9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

As a library, NLM provides access to scientific literature. Inclusion in an NLM database does
not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of
Health.
Learn more: PMC Disclaimer | PMC Copyright Notice

Ann Am Acad Pol Soc Sci. Author manuscript; available in PMC 2018 Nov 9. PMCID: PMC6226253
Published in final edited form as: NIHMSID: NIHMS968867
Ann Am Acad Pol Soc Sci. 2017 Nov 1; 674(1): 27–58. PMID: 30416198
Published online 2017 Oct 25. doi: 10.1177/0002716217730009

Education Outcomes of Immigrant Youth: The Role of Parental Engagement

Zhen Liu* and Michael J. White

Abstract

Using the 2009–2012 waves of the High School Longitudinal Survey, this article examines the role of
parental engagement on academic achievement in the United States. Specifically, we examine the influence
of parental engagement while also investigating the academic trajectories of racial/ethnic and immigrant
groups, controlling for other standard factors. Results suggest that the progression of students’ academic
performance varies substantially by race/ethnicity and by immigrant generational status. After controlling
for 9th grade test scores and family and other school-level characteristics, we find that first-generation im‐
migrant youth generally have higher 11th grade test scores and lower probability of dropping out compared
to native-born students who are second or third generation. Greater levels of parental engagement predict
superior test scores and lower rates of dropout for youth of various racial and immigrant generation back‐
grounds, even in the presence of a variety of controls.

Keywords: immigrant minority youth, educational outcomes, parental engagement

The proportion of children in immigrant families in the United States has increased dramatically during the
past two decades. The percent of children with one or more foreign-born parents has increased to account
for approximately 25.5 percent of all children under age 18 in 2015 (Migration Policy Institute 2017). As
more and more immigrants from Latin America and Asia have come to the United States and have settled
in new immigrant destinations that previously had very little immigration besides traditional immigrant
gateways, both the composition and geographic distribution of immigrants has become more diverse
(Fischer 2010; Donato et al 2008; Iceland 2009; Lichter and Johnson 2009; Lichter 2012; Logan and

[Link] 1/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Zhang 2010). The impact of increasing first and second-generation immigrant children on school systems
is also felt by communities across a broader set of geographic contexts, particularly those new destinations
in small-size cities or rural places that are adapting to absorb children from immigrant families (Fischer
2010). Consequently, the educational outcomes of children of recent immigrants has attracted increasing
attention from social researchers, since educational attainment is so closely linked to perceived upward im‐
migrant mobility and assimilation (Alba and Nee 2009; Hirschman 2001; Portes and Rumbaut 2001; White
and Glick 2009).

The original concerns of James Coleman, seen in the range of his work from the 1966 Coleman Report it‐
self (Coleman et al. 1966), through the longitudinal analysis of the 1980s, are also manifest in our work.
What has changed since that time is the composition of American schools, which now reflect the increas‐
ing diversity of the U.S. population, driven by immigration. In this article, our analysis and approach owes
much to that of Coleman; in fact, we use a longitudinal U.S. educational dataset—the High School
Longitudinal Study—a successor to the High School and Beyond study, the source for Coleman’s work in
the 1980s on schooling and achievement (Coleman, Hoffer, and Kilgore 1982; Coleman and Hoffer 1989)

This new immigration and the varied performance of first- and second-generation students in U.S. schools
further invigorates questions about the role of immigration and ethnicity in school achievement, and sug‐
gests a reexamination of the mechanisms through which families and schools translate resources from one
generation to the next. To be sure, how immigrants do, and particularly how well the children of immi‐
grants do in school or, in other words, whether they experience educational inequality, has sweeping conse‐
quences for educational policy and immigration policy in the United States and other immigrant-receiving
nations.

Previous research has suggested various mechanisms for how race, ethnicity, generational status, and fami‐
ly resources may lead to disparate educational outcomes among children. Following the early Equality of
Educational Opportunity study (Coleman Report) in 1996, there continued a long debate on how schools
and family resources affect child well-being (Parcel, Dufer, and Zito 2010). The original Coleman Report
(1966) argued for the importance of student background and family socioeconomic status in determining
school achievement, whereas much prior discussion had emphasized inputs to the school environment it‐
self. Coleman’s own later writing questioned the advisability of the phrase “equality of educational oppor‐
tunity,” but did argue for refocusing the role of schooling to reduce inequality (Coleman 1975). Since that
time many studies, including ours, have sought to determine the relative importance of various factors—
family background, demography, student’s in-school performance, peer characteristics, school organization
—in predicting achievement within and beyond school. Coleman is also noted for his early and extensive
attention to social capital (Coleman 1988). In an oft-observed phrase, Coleman argued, “… social capital
inheres in the structure of relations between actors and among actors” (Coleman 1988, S98). In our analy‐
sis, we give attention to the impact of “social capital” on educational outcomes while controlling for
school-level characteristics, but in doing so, our approach concentrates on specific indicators of the parent-
child relationship and parental school involvement (directly measured in the High School Longitudinal
Survey [HSLS]) to capture elements of parental engagement.

There remains uncertainty about whether first- and second-generation individuals outperform or, instead,
lag behind later-generation U.S.-born children in academic performance. Some researchers have found that
children of immigrants, particularly some subgroups of East and Southeast Asian immigrants, are doing as

[Link] 2/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

well as or even better than their U.S.-born counterparts despite their low initial socioeconomic status and
language barriers (Glick and White 2004; Kao 1995; Kao and Tienda 2005; Kao and Rutherford 2007;
White and Glick 2000, 2009). Other researchers have found that children of immigrants, particularly sec‐
ond-generation Hispanic immigrants from Caribbean origins, are disadvantaged in terms of both family re‐
sources and economic resources and thus are experiencing downward assimilation (Portes and Rumbaut
2001; Portes and Rumbaut 2006; Portes and Vickstrom 2015; Portes and Fernández-Kelly 2008; Haller,
Portes, and Lynch 2011). Moreover, it is unclear to what degree the variation in educational outcomes
among youth of differing race/ethnicity/generation groups is due to differential family social capital.

Coleman (1988) argues that family background can be divided into three components: financial capital, hu‐
man capital, and social capital (Coleman 1988, 109). Coleman (1988) also suggests that parental social
capital reflected through time and effort that parents spend on children, or social capital of the family de‐
fined as (strong) relations between children and parents is essential for transferring human capital of par‐
ents to children and that human capital such as parents’ education is relevant to children’s educational
growth only when it is complemented by social capital in the family. Parcel and Menaghan (1993) and
Parcel and Dufur (2001) also provide empirical evidence for the positive impact of family social capital on
child well-being. However, there might be significant differences in parental practices among race/ethnic
groups, as well as differences between immigrant and native populations (Kao 2004a). Children in immi‐
grant families might experience different levels and forms of family social capital in terms of parental in‐
teraction and parental school involvement due to cultural differences or language proficiency of immigrant
parents (Kao 2004a; Kao and Rutherford 2007).

Using the 2009 and 2012 waves of the HSLS, we examine the extent to which parental engagement im‐
pacts high school students’ academic performance, and further the extent to which that impact varies by the
race/ethnicity and generational status. In our approach, we consider the mediating effect of parental en‐
gagement on educational outcomes along race/ethnicity and immigrant generation lines. We consider the
educational outcomes of high school dropout and standardized test score progression across waves of the
survey. Parental engagement, which can be seen as a form of or linked to family social capital, is measured
here through parental school involvement, parent-child communication, and parental participation in activi‐
ties with students. This article refers to first-generation students as students who are born outside of the
United States with foreign-born parents. Second-generation students refers to students who are born within
the United States but with at least one foreign-born parent. Third (or higher order)-generation students
refers to all students who are born in the United States to native-born parents. The use of two waves of the
HSLS gives us the considerable advantage of longitudinal, nationally representative data—collected with
the dedicated goal of measuring educational achievement—to investigate these issues.

Previous research has shown that the assimilation of immigrants and children of immigrants may not fol‐
low the same paths for all racial/ethnic groups. Parent engagement has long been identified as an important
factor influencing educational achievement, but whether parental engagement, in terms of parent-child in‐
teraction and school involvement, operates differently by race/ethnicity and generation status remains un‐
clear (Kao 2004a; Kao and Rutherford 2007). Thus, we also extend previous research by examining
whether there are systematic differences in levels of parental engagement among racial/ethnic/generation
groups and whether such groups of students experience differential returns to levels of parental school in‐
volvement, parent-child communication, and parent participation in activities with students.

[Link] 3/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

In sum, this article aims to answer the following research questions: 1) What is the role of parental engage‐
ment in affecting educational performance (as measured by change in standardized math test scores from
9th to 11th grade and dropout status)? 2) Does the effect of this engagement on educational outcomes differ
by race/ethnicity and generational status? Figure 1 shows key presumed determinants of educational out‐
comes, with attention to the pathways through which race/ethnicity and generational status might affect
those outcomes. We aim to understand the links between race/ethnicity/generational status and educational
performance as well as the distinct and mediating effect of parental engagement on educational outcomes.

