit measure, immigrant group educational selectivity, pro-
vides only a slightly better fit according to AIC, but not BIC. Moreover, immi-
grant group post-migration SES does not significantly predict educational
attainment once educational selectivity is introduced (see Supplemental
Table 2). According to both BIC and AIC, Model 5 (with immigrant group selec-
tivity only) fits the data better.
Models 11–13 consider school-level factors. Model 11, with the per cent
low-SES students, provides a slightly better fit than Model 1 per AIC, but
not BIC. Models 12 and 13, adding minority and inner city school respectively,
fit the data poorly, as did models with school racial composition measures
(not shown).
Finally, I considered models combining school factors and contexts of exit
and reception (not all shown). In the end, Model 14 with immigrant group
educational selectivity, a context of exit measure, and school SES, a school
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Table 3. Model comparison tests of macro factors predicting years of schooling (n = 2777).
Types of variables included
Individual/ Contexts of Contexts of School Wald Log restricted-
family reception exit contexts chi2 DF likelihood AIC BIC
Model 1: Age, sex, nativity, family SES and x 276.8 9 −5518.8 11,055.5 11,108.8
structure
Model 2: Group post-migration SES x x 286.0 10 −5515.4 11,050.9 11,110.1
Model 3: per cent of citizens x x 278.5 10 −5522.5 11,064.9 11,124.2
Model 4: Group size x x 278.2 10 −5524.0 11,067.9 11,127.1
Model 5: Group selectivity x x 292.8 10 −5512.4 11,044.7 11,104.0
Model 6: Distance from U.S. x x 288.0 10 −5516.9 11,053.8 11,113.1
Model 7: GDP x x 276.9 10 −5521.4 11,062.8 11,122.0
Model 8: Political suppression x x 280.0 10 −5519.2 11,058.4 11,117.7
Model 9: Income inequality x x 280.8 10 −5491.6 11,061.5 11,120.7
−5511.3

ETHNIC AND RACIAL STUDIES


Model 10: Group selectivity and post-migration x x x 295.4 11 11,044.5 11,109.7
SES
Model 11: per cent of free/reduced lunch x x 304.0 10 −5517.2 11,054.4 11,113.6
Model 12: Minority school x x 279.3 10 −5518.9 11,057.8 11,117.1
Model 13: Inner city school x x 282.6 10 −5517.7 11,055.4 11,114.6
Model 14: Group selectivity and per cent of x x x 320.0 11 −5510.7 11,043.5 11,108.6
free/reduced lunch
Notes: Full models shown in Supplemental Table 2.

13
14 C. FELICIANO

context measure, provided the best data fit per the AIC statistic. While Model 5
remains the best-fitting model per BIC, I choose Model 14 for subsequent
results since group educational selectivity and school SES are the most
highly significant macro-level variables, and the school context’s importance
confirms existing theories.

Macro and micro-level effects on second-generation educational


attainment
Table 4 presents the multi-level analysis of family, immigrant group, and
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

school contexts on adult children of immigrants’ educational attainment.


The analysis of variance components in the null model show that most var-
iance is at the individual level. The intraclass correlation coefficients (ICC)
show the proportion of total variance attributed to immigrant groups is
0.07, and the proportion of the total variance attributed to schools is similar
at 0.066.
Model 1 includes individual-level demographics and family contexts.4 As
expected, family SES and two-parent home background positively influence
education. Introducing demographic and family contextual factors reduces
the ICC for immigrant groups from 0.07 to 0.05, explaining approximately
29 per cent of the variance between immigrant groups. In contrast, family

Table 4. Macro and micro-level effects on children of immigrants’ years of schooling.


Null Model
model Model 1 Model 2 Model 3 3A
Constant 15.101 13.993 14.739 13.999
Individual-level demographics
Age −0.186*** −0.185*** −0.205*** −0.196***
Female 0.445*** 0.446*** 0.436*** 0.453***
U.S. born 0.102 0.115 0.096 0.101
Family context, time 1
Family SES 0.265*** 0.261*** 0.250*** 0.382***
Family structure (two-parent) 0.407*** 0.404*** 0.394*** 0.387***
Immigrant group context of exit
Immigrant group educational 2.049*** 2.061*** 2.057***
selectivity
School context
Per cent of students receiving free/ −0.009*** 0.004
reduced lunch
Family SES × School SES −0.003***
Variance components
Immigrant groups 0.270 0.169 0.107 0.110 0.104
Schools 0.254 0.101 0.101 0.059 0.040
Individuals 3.335 3.082 3.076 3.074 3.067
ICC immigrant groups/residual ICC 0.070 0.050 0.033 0.034 0.033
groups
ICC schools/residual ICC schools 0.066 0.030 0.031 0.018 0.012
N of observations
Immigrant groups 29 29 29 29 29
Schools 42 42 42 42 42
Individuals 2,777 2,777 2,777 2,777 2,777
ETHNIC AND RACIAL STUDIES 15

