Does Capital at Home Matter More Than Capital at School? Social Capital Effects On Academic Achievement

Available online at www.sciencedirect.
com
Research in Social Stratification and Mobility 31 (2013) 1–21
Does capital at home matter more than capital at school?

Social capital effects on academic achievement
Mikaela J. Dufur a,∗ , Toby L. Parcel b , Kelly P. Troutman c
a Department of Sociology, 2008 JFSB, Brigham Young University, Provo, UT 84602, United States
b Department of Sociology and Anthropology, Campus Box 8107, North Carolina State University, Raleigh, NC 27603, United States
c Department of Sociology, 3151 Social Science Plaza, University of California, Irvine, CA 92697, United States
Received 7 December 2011; received in revised form 17 August 2012; accepted 29 August 2012
Abstract
A relatively neglected problem is how individuals derive social capital from more than one context and the extent to which they
benefit from the capital in each. We examine whether social capital created at home and at school has differing effects on child
academic achievement. We hypothesize that children derive social capital from both their families and their schools and that capital
from each context promotes achievement. Using data from the National Longitudinal Education Study and structural equation
modeling, we show that capital from each context is helpful, with social capital in the family more influential than social capital at
school. We discuss the implications of these findings for research on child achievement and for studies of inequality generally.
© 2012 International Sociological Association Research Committee 28 on Social Stratification and Mobility. Published by Elsevier
Ltd. All rights reserved.
Keywords: Social capital; Academic achievement; Families; Schools; Structural equation modeling
1. Introduction It is also well known that resources from multiple

