You are on page 1of 20

J Youth Adolescence (2011) 40:322

DOI 10.1007/s10964-010-9579-5

EMPIRICAL RESEARCH

The Relationship Between School Engagement and Delinquency


in Late Childhood and Early Adolescence
Paul J. Hirschfield Joseph Gasper

Received: 8 October 2009 / Accepted: 23 July 2010 / Published online: 13 August 2010
 Springer Science+Business Media, LLC 2010

Abstract Engagement in school is crucial for academic


success and school completion. Surprisingly little research
has focused on the relationship between student engagement
and delinquency. This study examines whether engagement
predicts subsequent school and general misconduct among
4,890 inner-city Chicago elementary school students (mean
age: 11 years and 4 months; 43.3% boys; 66.5% black;
28.8% Latino). To improve upon prior research in this area,
we distinguish three types of engagement (emotional,
behavioral, and cognitive), examine whether the relationship between engagement and misconduct is bidirectional
(misconduct also impairs engagement), and control for
possible common causes of low engagement and misconduct, including peer and family relationships and relatively
stable indicators of risk-proneness. Emotional and behavioral engagement predict decreases in school and general
delinquency. However, cognitive engagement is associated
with increases in these outcomes. School and general
delinquency predict decreased engagement only in the
cognitive domain. Suggestions for future research and
implications for policy are discussed.
Keywords Urban education  School disengagement 
Problem behavior  Delinquency

P. J. Hirschfield (&)
Department of Sociology, Rutgers University,
26 Nichol Avenue, New Brunswick, NJ 08901, USA
e-mail: phirschfield@sociology.rutgers.edu
J. Gasper
Westat, 1600 Research Boulevard, Rockville, MD 20850, USA
e-mail: josephgasper@westat.com

Introduction
School performance and behavior are among the most
frequently cited factors in the etiology of delinquency
(Cloward and Ohlin 1960; Cohen 1955). A solid research
tradition in criminology has examined delinquency as a
function of academic performance (e.g., grades and test
scores; for a review, see Maguin and Loeber 1996). Likewise, delinquency is often linked empirically to school
bonds (Cernkovich and Giordano 1992; Liska and Reed
1985), which encompass the myriad ways in which students
are connected to and invested in their schools (attachment
involvement, conventional school beliefs, and commitment;
Cernkovich and Giordano 1979; Hirschi 1969; Liska and
Reed 1985). In short, a large body of research has consistently established that students who perform poorly in
school and who feel alienated from school are more likely to
engage in delinquency.
These relationships have attracted less attention within
the fields of education and adolescent development, where
a close conceptual cousin to Hirschis school bond is a
cluster of school attitudes and behaviors known as
engagement. Whereas the concept of social bonds was
developed to explain restraints on deviance, engagement
encompasses active, goal-oriented, focused, and constructive behaviors and interactions (Furrer and Skinner 2003).
Disengagement from school, which may take root as early
as first grade, is widely regarded as the primary developmental process underlying school failure and dropout
(Alexander et al. 2001; Finn 1989; Marks 2000). Failure
and dropout, in turn, exact many economic and social
costs, especially in urban America (Manlove 1998; Orfield
2004; Pettit and Western 2004). Accordingly, student disengagement is increasingly recognized as a serious social
problem (National Research Council and Institute of

123

Medicine 2003) and is the focus of social, curricular, and


instructional interventions (Fredricks et al. 2004).
There is no universally accepted definition of engagement. Scholarly definitions variably incorporate such
overlapping concepts as effort, attention, belonging, commitment, and motivation (Jimerson et al. 2003). Engagement is viewed as the antidote to student boredom,
alienation, withdrawal, and apathy (Finn 1989). Nevertheless, engagement, which is not tethered to social control
theory (Hirschi 1969) and is favored among psychologists
and educational researchers, may progress more quickly
toward scientific validation than measures rooted in control
theory. Like Jimerson et al. (2003), Fredricks et al. (2004)
recently specified three distinct but mutually reinforcing
dimensions of engagement. Behavioral and emotional
engagement roughly correspond, respectively, to the
involvement and attachment components of the school
social bond, whereas cognitive engagement lacks a close
social bond counterpart. Behavioral engagement refers to
participation in school-related activities, both academic
and extra-curricular. Whereas school attachment connotes
affection for and care about the wishes and expectations
of school actors (e.g., teachers) and the institution as a
whole (Hirschi 1969), emotional engagement emphasizes
positive emotional dispositions and affective responses
toward educational processes and practices, as well as
actors. Cognitive engagement is the mental labor that
one invests or is motivated to invest in academic tasks
(Fredricks et al. 2004). Psychological investment may be
manifest as flexibility in problem solving, preference for
hard work, and positive coping in the face of failure
(Fredricks et al. 2004, p. 64). In essence, cognitive
engagement taps a students willingness to engage in the
difficult mental labor necessary to comprehend complex
ideas and master difficult skills (p. 60). These three types
of engagement have been the focus of most theoretical and
empirical research on academic engagement.
The primary purpose of this study is to assess competing
interpretations of the link between engagement and adolescent misbehaviorboth school and generalin late
childhood and early adolescence. These alternate interpretations, in simplified form, are that engagement decreases
delinquency, that delinquency hinders engagement, and that
engagement and delinquency are manifestations or consequences of the same underlying conditions (e.g., personality). A secondary purpose is to examine which specific
form(s) of engagement are most important in forging this
link. Given the elevated rates of academic disengagement
among disadvantaged urban minorities (Felice 1981) and
the theorized importance of engagement during the elementary school years for later development (Alexander
et al. 2001), we seek to characterize the relationship
between engagement and misconduct among a panel of

123

J Youth Adolescence (2011) 40:322

elementary school children and early adolescents living in


inner-city Chicago.

Prior Theory and Research on Engagement


and Delinquency
Disengagement as a Cause of Delinquency
There are several processes through which engagement may
lower delinquency. Emotional engagement in school reduces school misconduct to the extent that it helps youth forge
strong emotional connections with school actors, especially
teachers and similarly engaged peers (Steinberg and
Avenevoli 1998). Within a social control framework,
weakened attachments to these actors reduces students
concern about these actors social disapproval in response to
misconduct (Hirschi 1969). Similarly, social bonding theory
would posit that behavioral engagement lowers the time and
energy available for deviant activities. A weakness of control theory is that it treats all conventional attachments and
behavioral diversions as equally protective (Huebner and
Betts 2002). Some school activities (e.g., team sports)
appear to increase some forms of delinquency (Cernkovich
and Giordano 1992; Eccles and Barber 1999). When theorized through a wider lens, self-directed, strategic, and
challenging academic behaviors like homework and studying, with the requisite emotional and cognitive engagement,
seem more effective in curbing delinquency. They not only
divert deviant energies but also foster individual skills,
confidence, and rewards (Fredricks et al. 2004; Schmidt
2003), which may autonomously influence delinquency, as
emphasized elsewhere (Catalano et al. 1999).
School engagement may also decrease off-campus
delinquency through altering parental and peer interactions
(Jang 1999). For example, if engagement leads to positive
feedback from the school, it may strengthen parentchild
bonds (Liska and Reed 1985; Menard and Morse 1984).
Conversely, school-related parental frustrations and conflicts may weaken parental discipline and supervision and
their effectiveness, which, in turn, may increase time with
peers and delinquency (Rankin and Wells 1990; Steinberg
and Avenevoli 1998). Likewise, if disengagement and
school failure result in social rejection at school and specialized classroom placement, they may foster and intensify associations with deviant peer groups (Cohen 1955;
Padilla 1992; Wong 2005). Thus, school engagement may
affect both on campus and off-campus delinquency.
Few studies have examined the impact of engagement,
per se, on delinquency. Tapping the emotional side of
school engagement (Steinberg et al. 1996, p. 70), Steinberg and Avenevoli (1998) found that bonding to high
school teachers and commitment to school strongly

J Youth Adolescence (2011) 40:322

predicted decreased substance use, delinquency, and


parental permissiveness a year later. The attrition of the
most disengaged and delinquent students somewhat
depressed these effects but was likely offset by the omission of most aspects of psycho-social functioning and
social context from their models. Therefore, rather than
capturing the effects of academic engagement, their
results may reflect the influence of some other, unmeasured factor whose appearance precedes or provokes both
disengagement from school and a range of other problems
(p. 421).
With respect to behavioral engagement, at least three
studies isolated the impact of academic activity from other
activities (Barnes et al. 2007; Huebner and Betts 2002;
Wong 2005). Each found that time spent doing homework
or studying predicts less delinquency, net of time spent in
other activities. Except Barnes et al. (2007), they also
controlled for other social bonding indicators. The temporal order of behavioral engagement and delinquency
remains unclear from these cross-sectional analyses, as is
whether these effects obtain in urban contexts.
Whereas few studies assessed the impact of engagement
on delinquency, voluminous research has examined the
effects of school social bonds. Because scholars conceptualizations and measurements of school bonds often
overlap with conceptions and measures of engagement, a
brief review of this prior research is warranted. Several
studies observed an influence of teacher and school
attachment. Sampson and Laub (1993) reported a strong
temporal link between school attachment and delinquency
for males. Crosnoe (2006) found beneficial effects of teacher attachment on adolescent drinking, and Crosnoe et al.
(2002), on delinquency and drug use (among females only).
However, Dornbusch et al. (2001), using the same dataset,
included baseline outcome measures and found null or
modest effects of school attachment that varied by delinquency type and measurement level. Other studies reported
weak or null effects of school attachment, perhaps due to
inclusion of other dimensions of social bonding either
within (Liska and Reed 1985; Ford 2005) or alongside
(Agnew 1991; Cernkovich and Giordano 1992; Payne
2008) measures of school attachment.
No known research has assessed the effects of cognitive
engagement on delinquency. Although not a study of
cognitive engagement, per se, Schmidt (2003) found that
both time spent in challenging activities and perceived
success in these activities independently and inversely
predicted concurrent school misconduct, controlling for
grades, family and school adversity, and demographics.
Perceived success predicted decreased school misconduct
2 years later. The impact of challenging activity on offcampus delinquency, along with whether these effects are
driven by cognitive engagement in academic activities

