Content uploaded by Krista S Gehring
Author content
All content in this area was uploaded by Krista S Gehring on Nov 14, 2015
Content may be subject to copyright.
EMPIRICAL RESEARCH
Sex Differences in Sources of Resilience and Vulnerability to Risk
for Delinquency
Jamie Newsome
1
•Jamie C. Vaske
2
•Krista S. Gehring
3
•Danielle L. Boisvert
4
Received: 17 August 2015 / Accepted: 25 October 2015
ÓSpringer Science+Business Media New York 2015
Abstract Research on adolescent risk factors for delin-
quency has suggested that, due to genetic differences,
youth may respond differently to risk factors, with some
youth displaying resilience and others a heightened vul-
nerability. Using a behavioral genetic design and data from
the National Longitudinal Study of Adolescent to Adult
Health, this study examines whether there are sex differ-
ences in the genetic and environmental factors that influ-
ence the ways in which adolescents respond to cumulative
risk for violent, nonviolent, and overall delinquency in a
sample of twins (152 MZ male, 155 MZ female, 140 DZ
male, 130 DZ female, and 204 DZ opposite-sex twin pairs).
The results revealed that males tended to show greater
vulnerability to risk for all types of delinquency, and
females exhibited greater resilience. Among males, addi-
tive genetic factors accounted for 41, 29, and 43 % of the
variance in responses to risk for violent, nonviolent, and
overall delinquency, respectively. The remaining propor-
tion of variance in each model was attributed to unique
environmental influences, with the exception of 11 % of
the variance in nonviolent responses to risk being attributed
to common environmental factors. Among females, no
significant genetic influences were observed; however,
common environmental contributions to differences in the
ways females respond to risk for violent, nonviolent, and
overall delinquency were 44, 42, and 45 %, respectively.
The remaining variance was attributed to unique environ-
mental influences. Overall, genetic factors moderately
influenced males’ responses to risk while environmental
factors fully explain variation in females’ responses to risk.
The implications of these findings are discussed in the
context of improving the understanding of relationships
between risks and outcomes, as well as informing policy
and practice with adolescent offenders.
Keywords Add Health Behavioral genetics
Resilience Risk Sex differences Vulnerability
Introduction
Adolescence is a critical period of development, during
which individuals encounter a variety of individual, social,
and environmental factors that shape their behavioral tra-
jectories. As a result, there has been substantial scholarly
interest devoted to identifying risk factors for delinquency.
More recently, however, there has been a recognition that
the relationships between risks and behavioral outcomes
are quite complex (Wikstro
¨m2008). Individuals may be
differentially impacted by risk factors, with some dis-
playing a heightened vulnerability and others remaining
seemingly unaffected, or resilient. It is further possible that
youth influenced to a similar extent may elect to respond to
risks in different ways. Though some individuals may
respond to risks by engaging in mild forms of antisocial
behavior, others may become aggressive or violent. The
&Jamie Newsome
Jamie.Newsome@utsa.edu
1
Department of Criminal Justice, University of Texas at San
Antonio, 501 W. Ce
´sar E. Cha
`vez Blvd., San Antonio,
TX 78207, USA
2
Department of Criminology and Criminal Justice, Western
Carolina University, Belk 411F, Cullowhee, NC 28723, USA
3
Department of Criminal Justice, University of Houston-
Downtown, 1002 Commerce St. Suite C340, Houston,
TX 77002, USA
4
College of Criminal Justice, Sam Houston State University,
Huntsville, TX 77341, USA
123
J Youth Adolescence
DOI 10.1007/s10964-015-0381-2
heterogeneity in responses to risks is particularly evident
when comparing male and female offendingpatterns—both in
terms of overall participation in crime and by type of delin-
quent involvement (Federal Bureau of Investigation 2015).
Such variation, both generally and by sex, suggests that lim-
iting investigations to identifying risk factors may stifle our
capacity to fully understand behavioral development.
Researchers have now begun to shift focus in order to
uncover the sources of variation in individual responses to
risks. Although some researchers have proposed that this
variation may be due to genetic differences between indi-
viduals (Belsky 1997,2005; Boyce and Ellis 2005; Ellis
et al. 2011), few studies have examined the extent to which
genetic factors may account for differences in responses to
risk. Much of the research has instead emphasized the
importance of identifying protective or promotive factors,
which often represent the opposite extremes of traditional
risk factors, as a possible explanation for resilient outcomes
(Farrington and Welsh 2007). Moreover, studies tend to
focus on isolated groups, such as delinquent or resilient
youth only, and do not consider the full range of responses
at varying levels of risk. The present study extends this
body of research by investigating potential sex differences
in genetic and environmental influences on individual
responses to risk while considering the full range of
responses (i.e., resilience to vulnerability) across multiple
forms of delinquency.