social contexts influence academic attainment among
It is well known that there is variation across ado- adolescents. After classic studies of occupational status
lescents in tested levels of achievement (Farkas, 2003; attainment demonstrated the importance of family back-
Fischer et al., 1996; Parcel & Menaghan, 1994) and that ground and the intervening effects of education on adult
differences in achievement are associated with variations attainment (Blau & Duncan, 1967), scholars developed
in school success (Jencks et al., 1972; Sewell & Hauser, increasingly sophisticated models of why and how edu-
1972). School success, in turn, is a critical predictor of cation was so consequential (Breen & Jonsson, 2005;
occupational and earnings attainment (Farkas, 1996), an Shavit & Blossfeld, 1993). Others studied why various
important component of life in meritocratic society, and aspects of families, such as family structure (McLanahan
consequential to the transmission of inequality across & Sandefur, 1994) and family process (Conger, Conger,
generations. & Martin, 2010) were influential. Theorists such as
Coleman (1988, 1990) pointed to social capital as an
underlying construct influential in both families and
schools that influenced attainment, as well as to the gen-
∗ Corresponding author. Tel.: +1 801 422 1720; erality of the social capital concept in explaining other
fax: +1 801 422 0625. social outcomes. Despite these accomplishments, addi-
E-mail address: mikaela [email protected] (M.J. Dufur). tional questions remain.
0276-5624/$ – see front matter © 2012 International Sociological Association Research Committee 28 on Social Stratification and Mobility. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.rssm.2012.08.002
2 M.J. Dufur et al. / Research in Social Stratification and Mobility 31 (2013) 1–21
Specifically, relatively little research explicitly 1991). Social capital can also be used in the produc-
addresses how family and school contexts affect the tion of group-level goods, as when immigrant groups
accumulation and use of social capital as it is transmit- use within-group ties to promote entire communities
ted to educational outcomes. For example, does social (Portes, 1998). We focus on individual-level creation of
capital pay off regardless of where it is generated? Or and returns to social capital, with investments from the
is the benefit of social capital dependent on the context family and the school serving as key contexts from which
in which the investment is made? In addition, can social children may draw social resources.
capital reflect resources in more than one context? Social capital theory implies purposeful investment
In this paper we address these issues with a study of on the part of the actors. Considering children’s academic
family and school social capital effects on adolescent development, Coleman (1988) argued that the mere pres-
academic achievement. We hypothesize that children ence of parental stores of knowledge is insufficient
derive social capital from their families and from their to ensure transmission of that knowledge to children.
schools and that capital from each context is influential in Instead, parents must make specific choices to invest in
promoting academic achievement. We demonstrate that their children’s development and engage in interactions
social capital at home and social capital at school are with a specific child to create the bonds along which
distinct constructs. We also show that capital from each information can pass. Families who invest expect to see
context is helpful in promoting child academic achieve- higher levels of academic achievement in their children;
ment. Our empirical strategies differ from previous work they undertake these investments with the expectation
in allowing us to evaluate the extent to which specific that strong academic achievement during childhood will
measures of capital actually reflect more than one con- translate into higher levels of education and occupational
text simultaneously, as well as the overall effectiveness attainment in later years, thus promoting upward mobil-
of capital from each context. ity in our stratification system. For this reason, family
We first review literature on social capital that iden- social capital investment has intergenerational conse-
tifies which aspects of families and schools may be quences.
important to study. We then review literature pointing The creation of social capital is not limited to the
to limitations in prior studies of social capital effects family. Social capital associated with schools refers
on academic achievement, followed by details of our to investments between students and schools that can
empirical strategy and findings. We conclude by dis- facilitate educational outcomes. These bonds can reflect
cussing how our strategies can inform questions in the community ties, but typically refer to the relationships
study of inequality more generally by identifying how that parents and children form with school teachers and
resources in other contexts, such as neighborhoods and personnel. For example, arguments regarding the supe-
workplaces, also differentially affect stratification out- riority of Catholic schools are based on the notion that
comes. the religious ties that link many attendees, their families,
and their teachers promote common norms useful in sup-
2. Background porting academic achievement (Coleman, 1988, 1990;
Morgan, Sorensen, & Todd, 2009). In addition, attend-
2.1. Social capital as a concept and a construct ing Catholic schools may promote the formation of social
capital through social closure; parents are more likely to
By social capital we refer to resources that inhere have relationships with other parents whose children also
in the relationships among actors and that facilitate a attend the school. Adults in non-family settings, such as
range of social outcomes (Coleman, 1990). Social cap- coaches and church leaders, also incur costs of investing
ital is contained in relationships among individuals as in children as they transmit knowledge and norms that
well as among corporate actors; it reflects the presence support academic achievement.
of objective ties, as well as subjective relationships con- In addition, Putnam (2000) distinguished between
tingent upon trust, reciprocity, or other positive emotions bonding and bridging social capital. Bonding capital
(Paxton, 1999). In describing social capital, Coleman includes the intra-family connections we have described
(1988) focuses specifically on the information, obliga- above, connections that we argue facilitate academic
tions, and norms that are transmitted through social ties, achievement for children and adolescents. In addition,
resources that help children acquire knowledge from there is considerable evidence that children benefit from
adults. Social capital can be used in the production of the social connections that parents have with others
individual-level goods, such as when a worker recom- such as neighbors, school personnel, and work col-
mends a friend for a job (Kirschenman & Neckerman, leagues (Crosnoe, 2004; Dufur, Parcel, & McKune,
M.J. Dufur et al. / Research in Social Stratification and Mobility 31 (2013) 1–21 3
2008; Johnson, Crosnoe, & Elder, 2001; Parcel & Dufur, (2011) showed that such “shadow education” was tied
2001a, 2001b). These connections illustrate bridging to students’ desires to attend prestigious higher educa-
social capital; the stronger these connections are, the tional institutions. Bassani (2006) studied the effects of
greater the resources to which children have access. social capital at home and at school on math achieve-
Because parents and schools share the responsibil- ment in Canada, Japan and the United States. She found
ity of educating children, both bonding and bridging that supportive family interaction had positive effects in
social capital may be important in promoting academic all three societies, while the effects of social capital at
achievement. school were stronger than those at home in Japan, but
The United States is not the only locale in which weaker than those at home in the United States. More
scholars have studied these issues, often in the con- generally, Ferguson’s (2006) review of studies from sev-
text of analysis of cultural capital as well as social eral countries highlighted the consistency across these
capital.1 For example, using the 1992–1993 Netherlands studies, with investment in social capital tied to positive
Family Survey, De Graaf, De Graaf, and Kraaykamp measures of children’s well-being.
(2000) demonstrated that parental reading behavior has Social capital theory, then, explains the mechanisms
a stronger effect than parental beaux arts participation and processes by which bonds between children and
on children’s educational attainment, particularly when other actors, such as their parents or their teachers,
the parents themselves have low levels of education. produce academic achievement. Adult investment in
Pensiero (2011) argued that it is child engagement in children is more than supervision; it creates the mech-
cognitively stimulating activities, rather than participa- anisms via which children are socialized and educated.
tion in organized activities, that predicts reading ability Family and school environments characterized by low
and locus of control. In addition, parental expectations, levels of social capital will be insufficient to transmit nec-
direct stimulation, parental interactions with the school essary information and knowledge to children, leading
and children’s engagement in cognitively stimulating to lower levels of achievement.
activities mediated more than half of the SES effect on
these two dependent variables, even in the presence of 2.2. Promoting academic achievement: families
strong controls. Notten and Kraaykamp (2010) studied versus schools
over 8300 respondents from the Family Survey Dutch
Population to investigate the role of parental reading There are theoretical reasons to distinguish between
and television viewing, as well as parental guidance on the capital created in the family and that created at school.
children’s reading and television viewing, on child edu- The ties created between parents and children are strong,
cational attainment. They find that parental reading paid the result of repeated and frequent interactions; because
off for children, while parental television viewing was the parent–child bond is one of the most intimate rela-
disadvantageous. In addition, active parental investment tionships in early life, we would expect the social capital
in children’s reading, such as giving children books for created in families to exert a heavy influence on child
holiday gifts, reading to young children and discussing academic outcomes, even into adolescence.
books with them, had a positive effect on children’s edu- The ways parents choose to invest in their connections
cational attainment. with their children, then, can have powerful and long-
Studies of Asian samples take a somewhat different lasting effects. Lareau (2011) argues that middle class
approach. Park, Byun, and Kim (2011) found that parents and working class or poor children experience “unequal
invested considerable resources in locating and moni- childhoods” as a function of differing socialization
toring private tutors for their children. Lee and Shouse strategies at home. Middle class parents use concerted
cultivation, creating a full schedule of activities for their
children to encourage academic development through
1 Some international research studied cultural and social capital
intense parent–child interaction and activities coordi-
together, often seeking to parse the effects of these respective forms
of capital on academic outcomes of children and adolescents (Jager &
nated with schools. In contrast, working class and poor
Holm, 2007; Notten & Kraaykamp, 2010). Our focus differs, in that parents schedule far fewer activities and instead view
we seek to untangle the effects of social capital at home and at school. child development as accomplishment of natural growth.
These lines of research are related, however, in that some studies used They are less likely to actively play with their children,
indicators of what they call cultural capital that others might identify leaving offspring more time to spend in free play. They
as social capital (e.g., Yamamoto & Brinton, 2010). While addressing
cultural capital effects on child cognition is not our purpose here, these
also talk with their children less (Hart & Risley, 1995).
related approaches both offer important insights into the reproduction Still, while parental influences remain strong through-
of social inequality across generations. out the teen years (Amato & Booth, 2000), relationships
with people outside the kin circle become increasingly In addition, the debate about how to empirically
important as children age. We expect that students who handle the potential differences between family and
make additional investments in social capital at school school resources continues. For example, some scho-
by participating in activities that introduce closer rela- lars have argued that social capital as measured by the
tionships to peers, teachers, and other parents would social ties between parents and children’s friends and
see returns to that investment in the form of greater between parents and other adults should be considered an
achievement. In addition, parents who become involved individual-level resource (Carbonaro, 1999; Hallinan &
in their child’s school or who get to know other par- Kubitschek, 1999) while others view it as a school-level
ents are investing in a set of weak ties (Granovetter, resource (Morgan & Sorensen, 1999). While proponents
1983) or in bridging social capital (Putnam, 2000) of both positions make compelling arguments, their
that provide broader access to social resources pro- divergent analytic choices lead to divergent findings,
moting academic achievement than parents who rely with individual-level social capital sometimes positively
only on the strong ties of close-knit kin and neigh- related to academic outcomes and group-level social
borhood circles. Lareau (2011) adds to this argument capital potentially a negative predictor of academic
when she notes that working-class parents are less outcomes. Parcel, Dufur, and Zito’s (2010) review doc-
effective in dealing with schools than middle-class uments these arguments.
parents because they are not as assertive in their chil- This lack of consensus provides little guidance for
dren’s education in making requests of teachers and researchers trying to model the effects of capital on chil-
other school personnel. The discordance between the dren’s achievement. We take this debate as a point of
schools’ expectations for parental involvement and departure. Our approach allows a more explicit test of
the lower levels of this involvement among working whether social capital created at school can be distin-
class/poor families are a hindrance to child academic guished from social capital created in families. Further,
achievement. we can examine whether family or school social capital
However, debate continues as to the degree to which is stronger in its effects on academic achievement.
schools actually influence children’s learning. Heckman
(2008) argues that ability gaps across children develop 2.3. Social capital effects on academic achievement
early in their lives, and that the family environments
of young children are major predictors of later suc- Several studies deal obliquely with the notion that
cess; he also argues that early interventions, including capital from more than one context may be influential
those within families, are more efficacious in promot- in predicting academic outcomes. Teachman, Paasch,
ing positive school outcomes, while later interventions and Carver (1996) posit that social capital protects stu-
such as improved pupil–teacher ratios have weaker dents against dropping out of school. Their analyses
economic returns. Similarly, Grubb (2009) argues that show strong protective effects of social capital, but
school factors vary in their effects on student learning, they consider attendance at a Catholic school and liv-
and favors policies that bolster both families and schools ing in a two-parent family as two indicators of the
in order to promote student academic outcomes. Oth- same idea—social capital—rather than investigating
ers argue that the ways families shape students’ lives whether the site (in a school or at home) or type of
may not be something that even very good schools can investment (bonding between teachers and students or
overcome (cf. Hanuschek, 1994; Morgan & Sorensen, bonding between parent(s) and children) makes a differ-
1999). Some of Coleman’s evidence for the impor- ence. Similarly, McNeal (1999) explicitly looks at both
tance of social capital, such as the effects of attending parent–child discussion and PTA participation as social
Catholic schools or schools with low teacher–student capital in direct contrast to human and financial capital,
ratios, diminish notably once background controls are but does not investigate whether these indicators might
introduced (Alexander & Pallas, 1985; Hallinan & be substantially different from each other because one
Kubitschek, 1999). At the same time, other studies link is created at home and one is created at school. Pong
school social and financial resources to greater achieve- (1998) demonstrates the influence of social capital on
ment (cf. Elliott, 1998; Greenwald, Hedges, & Laine, 10th-grade mathematics and reading achievement as a
1996; Parcel & Dufur, 2001a). The lack of consensus macro-level aggregation of the effects of individual stu-
on school effects underscores the need to more carefully dents’ family capital, but this analysis focuses on the
consider the context from which students receive social need to control statistically for nested levels of data rather
capital when contemplating how to improve educational than distinguishing between contexts in which capi-
outcomes. tal might be created. Family-based indicators, such as
percent of students from a single-parent family or per- outcomes; for example, more positive school social envi-
cent of students who say their parents know their friends ronments help blunt the negative effects of very high
well, and potentially school-based indicators, such as maternal work hours. Similarly, Crosnoe (2004) uses
principal-reported measures of parental involvement, are data from the National Survey of Adolescent Health
all modeled at the school level to try to capture systemic to argue that social capital may be housed in both the
effects. In contrast, we view family-based indicators and family and the school and that those institutions and
school-based indicators as potentially reflecting separate the capital they create may be intertwined to facilitate
constructs. academic achievement. His findings provide additional
Other scholars speculate that social capital exists evidence for the idea that family and school capital
in more than one context but have not fully tested work both independently and together to affect child
this supposition. Swanson and Schneider (1999) sug- outcomes. Finally, Hoffmann and Dufur (2008) find
gest that moving to a new school and moving to a that high-quality schools partly compensate for poor
new home have different consequences. Using NELS parental attachment and low parental involvement in
data and distinguishing among students who change school in preventing juvenile delinquency, especially for
only houses, only schools, or both houses and schools, low-achieving youths.
they find that the type of mobility matters, with What these studies could not demonstrate, however,
school changes exerting important effects. However, is the extent to which a given indicator of capital may
they frame these differences as mobility issues rather actually reflect capital from more than one context. For
than focusing on the varying contexts in which social example, when a parent participates in school organiza-
capital might be accrued. The effects of changing tions, does this reflect family capital? Or does it reflect
schools may be due to the loss of school social the creation and strengthening of capital at school, given
capital, while family social capital in their analyses that the associations parents are building are derived
remains largely intact. Furstenberg and Hughes (1995) from the school context? Or does it reflect both forms of
attempt to examine social capital in two separate spheres, capital, with the potential that such involvement could
within and outside the family, by examining 252 children provide two pathways of influence to promote child
of young mothers. They include both measures of what achievement?
they call “within-family” ties (such as parents and chil- Another limitation of these analyses is that their mod-
dren sharing activities) and “family’s embeddedness in els could not produce a summary measure of a complex
the community” (such as school quality). They find pos- concept like social capital. Previous work, particularly
itive effects of social capital, but conclude in the end studies looking at the effects of family and school social
that more research is required to distinguish adequately capital on child well-being, has tended to use sets of sin-
among potential types of social capital and how those gle indicators to try to capture the effects of social capital.
types affect different outcomes. In sum, these studies For example, Teachman et al.s’ (1996) excellent paper
show that social capital may be tied to many positive explicitly distinguishes among family and school cap-
child outcomes, but do not clearly estimate the relative ital, but uses sets of single indicators in a multivariate
strength of capital effects across contexts. We agree with setting to test that rather than measuring social capi-
Furstenberg (2005), who called upon researchers to more tal as a more global construct that might encompass a
thoroughly study the embeddedness of actors in larger number of investment approaches (see Parcel & Dufur,
social systems in order to measure more accurately social 2001a, 2001b, and others for similar strengths and weak-
capital and understand how actors use it. Our empirical nesses). Such an approach may introduce measurement
strategies allow us to assess whether capital in one con- error and may not be sufficient to investigate the effects of
text is more important than capital in another in affecting the underlying concept of social capital (Paxton, 1999).
child achievement, thus suggesting where interventions In addition, using individual variables restricts the abil-
might be more effective. ity to distinguish clearly between capital at home and
Several analyses that have attempted to model chil- at school, limiting the ways the findings may be used
dren’s social capital guide the current investigation. to craft intervention programs. While previous research
Parcel and Dufur (2001a) argue that capital at home and reveals that more family indicators than school indica-
at school may operate in parallel ways, demonstrating tors were predictive of cognitive development (Parcel &
that both are helpful in promoting reading and mathemat- Dufur, 2001a), the findings do not allow us to conclude
ics achievement among 5–13-year-old children. They definitively that family social capital is more important
find that capital acquired in families sometimes inter- than school social capital in creating positive academic
acts with capital acquired at school to produce desired outcomes. Through use of structural equation modeling,
we address the issue of measurement error and are able to information in student, parent, and teacher behaviors,
include more—and more accurate—information about an especially important point in examining bonding and
the form and function of children’s social capital (Paxton, bridging social capital created across different actors
1999). Our approach to testing whether social capital is and in different settings (Coleman, 1988, 1990; Putnam,
created and used in different contexts also allows a com- 2000). In contrast, other data sets that include data on
parison of the strength of capital effects at home and at both family and school contexts, such as the NLSY
school. This approach has proved useful in the study of Child–Mother data set, contain data from fewer infor-
child behavior problems (Dufur et al., 2008); these strate- mants, thus limiting the richness of the measures. In
gies may also illuminate analyses of child achievement. addition, the data attached to the NLSY Child–Mother
data set lack measures of child or parental attachment
2.4. Hypotheses to the school that are available in the NELS data. Given
that our central focus is on theory testing, the age of the
Our first hypothesis is that social capital is created in data set need not be a major consideration; we address
both the family and at school. We expect that actors will this issue further in Section 7.
create important ties and exchange information with both
family members and school personnel, but that social
4. Methods
capital at school and social capital at home are dis-
tinct constructs. Nonetheless, social capital created in
4.1. Overall analytic strategy
disparate contexts may still be interrelated. Thus, our
second hypothesis is that there are inter-relationships
To assess the structure of social capital, we first
across measures of each form of capital that reflect
perform confirmatory factor analyses using structural
both family and school influence on capital creation.
equation models in AMOS. Confirmatory factor analysis
Such inter-relationships would indicate that social capi-
is the most suitable method to use in this study because
tal created in multiple contexts is closely connected and
it allows us to determine (1) the extent to which each
would work together in some circumstances to promote
individual indicator does or does not reflect social capi-
achievement. Third, we hypothesize that social capital in
tal at home or social capital at school; (2) the frequency
the family and at school will each have positive effects on
and extent to which individual indicators reflect capital
child academic achievement. Fourth, given that several
in both contexts, the family and the school. In addition,
investigations of child outcomes have found a greater
this strategy allows us to use the best fitting models to
number of family than school predictors significant, we
estimate capital effects on academic achievement within
believe that the overall effect of family social capital will
a structural equation format. We also test for correlated
be stronger than that of overall school social capital in
errors between similar items to provide unbiased models.
predicting child academic achievement.
After determining the structure of latent constructs tapp-
ing family social capital and school social capital, we
3. Data investigate the extent to which each of these influences
youth academic achievement. We describe the models
We use data from the second follow-up wave of the in greater detail below. We then add demographic con-
National Education Longitudinal Study (NELS), controls to determine whether the effects of social capital
ducted by the National Center for Educational Statistics. remain strong when additional factors that have been
The NELS is a nationally representative study that gath- shown to influence academic outcomes are considered.
ered data from students, parents, teachers, and school Finally, we subject these models to additional analyses
administrators. The first wave of the study was conducted that demonstrate the robustness of the findings.
in 1988, drawing random samples of approximately
25 eighth-grade students from each of 1000 randomly
selected schools. We employ data from the 12th grade 4.2. Measures
surveys, yielding a sample of 10,585 students. The
NELS has extensive data on schools and schooling for 4.2.1. Dependent measure
a large, nationally representative sample over time, and Table 1 describes the variables used in the analy-
includes many indicators of social capital. The NELS is ses. Our dependent variable, academic achievement, is a
an especially useful data set for this inquiry as it pro- composite standardized test score variable of knowledge
vides potential indicators of social capital from multiple in math, reading comprehension, and science. Students
types of actors. Using the NELS, we can incorporate completed the test during their senior year of high school.
Table 1
Measures and variables.
Variable Description
Dependent variable
Composite test scores Examines math, reading, science, and history knowledge. Scale = 0–100
Capital indicators
Family social capital
Parental trust in child Student report; 0 = only rarely to 6 = all the time
Discuss issues with Three items measuring how often students report discussing (1) school programs, (2) school activities, and (3)
parents school classes with parents (0 = only rarely to 6 = all the time)
Parent checks student’s Parental report; 0 = only rarely to 6 = all the time
homework
Parents attend school Parental report on total number of times; range from 0 to 5 and above
meetings
Parents attend school Parental report on total number of times; range from 0 to 5 and above
events
School social capital
Student participation in Student report; four-item scale that includes athletics, academic clubs, and other school extracurricular activities
extracurricular (α = .74)
activities
School contacts parent Parental report; 0 = never to 5 = five or more times
High teacher morale Teacher report; 0 = low to 5 = high. Higher scores = higher morale
Low conflict between Teacher report; 0 = frequent conflict to 5 = no conflict (reverse coded to reflect capital accumulation)
teachers and
administrators
Teachers respond to Parental report; 0 = never to 5 = often
individual needs
School environment 14-Item scale containing items asking students about the degree to which various school problems, such as
delinquency, violence, and absenteeism, are a problem in their schools (α = .88). Coded so that higher
scores = more positive environment
Controls
Individual-level controls
Child sex 1 = male; 0 = female
Child race 1 = white; 0 = nonwhite
Family socioeconomic Composite of parental education and income, reported by NELS
status
Maternal marital status 1 = married; 0 = not married
Family size Parental report of number of children in sibship
School-level controls (Section 5.3)
Percent free or Administrator report of the percent of students in the school receiving free or reduced-price lunch
reduced-price lunch
Percent low-income Administrator report of the percent of students in the school who qualify as low-income
students
Teacher education Aggregation of teacher reports of level of education received, recoded to years of post-secondary education prior
to aggregation
Highest teacher salary Administrator report of the highest salary paid to any full-time teacher at the school. Responses range in dollars
from 0 to 24,999; 25,000 to 29,999; 30,000 to 34,999; 35,000 to 39,999; 40,000 to 44,999; 45,000 to 49,999; and
50,000+
Teacher experience at Aggregation of teacher reports of how of how many years they have taught in their current school
school
4.2.2. Independent measures: social capital form and function of that collective good is beyond the
In this paper, we focus on individual-level capital scope of this paper. Instead, we utilize variables that indi-
that individual students may accrue through their con- cate potential capital provided either by the adolescent or
tact with key actors in their families and schools. While an adult attached to that specific adolescent. We investi-
we agree that a global or collective form of social capital gated multiple specifications of social capital using more
that students could draw on could be created in an orga- than 65 potential measures of family social capital and
nization like a school, a thorough investigation of the more than 45 measures of school social capital derived
from previous theoretical and empirical work. We used how often the parent checks homework, allowing them
four criteria to select indicators for our models. First, to transmit information about specific academic sub-
the indicators had to reflect at least one aspect of either jects to children; higher scores indicate more contact.
bonding or bridging social capital as outlined in previ- Parental contact with other adults or institutions creates
ous theoretical treatments. Second, they had to exhibit the kinds of closed networks Coleman argues are partic-
acceptable factor loadings on the respective constructs of ularly valuable in encouraging child development. Two
either family or school social capital. Third, their load- items, parental attendance at parent–teacher meetings
ings had to be statistically significant. Finally, inclusion and parental attendance at school events, tap parental
of the indicator had to improve the overall statistical fit reports of interaction with schools. Although we use
of the model. some ordinal variables, all such variables have five or
The indicators included in the models reported in Sec- more response categories, and none are notably skewed;
tion 5.2 generally meet these four criteria. While we find they are therefore appropriate for use in structural equa-
the exclusion of some variables surprising (for exam- tion modeling.
ple, family attendance at cultural events), such variables We model school social capital in the NELS with six
simply were not a good fit to models specifying social items. Looking to tap the networks that connect students,
capital.2 In addition, while data sets examining youth parents, and teachers to each other, building the social
outcomes often include a myriad of school characteris- capital that will allow adults to transmit information
tics, a school characteristic does not necessarily indicate and norms to youth, we measure student participation
school social capital. School characteristics such as aver- in extracurricular activities with a four-item scale that
age standardized achievement scores are likely outcomes includes athletics, academic clubs, and other school
of interest that social capital might predict, but they extracurricular activities (α = .74). To indicate the closed
would not directly reflect the social ties, norm transmis- networks Coleman describes, we also include a measure
sion, or interconnectedness that would be characteristic of how often parents say the school contacts the students’
of social capital. In the case of two variables (noted parents; although it is possible that schools contact par-
below), the indicators meet three of the four criteria; all ents in response to poor performance or inappropriate
others meet all four criteria we establish above. behavior, this variable is positively correlated with our
We measure family social capital with seven indi- measure of academic performance and so we use it as
cators. Previous theory-building in social capital has an indicator of positive contact. Higher scores indicate
argued that trust between individuals and institutions more contact.
is a key indicator of social capital (e.g. Paxton, 1999). We also examine the connections among other actors
We include a variable asking parents how much they at school that could create a store of social capital indi-
trust their children; higher scores indicate greater trust. vidual students could access (Paxton, 1999; Putnam,
Coleman (1988, 1990) argued that adult stores of human 2000). We note again that although the following vari-
capital would have little effect on children’s develop- ables might indicate a more general store of social
ment without the availability of social capital to transmit capital at school that connects students, teachers, and
that knowledge to the children. We are able to tap administrators together, these variables are reported by
parent–child interconnectedness, allowing the flow of individual students and teachers and therefore differ for
information from adult to child, in these data. We each respondent, representing the individual’s percep-
include variables on how often students say they dis- tion of the available social capital. Reports of morale
cuss (1) school programs, (2) school activities, and (3) among school employees and conflict between school
school classes with their parents; higher scores indi- employees, reported by one of the target student’s tea-
cate more discussion. We also use a variable tapping chers, provide an additional indicator of social capital;
high morale and low conflict are linked to greater ties
among adults in the school, providing greater overall
2 Previous examinations of the effects of social capital on adolescent school social capital into which students can tap. We
outcomes using the NELS data have used multiple combinations of reverse-code the original conflict measure so that higher
similar measures (cf. Carbonaro, 1999; Hallinan & Kubitschek, 1999; scores on both variables would indicate greater social
Morgan & Sorensen, 1999; Teachman et al., 1996). Our contribution capital. An indicator of student–teacher contact asks
is not in identifying additional variables that could be used to measure parents whether they feel teachers meet the needs of
social capital; rather, we use previously studied indicators to examine
the ways that capital may be created and used in different contexts
individual students. Higher scores indicate greater con-
and to examine whether the modeling of said indicators in different tact. We also utilize a 14-item scale containing variables
contexts is important in determining the effects of social capital. asking students about their school environment. Better
school environments reflect more general adoption on Findings for both strategies are very similar; we report
the part of students of pro-social norms taught by adults the latter here.
in the school. In addition, we argue that higher scores on
these scales indicate schools where adults and students 4.4. Models
work to create a positive environment and may be more
likely to remain on school grounds and build social ties, An issue of interest in creating the measurement mod-
an approach commonly used in the study of social capital els for social capital in the family and at school was how
in neighborhoods (Putnam, 2000; Sampson, Morenoff, to deal with interaction between parents and their chil-
& Earls, 1999). These include items concerning the dren’s schools. Is such interaction family social capital,
degree to which various issues are a problem in the school social capital, or something else? We model sev-
school: (1) school-related behaviors (such as absen- eral possibilities (Fig. 1): (A) children’s social capital
teeism), (2) delinquent behaviors (such as violence), (3) is most appropriately modeled as one factor, regardless
and social issues (such as racism) (α = .88). We reverse- of where the capital is created or used. If this model
code many of the included items so that higher scores best fits the data, our findings will provide support for
indicate more positive school environments. the approaches most often taken in previous research
that does not distinguish the site in which capital is
4.2.3. Controls created. That is, if this model fits best, we can con-
After the initial model predicting academic achieve- clude that it does not matter where capital is created.
ment, we estimate a second model that includes basic (B) Children’s social capital is created and used in two
demographic controls to ensure the effects of the social distinct contexts, families and schools. (C) Children’s
capital constructs are not spurious. These controls are social capital is most appropriately modeled in multi-
for child sex and race (Parcel & Dufur, 2001a); fam- ple contexts, with one latent construct tapping family
ily size (Downey, 1995; Parcel & Dufur, 2001a); family social capital, one latent construct comprising school
socioeconomic status (Duncan & Brooks-Gunn, 1997); social capital, and a third latent construct reflecting social
and maternal marital status (Kim, 2011; McLanahan & capital created jointly between families and schools.
Sandefur, 1994). Indicators that tap substantial parent–school interaction,
such as student participation in extracurricular activities
4.3. Missing data or parental participation in PTA meetings, might reflect
this joint capital. D) Child social capital is created and
Initial analyses indicated that missing data on some used in two distinct contexts, families and schools, but
of the school variables was of non-trivial proportions, some parent–school interaction variables “cross-load”
with as much as 15% missing on one school administra- onto both latent variables, indicating both separate con-
tor variable. Because many commonly used methods of texts for capital and that capital is created and used jointly
dealing with missing data produce substantial distortion across two institutions. We test all four of these models
in variables’ distributions and unrealistic variances, we and report the best-fitting solutions in Section 5.1.
use two strategies to deal with missing variables. First, We then use the social capital latent variables iden-
we use the maximum likelihood estimation option in tified in the confirmatory factor analyses to predict
AMOS, as maximum likelihood estimation improves the academic achievement. We are able to consider family
accuracy and the power of analyses relative to methods and school social capital as single effects rather than
such as mean substitution or listwise deletion (Enders, examining individual coefficients associated with multi-
2010; Schafer & Graham, 2002). Second, to exert more ple variables, allowing us to use standardized coefficients
control over the utilized equations, we follow Little and to compare the actual strength of capital created in the
Rubin (1987) in using interpolation and imputation to family to that created at school.
deal with cases that have missing data. If a variable is Although our models include some categorical vari-
missing data on fewer than 10% of cases, we use linear ables, most have enough response categories to assume
interpolation to replace missing values. If a variable was unbiased estimates; in addition, estimates we derived
missing data on more than 10% of cases, we interpolate using polychoric correlations did not differ significantly
scores on that variable through regression imputation, (Joreskog & Sorbom, 1993). We therefore report the
creating a regression equation to predict the variable findings from the non-polychoric analyses here as this
that has missing data; we could then use the coefficients allows us to use typical standardized coefficients to indi-
generated by that equation to create unique new values cate size of effect, and most models used in similar
for each case based on other characteristics of that case. literature employ the non-polychoric models, allowing
: Tested
Measurement Models
Model B
Model A
School Social Capital
Family Social Capital
Social Capital
Model C Model D
Fig. 1. Tested measurement models.
for comparison across findings. We report standardized imposes greater penalties for a lack of parsimony than
coefficients, allowing us to compare the size of effects does the AIC. Lower values on the BCC indicate superior
across variables. fit. We also report the root mean-squared error of approx-
imation (RMSEA) to compare observed and predicted
5. Findings covariance matrices and to take into account model com-
plexity; for models such as ours, an RMSEA below .05
5.1. Measurement models indicates a good fit. We also report chi-square statistics
to test across models to determine whether a model fits
Table 2 reports fit statistics for each of our four better than others.
potential models. We use multiple fit indices to more Our findings indicate children’s social capital is best
fully evaluate the model (Browne & Cudeck, 1993; represented as two separate latent factors, one reflecting
Tanaka, 1993). We report the comparative fit index social capital created in the family and one reflecting
(CFI), which measures the fit of the models as compared social capital created at school. Although this finding
to the null model; a CFI approaching one indicates a would be consistent with either Model B or D, the best
good fit. The Akaike Information Criterion (AIC), by fitting model is Model D, or the model in which indi-
contrast, can make comparisons across nonnested mod- cators could “cross-load” onto both latent social capital
els. Smaller AICs indicate a better fit. Similarly, the constructs. In this case, one indicator—student participa-
Bayesian Information Criterion (BIC) (Raftery, 1996) tion in extracurricular activities—cross-loads onto both
is an approximation to the log of a Bayes factor for the family social capital and school social capital. As Table 2
experimental model compared to the saturated model, indicates, this cross-loading model is a better fit to the
favors parsimonious models and is recommended with data, with a lower RMSEA (.045), a CFI closer to one
large sample sizes, as we have here. Again, lower val- (.877), and lower values on AIC, BCC, and BIC scores
ues on the BIC indicate superior fit. Because the AIC than competing models. Although we note that each of
and BIC may be sensitive to large sample sizes, we also the models has a significant chi-square, potentially indi-
include the Browne–Cudeck Criterion (BCC), which cating poor fit, because the chi-square is a comparison of
Table 2
Comparison of social capital measurement models’ fit statistics for Model A (one latent variable), Model B (two latent variables; family social
capital and school social capital), Model C (three latent variables; family social capital, school social capital, and joint family–school capital), and
Model D (two latent variables—family social capital and school social capital—with shared indicators)—NELS.
Model X2 CFI RMSEA AIC BCC BIC
Model A 7329.666*** (df = 75) .476 .072 7417.67 7417.78 7609.82