Figure 1

What are the possible mechanisms for race/ethnicity, generational status and parental engagement to affect educational
outcomes?

Parental Engagement and Education Outcomes

Parental engagement as a form of social capital

The concept of social capital has gained broad currency throughout the research and policy worlds. At the
same time, what exactly constitutes social capital has been broadly interpreted (Bankston and Zhou 2002a
and 2002b; Kao 2004b; Kao and Rutherford 2007; Turney and Kao 2009; Portes 2014). What is more, the
application of the concept of social capital to immigrant and minority populations, while quite relevant,
also enjoys no singular interpretation. The concept of social capital is widely recognized to originate in the
work of Bourdieu and of Coleman (Portes 2000a and 2000b; Kao and Rutherford 2007). Both Bourdieu’s
and Coleman’s definitions of social capital focus on the benefits of ties between individuals, though there is
significant variation (Bourdieu 1985; Coleman 1993; Portes 2000b). As summarized by Portes (2000b, 2),
sociologists have focused on exploring three aspects of social capital: social capital as “a source of social
control”; “a source of family-mediated benefits”; and as “a source of resources mediated by nonfamily net‐
works.” It is noteworthy that Coleman’s own theoretical treatise admits a key role for family ties among the
features of social capital (Coleman 1990, 300ff). What is more, Coleman’s empirical work on educational
achievement quite explicitly discussed family social capital, even going so far as to provide an example of
immigrant families harnessing family social capital to help a child do well in school (Coleman and Hoffer
1987, 223).
[Link] 4/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

As a likely consequence of these various interpretations, there is no standard measure of social capital. The
concept has been operationalized with various measures, such as community level networks and interaction
within families (Portes 2000b; Kao and Rutherford 2007). In studies of education outcomes, researchers
have measured social capital using parent-school involvement (Kao and Rutherford 2007; Sun 1998; White
and Glick 2000; Yan 1999), intergenerational closure (Carbonaro 1998; Morgan and Sørensen 1999; Kao
and Rutherford 2007; Portes 2000b), the parent-child relationship (Muller and Ellison 2001, Kao 2004a),
intact families (White and Kaufman 1997), and networks in the community (Bankston and Zhou 2002a,
Zhou and Kim 2006).

Our approach follows the reasoning of several scholars who suggest that parental engagement is likely to
impact education outcomes, especially for children of immigrant families that are economically disadvan‐
taged. For those groups faced with economic disadvantage, upward mobility and successful assimilation
likely depends more on the availability of parental involvement and the wider set of social ties. Coleman
(1988, 1991) argues that parent-child interactions and parental involvement help parents to transmit their
human capital to children. The positive effect of parent-child interaction is also supported in other empiri‐
cal work (Valenzuela and Dornbusch 1994; White and Glick 2000; Kao 2004a; Kao and Rutherford 2007).
The type of educational outcomes and social capital measured also matters. For instance, parental involve‐
ment is found to have a positive impact on students’ GPA and standardized test scores, continuation in high
school, and enrollment in postsecondary education (Goyette and Conchas 2002; Sun 1999; Teachman,
Paasch, and Carver 1997; White and Kaufman 1997; White and Glick 2000; Kao and Rutherford 2007).
Yet Morgan and Sørensen (1999) found that a high level of intergenerational closure is negatively associat‐
ed with gains in math test scores of students from 10th to 12th grades. McNeal (1999) found that parents’
participation in PTO meetings in school is negatively related with children’s science achievement though it
seems to help with preventing dropout. Overall, previous studies provide mixed evidence on the impact of
parental engagement on education outcomes, depending on both the type of parental engagement and the
indicator of academic achievement used.

Varied levels of parental engagement and parenting styles by race/ethnicity and immigrant status

Previous studies not only suggest that parental engagement could impact education outcomes, but they also
find that there are significant differences in parental practices among race/ethnic groups, as well as differ‐
ences between immigrant and native populations. Many studies suggest that minority students, particularly
minority first- and second-generation students, have lower access to parental social capital compared to
white third- and higher generation students with native born parents (Sui-Chu and Willms 1996; Kao and
Rutherford 2007). But at the same time, there is controversy on whether minority immigrant children in‐
deed have less access to social capital, or they in fact possess more ethnic-specific social capital (Bankston
and Zhou 2002a; Fuller and Hannum 2002; Kao 2004b; Lee and Zhou 2015; Zhou and Kim 2006). Zhou
and Kim (2006) argue that ethnic social structures, namely ethnic supplementary education in the Chinese
and Korean immigrant communities and the associated ethnic social environment are important factors
contributing to the educational success of Asian immigrant children. Lee and Zhou (2015) suggest that
high educational attainment among second-generation Asian students is partly due to a specific “success
frame” that put great emphasis on getting into elite universities and obtaining higher education. Also, this
kind of success frame is reinforced within Asian communities through supporting environments and re‐
sources such as college preparation courses and tutoring that might lead to better academic outcomes
among Asian students. Lee and Zhou argue further that hyper-selectivity, the very high degree of selection

[Link] 5/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

for favorable educational characteristics compared to those left behind, characterizes contemporary Asian
immigration (Lee and Zhou 2015, 6). While Lee and Zhou concentrate on the case of Asian-origin persons,
and while they argue that hypo-selectivity may operate for migrants from Mexico, some degree of positive
selectivity for success, even across unobserved characteristics, may operate among several immigrant sub‐
populations. Both Sui-Chu and Willms (1996) and Kao (2004a) found that Asian parents have lower levels
of involvement in school than whites, but they do not have significantly lower levels of school-related dis‐
cussions. While direct measurement is not possible, one might speculate that those parent-child discussions
convey the success frame, even when direct parental engagement in school itself is somewhat muted. Once
again, the literature provides mixed evidence about whether immigrant and minority groups are disadvan‐
taged in terms of levels of parental engagement, depending on the form of parental engagement being
measured.

More pertinent to our current investigation, we know less about whether parental engagement has a more
powerful mediating impact on education outcomes for one immigrant and/or minority group over others.
Researchers have started to examine whether particular parenting practices are contributing to the seeming‐
ly remarkable academic performance of immigrant children, especially those of Asian origin, given their
generally less advantaged socioeconomic status.

Although the literature on parenting styles is vast and there are numerous dimensions of parenting prac‐
tices, in this article we focus on three aspects of parenting styles that are frequently used in literature and
can be operationalized readily in the 2009–2012 HLLS: 1) Parent school involvement (as measured by par‐
ticipation in school meetings and events); 2) parent-child interaction (measured as parent extracurricular
participation in activities with their child); and 3) parent-child communication (as measured by discus‐
sions about school work, college, or careers; or personal problems with their child). These three indicators
of involvement are all expected to help translate parental aspiration and resources into student education
outcomes. Direct involvement in school programming both shows support for the value of education itself
(joining a school fundraiser) and provides a conduit for information about student performance (PTO meet‐
ings). Direct communication between parent and child is beneficial for students’ academic outcomes
(Muller 1998; Glick and White 2004; Kao 2004a). Through open communication with parents, children
can learn essential social skills to deal with problems at school or in relationships with peers. Kao (2004a)
also argues that frequent discussion about school or college helps parents to communicate to children their
educational aspirations. If parental aspiration is to positively influence academic performance, then discus‐
sion between parent and child is most likely essential for that aspiration to be realized. Interaction through
participation matters as well. This aspect of parental engagement, such as attending museums or school
science fairs, is shown to have a positive impact on students’ educational outcomes by providing a support‐
ive environment (Muller 1993).

Variation in parenting styles such as parent-child communication, parent participation in activities with stu‐
dents, and parental aspiration has been observed among racial groups and by generational status.
Valenzuela and Dornbusch (1994) suggest that cultural differences in parenting style may account for some
of the racial and ethnic disparities in educational achievement. Immigrant parents are often found to have
high educational aspirations for their children compared to their native counterparts (Glick and White
2004; Kao 1995; Kao and Tienda 2005; Goyette and Xie 1999). But notably, Asian-origin parents are also
less likely to share decision-making power with children or to communicate with children compared to
white parents (Kao 2004a). It also may be the case that children of different racial groups respond differ‐

[Link] 6/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

ently to parenting styles. Dornbusch (1989) suggests that while authoritarian parenting is negatively associ‐
ated with white students’ grades, it is positively related to grades for Asian students. However, there is lim‐
ited evidence on how parent-child discussion and parent participation in activities with students may oper‐
ate differently among racial/ethnic groups and by generational status. The study by Kao (2004a) is one of
the few that has attempted to examine whether children of different race/ethnicity and generation status
may respond differently to a particular parent-child interaction. Using the National Education Longitudinal
Survey of 1988 (NELS), Kao (2004a) found that while black and Hispanic students respond favorably to
general discussions about school, Asian students respond negatively to general discussions about school.
Also, discussions about college have a significant positive impact on GPA for Asian students but not for
Hispanic or black students. Using the same NELS data, Kao and Rutherford (2007) found that immigrant
and minority students experience different returns to parental-school involvement and intergenerational
closure. However, neither of two studies examines the impact of parental social capital on educational out‐
comes net of school effects, Therefore, our work aims to update and expand on Kao (2004) and Kao and
Rutherford (2007)’s study by incorporating more aspects of parental involvement and also including
school-level contextual factors.