Risk, Delinquency, and Errors in Prediction
Prior research on crime has consistently revealed two
patterns. First, a substantial proportion of all crimes are
committed by youthful offenders. A small group of serious,
persistent offenders begin engaging in crime prior to ado-
lescence and persist beyond early adulthood; however, the
majority of offenses are nonviolent crimes committed by
adolescents (Blumstein and Cohen 1987; Farrington 1986;
Moffitt 1993). A second pattern that has reliably been
observed is that the vast majority of offenders—particu-
larly violent offenders—are male (Moffitt et al. 2001). The
most recent statistics from the Uniform Crime Reports
indicate that males accounted for 79.8 % of arrests for
violent offenses and 61.8 % of arrests for property crimes
in 2014, statistics that echo the trends that have been
reported year after year (Federal Bureau of Investigation
2015). These general patterns have received a great deal of
attention from scholars seeking to understand the factors
associated with delinquency.
Following the developmental-ecological framework
(Bronfenbrenner 1979), the effects of several risk factors
for delinquency across individual, family, peer, school, and
neighborhood domains have been investigated. Examples
of such factors include low intelligence,
neuropsychological deficits, low parental supervision and
involvement, low attachment to parents and school, low
academic achievement, delinquent peers, and disadvan-
taged and high crime neighborhoods (Farrington and Welsh
2007). The relationship between risks and adverse out-
comes, however, is complex. Individuals who experience
similar outcomes may have encountered different constel-
lations of risks during development (i.e., equifinality); and,
alternatively, youth exposed to the same risk factors may
manifest very different outcomes (i.e., multifinality; Cic-
chetti and Rogosch 1996). Recognizing that the develop-
mental trajectories of delinquent youth may differ, many
studies have begun investigating the effects of the accu-
mulation of risk factors on different behavioral outcomes.
The creation of a cumulative risk score typically
involves dichotomizing each risk factor (0 =not at-risk,
1=at-risk) at a specified threshold, often the upper or
lower quartile of the distribution that is associated with
greater risk, and then summing the dichotomized risk
factors. This approach acknowledges equifinality in
developmental trajectories, and focuses on the extent of
exposure to risks rather than the specific risks encountered.
From a theoretical perspective, it has been argued that
psychological, cognitive, and social development may be
compromised as the number of risks increase because
coping with multiple challenges can heighten mental and
physiological demands on youth (Evans 2003; Evans et al.
2007). Individuals may be particularly vulnerable to the
negative effects of these demands during adolescence
because sensitivity to stressors may be elevated during the
teenage years (Spear 2009). Indeed, prior research suggests
that internalizing and externalizing behaviors (Appleyard
et al. 2005; Buehler and Gerard 2013; Dekovic
´1999;
Gerard and Buehler 2004), nonviolent and minor forms of
delinquency (Sprott et al. 2005; van der Laan et al. 2010),
serious delinquency (Sprott et al. 2005; Stouthamer-Loeber
et al. 2002; van der Laan et al. 2010), and violence and
aggression (Herrenkohl et al. 2000; Ribeaud and Eisner
2010; Stoddard et al. 2012) have all been found to increase
as the number of risks to which one is exposed increases. In
other words, cumulative risk and adverse outcomes appear
to be linearly related, and these findings have been
observed in both males and females (Fergusson and
Woodward 2000; Sameroff et al. 1998).
There is considerable overlap in the risk factors asso-
ciated with various forms of delinquency among males and
females, but some evidence indicates that there may be
some differences in their pathways to delinquency (Zheng
and Cleveland 2013). The growing body of research into
female delinquency has largely investigated the potential
sex differences in exposure and responses to environmental
risk factors (Belknap and Holsinger 2006; Daigle et al.
2007; Steketee et al. 2013). Although relatively fewer
J Youth Adolescence
123
studies have investigated biological differences that may
account for differences in risk and delinquency by sex, it
has been suggested that the persistent, serious, and violent
forms of delinquency more commonly observed among
males may be the result of the combined influences of
biological and environmental risk factors (Moffitt 1993;
Moffitt et al. 2001; Raine 2002).