Model B 5483.571*** (df = 129) .847 .050 2281.99 2282.11 2474.14
Model C 2419.817*** (df = 76) .583 .116 2380.00 2380.32 2690.63
Model D 1783.16*** (df = 74) .877 .045 1873.16 1873.28 2072.65
N = 10,585.
*p < .05 two-tailed tests.
**p < .01 two-tailed tests.
*** p < .001 two-tailed tests.
a model to a “perfect fit,” even a slight deviation from a criteria. One correlation between error terms is signif-
perfect fit can induce a significant score, especially in the icant: the errors terms associated with participation in
presence of a large sample size (cf. Paxton, 1999). Likeli- school activities and school contacting the parent are
hood ratio tests between Model B and Model D, in which significantly correlated, possibly because participation
Model B is nested, indicate that Model D fits the data sig- in an extracurricular activity requires substantial coor-
nificantly better than does Model B (F = 410.83***; df dination between parents and school personnel. These
1). Thus, the model that best fits the data suggests two results provide support that all of these measures reflect
separate forms of social capital, one in the family and school social capital.
another at school, with variables able to indicate capi- The indicator of participation in extracurricular
tal in both settings, providing affirmative evidence for events loads onto both the family social capital measure
Hypotheses 1 and 2. and the school social capital measure when both forms of
In examining family social capital (see Fig. 2), capital are included simultaneously. Although the factor
the standardized coefficients are all significant and in loading for extracurricular participation on school social
the expected direction: more discussion with offspring, capital is small, it is significant, the model fit is superior
attendance at school events, and supervision of home- when it is included, and the connection is theoretically
work can be interpreted as reflecting greater social supported. In other words, when modeled together, fam-
capital in the family. Parental trust that the student will ily social capital and school social capital as measured
do what is expected has a small but significant effect; in the NELS share an indicator, notably one that might
models without this indicator are not as good a fit to the require active involvement of both parents and schools in
data. Error terms are significantly correlated between order for successful student participation. These findings
parental attendance at school meetings and at school provide some support for the idea that although the mea-
events, a finding we would expect given that the ques- surement of youths’ social capital must take into account
tions are structured in an almost identical fashion and multiple contexts, indicators of social capital may be
any error associated with how the first item is answered shared across contexts.
is likely to be similar to the error associated with answer-
ing the second. These findings provide support for our 5.2. Explanatory models predicting academic
arguments that these indicators are all good measures of achievement
family social capital.
Similarly, standardized coefficients are all significant We now estimate the effects of these respective forms
and in the expected direction when looking at school of capital on student academic achievement by using
social capital: more parental contact with the school; these constructs within path analytic models. The find-
better school environments; and better relations among ings in Fig. 3 show that both family social capital and
parents, teachers, and administrators all can be inter- school social capital are positive, significant predictors
preted to reflect greater social capital at school. Although of test scores. As predicted, the standardized effect of
the effect of school contacting parent is small as indicated family social capital (.371) is stronger than the effect
by the standardized coefficients, the effect is significant, of school social capital (.106). This difference provides
and the model better fits the data with that indicator evidence in favor of the hypothesis that capital in the fam-
included. Thus, this measure meets three of our four ily is more influential for children’s achievement than is
Figure 2: Measurement Model with