(as opposed to non-school activities) or by other dimensions of engagement or social bonds remain unclear.
Delinquency as a Cause of Disengagement
A conflicting interpretation of the relationship between
engagement and delinquency is that delinquency lowers
engagement, or that they have a reciprocal relationship
(Thornberry et al. 1991). School delinquency can prompt
peer rejection, exclusionary discipline, and negative valuations at school and at home (Ford 2005; Sampson and
Laub 1997). These reactions may reduce enthusiasm for
school (Finn 1989; French and Conrad 2001; Natriello
1984). An impact of off-campus delinquency on school
engagement is surely less common and direct. However,
some offenses may deplete the time or energy available for
school work. Juvenile justice involvement may also divert
time and energy while promoting stigmatization, social
exclusion, and deviant peer affiliation (Hirschfield 2009;
Hjalmarsson 2008).
Relatively serious delinquency and substance use appear
to increase dropout (Fagan and Pabon 1990; French and
Conrad 2001; Mensch and Kandel 1988)perhaps the final
act of disengagement. Liska and Reed (1985) conducted
one of three known studies that estimate both the impact of
school bonds or engagement on delinquency and vice
versa. They reported that, among a national sample of high
school boys, school bonds (a 19-item mega-construct that
taps attachment, commitment, and distinct aspects of
emotional and cognitive engagement) had null to very
weak effects on subsequent delinquency. However, violent
and property offending markedly reduced school bonds, net
of parental attachment. The second study similarly found
that delinquency weakened school bonds (a composite of
attachment, effort, and commitment; Thornberry et al.
1991). Both studies, however, omitted factors like family
adversity/disruption and peer behavior that predispose
youth to both disengagement and delinquency. Steinberg
and Avenevoli (1998), by contrast, found that delinquency
predicted increased high school engagement. This anomalous result likely occurred because the measure of delinquency indexed mainly illicit recreational pursuits (e.g.,
marijuana, alcohol, fake ids) rather than anti-social
behavior. The observed residual positive association
between risky behavior and subsequent enjoyment of
school may attest to the social benefits of participation in
the party subculture (Hagan 1991).
Delinquency and Disengagement Have a Common
Etiology
A third perspective holds that disengagement and delinquency are not causally related but, rather, share common

123

origins. According to problem behavior theory (Donovan


and Jessor 1985; Jessor and Jessor 1977), an array of
undesirable behaviors, including precocious sexuality,
delinquency, smoking, and drinking co-occur in adolescence as part of an underlying tendency toward deviance or
youthful rebellion. Problem behaviors are functionally
equivalent; they all represent an assertion of independence
and adult status. Such orientations often take root in
unconventional or dysfunctional contexts (Jessor and
Jessor 1977), such as harsh or lax parental controls
(DeBaryshe et al. 1993). Problem behavior theory expects
disengaged youth to be prone to delinquency, but not
because the behaviors are causally related. Rather, disengagement and delinquency manifest a syndrome of
adolescent maladjustment that has its roots in an underlying proneness to deviance (Steinberg et al. 1996; Steinberg
and Avenevoli 1998).
A proneness to deviance or rebelliousness is but one
latent trait that has been linked to both delinquency and
low school effort. Other constitutional factors like low selfcontrol allegedly explain both behaviors (Felson and Staff
2006; Gottfredson and Hirschi 1990). Similar claims have
been advanced with respect to antisocial traits (Ramsey,
Patterson, and Walker 1990) and intelligence (Wilson
and Hernstein 1985). Direct assessments of latent trait
explanations of the association between engagement/
bonding and delinquency are surprisingly rare. Most studies on the effects of school bonding do not control for
stable or innate propensities. Felson and Staff (2006) treat
teacher reported effort, independent of attachment and
commitment, as a proxy for self-control. This conceptualization overlooks educational theory that construes such
measures as indicative of behavioral or cognitive engagement. Several other relatively stable and time-varying
factors including SES and living arrangements (or disruptions therein) could also lead to both disengagement and
delinquency and should therefore be controlled (Astone
and McLanahan 1994; Cohen 1955). Likewise, delinquent
peers, discussed above as a potential mediator of reciprocal
relationships between engagement and delinquency, may
also help explain both disengagement and delinquency. To
illustrate, peer rejection, which strongly predicts subsequent delinquent peer group involvement and individual
delinquency (Laird et al. 2001), is a strong predictor of
subsequent academic engagement in late elementary and
middle school (Buhs 2005; Lubbersa et al. 2006).
On balance, rigorously designed longitudinal studies
suggest or leave open the possibility that increased levels
of delinquency among students who are behaviorally,
emotionally, and cognitively disengaged stem largely from
antecedent, adverse circumstances and behavioral problems or from the impact of delinquency on disengagement.
Furthermore, these relationships may be ethnically

123

J Youth Adolescence (2011) 40:322

variable. Liska and Reed (1985) found that among AfricanAmericans, but not among whites, school bonds predicted
strong reductions in violent and property offenses. They
suggested that structural disadvantages (i.e., fewer opportunities outside of school) may lead black youth to invest
more of their aspirations in school and, therefore, to be
more responsive to variations in teacher support and
control. Although these effects could be due to omitted
variable bias or to their relatively small African-American
sub-sample, subsequent studies also suggested that emotional or cognitive engagement exert larger effects in disadvantaged urban communities (Cernkovich and Giordano
1992; Dornbusch et al. 2001; Schmidt 2003). This possibility begs examination with larger samples and rigorous
longitudinal methods.
Another possibility that motivates the present inquiry is
that engagement and delinquency begin to affect each other
prior to high school (Liska and Reed 1985; Steinberg and
Avenevoli 1998). High school engagement is often an
extension of similar patterns during middle school or even
early in elementary school (Alexander et al. 2001). However, affirming Thornberrys (1987) predictions on a 1977
national longitudinal study, Jang (1999) found weaker
effects of school commitment (a composite of studying/
homework, attachment to school/teachers, and grades) at
ages 12 and 13 than during high school. He argued that,
early in high school, as parental influence declines, schools
assume a larger role in mediating peer associations and
equipping children to cope with developmental challenges.
Nonetheless, the roles the specific dimensions of school
engagement play in delinquency, and vice versa, particularly in the context of disadvantaged urban elementary
schools in more recent times remain unsettled.
This study assesses whether there is a direct relationship
between delinquency and engagement and, if so, this
relationships directionality. Addressing several limitations
of prior research, this study makes some important contributions. First, following Steinberg and Avenevoli
(1998), this study focuses on engagement, a multi-dimensional construct that has greater conceptual clarity and
practical importance than the school bonding constructs
derived from social control theory. Second, it uses data
from a large panel of elementary school students (grades 5
through 8) from disadvantaged inner-city Chicago neighborhoods. Third, it controls for several potential common
causes of disengagement and delinquency that are generally omitted from prior studies, such as disruptive life
events, anger levels, and peer deviance. Foregoing theories
of the behavioral impact of academic engagement and of
the underlying sources of problem behavior yield conflicting notions as to whether family and peer problems
confound the impact of school engagement on delinquency
or whether they are precursors or sequelae (i.e., mediators)

J Youth Adolescence (2011) 40:322

along the same causal pathways as engagement. Therefore,


both conservative models that treat them as potential
confounds and liberal models that exclude them as causally
synonymous are presented. Estimating both models
explores the possible confounding or mediating influence
of parents and peers. Finally, this study distinguishes
between in-school deviance and general misconduct. Most
studies examine one or the other.

Hypotheses
The goal of this study was to determine whether engagement predicts delinquency, delinquency predicts engagement, or both. It was hypothesized that emotional,
behavioral, and cognitive engagement would each negatively predict school misconduct and general delinquency
in disadvantaged urban contexts, although the inclusion of
parental and peer relationship dynamics weakens these
relationships. This hypothesis was largely based on social
control theory and prior studies showing an impact of
attachment and involvement on delinquency. There are
also theoretical and empirical reasons for examining whether the relationship runs in the opposite direction. It was
therefore hypothesized that school and general misconduct,
especially the former, negatively predict each of the three
dimensions of school engagementrelationships partially
explained by parental and peer relations.