These empirical advances have enhanced the under-
standing of the development of delinquency; yet, the
challenge of predicting behavioral outcomes endures as
evidenced by the considerable number of errors in pre-
diction that remain. A number of youth who are exposed to
multiple risk factors are able to remain prosocial (i.e., false
positives), and others who are exposed to little or no risks
become involved in delinquent behaviors (i.e., false nega-
tives). Loeber and Dishion’s (1983) review of over 30
studies found that the false positive rate ranged from 1.9 to
51 % across studies, and the false negative rate was 1.9 to
40.2 %. Along similar lines, Farrington (1995) reported
that approximately 32 % of boys with the highest scores on
a predictive risk factor index were not convicted by their
25th birthday. Other studies have also found that there is a
notable amount of error in efforts to correctly predict
delinquency (especially non-violent delinquency), and to
identify serious juvenile offenders (Giordano 1989; van der
Laan et al. 2010; Welsh et al. 2008).
These findings suggest that, although the development
of some youth follow a predictable trajectory based on
exposure to risks (or a lack thereof), the behavioral out-
comes of others are not fully understood. Errors in pre-
diction are often dismissed as being the result of a number
of methodological limitations that may be present in
empirical investigations; however, some have suggested
that errors in prediction may be indicative of unique groups
of interest (Newsome and Sullivan 2014; Stein et al. 1970).
Examining differences in responses to risks may reveal that
a portion of the errors in prediction are not sporadic or
trivial. Cases that are identified as false positives may
represent resilient youth and, similarly, false negatives
could include youth that are more vulnerable to the effects
of even minimal risk exposure. Examining differences in
responses to risk—including both resilient and particularly
vulnerable youth—may prove useful in understanding
differences in violent and nonviolent offending, explaining
behavioral differences in males and females, reducing
errors in prediction, and providing insights to guide inter-
vention strategies (Sullivan 2011).
Explaining Differential Responses to Risk
Research attempting to understand the development of
youth whose outcomes do not confirm predictions has
focused predominantly on resilience in response to
adversity. A number of studies have highlighted the role of
promotive and protective factors may play in the rela-
tionship between risk factors on psychopathological out-
comes (Fergus and Zimmerman 2005). Similar to risk
factor research, protective and promotive factors are gen-
erally divided into three domains: individual characteris-
tics, attributes of the family or parent–child relationship,
and systems outside of the family (Masten et al. 1990;
Rutter 1990). Studies that have considered the potential
influences of these factors have explored multiple ways by
which promotive and protective factors may impact
development. One possibility is that the relationships
between particular risks and outcomes may be curvilinear,
with one end of the distribution being associated with a
heightened risk and the other with reductions in risk. An
alternative possibility is that some factors have purely
promotive effects and only work to decrease risk for
adverse outcomes. Further still, another mechanism by
which factors could influence development is by moder-
ating the effects of existing risk factors (Rutter 1985).
While the empirical research on the role of protective and
promotive factors is still in the early stages, the available
studies suggest that these factors can influence develop-
ment in meaningful ways (Loeber et al. 2008; Stoddard
et al. 2012; Stouthamer-Loeber et al. 2002; van der Laan
et al. 2010).
In contrast to the widespread interest in investigating the
sources of resilient outcomes, very little scholarly attention
has been given to investigating the causes of delinquency
among youth whose risk profiles suggest they are exposed
to few risks. These cases may indicate a need to expand
existing investigations into the causes of delinquency in
order to achieve a more complete understanding of devel-
opment (Newsome and Sullivan 2014). Much of the
research on risk, protective, and promotive factors to date
has focused on the effects of individual traits and envi-
ronmental factors; however, there are reasons to believe
that genetic and biological factors may also help to explain
the observed differences in responses to risks among
individuals. Theoretically, genetic factors may either
exacerbate or weaken the impact of risk factors on anti-
social behavior through diathesis-stress or protective
biosocial effects, respectively (Monroe and Simons 1991;
Raine et al. 1997). That is, genetic factors provide the
scaffolding for how certain brain components and regions
should be developed and how they should respond to
environmental stimuli. Indeed, candidate gene studies
show that individuals’ resilience and vulnerability may be
linked with their genetic composition and expression
(Caspi et al. 2002; Fergusson et al. 2012; Guo et al. 2008;
Lee 2011).