SES and structure explain 55 per cent of the between-school variance (0.07 to
0.03). Thus, family differences explain more variation between schools than
between immigrant groups.
Model 2 adds educational selectivity, an immigrant group characteristic
shaped by contexts of exit. A one-unit increase corresponds to a net increase
of about 2.1 years of schooling. This translates into, for example, children of
Jamaican immigrants, whose education exceeds nonmigrants 68 per cent
more often than nonmigrants exceed immigrants, completing about one
more year of schooling than children of Mexican immigrants, whose edu-
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

cation exceeds nonmigrants only 20 per cent more often. Educational selec-
tivity explains an additional 34 per cent of the variance between immigrant
groups (ICC declines from 0.05 to 0.03). Considering immigrant group selectiv-
ity along with demographics and family contexts leaves only 3.3 per cent of
educational variation remaining between immigrant groups.
Model 3 adds school SES. The coefficient of −0.009 translates into a differ-
ence of 0.63 years of schooling between immigrants’ children from schools
where no one receives free/reduced lunch vs. those from schools where 90
per cent of students receive free/reduced lunch. While variation between
schools declines from 3.1 to 1.8, school SES explains no variation between
immigrant groups.
Do some characteristics matter more in some contexts than others? The
influence of immigrant group selectivity does not vary by family or school
SES (results available upon request). However, Model 3A shows, consistent
with research on grades and test scores (Portes and Hao 2004), more pro-
nounced family SES effects for immigrants’ children who attended high-SES
schools. The interaction between family and school SES explains more vari-
ation between schools, such that only 1.2 per cent remains.

Comparing contexts: family SES, immigrant group selectivity, and


school SES
Figure 2 calculates predicted years of schooling in varying immigrant group,
family and school contexts5 (from Table 4, Model 3A) to show how these con-
texts compare in importance for educational attainment. Not surprisingly,
immigrants’ children from low selectivity groups, low-SES families, and low-
SES schools (bar 1–13.1 years of schooling), complete the least education,
while those from high selectivity groups, high-SES families, and high-SES
schools (bar 8–16) complete the most.
The figure illustrates how the effect of family SES varies by school SES.
Among low-SES families, predicted education varies little by school SES, a con-
trast to the pattern among high-SES families. For example, those from high-
SES families and highly select groups complete only 14.9 years of schooling
in low-SES schools (bar 7) compared with 16 years in high-SES schools (bar 8).
16 C. FELICIANO

Figure 2. Comparing average predicted years of schooling in varying immigrant group,


family, and school contexts.
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Notes: Based on Model 3A, Table 4. Predicted values for U.S. born females from two-parent homes of
average age.

Figure 2 shows that young adults from highly educationally select groups
complete about one more year of schooling than those from low selectivity
groups, regardless of family or school SES contexts. For example, among
youth from low-SES families who attended low-SES schools (bars 1 and 3),
those from highly selected immigrant groups complete about one more
year of schooling than those from less selected immigrant groups (14.1 vs.
13.1). Thus, group-level immigrant selectivity strongly influences educational
attainment regardless of post-migration family or school context.
The relative importance of immigrant group educational selectivity as com-
pared with school SES depends upon family SES. Immigrants’ children from
low-SES families benefit more from membership in a highly select immigrant
group than attending a high-SES school: youth in highly select groups from
low-SES schools (bar 3) attain about 14.1 years of schooling while those in
less select groups from high-SES schools (bar 2) attain only about 13.3 years
of schooling. In contrast, predicted attainment is similar among those from
high-SES families, whether from low selectivity groups and high-SES schools
(15-bar 6) or high selectivity groups and low-SES schools (14.9-bar 7). For
the many children of immigrants who come from low-SES families (left
panel), immigrant group educational selectivity matters more than school
SES for their educational attainment.