Methods
Design and Sample
This study is set in inner-city Chicago from fall 1992
through spring 1997. It uses survey data from a sub-sample
of the 5th through 8th grade students in the 22 public
elementary schools (mean = 92% low-income in
19931994) that participated in Comers School Development Program Evaluation (SDP). SDP was a randomized
whole school reform to improve disadvantaged urban
elementary schools academic and social climates through
building cohesion and integrating parents and mental
health professionals into school operations and management (Cook et al. 2000). Students were surveyed twice
annually from the 19921993 school year through the
19961997 school year. About 95% of eligible students
participated in the surveys each year (see Cook et al. 2000
for full details of sampling design). First, an attitudes and
behavior survey measured academic and psycho-social
adjustment including delinquency and cognitive and
behavioral engagement. Later (1.6 months on average), a
school climate survey gauged perceptions and attitudes

regarding schools and teachers, including emotional


engagement.
Only a portion of the over 11,000 youth who participated
in SDP are included. While not designed to be a longitudinal study, embedded within the SDP is an incidental
panel of 4,890 students who participated in two or more
consecutive waves of the adolescent attitude survey. The
sample is restricted as such, because statistical models for
panel data require at least two waves of data per student.
Two overlapping sub-samples were examined in the
present study due to the use of measures drawn from two
different surveys. The first sub-sample (n = 3,580), which
was used to estimate the relationship between behavioral
and cognitive engagement and misbehavior, consisted of
students with valid measures, in two consecutive survey
waves, of behavioral engagement, cognitive engagement,
and the delinquency scales, all of which were drawn only
from the attitude survey. Table 1 presents descriptive statistics for all variables in the analysis at Time 1 and Time 2
for this sub-sample. This sample is predominantly AfricanAmerican (67%) and Latino (26%). The average age at
Wave 1 was 11.3 years. At wave one, two-thirds of the
sample resided with only one or neither biological parent
and they shared a household with four other children, on
average. Compared to the panel of 4,890 youth (52%
female, 22% Latino) the sub-samples are disproportionately female and non-black owing to males greater rates of
attrition and males and blacks higher rates of missing data
on delinquency and behavioral engagement. However,
there were no significant differences between these samples in Wave 1 delinquency or engagement.
The second sub-sample, which was used to estimate the
relationship between emotional engagement and misbehavior, consists of 2,768 youth who have valid measures of
emotional engagement, which is measured on the school
climate survey, and the delinquency scales. Inclusion in the
first sub-sample requires completion of only two consecutive attitude surveys, whereas inclusion in the emotional
engagement sub-sample requires consecutive participation
in both annual surveys. Because 2,480 youth are in both
samples, the two sub-samples are virtually identical with
respect to all this studys examined variables in wave 1 and
wave 2.
For students with more than two consecutive waves of
outcome data, the earliest two waves are used, because
observed patterns of disengagement and misconduct often
begin in early grades, and, accordingly, disengagement
may exert a formative influence on delinquency during
these grades (Alexander et al. 2001; Steinberg and
Avenevoli 1998). Missing data on all control variables
were imputed using multiple imputation (Allison 2002).
This strategy takes into account the dependence among the
multiple observations for each individual. Imputation

123

J Youth Adolescence (2011) 40:322

Table 1 Means and standard deviations of all variables in the


analysis, by survey wave (N = 3,580)

Measures

Variable

Male

43.3%

Female

57.6%

Measures of adolescent misconduct, peer delinquency,


time use (used for the behavioral engagement scale), and
disruptive life events were derived from the National
Youth Survey and used extensively in prior literature
(Armstrong et al. 2009). Thomas D. Cook, in a theoretically-guided fashion, developed the emotional engagement, parental attachment, parental control, and anger
scales, by combining items selectively drawn from a
variety of extant scales with new items. These reliabilityadjusted scales, along with positive coping and academic
efficacy scales from which the cognitive engagement scale
was derived, were employed in two large-scale evaluations
of SDP in Prince Georges County Maryland and Chicago
(see Cook et al. 2000). The Appendix provides detailed
information on all multi-item scales.

Latino, White, other

28.8%

School Misconduct

Black
Asian

66.5%
4.7%

Wave 1

Wave 2

Mean
or %

SD

Mean
or %

SD

School misconduct

1.57

0.67

1.63

0.69

General delinquency

1.39

0.62

1.43

0.63

Emotional

3.38

1.13

3.17

1.13

Behavioral

0.17

0.09

0.16

0.08

Cognitive

3.32

0.56

3.24

0.42

Problem behavior

School engagement

Demographics
Gender

Race

Age

11.33

1.02

12.44

1.08

Family structure
Both biological parents

32.6%

29.0%

Only one biological parent

46.0%

53.4%

No biological parent

21.4%

Parental education (Wave 1 only)

2.36

17.6%
1.23

Parental work status


Full-time

74.6%

79.3%

Part-time

14.9%

12.5%

Unemployed

10.5%

8.2%

Kids in household

3.99

2.94

3.73

2.70

Disruptive life events

0.74

0.77

0.76

0.75

1.15
1.78

0.68
0.98

1.24
1.73

0.79
0.98

Parental attachment

4.00

0.77

3.99

0.82

Parental control

3.89

0.49

3.80

0.55

Peer delinquency

1.41

0.63

1.44

0.62

Latent traits
Cigarette smoking
Anger
Social control

Descriptive statistics are for the behavioral and cognitive engagement


sample
SD Standard deviation

models included all of the variables in the model, at multiple time points, as predictors of all of the other variables.
Five imputed data sets were generated and each model was
estimated five times. The results of each set of five models
were averaged to generate the regression results presented
herein (correcting the standard errors to account for the
random variation introduced into our imputed values).

123

Two measures of the frequency of delinquency during the


past year were derived from the adolescent attitude survey
(1 = Never; 5 = 10 or more times). First, school
misconduct (a = .73 at wave 1) indicates the mean frequency of cheating on tests, bringing drugs to school, being
sent to the principals office, and skipping school. Skipping
school has also been conceived as a measure of behavioral
disengagement (see Jimerson et al. 2003 for a review).
However, in the present data, it clusters much more
strongly with measures of delinquency than measures of
engagement. This reflects the fact that truancy, a status
offense, often co-occurs with other delinquent behaviors.
General Delinquency
General delinquency (a = .82 at wave 1) includes seven
illegal or deviant behaviors that generally occur off campus
(e.g., gang fight participation, vandalism, shoplifting).
Emotional Engagement
Emotional engagement (a = .87), from the school climate
survey, is the mean of six Likert response scales on which
students project how much they would miss aspects of the
school experience including the principal, teachers, students, and the treatment of students (1 = not at all;
5 = a lot).
Behavioral Engagement
Based on prior literature suggesting that only academic
school activities reduce delinquency, behavioral engagement is the ratio of the number of hours on a typical

J Youth Adolescence (2011) 40:322

weekday spent doing homework relative to the number of


hours spent in six leisure activities (e.g., hanging out with
friends, talking on the phone). Less than 5% of respondents
indicated that they spent 8 or more hours a day on homework. These students inflate mean homework time enough
to reverse the negative correlation between behavioral
engagement and acting out. Since these students likely
overstated the time they spent on homework, their
responses were set to missing and multiply imputed.
Cognitive Engagement
Drawing from Connell and Wellborn (1991), cognitive
engagement (a = .71) is the mean of eight items gauging
psychological investment in school (e.g., school is boring), positive strategies to cope with poor school performance (e.g., studying harder to do better next time), and
how well students report that they can learn various
subjects (1 = not at all well; 5 = very well). Factor
analyses aided in the selection of these eight items from
among a larger set of items believed to tap cognitive
engagement. Consistent with prior literature, behavioral
and cognitive engagement are significantly correlated with
academic achievement (i.e., composite reading and
mathematics scores on the Illinois Test of Basic Skills or
ITBS) and emotional engagement is not (Fredricks et al.
2004).
Parental Control
Time-varying parental and peer relationships at Time 1 and
Time 2 may confound or mediate the relationships between
disengagement and delinquency. Parental control is the
mean of six Likert scales (a = .66) that gauge parental
specification and enforcement of rules about how and with
whom respondents spend their time.
Parental Attachment
Another six-item (Likert) scale (a = .81) regarding my
parents (1 = strongly disagree; 5 = strongly agree)
taps some dimensions of parental attachment (see Jang
1999; Liska and Reed 1985) like closeness, communication, and involvement (discuss my problems with me,
discuss important family matters with me) and other
aspects of effective, responsive parenting (Steinberg
et al. 1996) like acceptance and autonomy (trust me,
approve of my friends).
Peer Delinquency
Finally, peer delinquency (a = .85) averages five items
reporting the share of students friends who committed

various infractions in the last year, including stealing and


selling or taking drugs (1 = none; 5 = all).
Demographics and Family Circumstances
Although baseline outcome measures effectively fix the
effects of stable characteristics, estimated models, following Jang (1999), also control for several demographic
factors to capture unmeasured, structurally-conditioned
(Thornberry et al. 1991), social and cultural processes that
could affect both disengagement and delinquency. Sex is
binary (1 = male), and ethnicity is measured by dummy
variables for African-American and Asian. Latinos (26%),
whites, and others (2%) comprise the reference category.
Age (in years) is measured at the time of each attitude
survey. Including age is important, because of the plausibility that changes in delinquency and school bonds and the
influence of peer and family dynamics vary over the early
life course (Jang 1999). SES is not directly measured in the
CSDP, so several likely correlates of economic and family
adversity are included. Parental education is the highest
education level among both parents (1 = did not graduate
high school; 4 = graduated from college). Two dummies
indicate parental work status (1 = unemployed parents;
1 = only part-time working parents; at least one full-time
working parent is the reference category). Parents are
defined as the people who most act like a mother and
father, respectively. If the educational or work status of
only one parent is reported, only these values are used.
Family structure is indicated by dummy variables for one
natural parent households and no natural parents (two
parents is the reference category). The number of minor
children in the household is included as an indicator of
economic hardship among the poor that is strongly predictive of delinquency (Farrington and Welsh 2007). We
also include a measure of the number (among four) of
disruptive life events (e.g., parents new spouse or boyfriend, a parent moving out or losing job) the youth
experienced in the past year. Parental work status, family
structure, age, children in the household, and disruptive life
events are time-varying covariates. Sex, ethnicity, and
parental education are measured only at Wave 1.
Latent Traits
Unfortunately, a direct measure of the underlying deviance
proneness that may drive a developmental syndrome of
adolescent problem behavior is lacking. To address the
possibility that changes in engagement and delinquency
manifest such a syndrome, the frequency of cigarette use
in the month preceding the attitude survey (1 = none;
4 = About 1/2 pack per day; 7 = 2 or more packs per
day) is included. If smoking is integral to a cluster of