When tested empirically, twin and family studies have
shown that both genetic and environmental factors may
J Youth Adolescence
123
explain a significant proportion of variation in individuals’
responses to risks. With regard to resilience specifically,
Waaktaar and Torgersen’s (2012) analysis of adolescent
twins showed that additive genetic factors explained
approximately 70 % of the variation in a latent construct of
resilience, and non-shared environmental factors accounted
for approximately 30 % of the variation. Moreover, genetic
influences on resilience was statistically higher among
males (77 %) than females (70 %). Another study by
Boardman et al. (2008) showed that genetic factors
explained 58 % of the variation in psychological resilience
for men, but only 38 % of the variation for women. Finally,
a recent study of a latent construct of psychiatric resilience
to risk also found that genetic and environmental factors
contributed almost equally to variation in resilience (Am-
stadter et al. 2014). This study also investigated possible
sex differences in sources of resilience and found that,
although the magnitude of genetic and environmental
effects did not differ between males and females, the genes
contributing to resilience were qualitatively different.
Other studies have considered a more expansive range
of responses to risk for antisocial outcomes, specifically.
To capture the full range of variation, these studies have
conceptualized differential response to risk as the differ-
ence between individuals’ predicted and observed scores
generated from ordinary least squares regression models
(i.e., standardized residuals). The resulting distribution is
then interpreted such that cases with negative residuals are
considered resilient (less antisocial than predicted), and
those with positive residuals are considered vulnerable
(more antisocial than predicted). In one such study, Kim-
Cohen et al. (2004) examined the association between
socioeconomic status (SES) and antisocial behavior among
5 year-old twins, and found that additive genetic effects
explained 71 % of the variation in children’s responses to
low SES, and non-shared environmental factors explained
the remaining 29 %. A recent study by Newsome and
Sullivan (2014) reported that adolescents showed a range
of responses to cumulative risk for delinquency, and that
differential response to risk was a function of genetic and
non-shared environmental factors. Further analyses, how-
ever, revealed that extreme vulnerability in particular was
predominantly explained by genetic factors, while extreme
resilience was a function of environmental factors.
Taken together, these studies suggest that there is vari-
ation in how individuals respond to the risks to which they
are exposed, both genetic and environmental factors may
contribute to this variation, and the genetic and environ-
mental influences may vary between males and females.
Theoretically, it is reasonable to believe that there may be
sex differences in the genetic and environmental contri-
butions to adolescents’ differences in responses to risk
because: (a) there are sex differences in biological
processes (e.g., sex hormones, X chromosome inactivation)
that may enhance the expression or importance of a genetic
and/or environmental factor, respectively (Amstadter et al.
2014; Sjo
¨berg et al. 2008; Skuse 2006); (b) there may be
substantive sex differences in the composition of risk fac-
tors that are relevant to antisocial behavior (i.e., sex dif-
ferences in the type of stressful event that is most relevant
to crime for each sex, or how the composition of peer
groups promotes delinquency for males and females)
(Broidy and Agnew 1997; Haynie et al. 2007); and (c) there
may be significant differences in the social context in
which risk factors arise and play out (Miller 1998).
Current Study
While previous twin research has examined the impact of
genes and the environment on psychological resilience
across sexes, there is far less research on whether there are
sex differences in the genetic and environmental influences
of behavioral resilience, in particular adolescent antisocial
behavior. Moreover, there are even fewer studies that
investigate the sources of vulnerability for delinquency
among youth who do not encounter traditional risk factors.
Given the typical onset of delinquency and the emergence
of sex differences in antisocial behavior (especially violent
behavior) during adolescence (Eley et al. 1999; Vaske et al.
2014; Wang et al. 2013), examining the sources of varia-
tion in responses to risks for males and females could
enhance our understanding of the relationships between
risks and behavioral outcomes.
In light of this, the current study examines variation in
responses to cumulative risk among 1462 male and female
adolescent twins from the National Longitudinal Study of
Adolescent to Adult Health. Specifically, this study (1)
investigates the genetic and environmental sources of
variation across the full range of responses to cumulative
risk, ranging from vulnerability to resilience; (2) includes
three measures of delinquency (i.e., overall, nonviolent,
and violent delinquency) as youth may further vary by type
of delinquent response to risk; and (3) tests for significant
sex differences in genetic and environmental sources of
variation in differential response to risk. Given that males
tend to engage in delinquent acts more frequently than
females, we hypothesize that greater levels of vulnerability
will be observed among males, while resilience will be
observed more often in females. Additionally, because
prior research has shown stronger heritability estimates for
both vulnerability (Newsome and Sullivan 2014) and
resilience among males specifically (Boardman et al. 2008;
Waaktaar and Torgersen 2012), we hypothesize that there
will be significant sex differences in the genetic and
environmental sources of variation in differential response
J Youth Adolescence
123
to risk with stronger genetic influences observed among
males.