Data and Methods

Data source

We use data from the 2009 and 2012 waves of the HSLS (HSLS:2009–2012; NCES 2013). The HSLS is a
national representative study that began in fall 2009 with a cohort of more than 20,000 high school stu‐
dents in the 9th grade. The first follow-up was conducted in 2012. The HSLS is a superior data source for
two reasons. First, many previous studies have used the earlier rounds of similar national educational sur‐
veys conducted by NCES such as National Education Longitudinal Survey (NELS 1988–2000) to study the
relationship between education outcomes and parental engagement (Glick and White 2004; Kao 2004a;
Kao and Rutherford 2007). By using the most updated nationally representative survey on education out‐
comes conducted by the NECS, this article complements those earlier studies while also expanding the lit‐
erature using the most recent evidence. Moreover, the HSLS 2009 wave collected richer information on
parental involvement and participation than previous waves of national educational surveys. The goal of
doing so was to understand students’ educational trajectories such as how and why students choose certain
college majors and careers; at the same time, such data provide a great resource for studying parental en‐
gagement and its effects.

Since our central concern is how parental engagement influences adolescents’ educational performance, we
limit our analysis to participants in the 2009 wave for whom their parents completed the companion parent
survey. Many of the key independent variables of interest (parent-school involvement, parent-child commu‐
nication, parent participation in students’ extracurricular activities) are only available from the parent ques‐
tionnaire. The base sample used for this study therefore includes 13,000 respondents (rounded) for whom
values of all key variables are non-missing. Our analyses are weighted in accordance with NSLS documen‐
tation recommendations. Summary statistics of the analytical sample are presented in Table 1.

[Link] 7/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Table 1

Descriptive Statistics of Analytical Sample

Weighted meansa

Key outcome variables

Student ever dropped out of school (by follow up year 2012) 7.1%

Follow-up year math score (standardized theta score) 52.2 (10.0)

Base-year year math score (standardized theta score) 51.9(9.8)

Main independent variables

Generation Status

First Generation 5.2%

Second Generation 19.0%

Third + Generation 75.8%

Gender

Male 49.4%

Race/Ethnicity

Asian 4.2%

Black 10.7%

Latino 20.9%

Whites and Others 64.2%

Family Characteristics

Family socioeconomic status (standardized) 0.03(0.8)

Family structure (in base year 2009)

Intact family 63.0%

Single mother (with or without partner) 28.1%

Single father (with or without partner) 6.2%

No parent 2.7%

Parental Engagement

Parent school involvement

item: Attended a general school meeting since start of 2009-10 school year 82.9%

item: Attended a PTO meeting since start of 2009-10 school year 35.8%

item: Attended parent-teacher conference since start of 2009-10 school year 53.7%

item: Attended school event since start of 2009-10 school year 65.2%

Source: High School Longitudinal Survey 2009 and 2012 panels (N~=13,000). The analytical sample used in this paper
only include students that participated in both waves and also have parent interviewed in the base year who have answered
questions on parental engagement. Key information such as generation status, parental engagement are only available from

[Link] 8/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

the parent data.

Note:
a
The means and proportions are weighted by W1PARENT, which is defined as “ parent weight for studentlevel analyses
involving the parent questionnaire”. According to NCES, because of low response rate in parent questionnaires, analysis
involving parent information should be adjusted by parent weight.
b
Parent-child communication is measured based on students’ responses to a series of 10 questions regarding whether they
have ever discussed school work, college, jobs/careers and personal problems with mother/father during the academic year.
We collapsed the 12 items to 4. Answers from both mother/father are pooled together due to space constraints.

Descriptive Statistics of Analytic Sample

Key variables

Dependent variables: Educational outcomes We examine two educational outcomes in this article: stan‐
dardized math test scores (in the 11th grade follow-up wave) and drop out status (follow-up wave). The
standardized math test score is chosen because it is comparable for measuring students’ achievement across
schools and states, since it is based on the same computer test given to all interviewees during the time of
the survey, and it is standardized to a 100-point scale by the NCES, with a mean of 52 and standard devia‐
tion of 10. Drop out status is ascertained at the time of the follow-up survey. Students are asked whether
they ever dropped out between the time of the first wave (when students are in 9th grade) and the follow up
(when students are or would be in 11th grade). We argue that it is important to measure both test scores
and enrollment status, since they are indicators of different aspects of academic performance. Standardized
test scores have been widely used in studies of educational achievement. They have been generally treated
as an indicator of both intelligence and learned information (Kao and Thompson 2003), and also a “more
objective measures of what students have learned” (Kao and Rutherford 2007, 33). We follow a long-stand‐
ing tradition of examining drop out status, for its obvious value as an academic outcome and the associated
public policy concern, since high school dropouts are generally more likely to be unemployed or in contact
with the criminal justice system. Math scores are used instead of reading scores because they are less likely
to be affected by language barriers.

Independent variables: Race/ethnicity and immigrant status We categorize students by race/ethnicity (non-
Hispanic Asian, non-Hispanic black, Hispanic and non-Hispanic white [reference]), and by generational
status (first-generation, second-generation, and third-generation and above [reference]). First-generation
youth are classified as those born outside of the United States, while second-generation students are classi‐
fied as those born in the United States but with at least one foreign-born parent. Third+ generation youth
are those who are born in the United States with both parents also born in the United States. (For ease of
exposition we sometimes refer to this final group as “3rd” generation, recognizing that it also includes all
higher-order generational statuses; similarly, we often drop the “non-Hispanic” qualifier for the race-ethnic
categories). In our descriptive statistics (means comparison), we represent 12 categories: cross classifying
four race/ethnic groups by three generation groups.

Measures of parental engagement

[Link] 9/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

The score of parental engagement is constructed through principal-component factor analysis as a factor
score based on twenty-five items asked in the HSLS 2009 survey. Among the twenty-five items, seven
items tap parental school involvement: whether parent reported attending a general school meeting, a PTO
meeting, or school event; served as a school volunteer; participated in a school fund raiser; met with a
school counselor; or attended parent teacher conference since the start of the 2009–2010 school year (when
students are in 9th grade). Parental extracurricular participation in students’ activities is measured based
on parents’ responses to a series of eight questions regarding whether they went to a science or engineering
museum; worked or played on a computer; built or fixed something; attended a school science fair; helped
their 9th grader with a school science fair project; discussed a STEM program or article; visited a library;
or went to a play, concert, or live show with 9th grader in last year (prior to the time of the base-year sur‐
vey). Parent-child communication is measured based on students’ responses to a series of ten questions re‐
garding whether they have ever discussed school work, college, jobs/careers, and personal problems with
mother/father during the academic year 2009– 2010. Table 1 collapsed the ten items to four. Answers from
both mother/father are pooled together due to space constraints. For example, discussion about school work
includes four questions: whether 9th grader talked to mother about science, whether 9th grader talked to fa‐
ther about science, whether 9th grader talked to mother about other courses, and whether 9th grader talked
to father about other courses. Discussion about college includes two items: whether 9th grader talked to
mother about going to college and whether 9th grader talked to father about going to college. Discussion
about jobs/careers also includes two items: whether 9th grader talked to mother about adult jobs/careers
and whether 9th grader talked to father about adult jobs/careers. Discussion about personal problems also
includes two items: whether 9th grader talked to mother about personal problems and whether 9th grader
talked to father about personal problems.

In exploratory analysis we considered developing factor scores based on each of the three a priori group‐
ings. Given the cross-correlation of the three such factors and limited additional information the separation
conveyed, we opted for parsimony. We thus conducted factor analysis with all twenty-five items and ex‐
tracted a single over-arching factor to capture parental engagement.

Responses for each item are originally coded as dummy variables, and principal component factor analysis
is conducted to check the consistency among items and to construct the final factor score of social capital.
Principal-component analysis is used here because it is believed to be useful for reducing correlated ob‐
served variables to a smaller set of important independent composite variables (Abdi 2003; Suhr 2005).
Analytical weights are applied for item responses at the individual level. The complete list of items used
for constructing the composite score is presented in Table A1 in the appendix. Only one factor index of
parental engagement is retained because of its highest eigenvalue of 4.6, which is greater than the conven‐
tional criterion (>1). All of the response items are positively related to the final index, while attending
school events (fundraising or volunteering) and parent-child discussion about school have the greatest
weights in the parental engagement score.