Discussion and conclusion


I find that divergent educational outcomes between U.S. immigrants’ children
from different origin countries can largely be explained by family compo-
sitional differences, especially SES differentials, and a macro-level character-
istic related to immigrants’ context of exit – group educational selectivity.
Immigrant group educational selectivity explains over one-third of the
national-origin group differences in years of schooling that remain after
including family factors. In contrast, school contextual differences explain
no variation between national-origin groups. However, individual variation in
ETHNIC AND RACIAL STUDIES 17

educational attainment within immigrant groups is far greater than variation


between immigrant groups.
Unlike most previous studies of second-generation educational outcomes
that include country of origin as a set of dummy variables (Haller, Portes, and
Lynch 2011; Portes and Rumbaut 2001), I unpack ethnicity by examining
specific characteristics of national-origin groups. Prior research considered
how aspects of immigrant groups’ contexts of reception and contexts of
exit shape educational achievement (Levels, Dronkers, and Kraaykamp
2008), but did not include direct measures of immigrant group selectivity or
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

examine educational attainment. Other research has shown that immigrant


group selectivity predicts educational attainment (Feliciano 2005a), but did
not consider other contexts of exit and reception measures. By examining
several immigrant group characteristics within the same study, I show that
after including observed immigrant selectivity, other characteristics of immi-
grant groups’ contexts of exit have no additional explanatory power. Most sur-
prisingly, I find that measures related to immigrant groups’ contexts of
reception – post-migration SES, citizenship rates and group size – do not
predict educational attainment after including immigrant group educational
selectivity. This is surprising since existing literature argues that differences
in contexts of reception drive national-origin group disparities (Portes and
Rumbaut 2001; Portes and Zhou 1993). This study suggests, not that contexts
of reception are irrelevant, but that they are ultimately driven by contexts of
exit shaping which segment of an origin country’s population migrates.
While beyond this paper’s scope to evaluate the mechanisms linking immi-
grant group selectivity to the next generation’s education, existing literature
suggests two possibilities. First, immigrants’ educational selectivity may
shape group resources. Highly select immigrant groups may have knowledge
and skills that they mobilize to develop supportive social structures, such as
supplementary education centres, within ethnic communities (Zhou and
Kim 2006). Second, immigrant groups’ collective identities may be based on
their class status prior to migration. Collective outlooks may influence
second-generation youths’ desire for educational success through high
expectations, regardless of family or school contexts (Feliciano 2006; Zhou
and Bankston 1998).
Immigrant group advantages or disadvantages are often compounded by
advantages or disadvantages within families and schools. For example, the
least educationally select group, Mexicans, also have the lowest average
family SES and attend high poverty schools. Consistent with prior research,
family SES powerfully shapes educational attainment and helps explain
ethnic differences (Kasinitz et al. 2008; Rumbaut 2005b). In contrast, school
poverty differences do not help explain national-origin group differences in
educational attainment. However, school context do help explain individual
variation in educational attainment, especially among those from high-SES
18 C. FELICIANO

families. Consistent with prior research on grades (Portes and Hao 2004), I find
that family SES more strongly influences educational attainment among immi-
grants’ children who attended high-SES schools, and school context matters
little for adult children of immigrants from low-SES families. These findings
underscore both the disadvantages of growing up in a low-SES family –
which are not overcome by attending high-SES schools – and the disadvan-
tages of attending low-SES schools in adolescence, which negatively impact
even those from high-SES families.
Across the full range of family and school contexts, however, immigrant
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

group selectivity shapes educational attainment. Most importantly, for the


large proportion of immigrants’ children from low-SES families who attended
low-SES schools, high immigrant group educational selectivity boosts attain-
ment. These findings help clarify why ethnographic work on groups such as
Vietnamese has shown high achievement despite disadvantaged family and
school contexts (Zhou and Bankston 1998).
Since this study is limited to 29 national-origin groups in two metropolitan
areas, future research is needed to examine whether the conclusions hold in
other geographic contexts and for other ethnic groups. It may be, for instance,
that immigrant group selectivity only influences attainment in destinations
with a large population of co-ethnics. Nevertheless, this study yields important
conclusions. First, a characteristic associated with immigrants’ context of exit
relates more strongly to children of immigrants’ education than measures of
contexts of reception. Thus, debates about children of immigrants’ edu-
cational performance should not ignore pre-migration group differences
that contribute to advantages or disadvantages after migration. Second,
family, school, and immigrant group contexts intersect to shape the edu-
cational attainment of the second generation. Future research should
further examine how these factors shape the adaptation of immigrants’ chil-
dren growing up in different geographic contexts.