Family Social Capital and School
Social Capital (NELS)
0.051***
e e e e e
e
Extracurricular School contracts School Student/ Teacher/ Teacher morale Teachers meet
activities parent environment teacher ratio administrator conflict individual needs
Discuss school Discuss school Parents check Parents attend Parents attend
Discuss classes Parental Trust
programs activities homework meetings school events
e e e e e e e
0.154***
CFI 0.877 RMSEA 0.045
Fig. 2. Measurement model with family social capital and school social capital (NELS).
capital at school; however, school capital is still a signifi- social capital is a more important predictor of child aca-
cant positive factor in promoting academic achievement demic achievement than is school social capital, net of
even in the presence of family social capital. The background controls.
remaining epistemic correlations are similar across the
measurement and explanatory models. Goodness of fit 5.3. Additional analyses
statistics indicate that this model is a moderately good
fit to the data (CFI = .850; RMSEA = .049); we note, We also ran several robustness checks to strengthen
however, that our goal in this paper is testing whether our confidence in the findings showing that youth social
contextualized measures of social capital can be used to capital should be measured in multiple contexts and that
explain outcomes such as academic achievement rather social capital at home is more important than social cap-
than explaining all variance in that outcome. ital at school in predicting academic achievement. First,
We test models that introduce basic demographic we consider the possibility that demographic statuses
controls (race, family socioeconomic status, child sex, such as family SES or race might affect the creation of
family size, and maternal marital status) (Fig. 4). In family social capital. While previous work has shown
these models, the effects of family social capital and that the effects of family and school social capital per-
school social capital diminish, in the case of school sist net of family and school human and financial capital,
social capital by more than half, but remain positive, it is unclear to what extent other resources may have
significant predictors of test scores, net of controls. indirect effects that operate through social capital (cf.
These findings provide evidence that modeling social Parcel & Dufur, 2001a). For example, if modeling an
capital as reflecting separate, distinct constructs can be indirect effect of family socioeconomic status through
useful in explaining academic achievement. They also family social capital renders the direct effect of fam-
suggest that controlling for these background character- ily social capital on test scores nonsignificant, then our
istics, while important for understanding both academic conclusions should stress the indirect effect of family
achievement and how social capital affects it, does not socioeconomic status on test scores working through
interfere with our major conclusion. Specifically, family family social capital. In this case, significant indirect
Explanatory Model with