123

10

J Youth Adolescence (2011) 40:322

problem behaviors that includes disengagement and


delinquency, it should be a proxy for rebelliousness. Also,
a four-item scale taps feelings of anger (a = .81).
Research underscores not only the centrality of a volatile
temper among dimensions of self-control but also its relatively robust relationship with delinquency (Conner et al.
2009). These traits are time-varying. The wave 1/wave 2
correlation is r = .269 for smoking and r = .440 for anger.
Statistical Models
We use structural equations models (SEM) with reciprocal
effects to estimate bidirectional relationships between
school disengagement and delinquency. Due to uncertainty
about the lag structure between disengagement and delinquency, we estimated models with variable temporal lags.
The first was a cross-lagged model, which assumes that the
effects of disengagement and delinquency on each other
are still observable up to 1 year later. The general system
of equations can be written as:
D 2 E1 X1 D 1 Z U 2
E2 D1 X1 E1 Z U2

In the first equation, D2 is delinquency (school and


general) at wave 2, E1 is engagement at wave 1. X1 is a
vector of time-varying covariates, Z is a vector of timeinvariant variables associated with each individual, D1 is
delinquency at Wave 1, and U2 is an error term which is
assumed to be normally distributed. Likewise, in the
second equation, E2 and E1 are engagement at wave 2 and
wave 1, respectively, while D1, X1, Z, and U2 have the
same meaning as they do in the first equation. The error
terms of the two equations are assumed to be correlated.
Under the assumption that the cross-lagged variables are
exogenous, the models are estimated using ordinary leastsquares (OLS) regression (Finkel 1995).
Because the temporal lag between engagement and
delinquency is uncertain, we also estimated contemporaneous effects models (Eq. 2) in which delinquency and
disengagement predict each other in the same wave. The
general model is:
D 2 E2 X2 D 1 Z U 1
E2 D2 X2 E1 Z U2

In contrast to the cross-lagged model (Eq. 1), engagement


and delinquency exert their effects on each other at the same
wave. The first equation models wave 2 delinquency as a
function of behavioral and cognitive engagement at wave 2,
net of the prior waves measure of the outcome (D1) and
wave 2 controls (X2, Z). The second equation similarly
predicts that delinquency at wave 2 (D2) affects these two
dimensions of disengagement at wave 2 (E2).

123

The contemporaneous impact of emotional disengagement on delinquency is not estimated because the school
climate surveys were administered after the attitudes and
behavior survey. To ensure proper temporal ordering of
these variables and to partially adjust for the impact of
emotional engagement on delinquency, an SEM is estimated that models emotional engagement (P2) as a function of wave 1 controls including, wave 1 emotional
engagement (E1). Given the strong association between
wave 1 and wave 2 emotional engagement (r = .382),
controlling for P1 reduces the share of the association
between delinquency and engagement a month or two later
that can be accounted for by the effects of prior engagement on delinquency.
Unlike the cross-lagged model, the contemporaneous
effects models cannot be estimated using OLS regression.
Because delinquency and disengagement are allowed to
affect each other, they correlate with the error terms of
each equation, violating one of the basic assumptions of
OLS. The maximum likelihood (ML) estimation methods
in SAS (PROC CALIS) are employed to address this
problem by taking into account the endogeneity of
engagement and delinquency.
Estimates of the impact of engagement on delinquency
and, especially, of the impact of delinquency on disengagement are conservative. All time-varying covariates
measured during the same wave as delinquency (except
disruptive events) reference a period that largely postdates
the measurement period for delinquency, which is the past
year. Thus, cross-lagged models technically examine the
impact of retrospective delinquency at wave 1 on engagement at wave 2, independent of concurrent engagement at
wave 1. As mentioned, peer and family dynamics may
mediate (and thereby attenuate) the impact of wave 1
engagement on wave 2 delinquency, and vice versa.
Therefore, both partially specified models, which include
only presumably exogenous controls, and full models,
which include parental control and attachment and peer
delinquency, are estimated. The contemporaneous effects
models (Eq. 2) are less clearly conservative, because the
sequential order of cognitive and behavioral engagement
and delinquency is less precise. Both partial and full synchronous effects models are presented to proximately
measure the potential confounding/mediating influence of
peer and parental factors.

Results
Descriptive Findings
As shown in Table 2, the three measures of engagement are
only modestly intercorrelated at both wave 1 and wave 2.

.420**

Peer delinquency

.289**

Anger

-.242**

.356**

-.178**

.092**

Disruptive life events

Cigarette smoking

Parental control

.091**

Parental attachment

.005

-.212**
-.152**

Behavioral engagement
Cognitive engagement

Parental education

-.113**

Emotional engagement

Kids in household

1
.722**

School misconduct

General delinquency

.721**

.481**

-.325**

-.203**

.411**

.373**

.118**

.086**

.018

-.271**
-.124**

-.183**

-.075**

.183**

.242**

-.162**

-.054**

.002

-.004

.009

.108**
.063**

-.147**

-109**

-.167**

.231**

.018

-.195**

-.120**

-.113**

-.059**

-.037*

1
-.014

.047**

-.200**

-.171**

-.115**

.130**

.226**

-.052**

-.094**

.057**

-.020

.102**

.016
1

.058**

-.134**

-.143**

5
.063**
.049**

.065**

.030

.072**

.020

.036*

.052**

-.017

-.036*
.079**

-.038*

6
.053**
.001

.046**

.047**

.006

-.039*

.044**

.044**

.019

-.025

-.062**
-.002

7
.067**
.115**

.067**

-.020

.007

.154**

.028

.030

.048**

-.087**
.037*

-.007

-.047**
-.020
-.167**
-.109**

.028
.037**
.030
1

.228**
.201**
-.290**

-.109**
-.168**
.317**

.054**

.191**
.054**

-.113**
-.076**

.146**

-.030**
.161**

.127**

-.057**

-.143**

.184**

-.146**

-.107**

.018

-.020

.219**

-.176**

-.393**

-.154**

-.284**

11

.344**

10

.324**

-.178**

.215**

-.143**

-.141**

.001

-.006

.055**

.135**
.174**

.159**

-.260**

-.202**

12

-.152**

-.138**

.244**

.343**

.082**

.054**

.053**

-.139**
-.107**

-.096**

.433**

.445**

13

* p \ .05; ** p \ .01

Correlations in the lower left half of the table are for Wave 2 and correlations in the upper-right half are for Wave 1. Correlations are presented for all variables expect for dichotomous
demographic variables and are computed as Pearsons R

13

12

11

10

4
5

Variable

Table 2 Correlations among Wave 1 and Wave 2 variables

J Youth Adolescence (2011) 40:322


11

Emotional engagement is only modestly correlated with


behavioral engagement (r1 = .047; r2 = .108) and cognitive engagement (r1 = .058; r2 = .063). Behavioral and
cognitive engagement are uncorrelated at both waves.
These patterns suggest that the emotional, cognitive, and
behavioral domains of school engagement are relatively
independent. They also underscore the independence of
school and home behaviors, especially during the earliest
grades when peers are less influential (Jang 1999). As
control theory emphasizes, homework may reduce delinquency by occupying students in a conventional activity,
independent of any relationship to motivation or achievement. Table 2 also shows consistently significant contemporaneous associations between engagement and deviance.
All these correlations exceed .10 (p \ .01).
Multivariate Results

Cross-Lagged Models

Figure 1 reports the results from the cross-lagged models


of disengagement and school misconduct. Figure 2 shows
these results for general delinquency. In each figure, the
left panel presents the results for the partially specified
models that omit parental and peer factors, and the right
panel depicts the fully specified models.
The estimated partially specified models (Fig. 1) suggest
a modest relationship between emotional engagement
and school misconduct about a year later, controlling for
prior levels of school misconduct and the other controls.

Fig. 1 Cross-lagged models predicting school misconduct and


school engagement, Wave 2

123

12

Fig. 2 Cross-lagged models predicting general delinquency and


school disengagement, Wave 2

However, this effect may be significant only due to the


large sample and is reduced to non-significance in the
fully-specified model. Wave 1 behavioral engagement, by
contrast, is strongly predictive of reduced school misconduct in wave 2 in both partial and full models. Cognitive
disengagement does not predict increased school misconduct. Reversing the causal arrows, school misconduct
appears to have no discernible long-term effects on any of
the three dimensions of engagement.
Figure 2 reveals the same patterns with respect to the
effects of emotional and behavioral disengagement on
general delinquency. Further examination reveals that the
inclusion of both parental attachment and control together
rather than peer delinquency explains the small but significant relationships between emotional engagement and
both measures of delinquency. Unexpectedly, cognitive
engagement predicts modest but significant increases in
general delinquency in both partial and full models. Like
school misconduct, general delinquency does not influence
engagement in the cross-lagged models.
Contemporaneous Effects Models
Table 3 presents the coefficients from the contemporaneous effects models of the reciprocal relationship between
school engagement and school misconduct. As in the crosslagged analyses, partially and fully specified models are
presented for emotional, behavioral, and cognitive
engagement separately, resulting in a total of six SEM
models. Each SEM model contains two equationsthe first