Methodology
Sample
Data for this study came from the National Longitudinal
Study of Adolescent to Adult Health (Add Health). The
Add Health study began in 1994 when more than 90,000
American youth in grades 7 through 12 across 132 schools
were asked to complete a questionnaire spanning several
topics (Harris et al. 2006). In addition to the in-school
questionnaire, a subsample of youth was randomly selected
to participate in an interview administered in their homes,
forming a core sample. Other subsamples of interest were
also selected to participate in the in-home interviews,
including a genetic subsample. Adolescents were selected
for the genetic subsample if they indicated during the in-
school questionnaire that they resided with a sibling (in-
cluding siblings that are not genetically related) who was
also in grades 7 through 12. Siblings were then automati-
cally solicited for participation in the Add Health study,
and both youth were added to the genetic subsample.
Twin pairs from the genetic subsample form the analytic
sample for this study. All opposite-sex pairs were classified
as dizygotic; however, the zygosity of each same-sex pair
was determined through two methods (Harris et al. 2006).
First, individuals in each pair were asked four questions
about their physical similarity and how often family
members, teachers, and strangers confuse them with their
co-twin (Rowe and Jacobson 1998). Responses from both
twins were averaged to create a confusability score that
was used to classify twins as monozygotic (MZ) or dizy-
gotic (DZ). Some pairs could not be classified using this
method because their average confusability scores excee-
ded the threshold for DZ classification but did not meet the
threshold for MZ classification. Saliva samples for DNA
analysis collected during Wave III were used to determine
the zygosity of the undetermined pairs. If a twin pair was
concordant on all genetic markers, they were classified as
MZ. The DNA analysis was used to classify 18 pairs that
were of unknown zygosity and to reclassify an additional
16 pairs that were initially incorrectly classified (Harris
et al. 2006). Taking a conservative approach, the remaining
25 pairs for which zygosity could not be determined were
classified as DZ because this may overestimate environ-
mental influences (Rowe and Jacobson 1998). The final
analytic sample is comprised of 155 MZ female (MZ
F
)
pairs, 130 DZ female (DZ
F
) pairs, 152 MZ male (MZ
M
)
pairs, 140 DZ male (DZ
M
) pairs, and 204 DZ opposite-sex
(DZ
O
) pairs aged 12 to 19 (
x¼16:05, SD =1.63;
white =62 %). In each same-sex pair, co-twins were
randomly assigned as twin one or twin two, while males
were classified as twin one in DZ
O
pairs.
Measures
Assessing the origins of differential response to risk requires
a single measure that accounts for variation in exposure to
risks and the outcome of interest, and previous research has
operationalized this construct as the standardized residuals
from regression analyses (Kim-Cohen et al. 2004; Newsome
and Sullivan 2014). The same strategy was used in the cur-
rent study by regressing three different measures of delin-
quency on cumulative risk at Wave I. The following
describes the measure of cumulative risk, each measure of
delinquency, and the three measures of differential response
to risk that are the key variables analyzed in this study.
Descriptive statistics for all measures of risk, delinquency,
and differential response to risk are shown in Table 1.
Cumulative Risk A total of 14 risk factors across
individual, familial, and environmental domains were
included in the cumulative risk index. Prior studies have
dichotomized (0 =not at risk; 1 =at risk) and summed
risk factors,
1
with thresholds for risk being associated with
the lowest or highest quartile of the distribution (Buehler
and Gerard 2013; Gerard and Buehler 2004; Newsome and
Sullivan 2014). The same approach was adopted here for
the majority of risk factors,
2
and to maintain the size of the
sample youth were required to have a valid score on 10 of
the 14 risk items to be included in the study.
Individual level risk factors Individual level risk factors
were created using items from the adolescents’ self-reports
during the in-home interviews at Wave I. These measures
include poor school performance, low attachment to
school, low intelligence, poor problem-solving skills, poor
coping skills, marijuana use, and cigarette use.
Poor school performance Each youth’s performance in
school was operationalized by averaging the grades each
adolescent reported to have earned during the most recent
grading period in English, mathematics, history and sci-
ence (1 =A, 2 =B, 3 =C, 4 =D or below). Higher
scores were indicative of greater risk, and the measure was
dichotomized at the 75th percentile (C2.75).