Social Capital Predicting Test
Scores (NELS)
0.050***
e e e e e
e
Standardized test
scores
e e e e e e e
0.031***
CFI 0.850 RMSEA 0.049
Fig. 3. Explanatory model with family social capital and school social capital predicting test scores (NELS).
effects might reveal that the relationship between fam- indirect effects simultaneously, so each effect was tested
ily social capital and the dependent variable is largely individually.
a mechanism through which basic status characteristics Table 3 shows the results of these analyses. The
influence academic outcomes, or even that the effect of first column lists the tested effects. The second column,
family social capital is spurious. Model 1, shows the effect of family social capital and
To test this possibility, we model indirect effects demographic controls on test scores without the inclu-
through family social capital of race, family SES, child sion of indirect effects (in other words, our original
sex, maternal marital status, and family size. Each of model). Each subsequent pair of columns represents the
these variables could potentially affect the creation effects of a different test of indirect effects; the first col-
of family social capital. For example, better educated umn in each pair displays the coefficients from a model
parents might have greater awareness of the impor- including one indirect effect of a demographic control,
tance of helping their offspring with homework, one while the second shows the change in coefficients com-
of our indicators of family social capital, and there- pared to the models without the indirect effect. For
fore might be more willing or able to build capital in example, the third column, Model 2, shows coefficients
that way. Greater income might be a proxy for more from the model where we test the indirect effects of race
flexible work hours, allowing parents to interact with on test scores through family social capital. The fourth
the school more freely. Unfortunately, model identifi- column shows the difference in the coefficients between
cation issues made it impossible to test all of these Model 1 and Model 2, and so forth. The first row of
14
Table 3
Tests of indirect effects of individual background control variables on composite test scores through family social capital; standardized coefficients.
M.J. Dufur et al. / Research in Social Stratification and Mobility 31 (2013) 1–21
Effect of control Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
variable
Model White → Change SES → Change Male → Change Mother Change Family Change
with no family from M1 family from M1 family from M1 married → from M1 size → from M1
indirect social to M2 with social to M3 with social to M4 with family to M5 with family to M6 with
effects capital indirect capital indirect capital indirect social indirect social indirect
effect effect effect capital effect capital effect
Control → family NA .167*** NA .395*** NA −.112*** NA .226*** NA −.198*** NA

social capital
Family social .209*** .206*** −.003 .216*** .007 .206*** −.003 .208*** −.001 .207*** −.002
capital → test
scores
White → test scores .102*** .089*** −.013 .099*** −.003 .102*** .000 .115*** .013 .101*** −.001
SES → test scores .407*** .403*** −.004 .361*** −.046 .405*** −.002 .382*** −.035 .391*** −.016
Male → test scores −.081*** −.082*** .001 −.087*** .006 −.085*** .004 −.087*** .006 −.083 .002
Mother .255*** .240*** −.015 .250*** −.005 .256*** .001 .266*** .011 .249*** −.006
married → test
scores
Family size → test .312*** .311*** −.001 .319*** .007 .309*** −.003 .318*** .006 .304*** −.008
scores
N = 10,585.
*p < .05.
**p < .01.
*** p < .001.
Explanatory Model with