123

J Youth Adolescence (2011) 40:322

predicts school misconduct and the second predicts emotional, behavioral, or cognitive engagement. The top panel
presents the results for the first equation predicting school
misconduct, and the bottom panel for the second equation
predicting each dimension of engagement. The final two
columns present the results of nested models (partial and
full, respectively) of the impact of wave 1 emotional
engagement on wave 2 school misconduct, net of wave 1
delinquency and controls.
As expected based on Fig. 1, wave 1 emotional
engagement exerts small effects on wave 2 school misconduct (model 1), which are no longer significant after
controlling for parental control and peer delinquency
(model 2). The protective effects of spending more time on
homework relative to leisure appear relatively substantial
and robust to model specification (models 3 and 4). Models
(not shown) that controlled for academic achievement in
the form of reading and mathematics test scores (which
address concerns that the engagement/delinquency relationship is due to confounding strain-induced school failure) did not change the findings. The effect of cognitive
engagement is significant and positive in the full and partial
models (model 5 and 6). Thus, all else being equal,
increases in cognitive engagement appear to increase
school misconducta reversal of the negative association
between these two variables at wave 2 (r = -.152).
The control variables hold few surprises. Males, smokers, and those living in households with a greater number of
other children all exhibit increased levels of school misconduct. Anger, parental attachment and control, and peer
delinquency also predictably influence school misconduct.
As expected, most selection factors that are relatively stable exert a weak or null influence on school misconduct
with a lagged measure of the outcome in the model.
The models in the second half of Table 3 examine
relationships in the opposite direction (Eq. 2). Models 1
through 4 suggest no effect of prior school misconduct on
either emotional or behavioral engagement. Consistent
with the cross-lagged results, school misconduct exhibits a
negative effect on cognitive engagement (Models 5 and 6).
Unexpectedly, however, African-American youth become
more cognitively engaged (but less emotionally and
behaviorally engaged) from school. Parental attachment
and control appear to improve cognitive engagement, as
expected.
Table 4 presents similar results for simultaneous equation models for engagement and general delinquency. The
impact of wave 1 emotional disengagement on general
delinquency is almost identical to its impact on school
misconduct (models 1 and 2). Similar to the results for
school misconduct, the inclusion of parental attachment
and control renders the relationship between emotional
engagement and general misconduct non-significant. The

J Youth Adolescence (2011) 40:322

13

Table 3 Parameter estimates from simultaneous effects models of the reciprocal relationship between school engagement and school
misconduct, ML estimates: Wave 2
Variable

Model 1

Model 2

Model 3

Model 4

-.09*

-.08*

Model 5

Model 6

.11*

.10*
.04*

Equation 1: School engagement predicting school misconduct


School engagement
Wave 2 emotional

-.09*

-.06

Wave 2 behavioral
Wave 2 cognitive
Demographics
Age

.06**

.03*

.07**

.04*

.07**

Male

.08**

.05**

.07**

.05**

.08**

.05**

Black

-.04*

-.03

-.04*

-.04

-.06*

-.05*

Asian

-.03

-.03

-.01

-.01

-.02

-.02

Only one biological parent


No biological parent
Parents education

.02
.06**

.02
.05*

.02
.06**

.02
.05*

.03
.08*

.02
.06*

-.01

-.01

-.01

-.01

-.02

.01

.00

.00

.00

.02

-.05*

-.05*

Kids in household

.04*

.03*

.03*

.03

.04*

.03*

Disruptive life events

.04**

.04*

.04*

.03

.04*

.03

Part-time
Unemployed

-.04

-.05*

-.03

-.02
.01
-.04*

Latent traits
Cigarette smoking

.25**

.19**

.25**

.19**

.26**

.19**

Anger

.18**

.12**

.19**

.11**

.20**

.12**

Social control
Parental attachment

-.04**

-.05**

-.06**

Parental control

-.09**

-.08**

-.11**

.23**

.23**

.24**

Peer delinquency
Prior school misconduct
Wave 1 school misconduct
Variable

.27**

.21

Emotional
Model 1

.26**

.21**

Behavioral
Model 2

Model 3

.28**

.22**

Cognitive
Model 4

Model 5

Model 6

.01

-.23**

-.21**
-.03

Equation 2: School misconduct predicting school engagement


School misconduct
Wave 2 school misconduct
Demographics
Age

.02

.02

-.03

-.03

-.01

.00

.02

-.05*

Male

.01

.01

-.06**

-.04*

.01

Black

-.03

-.06**

-.05*

-.04*

.11**

Asian

-.02

.00

.11*

.12**

-.04*

.00
.10**
-.02

Only one biological parent

.01

.01

-.01

-.01

.00

.00

No biological parent

.01

.03

.00

.00

.00

.01

Parents education

.03

.02

.01

.00

.03

.02

Part-time

.01

.02

-.03

-.03

-.03

-.02

Unemployed

-.01

-.01

-.01

-.01

-.06**

-.06*

Kids in household

.01

.02

-.02

-.02

-.02

-.02

Disruptive life events

.02

.02

-.04*

-.04*

.02

.01

Cigarette smoking

-.02

-.01

-.05*

-.03

.03

.04

Anger

-.13**

-.08**

-.10**

-.07**

-.00

.04

Latent traits

Social control

123

14

J Youth Adolescence (2011) 40:322

Table 3 continued
Variable

Emotional
Model 1

Behavioral
Model 2

Parental attachment

.17**

Parental control

.07**

Peer delinquency

.00

Model 3

Cognitive
Model 4

Model 5

-.03

Model 6
.14**

.16**

.06**

-.08**

.01

Prior school engagement


Wave 1 emotional

.35**

.32**

Wave 1 behavioral

.30**

.28**

Wave 1 cognitive
RMSEA

.04

.04

.03

.03

.31**

.30**

.04

.03

Standard errors are in parentheses. Control variables are measured at Wave 2 for models including behavioral and cognitive engagement and
Wave 1 for models including emotional engagement
* p \ .05; ** p \ .01

results for model 3 and 4 show that more homework time


robustly predicts less general delinquency. Likewise, cognitive engagement predicts off-campus delinquency as
strongly as it predicts school misconduct (models 5 and 6).
Interactions between engagement and gender, race, and
grade in OLS models (not shown) were not significant,
indicating that the effects of engagement on delinquency
do not vary across subgroups.
Similar to the corresponding models in Table 3, models
1 through 4 in the second half of Table 4 suggest that
general delinquency has no independent influence on subsequent emotional and behavioral engagement. Likewise, it
exerts substantial negative effects on cognitive engagement
(models 5 and 6).
We examined whether each type of engagement has
unique effects on delinquency when the other two are
included simultaneously. Unfortunately, an SEM model
that included reciprocal relationships between all three
engagement measures and delinquency would not converge. However, an SEM model that included behavioral
and cognitive engagement but not emotional engagement
would converge. The effect of behavioral engagement was
unchanged, and cognitive engagement had a positive a
highly significant effect on delinquency. Since emotional
engagement, unlike behavioral and cognitive engagement
failed to significantly predict delinquency, we would not
expect emotional engagement to markedly alter the impact
of the behavioral or cognitive dimensions.
To further examine this issue, all three measures were
included in an OLS regression model. Although this model
does not take into account the reciprocal relationships
among the engagement and delinquency measures, the
sizable impact of behavioral engagement on both measures
of delinquency is evident in supplementary OLS models

123

even after cognitive and emotional engagement (wave 1 or


wave 2) are controlled. Cognitive engagement is negatively
related to delinquency in these models (and in OLS models
without the other engagement measures) which apparently
reflects the negative contemporaneous influence of delinquency on cognitive engagement. Taken together, these
findings seem to suggest that behavioral and cognitive
engagement have independent effects on delinquency.

Discussion
Disengagement from school, which often begins during
elementary school, is widely recognized as the principal
long-term social-psychological process that turns motivated students into high school dropouts (Alexander et al.
2001; Finn 1989; Steinberg et al. 1996). Scholars also tend
to agree that this multifaceted problem, however defined
and wherever it originates, is more pronounced and more
costly in inner cities, where dropouts have a relatively slim
chance of a healthy and productive future (see Orfield
2004). The scientific community is still a long way from
consensus, however, regarding whether disengagement is
also a primary and direct cause of delinquency, especially
outside of school. The first obstacle to reliable causal
inference is that measures of school bonds rooted in
social control theory are not completely consonant with
prevailing conceptualizations of engagement among
scholars of education, which are themselves beset with
inconsistency or ambiguity. Second, research linking
school bonds/engagement to delinquency is vulnerable to
the criticism that latent factors explain both engagement
and delinquency. Third, delinquency may lead to disengagement rather than the other way around. Fourth, most

J Youth Adolescence (2011) 40:322

15

Table 4 Parameter estimates from simultaneous effects models of the reciprocal relationship between school engagement and school misconduct, ML estimates: Wave 2
Variable