1
An alternative approach involves trichotomizing risk factors to
account for potential promotive effects. Although the aim of this
study is to examine differences in response to risks specifically, an
alternate cumulative risk index was created in which the risk factors
were trichotomized to explore for the possible impact of promotive
factors. The alternative index was a slightly worse predictor of
delinquency, and the resultant measures of differential responses to
risk were strongly correlated (r=.98) with those used in this study.
2
Marijuana use, cigarette use, and neighborhood safety were initially
operationalized as dichotomous measures.
J Youth Adolescence
123
Low attachment to school Adolescents were asked to
report how often during the school year they had trouble
getting along with teachers; getting homework done; pay-
ing attention in school; and getting along with other stu-
dents (0 =never, 4 =everyday), and the degree to which
they agreed with statements about feeling close to people at
school; feeling like they are a part of the school; being
happy to be at the school; that teachers treat students fairly;
and feeling safe at school (1 =strongly agree). These nine
items were standardized and summed to create a measure
of attachment to school (a=.78), with scores ranging
from -10.3 to 21.52. The cutoff for being considered at-
risk with respect to attachment to school was the 75th
percentile (C3.16).
Low intelligence The level of intelligence for each youth
was measured using scores obtained using an abridged
version of the Peabody Picture Vocabulary Test-Revised.
The range of values for this measure was 0–87, with lower
scores indicating lower intelligence. Intelligence was
dichotomized at the 25th percentile, and youth scoring 57
or lower were considered at-risk.
Poor problem-solving skills Youth were asked to indi-
cate their level of agreement (1 =strongly agree) with four
statements regarding their general approach to solving
problems they may encounter. Specifically, the questions
asked if youth try to get as many facts as possible about a
problem they may have, if they think of different ways to
approach a problem, if they systematically compare alter-
native decisions, and if they try to analyze what went right
and wrong after carrying out a solution. Responses to these
four items were summed (a=.74), and scores ranged from
4 to 20 with higher scores indicating lower problem-solv-
ing skills. The threshold for risk was set at the 75th per-
centile (C10).
Poor coping skills In addition to the items reflecting
problem solving strategies, youth were also asked to indi-
cate their level of agreement (1 =strongly agree) with
three statements that were indicative of their emotional
responses to problems. Participants’ responses to items
reporting if they go out of their way to avoid dealing with
problems, if difficult problems make them very upset, and
if they go with their ‘‘gut feelings’’ when making decisions
Table 1 Descriptive Statistics
for Full Sample and by Sex Measures Full sample Males Females
Risk measures
Poor school performance 2.19 (.76) 2.32 (.77) 2.07 (.73)
Low attachment to school -.07 (5.39) .16 (5.16) -.32 (5.60)
Low intelligence*** 63.48 (10.71) 64.64 (10.74) 62.33 (10.56)
Poor problem-solving 8.89 (2.53) 8.76 (2.57) 9.00 (2.49)
Poor coping skills 8.35 (2.19) 8.42 (2.23) 8.29 (2.14)
Marijuana use 13.2 % 14.7 % 11.6 %
Cigarette use 24.9 % 27.7 % 22.1 %
Low attachment to parents** 15.86 (4.15) 16.22 (3.99) 15.54 (4.23)
Low parental involvement 5.69 (3.38) 5.68 (3.49) 5.72 (3.25)
Low parental engagement 15.85 (6.39) 15.61 (6.18) 16.01 (6.49)
Low parental supervision** 16.36 (5.28) 15.92 (5.18) 16.75 (5.30)
Delinquent peers* 2.46 (2.64) 2.62 (2.73) 2.29 (2.53)
Low social support 28.34 (4.05) 28.21 (3.98) 28.49 (4.13)
Neighborhood risk 12.1 % 10.9 % 13.4 %
Cumulative Risk 3.79 (2.56) 3.81 (2.49) 3.75 (2.64)
Delinquency measures (log transformed)
Overall delinquency*** .80 (.87) .96 (.91) .64 (.81)
Violent delinquency*** .45 (.63) .56 (.68) .34 (.56)
Nonviolent delinquency*** .53 (.75) .64 (.80) .42 (.68)
Measures of differential response to risk
Overall delinquency*** .00 (1.00) .20 (1.05) -.20 (.90)
Violent delinquency*** .00 (1.00) .18 (1.08) -.18 (.87)
Nonviolent delinquency*** .00 (1.00) .15 (1.07) -.15 (.90)
Mean values and standard deviations are reported for all continuous variables. Values reported for
dichotomous variables represent the percent of youth in the risk category. Significant differences observed
between males and females * p\.05; ** p\.01; *** p\.001
J Youth Adolescence
123
were summed to create a measure of coping skills
(a=.44). Scores ranged from 3 to 15, and were dichot-
omized at the 25th percentile (B7).