Control variables

Individual characteristics of gender, base-year math score, family socioeconomic status (SES), family
structure, and school-level characteristics are included as control variables in all models. Previous research
suggests that those are the measures of individual and family characteristics that influence educational out‐
comes (Glick and White 2004; Kao and Rutherford 2007). In the HSLS, family SES is a composite mea‐
[Link] 10/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

sure constructed from the education and occupation (prestige score) of parents and family income. Besides
family SES, we also control for family structure. Students living in single-parent households may face extra
barriers in accessing social capital and parental attention.

Crucially, we also control for students’ prior academic performance (math score in the 9th grade) to accu‐
rately capture the effects of parental engagement on achievement trajectory. School-level characteristics,
such as school type, share of immigrant students, and percent students in free-lunch program, are also in‐
cluded as controls to minimize the potential confounding impact of school characteristics on education
outcomes.

Methods

We begin with a descriptive comparison (Table 2) of educational outcomes, family background characteris‐
tics, parent-school involvement, parent-child communication, parent participation in students’ extracurricu‐
lar activities by generational status and race/ethnicity. We then employ mixed effects models to estimate
the effect of parental engagement on wave 2 math score (Table 3A) and probability of dropping out of high
school (Table 3B), after controlling for base year math score and other individual-level, family background,
and school context characteristics. We build our models sequentially, to determine how the predictive ef‐
fects of variables are altered in the presence of additional covariates. Last, stratified models (Tables 4A and
4B) are used to examine whether these processes appear to differ by race/ethnicity.

[Link] 11/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Table 2

Weighted Means of Measures of Academic Performance, Parental Engagement, Family Background, School Charcteristics,
by Race/Ethnicity and Generational Status

Asian Black Hispanic White

1st 2nd 3rd 1st 2nd 3rd 1st 2nd 3rd 1st 2nd

Key outcome
variables

Base-year year 57.57 60.23 55.47 49.20 49.98 46.43 50.40 49.52 49.50 55.58 55.9
math score
(standardized
theta score)

Follow-up year 58.18 61.55 52.94 50.96 48.36 46.93 49.45 49.24 50.02 57.65 55.9
math score
(standardized
theta score)

Student ever 0.03 0.05 0.09 0.03 0.02 0.12 0.06 0.10 0.11 0.14 0.04
dropped out of
school (by
follow up year
2012)

Parental −0.12 0.03 −0.39 −0.51 −0.24 −0.21 −0.36 −0.35 −0.27 −0.05 0.22
engagement
index

Household
characteristics

Intact 0.86 0.83 0.75 0.71 0.59 0.34 0.73 0.68 0.47 0.84 0.77
family(both
parent present)

Family 0.30 0.35 0.37 −0.15 0.11 −0.30 −0.72 −0.57 −0.12 0.12 0.47
socieconomic
status
(standardized
score)

School-level
characteristics

Percent in 91.39 93.38 93.01 94.66 89.26 96.83 98.11 96.00 94.06 93.53 86.8
public shool

[Link] 12/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC
*
Note: The number of students who are identified as third-generation Asian, first-generation or second-generation Black are
relative small, and thus the mean estimations have large standard deviations. They should be interpreted with caution.

[Link] 13/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Table 3a

Mixed-Effects Model Predicting 11th Grade Math Test Score

Model 1 Model 2 Model 3 Model 4 Model 5

Parental Engagement 2.262*** 0.636*** 0.607*** 0.310** 0.294*

(0.179) (0.142) (0.150) (0.155) (0.154)

Individual Characteristics

2009 (base year) standardized math score 0.684*** 0.671*** 0.640*** 0.638***

(0.0280) (0.0262) (0.0252) (0.0237)

Males (ref= Females) 0.329 0.239 0.275

(0.309) (0.312) (0.312)

Race/Ethnicity (ref=Non Hispanic whites)

Non-Hispanic Asian 2.032*** 1.944*** 1.944***

(0.585) (0.606) (0.615)

Non-Hispanic Black −2.145*** −1.702*** −1.686***

(0.660) (0.636) (0.628)

Hispanic −1.024* −0.492 −0.549

(0.562) (0.522) (0.537)

Generation Status (ref= Third+ Generation)

First Generation 1.106* 1.335* 1.167*

(0.669) (0.698) (0.686)

Second Generation 0.0466 −0.0324 −0.175

(0.434) (0.442) (0.424)

Family Background Characteristics

2009 (base year) composite SES score 1.438*** 1.317***

(0.235) (0.249)

Family Composition

Single mother −0.503 −0.467

(0.351) (0.350)

Single father −0.817 −0.769

(0.665) (0.658)

No parent −0.326 −0.272

***
p<0.01,
**
p<0.05,
*p<0.1

[Link] 14/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Note: Results presented in this table are based on complete cases. Weights are used to adjust for missing of parent question‐
aires in the survey, following documentation from NECS. Results are consistent when multiple imputation (20 imputations)
is used to impute missing values. Findings also hold when country of origin is added in the models.

[Link] 15/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Table 3b

Mixed-Effects Logistic Models Predicting Probability of Ever Drop Out

Model 1 Model 2 Model 3 Model 4 Model 5

Parental Engagement −0.694*** −0.564*** −0.557*** −0.432*** −0.427***

(0.0864) (0.0933) (0.0933) (0.115) (0.112)

Individual Characteristics

2009 (base year) standardized math score −0.0506** −0.0479** −0.0337** −0.0319***

(0.00915) (0.00906) (0.00975) (0.00957)

Males (ref= Females) 0.167 0.191 0.194

(0.162) (0.165) (0.164)

Race/Ethnicity (ref=Non Hispanic whites)

Non-Hispanic Asian 0.433 0.488 0.485

(0.390) (0.400) (0.382)

Non-Hispanic Black 0.327 0.141 0.184

(0.302) (0.282) (0.291)

Hispanic 0.182 −0.0178 −0.0114

(0.241) (0.255) (0.267)

Generation Status (ref= Third+ Generation)

First Generation −0.619** −0.745** −0.756**

(0.308) (0.315) (0.326)

Second Generation −0.357 −0.349 −0.350

(0.243) (0.253) (0.268)

Family Background Characteristics

2009 (base year) composite SES score −0.616*** −0.636***

(0.165) (0.153)

Family Composition

Single mother 0.110 0.132

(0.160) (0.158)

Single father 0.400 0.413

(0.297) (0.299)

No parent 0.717** 0.730**

***
p<0.01,
**
p<0.05,
*p<0.1

[Link] 16/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Note:
aResidual variance is assumed to be π^2/3. Results presented in this table are based on complete cases. Weights are used to
adjust for missing of parent questionaires in the survey, following documentation from NECS. Results are consistent when
multiple imputation (20 imputations) is used to impute missing values. Findings also hold when country of origin is added
in the models.

[Link] 17/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Table 4a

Mixed-Effects Model Predicting 11th Grade Math Test Score

Asian Black Hispanic White

Parental Engagement 0.646 −0.805 0.696** 0.246

(0.423) (0.635) (0.351) (0.206)

Individual Characteristics

2009 (base year) standardized math score 0.732*** 0.618*** 0.653*** 0.634***

(0.0372) (0.0392) (0.0393) (0.0403)

Males (ref= Females) 0.488 −1.346* 0.140 0.556

(0.802) (0.731) (0.676) (0.380)

Generation Status (ref= Third+ Generation)

First Generation 1.005 3.253** −0.698 4.131**

(1.784) (1.414) (0.843) (1.748)

Second Generation 1.646 −1.279 −0.678 −0.829

(1.620) (2.073) (0.613) (0.624)

Family Background Characteristics

2009 (base year) composite SES score 0.798* 1.444*** 0.747 1.603***

(0.428) (0.543) (0.471) (0.344)

Family Composition

Single mother −0.716 −1.579* −0.602 −0.256

(0.862) (0.830) (0.982) (0.515)

Single father −4.944 −1.335 −0.867 −0.472

(3.023) (1.230) (1.139) (0.879)

No parent −0.297 −1.749** 0.467 0.217

(1.116) (0.781) (1.247) (0.997)

School-level characteristics

School type (vs. Public school)

Non-Catholic private school 0.494 1.131 1.491 0.345

(0.941) (1.451) (1.259) (0.557)

Catholic private school −2.159** 2.590*** 1.431 0.907**

(1.092) (0.903) (0.969) (0.422)

***
p<0.01,
**
p<0.05,

[Link] 18/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC
*
p<0.1

[Link] 19/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Table 4b

Mixed-Effects Model Predicting High School Dropout Status

Asian Black Hispanic White

Parental Engagement −0.857*** −1.477*** −0.135 −0.487***

(0.241) (0.475) (0.175) (0.148)

Individual Characteristics

2009 (base year) standardized math score −0.0481** −0.00723 −0.0653** −0.0220*

(0.0230) (0.0347) (0.0179) (0.0116)

Males (ref= Females) −0.724 1.005 0.120 0.0442

(0.471) (0.684) (0.267) (0.194)

Generation Status (ref= Third+ Generation)

First Generation −1.928** −3.840*** −0.528 −0.0129

(0.849) (1.472) (0.413) (0.803)