Model 1

Model 2

Model 3

Model 4

-.12**

-.11**

Model 5

Model 6

.11**

.10**
.01

Equation 1: School engagement predicting general delinquency


School engagement
Wave 2 emotional

-.08*

-.05

Wave 2 behavioral
Wave 2 cognitive
Demographics
Age

.04**

.01

.05**

.01

.05**

Male

.10*

.07**

.09**

.06**

.10**

Black

.01

.01

.00

.01

Asian

.00

.00

.02

.02

.01

.00

Only one biological parent


No biological parent

.00
.01

.00
.00

.00
.01

-.01
-.01

.01
.03

.00
.00

-.01

-.01

-.01

-.01

-.01

-.01

.01

.00

.00

.00

.02

.01

-.02

-.02

-.02

-.02

.00

-.01

Kids in household

.03

.02

.02

.02

.03

.02

Disruptive life events

.05**

.04**

.04**

.03*

.05**

.04**

Cigarette smoking

.26**

.19**

.26**

.19**

.27**

.19**

Anger

.28**

.20**

.28**

.20**

.29**

.20**

Parents education
Part-time
Unemployed

-.01

.06**
.00

Latent traits

Social control
Parental attachment

-.04*

-.05**

-.06**

Parental control

-.14**

-.12**

-.15**

.29**

.28**

.29**

Peer delinquency
Prior general delinquency
Wave 1 general delinquency
Variable

.28**

.21**

Emotional
Model 1

.28**

.21**

Behavioral
Model 2

Model 3

.30**

.22**

Cognitive
Model 4

Model 5

Model 6

.03

-.21**

-.18**
-.03

Equation 2: General delinquency predicting school engagement


General delinquency
Wave 2 general delinquency
Demographics
Age

-.03

-.04

-.03

-.02

-.01

.00

.02

-.05**

Male

.02

.02

-.06**

-.04*

.01

Black

-.03

-.06**

-.05*

-.04*

.12**

Asian

-.02

.00

Only one biological parent

.01

.01

-.01

No biological parent

.02

.03

.00

Parents education

.03

.02

Part-time

.01

Unemployed

.11**

.12**

.00
.11**

-.03

-.02

-.01

-.01

-.01

.00

-.02

.00

.01

.00

.03

.03

.02

-.03

-.03

-.03

-.02

-.01

-.01

-.01

-.01

-.06*

-.05*

Kids in household

.02

.02

-.02

-.02

-.02

-.02

Disruptive life events

.02

.02

-.04*

-.04*

.02

.02

.01

-.05*

-.03

.03

.03

-.07**

-.10**

-.08**

.03

.05

Latent traits
Cigarette smoking

-.01

Anger

-.12**

123

16

J Youth Adolescence (2011) 40:322

Table 4 continued
Variable

Emotional
Model 1

Behavioral
Model 2

Model 3

Cognitive
Model 4

Model 5

Model 6

Social control
Parental attachment

.16**

-.03

.14**

Parental control
Peer delinquency

.06**
.01

.17**
-.09**

.06**
.01

Prior school engagement


Wave 1 emotional

.35**

.32**

Wave 1 behavioral

.30**

.28**

Wave 1 cognitive
RMSEA

.04

.04

.04

.01

.31**

.30**

.03

.01

Standard errors are in parentheses. Control variables are measured at Wave 2 for models including behavioral and cognitive disengagement
and Wave 1 for models including emotional disengagement
* p \ .05; ** p \ .01

studies on this topic focus on the high school period,


generally after trajectories of engagement and delinquency
are established (Alexander et al. 2001; Farrington and
Welsh 2007) and, perhaps, have impacted each other most
decisively. In short, although it is clear that engagement
and delinquency are linked, the exact causal nature of their
relationship is uncertain.
The present study attempted to reduce or minimize all
of these threats to causal inference. Drawing a panel of
Chicago fifth to eighth graders, it estimated the separate
influence of the emotional, behavioral, and cognitive
dimensions of engagement on delinquency, both on and off
campus. Improving the reliability of causal inferences, we
estimated models that accounted for both the possibility of
a reciprocal relationship between engagement and delinquency, and the influence of myriad potential confounding
factors like psychological, peer, and family problems. In
short, our findings qualify the insight that delinquency and
engagement affect each other. We find evidence that both
emotional and behavioral engagement have an effect on
delinquency, whereas the impact of delinquency on disengagement appears relatively short-term and limited to
cognitive engagement.
To test our first hypothesis that engagement affects
delinquency, two sets of models were estimatedcrosslagged and simultaneous. The cross-lagged models ensured
proper sequencing of disengagement and subsequent
delinquency but the imposition of a 1 year lag between
predictor and outcome likely depressed the actual relationship between these variables. Nevertheless, we found
that behavioral disengagement is significantly and independently predictive of increased school and general misconduct regardless of whether the outcome (past years

123

delinquency) is measured in the same survey wave or a


year later. This finding extends prior cross-sectional analyses on this issue (e.g., Huebner and Betts 2002), although
assessing whether homework reduces misconduct via
increased controls or reduced motivation to offend requires
further research.
The direct effects of emotional engagement at wave 1,
although modest, affirm and extend the findings of prior
longitudinal research with older student samples. They
suggest that, at least for African-Americans (Liska and
Reed 1985) and residents of disadvantaged communities
(Dornbusch et al. 2001), strengthening school attachment
can reduce delinquency. The diminished effect of attachment when peer and parental variables are included could
suggest that a decline in parental control and a corresponding increase in the influence of deviant peers results
in both emotional disengagement and delinquency (Jang
1999). Alternatively, parental control and peer delinquency
may be preceding or mediating forces along the same
causal pathway as emotional engagement. More elaborate
mediational analyses with at least three waves of data can
assess whether peer and family dynamics mediate the
deviance-reducing effects of emotional engagement, or
vice versa.
The most curious result is the positive effect of cognitive engagement (W2) on school and general misconduct (W2). This result seems counter-intuitive, because
most items, including cognitive effort in response to
failure, index commitment to school and its values
(Hirschi 1969). However, increased effort and high future
performance expectations may also result in frustration
and lowered school attachment if improved performance
does not result. Increased frustration at school, in turn,

J Youth Adolescence (2011) 40:322

may increase delinquency (Agnew and White 1992). A


less intriguing explanation is measurement errors on the
items asking about active coping in response to poor
school performance. The direct positive relationship
between cognitive engagement and misconduct may be
driven by students who rarely do poorly on school work
or misbehave, but who, nonetheless, accurately report that
they infrequently use active coping methodsbecause
they rarely have occasion to use them. Perhaps for similar
reasons, being African-American predicts more cognitive
engagement, although sampled black students are more
likely to drop out. However, models that controlled for
standardized test scores (not shown) did not alter the
conclusion about the positive effect of cognitive engagement on misconduct. Future research that directly examines the role of expectations and frustrations may lend
clarity to this issue.
Our second hypothesis was that misconduct, especially
school misconduct, lowers all three types of engagement.
This study is among the first to account for and directly
assess the bi-directionality of the relationship between
disengagement and delinquency. Unlike Liska and Reed
(1985) and Thornberry et al. (1991), this study attributes
very little of the negative relationship between engagement/school bonds and delinquency to the effects of the
latter on the former. Whereas these prior studies employ
composites of multiple dimensions of disengagement/
school bonds, the present findings suggest that misconduct
decreases only cognitive engagement. One possible
explanation for the difference in findings is that these
studies omitted potentially important risk factors for both
disengagement and delinquency, including stressful events,
peer relations, and relatively stable indicators of risk
proneness.
Although our results diverge from studies that find an
effect of delinquency on school bonds, the null effects of
delinquency on behavioral engagement are more consistent
with prior research. This finding may underscore the
gradual, cumulative nature of disengagement; future disengagement is largely a function of past disengagement
(Steinberg and Avenevoli 1998). Reflecting on similar
findings observed among male working and lower class
Whites in sixth and seventh grades, DeBaryshe et al.
(1993) suggested that, during these grades, children may
attend to schoolwork because of rules or structure imposed
by adults, rather than following their own preferences or
strategies (p. 802). Accordingly, DeBaryshe et al. suggested that parental disciplinary effectiveness may have an
overriding influence on early adolescents engagement in
school work. Future research should assess whether such
patterns persist during high school as parents lose influence
and sanctions for deviance grow more frequent and
exclusionary.

17

The present study also added cognitive disengagement


to the potential negative outcomes of delinquency. The fact
than school misconduct apparently impairs cognitive
engagement rather than behavioral engagement may be due
to the greater conceptual similarity that cognitive disengagement shares with school misconduct. Skipping school
and cheating are indicative of diminished academic motivation (Jimerson et al. 2003), so it is hardly surprising that
such students report that they less often invest cognitive
energy in school. Behavioral engagement references only
off-campus behavior, so it is naturally less responsive to
school misconduct.
However, the finding that general delinquency also
reduces cognitive but not behavioral engagement is surprising and may suggest the importance of an underlying
explanation of both general delinquency and cognitive
disengagement. We attempted to account for this possibility by including in our models cigarette smoking and
anger levels. Neither significantly predicts cognitive
engagement, and they are inconsistently related to misconduct. Future research in this area should consider other
latent factors (e.g., impulsivity) or contextually-conditioned developmental process (e.g., individuation and
rebellion) that may condition changes in both problem
behaviors.
Although an investigation of the relationship between
engagement and delinquency in early adolescence, this
study makes several contributions to research on the effects
of some control variables. Most noteworthy in this connection are the literatures on the impact of parental rulesetting and monitoring and that of parental attachment on
academic engagement. Considerable research, including
recent work on disadvantaged urban youth, has suggested
that firm and consistent monitoring and rule enforcement
(Steinberg et al. 1996) as well as parental closeness,
acceptance, and communication (Furrer and Skinner 2003;
Garcia-Reid et al. 2005; Murray 2009) promote academic
engagement. However, most studies do not distinguish
between the effects of these two dimensions of parenting or
measure their effects on distinct dimensions of engagement. In fact, one of the few rigorous longitudinal studies
on the topic finds, among African-American students, no
impact of a composite measure of acceptance, strictness/
supervision, and autonomy on a four-dimensional measure
of engagement (Steinberg et al. 1992). The present findings
suggest that these relationships may be domain-specific, at
least among African-American students. Specifically,
although parental attachment and control both significantly
predict greater emotional and cognitive engagement,
parental attachment exhibits a stronger impact. Conversely,
only parental control significantly increases behavioral
engagement. Our findings support the arguments of Furrer
and Skinner (2003) that supportive relationships with