Marijuana use Two questions were included on the in-
home questionnaire about the adolescents’ personal sub-
stance use. First, individuals were asked if they had ever
used marijuana. If youth answered affirmatively, they were
asked to report how many days they had used marijuana
during the past 30 days. This measure was dichotomized so
that those that reported using marijuana at least one time in
the last month were considered at-risk (13 %).
Cigarette use. Participants were also asked to report the
number of days in the past 30 that they had used cigarettes.
Youth who reported smoking at least 1 day (25 %) were
classified asat-risk for this factor on the cumulative riskindex.
Family level risk factors Within the family domain, four
risk factors were included in the cumulative risk index
including: low attachment to parents, low parental
involvement, low parental engagement, and low parental
supervision. Following prior research, items associated
with the mothers and the fathers were combined in each of
the four measures (Gerard and Buehler 2004; Newsome
and Sullivan 2014). When a parent was not present in the
home, a score of 0 was entered for each of the associated
items. This strategy acknowledges the importance of both
parents in adolescent development, and may provide a
more accurate assessment of the parenting received by the
youth.
Low attachment to parents The degree to which a youth
was attached to his or her parents was operationalized by
summing responses to four questions asking individuals to
indicate how close they felt to each parent and how much
they felt each parent cared about them (1 =not at all;
a=.68). Scores ranged from 4 to 20 with lower scores
indicated lower levels of attachment. Individuals with a
score less than or equal to 12 (25th percentile) were clas-
sified as at-risk.
Low parental involvement Youth were asked to report
whether or not they had participated in 10 different activ-
ities, such as playing sports, shopping, or talking about
various aspects of their life, with either parent in the past
4 weeks (0 =no, 1 =yes). All 20 items were summed and
values ranged from 0 to 19. Parental involvement was
dichotomized at the 25th percentile (B3).
Low parental engagement Adolescents were asked to
indicate their level of agreement (1 =strongly agree) with
eight statements related to their relationships with their
parents. These statements included items regarding whe-
ther the youth’s mother was encouraging of independence,
whether the mother talks to the child about why particular
behaviors are wrong, if the mother or father were warm and
loving, if the youth was satisfied with the ways he or she
communicates with his or her mother or father, and if the
youth was satisfied with his or her relationships with the
mother and father overall (a=.88). Responses to the
statements were summed, and values ranged from 7 to 35
with higher scores reflecting less engagement. The measure
was dichotomized at the 75th percentile (C20).
Low parental supervision Parental supervision was
operationalized by summing each youth’s responses to six
questions about how often his or her mother and father is
home before or after school and when the youth goes to bed
(1 =always; a=.61). Values ranged from 4 to 30, and
individuals with a score greater than or equal to 20 (75th
percentile) were considered at-risk.
Environmental level risk factors The final domain of
risk factors includes three more distal measures associated
with risk for delinquency: delinquent peers, low social
support, and neighborhood risk.
Delinquent peers During the Wave I in-home inter-
views, youth were asked how many of their three closest
friends use cigarettes, alcohol, or marijuana. An index of
delinquent peers was created by summing responses to all
three questions (a=.76). Scores ranged from 0 to 9, and
the measure was dichotomized at the 75th percentile (C4).
Low social support Participants were asked to indicate
the extent to which they felt adults, teachers, parents, and
friends cared about them; that people in their family
understand and pay attention to them; and that they have
fun with members of their family (1 =not at all).
Responses to these seven items were summed and values
ranged from 7 to 35. Additionally, because lower scores
were associated with greater risk, this measure was
dichotomized at the 25th percentile (B26).
Neighborhood risk Youth were asked to report whether
or not they felt safe in the neighborhood where they resided
(1 =no). Respondents that indicated that they did not feel
safe were classified as at-risk (12 %).
Delinquency Prior research suggests that there is mul-
tifinality in developmental outcomes for at-risk youth
(Cicchetti and Rogosch 1996). Moreover, it has long been
recognized that sex differences exist across different forms
of delinquency. To account for these differences, three
measures of delinquency taken at Wave I were used in this
study: violent delinquency, nonviolent delinquency, and
overall delinquency.