Second Generation −1.385*** −3.197** 0.0696 −0.807**

(0.439) (1.578) (0.381) (0.401)

Family Background Characteristics

2009 (base year) composite SES score 0.0464 −1.151** −0.920*** −0.456**

(0.359) (0.564) (0.258) (0.210)

Family Composition

Single mother 0.781 0.948 −0.0151 0.140

(0.521) (0.801) (0.352) (0.198)

Single father 0.388 2.427 0.307 0.653

(0.691) (1.754) (0.661) (0.406)

No parenta −0.527 1.953*** 0.836**

—
(1.272) (0.662) (0.416)

School-level characteristics

School type (vs. Public school)

Non-Catholic private school −0.0397 −6.217** 0.625 0.695*

(0.671) (2.547) (0.779) (0.398)

Catholic private schoolb −2.149 0.265 −0.942**

—
(1.652) (0.558) (0.465)

Standard errors in parentheses;

***
p<0.01,
**p<0.05,

[Link] 20/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC
*
p<0.1
Note:
a
There is no coefficient for no parent for the Asian subgroup because there are only 20 (rounded) Asian students without
any biological parent present in the sample and none has dropped out during Grade 9 and Grade 11.
Note:
b
There is no coefficient for Catholic private school for the Asian subgroup because there are 130 (rounded) Asian students
enrolled in Catholic private school in the sample and none has dropped out during Grade 9 and Grade 11.

We employ mixed-effects models to estimate the effects of parental engagement, race/ethnicity, and gener‐
ational status on education outcomes, while controlling for family background and school characteristics.
Mixed-effects models are advantageous for their ability to reflect and adjust for influences at the higher lev‐
el of aggregation (the school). Hearkening back to the early work of Coleman and colleagues, who used
High School and Beyond Data (Coleman, Hoffer, and Kilgore 1982), one would expect school type to mat‐
ter for students’ academic performance and students’ family background characteristics to impact the types
of schools that students attend. Using the Children of Immigrants Longitudinal Study, Portes and Hao
(2004) find that school-level SES is associated with higher GPA and lower probability of school attribu‐
tion. Mixed-effects models have further advantages, since students with similar background characteristics
are clustered within school and the errors for individual observations in the same school are likely to be
correlated, with education outcomes reflecting school-level context in addition to individual-level charac‐
teristics. Students are not at all randomly distributed across schools. In fact, there is substantial variation in
school type (and probably choice where able) by race/ethnicity and by generational status. As shown in
Table 2, first-generation students regardless of racial and ethnic origin are far more likely to enroll in
schools with higher shares of immigrants, compared to their native-born counterparts. The clustering of
peers of similar ethnicity and immigrant status might have varied implications for students of different
racial and ethnic origin. Portes and Hao (2004) suggest that the disadvantage in academic performance
among second-generation Mexican are significantly attenuated by attending schools with higher shares of
coethnics, while the advantage of Asian second-generation immigrants is also reduced by attending schools
with high concentration of Asian students. Also, minority students, particularly black and Hispanic stu‐
dents, are more likely to be clustered in public schools or schools with lower average SES compared to
their white counterparts. School-level characteristics such as learning resources (e.g., teacher quality or
teacher-student ratio) that are not measured in the model but could affect students’ education outcomes are
adjusted with a random constant at the school level to allow variation among schools. In sum, to examine
the effects of both individual characteristics and school contexts, while addressing the issue of correlated
errors at the school level, we employ two-level mixed-effects models, with the students as level 1 and
schools as level 2 observations and random intercepts at the school level. Using the math test score as an
example, the full mixed-effects model is specified as:

Yij = 𝛽0j + 𝛽1 X1ij + 𝛽2 X2ij + 𝛽3 X3ij + 𝛽4 X4ij + β5 Zj + 𝑒𝑖𝑗 ,

where i is the student level; j is the school level; X1 includes all combinations of race/ethnicity and genera‐
tion status; X2 refers to prior academic performance; X3 refers to family characteristics such as SES and
family structure; X4 refers to parental engagement; Z refers to school specific context (school type, percent
[Link] 21/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

student in free lunch program, percent taking AP courses, immigrant stock); and β0j = γ00 + u0j (random
intercept by school).

Results

Descriptive statistics

First, we present the descriptive statistics for the analysis sample in Table 1. The descriptive statistics show
that HSLS 2009 provides a diverse sample with about 5.2 percent of student respondents being first-gener‐
ation and 19.0 percent of them being second-generation. Table 1 also presents the weighted summary sta‐
tistics for item measures of the clusters of parental engagement: parent-school involvement; parent partici‐
pation in extracurricular activities with children; and parent-child communication. As shown in Table 1,
parents in general report a fair amount of parent-school involvement and participation. More than half of
parents report attending a general school meeting, parent-teacher conference and school event, as well as
participation in most extracurricular activities. But parents are less likely to serve as school volunteers, or
participate in fund raisers or PTO meetings, all of which are indictors of higher levels of involvement that
require more resources (temporal and perhaps monetary) from the parents. Meanwhile, students also report
quite high levels of communication with parents. More than 80 percent of the high school students inter‐
viewed report discussing going to college and future careers with parents, though the percent of them who
are willing to discuss personal problems with parents is lower (63.3 percent).

Table 2 presents the weighted means of academic performance, parental engagement, and household char‐
acteristics by race/ethnicity and immigrant status. A significant difference (symbols not indicated for rea‐
sons of space) is observed between minority first- and second-generation groups and third+ generation
whites in terms of education outcomes. With education outcomes measured as math test scores, first- and
second-generation Asians on average have higher math test scores in both 9th grade and 11th grade than
native third-generation whites, while immigrant blacks and Hispanics (both first and second generation)
perform worse in high school compared to third- or higher-generation whites. Notably, children of Asian
and black immigrants all seem to do better than their third-generation counterparts within the same
racial/ethnic group, but the generational ordering is not the same for all minority racial groups. For Asians,
the second-generation children perform best while for blacks, the first-generation seems to perform best in
terms of 11th grade math test scores. For Hispanics, the first- and second-generations seem to lag behind
compared to their third+ generation counterparts. Dropout rates increase with generation for all three mi‐
nority groups. Hispanic and black dropout rates for those in the third (and higher order) generation are al‐
most twice that for comparable whites.

Though immigrant children and children of immigrants have both higher base wave and follow-up wave
test scores for all groups except for Hispanics, they are not advantaged in terms of levels of parental en‐
gagement. First-generation or second-generation immigrant minority youth tend to have lower parental en‐
gagement compared to their third+ generation counterparts, except for Asians. For Asians, third-generation
youth have the lowest parental engagement, which could be because the Asian group is more heteroge‐
neous, and that the number of third-generation Asians is relatively small and thus the mean estimation has
a large standard error, so the mean estimation for third-generation Asians should be interpreted with cau‐
tion. In terms of racial/ethnic differences, minority groups all have lower levels of parental engagement
compared to whites, regardless of generation status.
[Link] 22/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Children of different racial/ethnic groups and immigrant status are also exposed to varied household and
school-level characteristics. As shown in Table 2, first-generation Hispanic children in general have the
lowest SES, though the condition improves across generation. Meanwhile, first- and second-generation
Asians are not economically disadvantaged compared to third+ generation whites and economic condition
improves over time for Asians. Black youth regardless of generational status are economically disadvan‐
taged compared to their third+ generation white counterparts, and the economic disadvantage among black
youth seems to amplify over time. Among whites, the second-generation youth seem to have the highest
family SES.

In terms of school-level characteristics, first-generation immigrant youth are most likely to enroll in
schools with the highest immigrant student stock and least likely to enroll in Catholic private schools, re‐
gardless of race and ethnicity. Both first- and second-generation youth are more likely to enroll in schools
with relative higher proportions of students taking AP courses compared to their third-generation counter‐
parts among all racial groups. Minority youth are more likely to enroll in less economically advantaged
schools (measured by percent student in free lunch program) overall, although there are some differences
along generation lines. First-generation Asians and Hispanics are more likely to enroll in less economically
advantaged schools, compared to their second- or third-generation same-race counterparts. Yet among
black youth, the third-generation students are most likely to end up in schools that are more economically
disadvantaged.

Mixed-Effects Regression Analysis

Impact of parental engagement on educational trajectories

Mixed-effects Generalized Linear Model(GLM) regression results predicting the effects of race/ethnicity,
immigrant status, and parental engagement on math test scores are presented in Table 3A, and logistic re‐
gression results predicting probability of dropping out are presented in Table 3B. All models adjust for
school-level clustering. In the tables we present here, we include parental engagement as the key indepen‐
dent variable first, to examine its effect solely, since race/ethnicity, generational status, and other family
characteristics are likely to have an impact on the level of parental engagement.