Social Capital Predicting Test
Scores (with controls) (NELS)
0.388***
e e e e e
e

e
Male
Parents e
married
Standardized test White e

scores
0.081***
Socio-
economic e
status
Family e
size
e e e e e e e
0.033***
CFI .801 RMSEA 0.056
Fig. 4. Explanatory model with family social capital and school social capital predicting test scores (with controls) (NELS).
coefficients shows that all five demographic variables we are included. The direct effects of the other controls on
report here exercise statistically significant direct effects test scores change only slightly in each model. Statisti-
on achievement, as expected and as noted previously in cally, then, inclusion of indirect effects does influence
Fig. 4. The second row shows that the effect of family the direct effect of family social capital on test scores;
capital on test scores remains strong in the presence of practically, however, these changes are negligible, sug-
modeled indirect effects, with only very small changes gesting that even after controlling for the ways family
once these indirect effects are taken into account. Family background influences family social capital, that form
social capital, then, exerts significant effects on academic of capital still has a positive association with academic
achievement even in the presence any indirect effects of achievement.
child race and sex, family SES or size, and maternal We conducted a second robustness check concern-
marital status. Subsequent rows display coefficients and ing school characteristics to strengthen confidence in
changes in them across models for the demographic con- our findings. Recall that our focus has been on school
trols themselves; changes across models are minimal. social capital measured at the individual level. How-
Thus, inclusion of these indirect effects creates very lit- ever, it remains a possibility that school-level factors may
tle change in the strength of the effect of family social influence academic achievement. For example, the per-
capital on test scores when any of the indirect effects centage of students receiving free or reduced-price lunch
and teacher qualifications/expectations have sometimes
to M6 with
from M1
Change
been found to influence test scores (Clotfelter, Ladd, &
school
effect
.002
.006
Vigdor, 2007; Roscigno & Crowley, 2001). It is possible
NA
that the effect of school social capital will no longer be
experience
significant net of such contextual variables. Thus, we test
at school
Model 6
Teacher
.081***
.205***
.047***
the possibility that the effect of individual-level school
social capital is spurious by including school-level con-
textual measures in the model.
to M5 with
Because school context variables were measured at
from M1
Change
school
the school, rather than the individual, level, these tests
effect
.009
.008
NA
require multilevel structural equation models, with the
dependent variable, family social capital and school
social capital as measured by individuals’ responses,
(highest)
Model 5
Teacher
.067***
.212***
.049***
salary
and individual-level controls (race, sex, etc.) at Level
1 and school characteristics at Level 2. We measure
school context with five variables: percent of students
to M4 with
receiving free or reduced-price lunch, percent of stu-
from M1
Change
school
effect
dents from low-income families, percent of teachers at
.005
.005
NA
the school with master’s degree or above, teacher pay
(highest salary), and teacher experience at current school
education
(see Table 1 for measurement details on school-level
Model 4
Teacher
.208***
.063**
.046**
variables).3 We note, as above, that model identification
issues make it impossible to include all of these variables
in the model simultaneously. In addition, inclusion of
to M3 with
school-level variables excludes a large number of cases from M1
Change
school
effect
.004
.004
from the sample; taken together, these limitations mean
NA
that the following findings should be interpreted with
caution. We use the TWOLEVEL analysis protocol and
Model 3
Tests of school-level control variables on composite test scores; standardized coefficients.
students
income
maximum likelihood estimation in MPlus to test these
.207***
.045***
% low
models. .038*
Results of these analyses are displayed in Table 4.
to M2 with
The first column lists the effects being tested. The sec-
from M1
Change
ond column, Model 1, shows the effect of family and

school
effect
.005
.002
NA
school social capital on test scores without the inclu-

sion of school-level controls (in other words, our original
% students
model). Each subsequent pair of columns represents the

on free or
reduced-
Model 2
.208***
.043***
effects of a different school-level social capital mea-

lunch
.044*
price
sure on achievement; the first column in each pair

displays the coefficients, while the second shows the
model (no
change in coefficients compared to the models without

controls)
Model 1
school-
Single-
.203***
.041***
the school-level variable. For example, the third col-

level
level
NA
umn, Model 2, shows coefficients from the model where

we include the percent of students in the school who
Family social capital → test scores
School social capital → test scores
School-level control → test scores
receive free- or reduced-price lunch. The fourth column
3 We ran the same tests using lowest teacher salary and teacher expe-
rience in any secondary school setting and found similar results, so we

do not report them here for the sake of parsimony. We also tested
*** p < .001.
urban school setting, type of school (public or private), and school size
* p < .05.
** p < .01.
N = 5675.
as measured by student/teacher ratio, but as none of these exerted a