123

18

parents can help children see other adults as trustworthy


and give children the confidence to respond to challenge
or difficulties with more vigor, flexibility, and constructive actions (p. 148). Likewise, they concord with
DeBaryshe et al. (1993) who, as noted earlier, attribute
engagement in homework to rules or structure imposed by
adults.
Some limitations bear mention and, hopefully, will be
remedied in future research. The true effects of engagement and delinquency on each other may be underestimated. If disengagement begins before fifth grade and is
cumulative, controlling for prior delinquency may absorb
some of the impact of earlier academic disengagement on
delinquency. Using data that also covers earlier grades and
measuring disengagement and problem behaviors at intervals of less than 1 year can address this limitation. On the
other hand, this study, despite its relatively expansive set of
control variables, may still be vulnerable to selection bias.
This bias may be reduced by controlling for intelligence,
competence, social integration, and parental disciplinary
methods. A fixed effects analysis or a quasi-experimental
design, which would address important, unobserved differences between students would also be helpful in this
connection.
Future research can also extend the present findings by
examining whether they obtain in other contexts such as
middle class and suburban schools, or in elementary
school and high school. Prior research provides ample
grounds to posit variable patterns by race, gender, grade,
and the type of delinquency examined (Cernkovich and
Giordano 1992; Payne 2008). Although this study found
no evidence of differences in the effects of engagement
by gender, race, or grade level, the urban nature of our
sample and limited timeframe (5th through 8th grade)
precludes us from drawing any firm conclusions about
these issues. Given the complex nature of the engagement
construct, it is also important to examine how reliable are
the observed patterns across alternate operationalizations
of engagement. Finally, research should systematically
examine mediation. Parental and peer relationships,
aspirations, strain, and achievement merit consideration as
mediators of the impact on disengagement on delinquency. Measurement can also be improved by employing
measures of key constructs obtained from multiple
informants.
As various effective school interventions attest,
engagement and delinquency are not fixed by constitutional factors (Gottfredson 2001; Lehr et al. 2004).
Consistent with such findings, the present study suggests

123

J Youth Adolescence (2011) 40:322

that improving engagement, especially behavioral engagement, will also reduce school and general delinquency.
Moreover, because the effects of engagement are direct,
reducing delinquency through increasing behavioral
engagement does not appear to require altering patterns of
peer and parental relationships. Our findings also caution
that that improving cognitive engagement without also
improving opportunities and resources for academic success may actually increase delinquency. Concordantly,
multi-modal approaches to improving student engagement
have met with some success. Effective or promising
interventions to improve elementary school student
engagement, too abundant to list here, include training
parents to reinforce homework completion (Debaryshe
et al. 1993), improving school social climate through
integrating parents and mental health professionals into
schools (Cook et al. 2000; shown to both increase school
attachment and reduce delinquency) and working directly
with students to build trust, communication, and problemsolving skills (Lehr et al. 2004).
Scholars of adolescent development have increasingly
defined the problem of high school dropout as essentially
a problem of academic disengagement. Fewer scholars,
however, have advanced similar claims with respect to
juvenile delinquency or other problem behaviors, such as
alcohol and drug use. Yet, disengagement, dropout, and
delinquency/problem behavior appear to go hand-in-hand
(Finn 1989). This lack of emphasis on disengagement
may reflect the prominence of strain, social bonding, and
problem behavior theories of schooling and delinquency.
It may also reflect the ambiguities and inconsistencies in
the measurement of disengagement and in their relationship with delinquency. This article lends much needed
clarity to these matters by distinguishing between
engagement and school bonding, by distinguishing among
dimensions of engagement, and by disentangling causal
order. It suggests that engaging early adolescents in
homework carries the greatest pay-off in terms of
reducing misbehavior both inside and outside of school.
Improving their emotional dispositions toward school is
also helpful but not as much, whereas improving cognitive engagement, by itself, is not helpful at all. Interventions designed to improve engagement should be more
attentive to such differences in the importance of the
various dimensions of engagement.
Acknowledgments The authors are grateful to Helene Raskin White,
the Editor, and three anonymous reviewers for their helpful feedback
on this manuscript and to Thomas Cook for providing the data.

J Youth Adolescence (2011) 40:322

19

Appendix
See Table 5.

Table 5 Scales/items from the comer school development program surveys


Scale
General delinquency

Items

Range

In the past year, how often have you?

1 = Never; 5 = 10 or more times

Hit someone because you didnt like something they said or


did (past year)
Damaged something that did not belong to you just for fun
Done some pretty dangerous things just for fun
Lied to parents about where you have been or who you
were with
Taken something from a store without paying for it
Stolen or tried to steal a motor vehicle such as car or
motorcycle
Gotten into a gang fight
School misconduct

In the past year, how often have you?

1 = Never; 5 = 10 or more times

Cheated on tests or exams


Brought alcohol or drugs to school
Skipped a day of school without a real good excuse
Been sent to the principals office because you had done
something wrong
Emotional engagement

If you had to stop going to this school how much would


you miss

1 = a lot; 5 = not at all

Your fellow students


Your teachers
The principal
Other school staff
The sense of school spirit
The way this school treats students
Behavioral engagement

On a school day, how many hours do you spend?

1 = none; 3 = 23 h; 5 = 67 h

Hanging out with a bunch of friends


Talking on the telephone
Just sitting and listening to music
Watching television or videos
Playing computer or video games
Playing pickup games outside of school
Doing homework
Cognitive engagement

When you dont do well on schoolwork (tests, answering


class questions), how often do you?

1 = all of the time; 5 = never

Tell yourself you will do better next time


Try to think of what you did wrong so it wont happen
again
Study harder to do better next time
How well can you learn math?

1 = not at all well; 5 = very well

How well can you learn reading and writing skills?


How well can you remember things taught in class and
school books?
School is boring (Reverse-coded)
It is important to do well in school

1 = strongly disagree; 5 = strongly


agree

123

20

J Youth Adolescence (2011) 40:322

Table 5 continued
Scale
Disruptive life events

Items

Range

In the last year which of these things happened

1 = yes; 0 = no

My mother got a new boyfriend


One or both my parent lost their job
Someone I knew well died
One of my parents or stepparents moved out.
Anger

In the last year, how often have you felt?

1 = never; 5 = everyday

You couldnt control your temper


Really mad at other people
So upset you wanted to hit or hurt someone
So angry you wanted to smash or break something
Parental attachment

My parents
Trust me

1 = strongly disagree; 5 = strongly


agree

Approve of my friends
Discuss my problems with me
Respect my ability to make decisions
Give me small jobs to do without checking on me a lot
Discuss important family matters with me
Parental control

In your home, how often do you have to follow rules


about

1 = never; 5 = all of the time

Letting your family know where you are going when you
go outside
Who you spend your free time with
Getting your homework done,
What time you should be home at night
What time you go to bed at night.
Having friends over when no grownups are home
Peer delinquency

Think of the friends you spend most of your time with.


During the last year, how many of them

1 = none; 5 = all

Asked you to do something against the law


Damaged or destroyed something that didnt belong to
them
Stole anything
Sold drugs of any kind
Took drugs

References
Agnew, R. (1991). A longitudinal test of social control theory and
delinquency. Journal of Research in Crime and Delinquency,
28(2), 126156.
Agnew, R., & White, H. R. (1992). An empirical test of general strain
theory. Criminology, 30(4), 475499.
Alexander, K. L., Entwisle, D. R., & Kabbani, N. S. (2001). The
dropout process in life course perspective: Early risk factors at
home and school. Teachers College Record, 103, 760822.
Allison, P. D. (2002). Missing date. Thousand Oaks, CA: Sage
Publications.
Armstrong, T. A., Lee, D. R., & Armstrong, G. S. (2009).
Criminological theory based on national youth survey data an
assessment of scales measuring constructs in tests of criminological theory based on National Youth Survey data. Journal of
Research in Crime and Delinquency, 46(1), 73105.

123

Astone, N. M., & McLanahan, S. S. (1994). Family structure,


residential mobility, and school dropout: A research note.
Demography, 31, 575584.
Barnes, G. M., Hoffman, J. H., Welte, J. W., Farrell, M. P., &
Dintcheff, B. A. (2007). Adolescents time use: Effects on
substance use, delinquency, and sexual activity. Journal of Youth
and Adolescence, 36, 697710.
Buhs, E. S. (2005). Peer rejection, negative peer treatment, and school
adjustment: Self-concept and classroom engagement as mediating processes. Journal of School Psychology, 43, 407424.
Catalano, R. F., Oxford, M., Harachi, T., Abbott, R., & Haggerty, K.
P. (1999). A test of the social development model to predict
problem behaviour during the elementary school period. Criminal Behaviour and Mental Health, 9, 3956.
Cernkovich, S. A., & Giordano, P. C. (1979). Delinquency, opportunity and gender. Journal of Criminal Law and Criminology,
70, 145151.