In alternative preliminary analyses (tables not presented here), we include race/ethnicity, generational sta‐
tus first before adding in parental engagement. We find that the mediating impact of parental engagement
on race/ethnicity and generational status is limited, but the mediating impact of parental engagement on
SES, as well as its independent impact, is substantial and statistically significant.

Model 1 in both Tables 3a and ​3B shows the effect of parental engagement has highly significant predictive
power on wave 2 math test scores and dropping out of high school. Model 2 adds in base-year (9th grade)
math, and the estimated effect of parental engagement decreases, but it is still significant. Base-year math
explains over half of the variation in the follow-up year math test score in all of the estimation models. (
Appendix Table A2 provides the regression on base-year math test score as the outcome from the equiva‐
lent set of covariates.) Models 3 in Table 3A and ​3B add in gender, race/ethnicity, and generational status.
Models 4 in Table 3A and ​3B add in family composition and SES. The results in model 4, Table 3A and
3B suggest that in terms of nativity difference, compared to third-generation and above youth, first-genera‐
tion youth are more likely to have higher 11th grade math scores and lower probability of dropping out af‐
[Link] 23/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

ter controlling for gender, race/ethnicity, base-year math score, parental engagement, and family SES
(p<.01). Meanwhile, the impact of being second-generation on math test score and probability of dropping
out of high school is not significant. (We do find that first- and second-generation youth in general have
higher 9th grade math test scores compared to their third-generation and above counterpart [see
appendix Table A2]).

In terms of racial/ethnicity differences, we find that Asian students tend to have higher 11th grade math
scores, while non-Hispanic black students tend to have lower 11th grade math scores compared to non-
Hispanic whites (p<.01), after controlling for base-year math and generational status. Hispanic students
also tend to have lower 11th grade math test scores compared to whites, but the effect is only marginally
significant, after controlling for base-year math, gender, and generational status. The racial differences ob‐
served for 11th grade math scores are also seen in 9th grade test scores. However, race/ethnicity is notably
not appreciably predictive of ever dropping out of high school, after controlling for family SES and
composition.

In keeping with the findings of much prior research, our results show that family SES (introduced in mod‐
els 4 and 5) is a highly significant predictor of academic performance regardless of whether the outcome is
measured as math test or dropout status. Children living in economically better-off families are more likely
to have higher 11th grade math scores and lower probability of dropping out of high school, even after con‐
trolling for 9th grade math test scores.

The lack of a two-parent family (family structure also introduced in models 4 and 5) points to inferior
schooling outcomes, although several specific differentials do not achieve statistical significance at conven‐
tional levels. Family structure effects are more strongly predictive at base year, as shown in
appendix Table A2. We do find that children living in households with no biological or adoptive parents
are much more likely to drop out of high school than children living in two-parent families. Meanwhile,
even after we include all the measures for family SES, family structure, the predictive effect of parental en‐
gagement is still statistically significant for improving math test scores and preventing youth from dropping
out of high school at p<.001.

To further understand how much school context could alter the measured effects of race/ethnicity, genera‐
tional status, and family characteristics, we introduce school-level characteristics into the regression mod‐
els. Model 4 (before the explicit introduction of school-level covariates) in both Tables 3A and ​3B suggests
that school-level clustering explains about 6 percent of the total variance in 12th grade math test scores and
about 8 percent of the residual variance in dropout probability, after we control for all individual and fami‐
ly-level characteristics. Model 5 introduces the several school-level measures: school organizational type;
poverty (percent free lunch); percent taking AP courses; and immigrant stock. Results in Model 5 indicate
that school type has a significant predictive power regarding education outcomes, net of individual and
family-level characteristics. Students in Catholic, private schools on average have higher 11th grade math
test scores and lower probability of dropping out, compared to students in public schools. Meanwhile,
school-level poverty measured by percent of students in the free lunch program is negatively associated
with 12th grade math scores and positively associated with probability of dropping out, but the associa‐
tions are not statistically significant. Other school-level characteristics are not statistically significant in
these models.

[Link] 24/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

These results are trajectories, net of base-year circumstance. When we examine base-year conditions (
appendix Table A2), we do see some important differences along these lines. Students in schools with a
higher percent of students in free lunch program have significantly lower 9th grade math test scores.
Student impoverished (large fraction free lunch) schools are predicted to perform less well initially, but
those in schools with more advantaged curricula and higher immigrant stock are predicted to do better on
the initial math test. Interestingly, students enrolled in private schools (catholic or non-Catholic) are pre‐
dicted to have lower base-year math test scores, net of the school-level and family background traits en‐
tered into the model.

Varied impacts of parental engagement by race/ethnicity and education outcomes

Tables 4a and ​4b look at within race/ethnic groups. We estimate these models, stratified by race/ethnicity,
to better understand the interactive effect of race/ethnicity, generational status, and parental engagement.
We present only models that include all substantive personal, family, and school covariates, as above. The
inclusion of parental engagement does not appreciably change the parameter estimates for other variables.
(All statistically significant coefficients retain their significance, sign, and approximate magnitude.)

Table 4a reveals that the predictive effect of prior math score is quite strong across all ethnic groups.
Generation status is generally non-significant in these models (except for immigrant white students), and is
in places inconsistent in sign. Family SES operates as expected to predict more favorable math trajectories,
although it is not significant in all cases. Students from non-dual-parent homes are predicted to fare less
well, although only a fraction of the coefficients achieve statistical significance. Table 4b echoes the predic‐
tive effect of early school performance (wave 1 math) on dropout probability. Students who did well in 9th
grade math are less likely to drop out (significant for all but blacks). First- and second-generation students
are generally less likely to drop out of high school, with the relationship most pronounced for Asians and
blacks. More favorable family background conditions—higher SES and dual-parent household—generally
pointed to lower dropout. (Interestingly, the SES effect was near zero for Asians.)

Parental engagement carries only substantively statistically significant predictive power over math trajecto‐
ry for Hispanics. For Asians and whites, the coefficient on engagement is smaller in size and non-signifi‐
cant. For blacks the coefficient is negative while non-significant. Asian, black, and white students from
homes with engaged parents were generally much less likely to drop out than their peers (p<.01). Oddly, a
parental engagement effect was in the same direction but not of statistical significance for Hispanic
students.

School-level effects, where they manifest, are inconsistent. For whites and blacks, Catholic school atten‐
dance predicts a superior math trajectory, but for Asians Catholic school attendance (and for peers in free
lunch programs) lowers the trajectory. School-level variables are not usually significant in the dropout
models. Estimation parameters from the multilevel math trajectory model also shed some interesting light
on the school achievement variation by race/ethnicity. Not surprisingly (and consistent with pooled mod‐
els) variance at the individual level dominates variance at the school level. We do find for blacks, however,
that this is the case, and in turn, the intraclass correlation is higher than for the other groups. This would be
consistent with higher error correlation (more similarity) within schools among blacks, perhaps in turn re‐
flective of more sorting or academic segregation not otherwise measured by included covariates in these
models. Due to the variation in sample size for the different racial groups and variances in standard errors,

[Link] 25/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

the results for cross comparison between racial groups should still be interpreted with caution. All told,
parental engagement comes through more strongly in predicting dropout than the trajectory of math. This
is consistent with a behavioral mechanism in which the parents are able to translate their aspirations more
into school attendance itself than into within-school performance.

Conclusion

While previous research has examined the effect of parental engagement on students’ educational outcomes
in general, we used the most recent nationally representative education survey to focus on both the overall
and conditional (including mediating) effects of parental engagement. We look at parental engagement
through a rich set of indicators which include parent-school involvement, parental participation in extracur‐
ricular activities, and parent-child discussions, and our analysis considers a host of family background
characteristics, including race/ethnicity, immigrant generation status, and characteristics of the school it‐
self. Our models take advantage of the longitudinal nature of the data, predicting 11th grade math out‐
comes while controlling for 9th grade math performance, thereby giving a better sense of how the covari‐
ates influence outcomes. Our approach is in the spirit of Coleman, in that we seek to assess the determi‐
nants of stratification in schools and improve our understanding of family social capital. Accordingly, our
models include individual demographic characteristics, conventional family background measures, direct
measures of parental engagement, and school contextual effects.

We find that higher levels of parental engagement predict improvements in students’ math test scores and
prevent students from dropping out of high school. This effect persists even when we control for conven‐
tional educational attainment variables such as race/ethnicity, family SES and school-level contextual fac‐
tors, as well as when interactions of parental engagement with other race/ethnicity/immigrant generation
variables are examined. In keeping with prior studies, we find that students from higher socioeconomic
backgrounds are less likely to drop out and are more likely to follow superior academic trajectories.
Further, we find that greater parental engagement predicts a superior achievement trajectory (as reflected in
standardized math test scores) across the high school years.

School-level effects are also discernable in our empirical results. This is evidenced both by direct covariate
measures (such as school organizational type and school socioeconomic composition) and by the portion of
the variance that remains at the school level. Including these school-level measures reduces the magnitude
of the parental engagement effect but notably does not remove it. All told, these findings add weight to ar‐
guments for the importance of family social capital.