Table 4
significant effect on composite test scores and their effects on family

and school social capital were negligible, we do not report them here.
shows the difference in the coefficients between Model investments on the part of parents, who often have to
1 and Model 2, and so forth. Row 1 shows that all arrange transportation, help raise funds, and chaperone
six school-level capital variables we report here exer- events, bringing the parents into greater contact with
cise statistically significant effects on achievement, as their own children, their children’s peers, and other par-
mentioned above. Rows 2 and 3 show that the effects ents and teachers. It is possible that activities or situations
of individual-level family social capital and individual- requiring heavy investment from two parties are best
level school social capital change only very slightly as considered as being created across multiple contexts.
each of the school-level controls is introduced to the Consistent with our third hypothesis, we also demon-
model. Changes in the effects of both family social cap- strate that capital from two distinct contexts, families
ital and school social capital are also much smaller than and schools, positively affects academic achievement.
when individual-level controls were introduced, as pre- Our confirmatory factor analysis reveals the extent to
vious research would suggest (see Figs. 3 and 4). Both which variables reflect the underlying concept of social
forms of capital, then, exert significant effects on aca- capital, better enabling scholars to tie the specifics of
demic achievement even in the presence of school-level social capital theory to everyday social interactions both
characteristics. Taken together, the robustness checks at home and at school. As Coleman (1988) predicted,
summarized in Tables 3 and 4 increase our confidence investment in the social ties that pass along information
that family and school social capital are built in different and create obligations and shared norms helps transmit
ways in different contexts, that both forms of capital are human capital to the next generation.
associated with higher academic achievement, and that In addition, this analytic strategy allows investigators
family social capital exerts stronger effects on student to assess the relative strength of social capital in various
achievement than does school social capital. contexts, an approach that has value for scholars and pol-
icymakers alike. Consistent with our fourth hypothesis,
6. Discussion we found that family social capital exerts stronger effects
on academic achievement than does school social capital,
Our goal in this paper is to investigate whether consid- a finding that clarifies ideas about sites of social invest-
ering the context in which social capital is created and ment and may aid practitioners in crafting intervention
used refines our understanding of how investment in policies that focus resources where they will have the
social resources affects child academic achievement. greatest benefit. These findings highlight the importance
Because test scores are associated with later school of the social capital children experience in the home as
success, inferences regarding where resources are most critical to promoting child development. In addition, we
influential are important to those concerned with how note that a comparable analysis of child behavior prob-
inequality is transmitted across generations. We argue lems also demonstrated that families exerted stronger
that social capital is not a monolith, but rather that invest- effects than schools; hence, the evidence appears to be
ments in different social sites may potentially lead to accumulating across these two dependent variables that
different returns. Our analytic strategies help us move the location in which social capital is created matters in
beyond prior research in several ways. terms of promoting positive child development (Dufur
Consistent with our first hypothesis, we demonstrate et al., 2008).
that children’s social capital is site-specific. In doing
so, we improve upon prior studies that were unable to 7. Limitations, future research, and implications
attend to the context from which social capital indica- for the study of inequality
tors are derived. There is less support for our second
hypothesis regarding social capital created jointly across A limitation to our analysis is that we have not studied
contexts. Only one item, participation in extracurricular the role of financial capital at home versus financial cap-
activities, acts as an indicator of social capital in both ital in schools, nor have we studied the relative effects
contexts. While this may suggest that children’s social of human capital at home compared to human capital at
capital is created and used in two very distinct con- school in their effects on child achievement. Although
texts, we note that this particular variable is indicative these analyses were beyond the scope of this project,
of shared effort between families and schools. Partic- we have controlled for aspects of parental human and
ipation in extracurricular activities requires clear and financial capital in our models, thus reducing the risk of
considerable investments by school personnel, including specification error. In addition, our models also provide a
time spent coaching, supervising, or otherwise training framework for studying the effects of these other forms
students. Similarly, these activities require considerable of capital in future investigations. Such studies would
provide additional information as to whether our con- both in terms of how much capital individuals may
clusions regarding the relative strength of social capital access and in where individuals might create their cap-
at home and at school would be replicated for these ital. While the investigation of how capital might shift
other forms of capital. If we were to find, for example, across time is beyond the scope if this paper, the baseline
that parental financial capital (e.g., parental earnings) we provide here—that youth capital is created in multi-
were more important than financial capital at school ple contexts—provides a stepping stone for examining
(e.g., per pupil expenditures), and that parental human those shifts in the future.
capital (e.g., parental education) were more important Our findings also have implications for scholars who
than schools’ human capital (e.g., teacher qualifica- are particularly concerned with the role of schools in
tions), these findings would provide additional evidence reproducing inequality. Schools in the United States have
regarding the relative importance of families and schools traditionally been asked to do many things, from integrat-
in reproducing inequality. Most likely, however, the story ing immigrants into society to ameliorating inequality
is much more complex (see Grubb, 2009). Finally, in our (Cheadle, 2008; Cuban, 2010). Current debates about
additional analyses, we were unable to simultaneously school accountability dramatize the tendency for many
control for the measures of indirect and school-level to argue that if student achievement is lacking, the solu-
effects included in Tables 3 and 4. It is possible that some tion is to “fix” schools, despite decades of evidence
of those specific findings could change if such analyses from social science suggesting that family influences are
could be conducted. very consequential (Coleman et al., 1966; Cuban, 2010;
Nonetheless, our theoretical model can be extended Parcel et al., 2010). In addition, because of the primacy
to contexts beyond families and schools to more collec- and privacy of family life, many may be reluctant to
tive stores of social capital. An obvious connection is advocate for strategies that would encourage the creation
to the excellent work by criminologists looking at the of greater social capital in the family, preferring instead
capital available in neighborhoods (cf. Sampson et al., to encourage investment in capital at school. Our findings
1999). Although the data we use here do not allow for a in conjunction with previous research (Dufur et al., 2008)
detailed examination of neighborhood social capital, it suggest that efforts to increase social capital at school,
is possible that youth may collect stores of social capital such as initiatives to reduce class size or attempts to
in families, schools, and neighborhoods, as well as other create parent–school programs and ties, would probably
settings such as churches or sports teams. Schools, for have a beneficial effects on students. However, we also
example, might be a site where collective stores of social find that family social capital has a stronger influence
capital are available for children to draw upon even if on child achievement than does school social capital,
they or their own teachers or parents do not help to cre- even after controlling for socioeconomic status and other
ate that capital. While beyond the scope of this paper, demographic characteristics. It remains possible that the
future research could examine the way social capital is relative roles of the family and schools have changed
created and used collectively at the school level (such as over the years since the NELS data were produced. This
through principal reports of global school capital) and is a possibility that future research should investigate.
whether such capital exerts a positive influence on aca- Nonetheless, looking at these implications highlights
demic achievement beyond the connections individual the importance of acknowledging how social capital may
students make at their schools. be constructed in different spheres and the consequences
We also speculate that families influence academic of investment in each of those spheres. Thus, although
achievement for both younger children and adolescents, investment in school social capital may be beneficial, our
but that other institutions such as schools exercise more results suggest that investment in family social capital
influence as children mature. In our study all children would reap greater returns. For example, social policies
were with a year or two of the same age, thus precluding and interventions focusing on increasing academic out-
our ability to investigate whether social capital at school comes could be tailored to allow greater investment in
would matter more for older as compared with younger both family and school social capital. One possibility
children, a worthy topic for future research. In addition, is expansion of supportive workplace policies, such as
peer groups likely become more important as children flextime, that would allow parents to attend school meet-
mature (Amato & Booth, 2000), another topic worthy ings and participate in extracurricular activities with their
of investigation. Again, the models we have estimated children without adversely affecting their jobs. Policies
are flexible enough to use in testing such hypotheses. such as job-sharing, or encouraging part-time work, that
Studying children across age groups also brings up the allow parents more time to discuss school with their
possibility that stores of capital may fluctuate over time, children or do homework together might build family
social capital and, therefore, have a greater effect on scores in Canada, Japan, and the United States. International Jour-
child academic achievement than policies focusing on nal of Educational Research, 45, 380–403.
school social capital. However, most families need full- Blau, P., & Duncan, O. D. (1967). The American occupational struc-
ture. New York: Free Press.
time work for one or two adults, thus rendering these Breen, R., & Jonsson, J. O. (2005). Inequality of opportunity in com-
strategies less practical than flextime in many cases. parative perspective: Recent research on educational attainment
Communicating to parents how efficacious family social and social mobility. In K. S. Cook, & D. S. Massey (Eds.), Annual
capital can be in promoting positive child outcomes may review of sociology (pp. 223–243). Palo Alto: Annual Reviews.
also be worthwhile (Tough, 2009). Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing
model fit. In K. A. Bollen, & J. S. Long (Eds.), Testing structural
Finally, our findings suggest the need to study the equation models (pp. 136–162). Newbury Park, CA: Sage.
effects of social capital in other contexts, particularly Carbonaro, W. J. (1999). Opening the debate on closure and schooling
when studying inequality among adults. For example, outcomes: Comment on Morgan and Sorenson. American Socio-
scholars studying labor markets note that weak ties, logical Review, 64, 682–686.
or those other than with family, can help explain who Cheadle, J. E. (2008). Educational investment, family context, and
children’s math and reading growth from kindergarten through the
gets good jobs (Granovetter, 1983; Kirschenman & third grade. Sociology of Education, 81, 1–31.
Neckerman, 1991). Studies of immigrant experiences Clotfelter, C. T., Ladd, H. F., & Vigdor, J. L. (2007). Teacher credentials
show how social capital in the destination community and student achievement: Longitudinal analysis with student fixed
helps new immigrants become successful more quickly effects. Economics of Education Review, 26, 673–682.
than those lacking ties and associations (Portes, 1998; Coleman, J. S. (1988). Social capital in the creation of human capital.
American Journal of Sociology, 94, S94–S120.
Raijman & Tienda, 1999). In both of these examples, Coleman, J. S. (1990). Foundations of social theory. Cambridge, MA:
the context in which social capital is located and used Harvard University Press.
appears to be a critical component in gaining desired Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J. M.,
outcomes. However, to date studies have not systemati- Mood, A. M., Weinfeld, F. D., et al. (1966). Equality of educational
cally demonstrated which contexts are most important in opportunity. Washington, DC: Government Printing Office.
Conger, R. D., Conger, K. J., & Martin, M. J. (2010). Socioeconomic
affecting which aspects of inequality, and under which status, family processes, and individual development. Journal of
conditions. Our study develops one model of how such Marriage and Family, 72, 685–704.
investigations might proceed. We look forward to the Crosnoe, R. (2004). Social capital and the interplay of families and
development of a broad body of evidence regarding the schools. Journal of Marriage and Family, 66, 267–280.
relative importance of several contexts in the reproduc- Cuban, L. (2010). As good as it gets: What school reform brought to
Austin. Cambridge, MA: Harvard University Press.
tion of social inequality. De Graaf, N. D., De Graaf, P. M., & Kraaykamp, G. (2000). Parental
cultural capital and educational attainment in the Netherlands: A
refinement of the cultural capital perspective. Sociology of Educa-
Acknowledgments tion, 73, 92–111.
Downey, D. B. (1995). When bigger is not better: Family size,
An earlier version of this paper was presented at parental resources and children’s educational performance. Amer-
the 2005 Annual Meetings of the American Sociolog- ican Sociological Review, 60, 746–761.
Dufur, M. J., Parcel, T. L., & McKune, B. A. (2008). Capital and
ical Association, Philadelphia. We thank the College of context: Using social capital at home and at school to predict
Family, Home and Social Sciences at Brigham Young child social adjustment. Journal of Health and Social Behavior,
University for partial support of this research, Ben McK- 49, 146–161.
une for assistance with graphs, and Tim Heaton, John Duncan, G. J., & Brooks-Gunn, J. (1997). Consequences of growing
Hoffmann, and Pamela Paxton for helpful comments. up poor. New York: Russell Sage Foundation.
Elliott, M. (1998). School finances and opportunities to learn: Does
Additional support was provided by the College of Social money well spent enhance student achievement? Sociology of Edu-
and Behavioral Sciences at the Ohio State University. cation, 71, 223–245.
Enders, C. K. (2010). Applied missing data analysis. The Guilford
Press: New York.
References Farkas, G. (2003). Cognitive skills and noncognitive traits and behav-
iors in stratification processes. In J. Hagan, & K. Cook (Eds.),
Alexander, K. L., & Pallas, A. M. (1985). School sector and cognitive Annual review of sociology (pp. 541–562). Palo Alto, CA: Annual
performance: When is a little a little? Sociology of Education, 58, Reviews Inc.
115–128. Farkas, G. (1996). Human capital or cultural capital? Ethnicity and
Amato, P. R., & Booth, A. (2000). A generation at risk: Growing up poverty groups in an urban school district. New York: Aldine de
in an era of family upheaval. Cambridge, MA: Harvard University Gruyter.
Press. Ferguson, K. M. (2006). Social capital and children’s well-being: A
Bassani, C. (2006). A test of social capital theory outside of the Amer- critical synthesis of the international social capital literature. Inter-
ican context: Family and school social capital and youths’ math national Journal of Social Welfare, 15, 2–18.
Fischer, C. S., Hout, M., Jankowski, M. S., Lucas, S. R., Swidler, A., Morgan, S. L., & Sorensen, A. B. (1999). Parental networks, social
& Voss, K. (1996). Inequality by design: Cracking the bell curve closure, and mathematics learning: A test of Coleman’s social cap-
myth. Princeton, NJ: Princeton University Press. ital explanation of school effects. American Sociological Review,
Furstenberg, F. F. (2005). Banking on families: How families generate 64, 661–681.
and distribute social capital. Journal of Marriage and Family, 67, Morgan, S. L., Sorensen, A. B., & Todd, J. J. (2009). Intergenera-
809–821. tional closure and academic achievement in high school: A new
Furstenberg, F. F., & Hughes, M. E. (1995). Social capital and success- evaluation of Coleman’s conjecture. Sociology of Education, 82,
ful development among at-risk youth. Journal of Marriage and the 267–286.
Family, 57, 580–592. Notten, N., & Kraaykamp, G. (2010). Parental media socialization
Granovetter, M. (1983). The strength of weak ties: A network theory and educational attainment: Resource or disadvantage? Research
revisited. Sociological Theory, 1, 201–233. in Social Stratification and Mobility, 28, 453–464.
Greenwald, R., Hedges, L. V., & Laine, R. D. (1996). The effect of Parcel, T. L., & Menaghan, E. G. (1994). Parents’ jobs and children’s
school resources on student achievement. Review of Educational lives. New York: Aldine de Gruyter.
Research, 66, 361–396. Parcel, T. L., & Dufur, M. J. (2001a). Capital at home and at school:
Grubb, W. N. (2009). The money myth: School resources, outcomes Effects on student achievement. Social Forces, 79, 881–912.
and equity. New York: Russell Sage. Parcel, T. L., & Dufur, M. J. (2001b). Capital at home and at school:
Hallinan, M. T., & Kubitschek, W. N. (1999). Conceptualizing and Effects on child social adjustment. Journal of Marriage and Family,
measuring school social networks: Comment on Morgan and 63, 32–47.
Sorensen. American Sociological Review, 64, 687–693. Parcel, T. L., Dufur, M. J., & Zito, R. C. (2010). Capital at home and at
Hanuschek, E. A. (1994). Money might matter somewhere: A response school: A review and synthesis. Journal of Marriage and Family,
to Hedges, Laine, and Greenwald. Educational Researcher, 23, 72, 828–846.
5–8. Park, H., Byun, S., & Kim, K. (2011). Parental involvement and stu-
Hart, B., & Risley, T. R. (1995). Meaningful differences in the every- dents’ cognitive outcomes in Korea: Focusing on private tutoring.
day experience of young American children. Baltimore: Paul H. Sociology of Education, 84, 3–22.
Brookes. Paxton, P. (1999). Is social capital declining in the United States? A
Heckman, J. J. (2008). Schools, skills, and synapses. Economic multiple indicator assessment. American Journal of Sociology, 105,
Inquiry, 46, 289–324. 88–127.
Hoffmann, J. P., & Dufur, M. J. (2008). Family and school capital Pensiero, N. (2011). Parent–child cultivation and child cognitive and
effects on delinquency: Substitutes or complements? Sociological attitudinal outcomes from a longitudinal perspective. Child Indi-
Perspectives, 51, 29–62. cators Research, 4, 413–437.
Jager, M. M., & Holm, A. (2007). Does parents’ economic, cultural, Pong, S. (1998). The school compositional effect of single parenthood
and social capital explain the social class effect on educational on 10th grade achievement. Sociology of Education, 71, 23–42.
attainment in the Scandanavian mobility regime? Social Science Portes, A. (1998). Social capital: Its origins and applications in modern
Research, 36, 719–734. sociology. In J. Hagan, & K. S. Cook (Eds.), Annual review of
Jencks, C., Smith, M., Acland, H., Bane, M. J., Cohen, D., Gintis, H., sociology (pp. 1–24). Palo Alto, CA: Annual Reviews Inc.
et al. (1972). Inequality: A reassessment of the effect of family and Putnam, R. D. (2000). Bowling alone: The collapse and revival of
schooling in America. New York: Basic Books. American community. New York: Simon and Schuster.
Johnson, M. K., Crosnoe, R., & Elder, G. H. (2001). Students’ attach- Raftery, A. W. (1996). Approximate Bayes factors an accounting for
ment and academic engagement. The role of race and ethnicity model uncertainty in generalised linear models. Biometrika, 83,
Sociology of Education, 74, 223–250. 251–266.
Joreskog, K. G., & Sorbom, D. (1993). LISREL 8: User’s reference Raijman, R., & Tienda, M. (1999). Immigrants’ socioeconomic
guide. Chicago, IL: Scientific Software International. progress post-1965: Forging mobility or survival? In C. Hirschman,
Kim, H. S. (2011). Consequences of parental divorce for child devel- J. DeWind, & P. Kasinitz (Eds.), International migration and the
opment. American Sociological Review, 76, 487–511. remaking of America (pp. 239–256). New York: Russell Sage Foun-
Kirschenman, J., & Neckerman, K. M. (1991). We’d love to hire them, dation.
but: The meaning of race for employers. In C. Jencks, & P. E. Roscigno, V. J., & Crowley, M. L. (2001). Rurality, institutional
Peterson (Eds.), The urban underclass (pp. 203–232). Washington, disadvantage, and achievement/attainment. Rural Sociology, 66,
DC: The Brookings Institution. 268–292.
Lareau, A. (2011). Unequal childhoods: Class, race and family life, Sampson, R. J., Morenoff, J. D., & Earls, F. (1999). Beyond social cap-
second edition with an update a decade later. Berkeley, CA: Uni- ital: Spatial dynamics of collective efficacy for children. American
versity of California Press. Sociological Review, 64, 633–660.
Lee, S., & Shouse, R. C. (2011). The impact of prestige orientation Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the
on shadow education in South Korea. Sociology of Education, 84, state of the art. Psychological Methods, 7, 147–177.
212–224. Sewell, W. H., & Hauser, R. M. (1972). Causes and consequences of
Little, R. J. A., & Rubin, D. B. (1987). Statistical analysis with missing higher education: Models of the status attainment process. Ameri-
data. New York: John Wiley. can Journal of Agricultural Economics, 54, 851–861.
McLanahan, S., & Sandefur, G. D. (1994). Growing up with a sin- Shavit, Y., & Blossfeld, H. (Eds.). (1993). Persistent inequality:
gle parent: What hurts, what helps. Cambridge, MA: Harvard Changing educational attainment in thirteen countries. Westview:
University Press. Boulder, CO.
McNeal, R. B., Jr. (1999). Parental involvement as social capital: Swanson, C. B., & Schneider, B. (1999). Students on the move: Res-
Differential effectiveness on science achievement, truancy, and idential and educational mobility in America’s schools. Sociology
dropping out. Social Forces, 78, 117–144. of Education, 72, 54–67.
Tanaka, J. S. (1993). Multifaceted conceptions of fit in structural Tough, P. (2009). Whatever it takes: Geoffrey Canada’s quest
equation models. In K. A. Bollen, & J. S. Long (Eds.), Testing to change Harlem and America. New York: Mariner
structural equation models (pp. 10–19). Newbury Park, CA: Sage Books.
Publications. Yamamoto, Y., & Brinton, M. (2010). Cultural capital in East Asian
Teachman, J. D., Paasch, K., & Carver, K. (1996). Social capital and educational context: The case of Japan. Sociology of Education,
dropping out of school early. Journal of Marriage and the Family, 83, 67–83.
58, 773–783.