Notes
1. “Second generation” here refers to U.S.-born individuals with immigrant parents
and those born abroad who migrated as children (the “1.5 generation” – see
Rumbaut 1994).
2. This subsample had a 64 per cent retention rate, like the overall study. I con-
structed a variable based on coefficients from predictive models of attrition
(age, family composition, grade point average), and included it in multivariate
analyses to correct for sample attrition (Berk 1983). Because results did not
differ with this variable’s inclusion, the simpler model is presented (see Portes
and Rumbaut 2005 for more detail).
3. The other origin countries are El Salvador, China, Hong Kong, India, Iran, Japan,
Korea, Thailand, Canada, Haiti, Guatemala, Ecuador, Peru, Greece, Hungary,
Ireland, Italy, Poland, Yugoslavia, and Puerto Rico. To assign national origin where
parents were not from the same country, I used one parent’s birthplace if the
ETHNIC AND RACIAL STUDIES 19

other was missing or U.S.-born. If parents were born in different foreign countries, I
used respondent’s birthplace or, if the respondent’s birthplace differed from either
parent, mother’s origin country. Fathers were often absent, and Rumbaut (1994)
shows that mothers’ characteristics are more important than are fathers’.
4. I considered parents’ time in the U.S. and English-language ability, but these
effects were explained by family SES.
5. I coded the top quarter in family SES, school SES, or immigrant group edu-
cational selectivity as “high” and the bottom quarter as “low”.

Acknowledgements
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

The author would like to thank Yader Lanuza and anonymous reviewers for helpful
feedback on earlier drafts of this article.

Disclosure statement
No potential conflict of interest was reported by the author.

References
Borjas, George J. 1987. “Self-Selection and the Earnings of Immigrants.” The American
Economic Review 77: 531–553.
Borjas, George J. 1992. “Ethnic Capital and Intergenerational Mobility.” The Quarterly
Journal of Economics 107: 123–150.
Crosnoe, Robert. 2005. “Double Disadvantage or Signs of Resilience? The Elementary
School Contexts of Children from Mexican Immigrant Families.” American
Educational Research Journal 42: 269–303.
Duncan, Otis Dudley. 1961. “A Socioeconomic Index for All Occupations.” In
Occupations and Social Status, edited by A. J. J. Reiss, 109–138. New York: Free Press.
Feliciano, Cynthia. 2005a. “Does Selective Migration Matter? Explaining Ethnic
Disparities in Educational Attainment among Immigrants’ Children.” International
Migration Review 39: 841–871.
Feliciano, Cynthia. 2005b. “Educational Selectivity in U.S. Immigration: How Do
Immigrants Compare to Those Left Behind?” Demography 42: 131–152.
Feliciano, Cynthia. 2006. “Beyond the Family: The Influence of Pre-migration Group
Status on the Educational Expectations of Immigrants’ Children.” Sociology of
Education 79: 281–303.
Fiel, Jeremy E. 2013. “Decomposing School Resegregation.” American Sociological
Review 78: 828–848.
Freedomhouse.org. 2010. “Combined Average Ratings – Independent Countries.” In
Freedom in the World. Washington, DC: Freedom House. https://freedomhouse.
org/report/freedom-world/freedom-world-2010.
Gonzalez, Gilbert G. 2013. Chicano Education in the Era of Segregation. Denton, TX:
University of North Texas Press.
Haller, William, Alejandro Portes, and Scott M. Lynch. 2011. “Dreams Fulfilled, Dreams
Shattered: Determinants of Segmented Assimilation in the Second Generation.”
Social Forces 89: 733–762.
20 C. FELICIANO

Kasinitz, Philip, John H. Mollenkopf, Mary C. Waters, and Jennifer Holdaway. 2008.
Inheriting the City: The Children of Immigrants Come of Age. New York and
Cambridge: Russell Sage Foundation and Harvard University Press.
Lee, Valerie E., and Anthony S. Bryk. 1989. “A Multilevel Model of the Social Distribution
of High School Achievement.” Sociology of Education 62: 172–192.
Lee, Jennifer, and Min Zhou. 2015. The Asian American Achievement Paradox. New York:
Russell Sage Foundation.
Levels, Mark, and Jaap Dronkers. 2008. “Educational Performance of Native and
Immigrant Children from Various Countries of Origin.” Ethnic and Racial Studies 31:
1404–1425.
Levels, Mark, Jaap Dronkers, and Gerbert Kraaykamp. 2008. “Immigrant Children’s
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