J Youth Adolescence (2011) 40:322


Cernkovich, S. A., & Giordano, P. C. (1992). School bonding, race,
and delinquency. Criminology, 30(2), 261291.
Cloward, R. A., & Ohlin, L. E. (1960). Delinquency and opportunity.
New York: The Free Press.
Cohen, A. K. (1955). Delinquent boys: The culture of the gang.
Glencoe, IL: The Free Press.
Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy, and
relatedness: A motivational analysis of self-system processes. In
M. Gunnar & A. Sroufe (Eds.), Minnesota symposium on child
psychology. Chicago: University of Chicago Press.
Conner, B. T., Stein, J. A., & Longshore, D. (2009). Examining selfcontrol as a multidimensional predictor of crime and drug use in
adolescents with criminal histories. Journal of Behavioral
Health Services & Research, 36(2), 137149.
Cook, T. D., Murphy, R. F., & Hunt, H. D. (2000). Comers school
development program in Chicago: A theory-based evaluation.
American Educational Research Journal, 37(2), 535597.
Crosnoe, R. (2006). The connection between academic failure and
adolescent drinking in secondary school. Sociology of Education,
79, 4460.
Crosnoe, R., Erickson, K., & Dornbusch, S. M. (2002). Protective
functions of family relationships and school factors on the
deviant behavior of adolescent boys and girls: Reducing the
impact of risky friendships. Youth and Society, 33, 515544.
DeBaryshe, B. D., Patterson, G. R., & Capaldi, D. (1993). A
performance model for academic achievement in early adolescent boys. Developmental Psychology, 29, 795804.
Donovan, J. E., & Jessor, R. (1985). Structure of problem behavior in
adolescence and young adulthood. Journal of Consulting and
Clinical Psychology, 53, 890904.
Dornbusch, S. M., Erickson, K. G., Laird, J., & Wong, C. A. (2001).
The relation of family and school attachment to adolescent
deviance in diverse groups and communities. Journal of
Adolescent Research, 16(4), 396422.
Eccles, J. S., & Barber, B. L. (1999). Student council, volunteering,
basketball or marching band: What kind of extracurricular
involvement matters? Journal of Adolescent Research, 14(1),
1043.
Fagan, J., & Pabon, E. (1990). Contributions of delinquency and
substance use to school dropout among inner-city youths. Youth
and Society, 21, 306354.
Farrington, D. P., & Welsh, B. C. (2007). Saving children from a life
of crime: Early risk factors and effective interventions. New
York: Oxford University Press.
Felice, L. G. (1981). Black student dropout behavior: Disengagement
from school, rejection, and racial discrimination. The Journal of
Negro Education, 50, 415424.
Felson, R. B., & Staff, J. (2006). Explaining the academic
performance-delinquency relationship. Criminology, 44,
229320.
Finkel, S. E. (1995). Causal analysis with panel data. Thousand Oaks,
CA: Sage Publications.
Finn, J. D. (1989). Withdrawing from school. Review of Educational
Research, 59, 117142.
Ford, J. A. (2005). Substance use, the social bond, and delinquency.
Sociological Inquiry, 75, 109128.
Fredricks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School
engagement: Potential of the concept, state of the evidence.
Review of Educational Research, 74, 59109.
French, D. C., & Conrad, J. (2001). School dropout as predicted by
peer rejection and antisocial behavior. Journal of Research on
Adolescence, 11, 225244.
Furrer, C., & Skinner, C. (2003). Sense of relatedness as a factor in
childrens academic engagement and performance. Journal of
Educational Psychology, 95, 148162.

21
Garcia-Reid, P., Reid, R. J., & Peterson, N. A. (2005). School
engagement among Latino youth in an urban middle school
context: Valuing the role of social support. Education and Urban
Society, 37, 257275.
Gottfredson, D. C. (2001). Schools and delinquency. Cambridge:
University Press.
Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime.
Stanford, CA: Stanford University Press.
Hagan, J. (1991). Destiny and drift: Sub cultural preferences, status
attainment, and the risks and rewards of youth. American
Sociological Review, 56, 567582.
Hirschfield, P. (2009). Another way out: The impact of juvenile
arrests on high school dropout. Sociology of Education, 82,
368393.
Hirschi, T. (1969). Causes of delinquency. Berkeley, CA: University
of California Press.
Hjalmarsson, R. (2008). Criminal justice involvement and high school
completion. Journal of Urban Economics, 63, 613630.
Huebner, A. J., & Betts, S. C. (2002). Exploring the utility of social
control theory for youth development. Youth and Society, 34,
123145.
Jang, S. J. (1999). Age-varying effects of family, school, and peers on
delinquency: A multilevel modeling test of interactional theory.
Criminology, 37, 642686.
Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial
development: A longitudinal study of youth. New York:
Academic Press.
Jimerson, S. R., Campos, E., & Greif, J. L. (2003). Toward an
understanding of definitions and measures of school engagement
and related terms. The California School Psychologist, 8, 727.
Laird, R. D., Jordan, K. Y., Dodge, K. A., Pettit, G. S., & Bates, J. E.
(2001). Peer rejection in childhood, involvement with antisocial
peers in early adolescence, and the development of externalizing
behavior problems. Development and Psychopathology, 13,
337354.
Lehr, C. A., Sinclair, M. F., & Christenson, S. L. (2004). Addressing
student engagement and truancy prevention during the elementary years: A replication study of the Check and Connect model.
Journal of Education for Students Placed At Risk, 9(3), 279301.
Liska, A. E., & Reed, M. D. (1985). Ties to conventional institutions
and delinquency: Estimating reciprocal effects. American Sociological Review, 50, 547560.
Lubbersa, M. J., Van Der Werfa, M., Snijdersb, T., Creemersc, B. P.,
& Kuypera, H. (2006). The impact of peer relations on academic
progress in junior high. Journal of School Psychology, 44,
491512.
Maguin, E., & Loeber, R. (1996). Academic performance and
delinquency. Crime and Justice, 20, 145264.
Manlove, J. (1998). The influence of high school dropout and school
disengagement on the risk of school-aged pregnancy. Journal of
Research on Adolescence, 82, 187220.
Marks, H. M. (2000). Student engagement in instructional activity:
Patterns in the elementary, middle, and high school years.
American Educational Research Journal, 37, 153184.
Menard, S., & Morse, B. J. (1984). A structural critique of the
IQ-delinquency hypothesis. American Journal of Sociology,
89(6), 13471378.
Mensch, B. S., & Kandel, D. B. (1988). Dropping out of high school
and drug involvement. Sociology of Education, 61, 95113.
Murray, C. (2009). Parent and teacher relationships as predictors of
school engagement and functioning among low-income urban
youth. The Journal of Early Adolescence, 29, 376404.
National Research Council and Institute of Medicine. (2003).
Engaging schools: Fostering high school students motivation
to learn. Washington, DC: National Academy Press.

123

22
Natriello, G. (1984). Problems in the evaluation of students and
student disengagement from secondary school. Journal of
Research and Development in Education, 17(4), 1424.
Orfield, G. (2004). Losing our future: Minority youth left out. In
G. Orfield (Ed.), Dropouts in America: Confronting the graduation rate crisis. Cambridge, MA: Harvard Education Press.
Padilla, F. M. (1992). The gang as an American enterprise. New
Brunswick, NJ: Rutgers University Press.
Payne, A. A. (2008). A multilevel analysis of the relationships among
communal school organization, student bonding, and delinquency. Journal of Research in Crime and Delinquency, 45(4),
429455.
Pettit, B., & Western, B. (2004). Mass imprisonment and the life
course: Race and class inequality in US incarceration. American
Sociological Review, 69, 151169.
Ramsey, E., Patterson, G. R., & Walker, H. M. (1990). Generalization
of the antisocial trait from home to school settings. Journal of
Applied Developmental Psychology, 11, 209223.
Rankin, J. H., & Wells, E. L. (1990). The effect of parental
attachments and direct controls on delinquency. Journal of
Research in Crime and Delinquency, 24, 140172.
Sampson, R. J., & Laub, J. H., (1993). Crime in the making: Pathways
and turning points through life. Boston, MA: Harvard University
Press.
Sampson, R. J., & Laub, J. H. (1997). A life course theory of
cumulative disadvantage and the stability of delinquency. In
T. Thornberry (Ed.), Developmental theories of crime and
delinquency. New Brunswick, NJ: Transaction Publishing.
Schmidt, J. A. (2003). Correlates of reduced misconduct among
adolescents facing adversity. Journal of Youth and Adolescence,
32(6), 439452.
Steinberg, L., & Avenevoli, S. (1998). Disengagement from school
and problem behavior in adolescence: A developmental-contextual analysis of the influences of family and part-time work. In
R. Jessor (Ed.), New perspectives on adolescent risk behavior.
New York, NY: Cambridge University Press.
Steinberg, L., Brown, B. B., & Dornbusch, S. (1996). Beyond the
classroom. New York: Simon and Schuster.

123

J Youth Adolescence (2011) 40:322


Steinberg, L., Lamborn, S. D., Dornbusch, S. M., & Darling, N.
(1992). Impact of parenting practices on adolescent achievement: Authoritative parenting, school involvement, and encouragement to succeed. Child Development, 63, 12661281.
Thornberry, T. P. (1987). Toward an interactional theory of delinquency. Criminology, 25(4), 863892.
Thornberry, T. P., Lizotte, A. P., Krohn, M. D., Farnworth, M., &
Joon Jang, S. (1991). Testing interactional theory: An examination of reciprocal causal relationships among family, school, and
delinquency. Journal of Criminal Law and Criminology, 82,
335.
Wilson, J. Q., & Hernstein, R. J. (1985). Crime and human nature.
Simon and Schuster: New York.
Wong, S. K. (2005). The effects of adolescent activities on
delinquency: A differential involvement approach. Journal of
Youth and Adolescence, 34(4), 321333.

Author Biographies
Paul J. Hirschfield is an Associate Professor in the Department of
Sociology, Rutgers University, Piscataway, NJ. He earned his
doctorate in Sociology from Northwestern University. His theoretical
and empirical work focuses on the control and criminalization of
youth in the contexts of schools and the juvenile justice system, along
with the effectiveness of schools in the prevention of delinquency and
recidivism. His current research centers on the reintegration of youth
from correctional facilities into schools.
Joseph Gasper received his Ph.D. in Sociology from Johns Hopkins
University. He currently works as a Research Associate at Westat in
Rockville, MD. His research interests include high school dropout,
student engagement, and adolescent delinquency and substance use.
In addition, he has conducted research on the effects of residential and
school mobility on youth educational and behavioral outcomes.

You might also like