Does Capital at Home Matter More Than Capital at School? Social Capital Effects On Academic Achievement

Uploaded by

Copyright:

Available Formats

Does Capital at Home Matter More Than Capital at School? Social Capital Effects On Academic Achievement

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Does Capital at Home Matter More Than Capital at School? Social Capital Effects On Academic Achievement

Uploaded by

Copyright:

Available Formats

Available online at www.sciencedirect.

Research in Social Stratification and Mobility 31 (2013) 1–21

Does capital at home matter more than capital at school?

1. Introduction It is also well known that resources from multiple

School Social Capital

Family Social Capital

Fig. 1. Tested measurement models.

Model A 7329.666*** (df = 75) .476 .072 7417.67 7417.78 7609.82

Figure 2: Measurement Model with

School Social Capital

Family Social Capital

CFI 0.877 RMSEA 0.045

Explanatory Model with

School Social Capital

Family Social Capital

CFI 0.850 RMSEA 0.049

Control → family NA .167*** NA .395*** NA −.112*** NA .226*** NA −.198*** NA

Explanatory Model with

School Social Capital

Standardized test White e

Family Social Capital

CFI .801 RMSEA 0.056

and teacher qualifications/expectations have sometimes

maximum likelihood estimation in MPlus to test these

ond column, Model 1, shows the effect of family and

school social capital on test scores without the inclu-

model). Each subsequent pair of columns represents the

effects of a different school-level social capital mea-

sure on achievement; the first column in each pair

change in coefficients compared to the models without

the school-level variable. For example, the third col-

umn, Model 2, shows coefficients from the model where

receive free- or reduced-price lunch. The fourth column

rience in any secondary school setting and found similar results, so we

as measured by student/teacher ratio, but as none of these exerted a

significant effect on composite test scores and their effects on family

You might also like

Control → family NA .167* NA .395* NA −.112* NA .226* NA −.198*** NA