Educational Achievement in Western Countries: Origin, Destination, and Community


Effects on Mathematical Performance.” American Sociological Review 73: 835–853.
Luthra, Renee Reichl, and Thomas Soehl. 2015. “From Parent to Child? Transmission of
Educational Attainment Within Immigrant Families: Methodological
Considerations.” Demography 52: 543–567.
Milanovic, Branko. 2005. Worlds Apart: Measuring International and Global Inequality.
Princeton, NJ: Princeton University Press.
Portes, Alejandro, and Lingxin Hao. 2004. “The Schooling of Children of Immigrants:
Contextual Effects on the Educational Attainment of the Second Generation.”
Proceedings of the National Academy of Sciences 101: 11920–11927.
Portes, Alejandro, and Rubén G. Rumbaut. 2001. Legacies: The Story of the Immigrant
Second Generation. Berkeley and New York: University of California Press and
Russell Sage Foundation.
Portes, Alejandro, and Rubén G. Rumbaut. 2005. “Children of Immigrants Longitudinal
Study (CILS), 1991–2006.” Inter-university Consortium for Political and Social
Research (ICPSR) [distributor].
Portes, Alejandro, and Rubén G. Rumbaut. 2006. Immigrant America: A Portrait.
Berkeley: University of California Press.
Portes, Alejandro, and Min Zhou. 1993. “The New Second Generation: Segmented
Assimilation and its Variants.” The Annals of the American Academy of Political and
Social Science 530: 74–96.
Quillian, Lincoln. 1995. “Prejudice as a Response to Perceived Group Threat: Population
Composition and Anti-Immigrant and Racial Prejudice in Europe.” American
Sociological Review 60: 586–611.
Raudenbush, Stephen, and Anthony S. Bryk. 1986. “A Hierarchical Model for Studying
School Effects.” Sociology of Education 59: 1–17.
Ruggles, Steven, and Matthew Sobek. 1997. “Integrated Public Use Microdata Series
(IPUMS): Version 2.0.” Edited by B. o. t. C. U.S. Department of Commerce.
Minneapolis: Historical Census Projects: University of Minnesota. http://www.
ipums.umn.edu.
Rumbaut, Rubén G. 1994. “The Crucible Within: Ethnic Identity, Self-Esteem, and
Segmented Assimilation among Children of Immigrants.” International Migration
Review 28 (4): 748–794.
Rumbaut, Rubén G. 2005a. “Children of Immigrants and Their Achievement: The Role of
Family, Acculturation, Social Class, Gender, Ethnicity, and School Contexts.” In
Addressing the Achievement Gap: Findings and Applications, edited by R. Taylor,
23–60. Charlotte, NC: Information Age Publishing.
ETHNIC AND RACIAL STUDIES 21

Rumbaut, Rubén G. 2005b. “Turning Points in the Transition to Adulthood:


Determinants of Educational Attainment, Incarceration and Early Childbearing
among Children of Immigrants.” Ethnic and Racial Studies 28: 1041–1086.
Schneider, Silke L. 2007. “Anti-Immigrant Attitudes in Europe: Outgroup Size and
Perceived Ethnic Threat.” European Sociological Review 24: 53–67.
Snijders, T., and R. Bosker. 1999. Multilevel Analysis: An Introduction to Basic and
Advanced Multilevel Modeling. London: Sage.
United States Department of Agriculture, Economic Research Service International
Macroeconomic Data Set. 2012. “Real GDP (2005 dollars) Historical”.
Valencia, Richard R. 2000. “Inequalities and the Schooling of Minority Students in Texas:
Historical and Contemporary Conditions.” Hispanic Journal of Behavioral Sciences 22:
Downloaded by [Therapeutic Goods Administration] at 06:56 07 September 2017

445–459.
van Tubergen, Frank, Ineke Maas, and Henk Flap. 2004. “The Economic Incorporation of
Immigrants in 18 Western Societies: Origin, Destination, and Community Effects.”
American Sociological Review 69: 704–727.
Zhou, Min, and Carl L. Bankston. 1998. Growing up American: how Vietnamese Children
Adapt to Life in the United States. New York: Russell Sage Foundation.
Zhou, Min, and Susan Kim. 2006. “Community Forces, Social Capital, and Educational
Achievement: The Case of Supplementary Education in Chinese and Korean
Immigrant Communities.” Harvard Educational Review 76: 1–29.

Vous aimerez peut-être aussi