We do find some variation in the impact of parental engagement by ethno-racial groups. Across all four
major groups that we can analyze in the HSLS:2009, we find that greater parental engagement is associated
with lower dropout rates. For this fundamental education outcome, our results are quite robust. Effects of
parental engagement on predicting math trajectory were weaker and inconsistent across the four groups.
Taken together we interpret these findings to mean that parental involvement translates more easily into
children continuing in school, rather than performing well. It also points to potential differences in the way
that school engagement is manifest across different race/ethnic groups and is correlated with other achieve‐
ment variables, such as family socioeconomic background and school type.

[Link] 26/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Parental engagement, as arguably one window into the broader landscape of family social capital, clearly
matters in ways that promote high school achievement. Still, future analysis must do more to understand
how different features of engagement may operate to promote success, and equally important, how parental
aspirations and their associated behaviors might differentially operate across various race/ethnicity/genera‐
tion groups. As schools come to have increasing shares of first- and second-generation minority children, it
is important to understand how the influences of parenting might operate differently among subpopula‐
tions, so that we can develop constructive policies that improve education outcomes for the growing num‐
ber of children of immigrant minorities in contemporary American society.

Acknowledgments

We are grateful to the Population Studies and Training Center at Brown University, which receives funding
from the NIH (P2C HD041020), for general support. A previous version of this paper was presented at the
2016 annual meetings of the Population Association of America in Washington D.C. We thank Margot
Jackson and Susan Moffitt for their helpful comments.

Biographies

Zhen Liu is a Ph.D. Candidate in the Department of Sociology at Brown University and also an adjunct re‐
search fellow at Asia Demographic Research Institute. Her current work examines assimilation of children
in immigrant families in the U.S. and also impacts of internal migration on children in China.

Michael White is Turner Distinguished Professor of Population Studies at Brown University, where he also
directs Spatial Structures in the Social Sciences. On US immigration White is author (with J. Glick) of
Achieving Anew (Russell Sage, 2009). He also edited the International Handbook of Population
Distribution and Migration (Springer 2016).

Appendix 1: Items used to construct index of parental engagement

[Link] 27/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Input Item Definition

Parent school involvement

P1SCHMTG Attended a general school meeting since start of 2009-10 school year

P1PTOMTG Attended a PTO meeting since start of 2009-10 school year

P1PTCONFER Attended parent-teacher conference since start of 2009-10 school year

P1SCHEVENT Attended school event since start of 2009-10 school year

P1VOLUNTEER Served as a school volunteer since start of 2009-10 school year

P1FUNDRAISE Participated in school fund raiser since start of 2009-10 school year

P1COUNSELOR Met with a school counselor since start of 2009-10 school year

Parental participation in extracurricular activities

P1MUSEUM Went to science or engineering museum with 9th grader in last year

P1COMPUTER Worked or played on computer with 9th grader in last year

P1FIXED Built or fixed something with 9th grader in last year

P1SCIFAIR Attended a school science fair with 9th grader in last year

P1SCIPROJ Helped 9th grader with a school science fair project in last year

P1STEMDISC Discussed STEM program or article with 9th grader in last year

P1LIBRARY Visited a library with 9th grader in last year

P1SHOW Went to a play, concert or live show with 9th grader in last year

Parent-Child discussion

S1MOMTALKS 9th grader talked to mother about science courses to take in 20092010

S1DADTALKS 9th grader talked to father about science courses to take in 2009-2010

S1MOMTALKOTH 9th grader talked to mother about other courses to take in 2009-2010

S1DADTALKOTH 9th grader talked to father about other courses to take in 2009-2010

S1MOMTALKCLG 9th grader talked to mother about going to college

S1DADTALKCLG 9th grader talked to father about going to college

S1MOMTALKJOB 9th grader talked to mother about adult jobs/careers

S1DADTALKJOB 9th grader talked to father about adult jobs/careers

S1MOMTALKPRB 9th grader talked to mother about personal problems

S1DADTALKPRB 9th grader talked to father about personal problems

Appendix 2: Mixed-Effects Model Predicting Base-year (9th Grade) Math Test Score

[Link] 28/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Model 1 Model 2 Model 3 Model 4

Parental Engagement 2.353*** 2.251*** 1.521*** 1.437***

(0.173) (0.183) (0.146) (0.146)

Individual Characteristics

Males (ref= Females) 0.105 −0.134 −0.0532

(0.341) (0.317) (0.314)

Race/Ethnicity (ref=Non Hispanic whites)

Non-Hispanic Asian 4.108*** 3.592*** 3.417***

(0.879) (0.831) (0.824)

Non-Hispanic Black −4.081*** −2.854*** −2.700***

(0.595) (0.564) (0.535)

Hispanic −2.545** −1.313 −1.598

(1.057) (0.975) (1.012)

Generation Status (ref= Third+ Generation)

First Generation 1.996** 2.787*** 1.981***

(0.818) (0.762) (0.753)

Second Generation 1.963*** 1.961*** 1.310***

(0.485) (0.463) (0.458)

Family Background Characteristics

2009 (base year) composite SES score 3.331*** 3.007***

(0.329) (0.303)

Family Composition

Single mother −0.311 −0.278

(0.463) (0.479)

Single father −1.738** −1.648**

(0.840) (0.836)

No parent −2.322*** −2.173***

(0.727) (0.705)

School-level characteristics

School type (vs. Public school)

***p<0.01,

**
p<0.05,
*
p<0.1

[Link] 29/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

Note: Results presented in this table are based on complete cases. Weights are used to adjust for missing of parent question‐
aires in the survey, following documentation from NECS. Results are consistent when multiple imputation (20 imputations)
is used to impute missing values. Findings also hold when country of origin is added in the models.

Footnotes

NOTE: This article and associated figures and tables have undergone disclosure review by the IES Data Security
Office, U.S. Department of Education. Approval for dissemination was received on 10 May 2017.

References

1. Abdi Herve. Factor rotations in factor analyses. In: Lewis-Beck M, Bryman A, Futing T, editors. Encyclopedia for research
methods for the social sciences. Thousand Oaks, CA: Sage; 2003. pp. 792–95. [Google Scholar]

2. Alba Richard, Nee Victor. Remaking the American mainstream: Assimilation and contemporary immigration. Cambridge, MA:
Harvard University Press; 2009. [Google Scholar]

3. Bankston Carl L. Social capital, cultural values, immigration, and academic achievement: The host country context and
contradictory consequences. Sociology of Education. 2004;77(2):176–79. [Google Scholar]

4. Bankston Carl L, Min Zhou. Social capital and immigrant children’s achievement. Schooling and Social Capital in Diverse
Cultures. 2002a;13:13–39. [Google Scholar]

5. Bankston Carl L, Min Zhou. Social capital as process: The meanings and problems of a theoretical metaphor. Sociological
Inquiry. 2002b;72(2):285–317. [Google Scholar]

6. Bourdieu Pierre. The social space and the genesis of groups. Theory and Society. 1985;14(6):723–44. [Google Scholar]

7. Carbonaro William J. A Little Help from My Friend’s Parents: Intergenerational Closure and Educational Outcomes. Sociology of
Education. 1998:295–313. [Google Scholar]

8. Coleman James S. What is meant by “an equal educational opportunity?” Oxford Review of Education. 1975;1(1):27–29. [Google
Scholar]

9. Coleman James S. Social capital in the creation of human capital. American Journal of Sociology. 1988;94:S95–S120. [Google
Scholar]

10. Coleman James S. Foundations of social theory. Cambridge, MA: Harvard University Press; 1990. [Google Scholar]

11. Coleman James S. Parental involvement in education: Policy perspective series. Washington, DC: Office of Educational Research
and Improvement of Practice; 1991. [Google Scholar]

12. Coleman James S. The rational reconstruction of society. American Sociological Review. 1993;58(6):898–912. [Google Scholar]

13. Coleman James S, Hoffer Thomas. Public and private high schools: The impact of communities. New York, NY: Basic Books;
1987. [Google Scholar]

14. Coleman James S, Hoffer Thomas. Public and private schools. New York, NY: Basic Books; 1989. [Google Scholar]

[Link] 30/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

15. Coleman James Samuel, Hoffer Thomas, Kilgore Sally. High school achievement: Public, Catholic, and private schools
compared. New York, NY: Basic Books; 1982. [Google Scholar]

16. Coleman James Samuel, U.S. Department of Health, Education, and Welfare . Equality of educational opportunity. Vol. 2.
Washington, DC: Office of Education; 1966. [Google Scholar]

17. Donato Katharine M, Tolbert Charles, Nucci Alfred, Kawano Yukio. Changing Faces, Changing Places: The Emergence of New
Nonmetropolitan Immigrant Gateways. New Faces in New Places. 2008:75–98. [Google Scholar]

18. Dornbusch Sanford M. The sociology of adolescence. Annual Review of Sociology. 1989;15:233–59. [Google Scholar]

19. Fischer Mary J. Immigrant Educational Outcomes in New Destinations: An Exploration of High School Attrition. Social Science
Research. 2010;39(4):627–641. [Google Scholar]

20. Fuller Bruce, Hannum Emily. Schooling and social capital in diverse cultures. Vol. 13. West Yorkshire, UK: Emerald Publishing;
2002. [Google Scholar]

21. Glick Jennifer E, Michael J White. Post-secondary school participation of immigrant and native youth: The role of familial
resources and educational expectations. Social Science Research. 2004;33(2):272–99. [Google Scholar]

22. Goyette Kimberly A, Gilberto Q Conchas. Family and non-family roots of social capital among Vietnamese and Mexican
American children. Research in Sociology of Education. 2002;13:41–72. [Google Scholar]

23. Goyette Kimberly, Xie Yu. Educational Expectations of Asian American Youths: Determinants and Ethnic Differences. Sociology
of Education. 1999:22–36. [Google Scholar]

24. Haller William, Portes Alejandro, Lynch Scott M. Dreams fulfilled, dreams shattered: Determinants of segmented assimilation in
the second generation. Social Forces. 2011;89(3):733–62. [PMC free article] [PubMed] [Google Scholar]

25. Hirschman Charles. The educational enrollment of immigrant youth: A test of the segmentedassimilation hypothesis.
Demography. 2001;38(3):317–36. [PubMed] [Google Scholar]

26. Kao Grace. Asian Americans as model minorities? A look at their academic performance. American Journal of Education.
1995;103(2):121–59. [Google Scholar]

27. Kao Grace. Parental influences on the educational outcomes of immigrant youth. International Migration Review.
2004a;38(2):427–49. [Google Scholar]

28. Kao Grace. Social capital and its relevance to minority and immigrant populations. Sociology of Education. 2004b;77(2):172–
175. [Google Scholar]

29. Kao Grace, Lindsay Taggart Rutherford. Does social capital still matter? Immigrant minority disadvantage in school-specific
social capital and its effects on academic achievement. Sociological Perspectives. 2007;50(1):27–52. [Google Scholar]

30. Kao Grace, Jennifer S Thompson. Racial and ethnic stratification in educational achievement and attainment. Annual Review of
Sociology. 2003;29:417–442. [Google Scholar]

31. Kao Grace, Tienda Marta. Optimism and Achievement: The Educational Performance of Immigrant Youth. The New Immigration:
An Interdisciplinary Reader. 2005:331–343. [Google Scholar]

32. Lee Jennifer, Zhou Min. The Asian American achievement paradox. New York, NY: Russell Sage Foundation; 2015. [Google
Scholar]

[Link] 31/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

33. Lichter Daniel T, Kenneth M Johnson. Immigrant Gateways and Hispanic Migration to New Destinations. International Migration
Review. 2009;43(3):496–518. [Google Scholar]

34. Lichter Daniel T. Immigration and the New Racial Diversity in Rural America. Rural Sociology. 2012;77(1):3–35. [PMC free
article] [PubMed] [Google Scholar]

35. Iceland John. Where We Live Now: Immigration and Race in the United States. Univ of California Press; 2009. [Google Scholar]

36. Logan John R, Charles Zhang. Global neighborhoods: New pathways to diversity and separation. American Journal of Sociology.
2010;115(4):1069–1109. [PMC free article] [PubMed] [Google Scholar]

37. McNeal Ralph B. Parental involvement as social capital: Differential effectiveness on science achievement, truancy, and dropping
out. Social Forces. 1999;78(1):117–44. [Google Scholar]

38. Migration Policy Institute. Stata immigration data profiles. 2017 Available from [Link]
profiles/state/demographics/US (accessed 7 April 2017).

39. Morgan Stephen L, Aage B Sørensen. Parental networks, social closure, and mathematics learning: A test of Coleman’s social
capital explanation of school effects. American Sociological Review. 1999;64(5):661–81. [Google Scholar]

40. Muller Chandra. Parent Involvement and Academic Achievement: An Analysis of Family Resources Available to the Child.
Parents, Their Children, and Schools. 1993:77–113. [Google Scholar]

41. Muller Chandra. Gender Differences in Parental Involvement and Adolescents’ Mathematics Achievement. Sociology of
Education. 1998:336–356. [Google Scholar]

42. Muller Chandra, Christopher G Ellison. Religious involvement, social capital, and adolescents’ academic progress: Evidence from
the National Education Longitudinal Study of 1988. Sociological Focus. 2001;34(2):155–83. [Google Scholar]

43. National Center for Educational Statistics (NCES) High School Longitudinal Survey 2009–2012. 2013 Available from
[Link]

44. Parcel Toby L, Mikaela J Dufur. Capital at home and at school: Effects on child social adjustment. Journal of Marriage and
Family. 2001;63(1):32–47. [Google Scholar]

45. Parcel Toby L, Elizabeth G Menaghan. Family social capital and children’s behavior problems. Social Psychology Quarterly.
1993;56(2):120–35. [Google Scholar]

46. Parcel Toby L, Dufur Mikaela J, Zito Rena Cornell. Capital at home and at school: A review and synthesis. Journal of Marriage
and Family. 2010;72(4):828–46. [Google Scholar]

47. Portes Alejandro. Social capital: Its origins and applications in modern sociology. In: Lesser Eric L., editor. Knowledge and social
capital. Boston, MA: Butterworth-Heinemann; 2000a. pp. 43–67. [Google Scholar]

48. Portes Alejandro. The two meanings of social capital. Sociological Forum. 2000b;15:1–12. [Google Scholar]

49. Portes Alejandro, Rumbaut Rubén G. Legacies: The Story of the Immigrant Second Generation. Univ of California Press; 2001.
[Google Scholar]

50. Portes Alejandro. Downsides of social capital. Proceedings of the National Academy of Sciences. 2014;111(52):18407–18408.
[PMC free article] [PubMed] [Google Scholar]

[Link] 32/33
9/28/24, 3:55 PM Education Outcomes of Immigrant Youth: The Role of Parental Engagement - PMC

51. Portes Alejandro, Patricia Fernández-Kelly. No margin for error: Educational and occupational achievement among disadvantaged
children of immigrants. The ANNALS of the American Academy of Political and Social Science. 2008;620(1):12–36. [Google
Scholar]

52. Portes Alejandro, Lingxin Hao. The schooling of children of immigrants: Contextual effects on the educational attainment of the
second generation. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(33):11920–
11927. [PMC free article] [PubMed] [Google Scholar]

53. Portes Alejandro, Rumbaut Ruben G. Immigrant America: A portrait. Berkeley, CA: University of California Press; 2006.
[Google Scholar]

54. Portes Alejandro, Vickstrom Erik. Diversity, social capital, and cohesion. In: Dustman Christian., editor. Migration: Economic
change, social challenge. New York, NY: Oxford University Press; 2015. pp. 461–79. [Google Scholar]

55. Sakamoto Arthur, Goyette Kimberly A, Kim Chang Hwan. Socioeconomic attainments of Asian Americans. Annual Review of
Sociology. 2009;35:255–76. [Google Scholar]

56. Suhr Diana D. Principal component analysis vs. exploratory factor analysis. SUGI 30 Proceedings. 2005:203–30. [Google
Scholar]

57. Sui-Chu Esther Ho, Willms J Douglas. Effects of parental involvement on eighth-grade achievement. Sociology of Education.
1996;69(2):126–41. [Google Scholar]

58. Teachman Jay D, Paasch Kathleen, Carver Karen. Social capital and the generation of human capital. Social Forces.
1997;75(4):1343–1359. [Google Scholar]

59. Turney Kristin, Grace Kao. Barriers to school involvement: Are immigrant parents disadvantaged? The Journal of Educational
Research. 2009;102(4):257–71. [Google Scholar]

60. Valenzuela Angela, Sanford M Dornbusch. Familism and social capital in the academic achievement of Mexican origin and Anglo
adolescents. Social Science Quarterly. 1994;75(1):18–36. [Google Scholar]

61. White Michael J, Jennifer E Glick. Generation status, social capital, and the routes out of high school. Sociological Forum.
2000;15:671–91. [Google Scholar]

62. White Michael J, Glick E Jennifer. Achieving anew: How new immigrants do in American schools, jobs, and neighborhoods.
New York, NY: Russell Sage Foundation; 2009. [Google Scholar]

63. White Michael J, Gayle Kaufman. Language usage, social capital, and school completion among immigrants and native-born
ethnic groups. Social Science Quarterly. 1997;78(2):385–98. [Google Scholar]

64. Yan Wenfan. Successful African American Students: The Role of Parental Involvement. The Journal of Negro Education.
1999;68(1):5–22. [Google Scholar]

65. Zhou Min, Susan Kim. Community forces, social capital, and educational achievement: The case of supplementary education in
the Chinese and Korean immigrant communities. Harvard Educational Review. 2006;76(1):1–29. [Google Scholar]

[Link] 33/33