Violent delinquency Responses to six items in which
adolescents were asked to report the frequency with which
they were involved in a serious physical fight, hurt another
person badly, used or threatened to use a weapon to get
something from someone, took part in a fight with a group
of friends, pulled a knife or gun on someone, or shot or
stabbed someone were summed (a=.74). Scores ranged
from 0 to 16 (
x¼1:01, SD =1.95). The values were log
transformed (ln(x ?1)) prior to analysis to adjust for the
skew of the distribution.
J Youth Adolescence
123
Nonviolent delinquency Nonviolent delinquency was
operationalized by summing responses to eight items in
which youth reported how often they painted graffiti,
damaged property, took something from a store without
paying for it, drove a car without permission, stole some-
thing worth less than $50, stole something worth more than
$50, gone into a house or building to steal something, or
sold drugs (a=.79). The range of values for nonviolent
delinquency was 0–24 (
x¼1:45, SD =2.79), and because
the scale was over dispersed, scores were log transformed.
Overall delinquency Overall delinquency was opera-
tionalized by summing responses to all 14 items included
in the violent and nonviolent delinquency scales (a=.84,
x¼2:46, SD =4.19). Similar to violent and nonviolent
measures of delinquency, overall delinquency was log
transformed to adjust for the skew of the distribution.
Differential Response to Risk To capture the full range
of responses to risk from resilience to vulnerability, dif-
ferential response to risk was operationalized by obtaining
the standardized residuals from the regression of each
measure of delinquency on cumulative risk. Following
previous research (Kim-Cohen et al. 2004; Newsome and
Sullivan 2014), the continuum of scores is interpreted such
that resilient individuals are those with negative residual
scores, or those less delinquent than predicted by cumu-
lative risk. At the other extreme, vulnerable individuals are
those with positive residual scores, indicating that they
were more delinquent than predicted.
Analysis
Prior to estimating sex differences in genetic and envi-
ronmental influences on differential responses to risk, three
separate ordinary least squares (OLS) regression models
were estimated in which violent delinquency, nonviolent
delinquency, and overall delinquency were each regressed
on cumulative risk. The standardized residuals from these
regressions were retained, and the variance in each was
analyzed using structural equation modeling techniques
based on the principles of biometrical genetics (Neale and
Cardon 1992). Building on the basic univariate model, the
variance in the phenotype of interest (P) is partitioned into
additive genetic (A), common environmental (C), and
unique environmental (E) influences. Additive genetic
influences are present when the individual effects of alleles
across loci are combined in an additive manner. Because
MZ twins are genetically identical, additive genetic effects
will be perfectly correlated between co-twins (r
g
=1.00),
but the correlation between DZ co-twins is half because
they share approximately half their discriminating genes
(r
g
=.50), on average. Common environmental influences
are those experiences shared by both twins in a pair, and
are perfectly correlated between co-twins regardless of
zygosity (r
c
=1.00). Unique environmental influences are
experienced by only one twin in a pair, and remain
uncorrelated across all pairs.
3
A nonscalar sex limitation model was estimated to test
for quantitative differences between males and females.
4
Quantitative differences may be observed when both sexes
share the same sources of variance, but the magnitude of
the genetic and environmental influences are not equal
(Neale and Cardon 1992). This is tested by first fitting a
model in which the a,c, and epathways are free to vary
between males and females, and then fitting a model in
which the pathways are constrained to be equal. The fit of
the model is then assessed by examining the Chi square
(v
2
) and Akaike Information Criterion (AIC) values. A
nonsignificant difference in v
2
indicates that the more
parsimonious submodel fits the data well. The significance
of the pathways can also be assessed by fixing the
parameters to zero and inspecting the fit of the model. The
model with the lowest AIC value and a nonsignificant
change in v
2
is considered the best-fitting model (Neale
and Cardon 1992).
Results
The results of the OLS regression models revealed that
cumulative risk was a significant predictor of violent
delinquency (b=.08, SE =.01, p\.000, R
2
=.10),
nonviolent delinquency (b=.11, SE =.01, p\.000,
R
2
=.14), and overall delinquency (b=.14, SE =.01,
p\.000, R
2
=.16). As noted previously, the standardized
residual scores from each of the regression models were
used to operationalize differential response to risk with
negative scores indicating resilience and positive scores
indicating greater vulnerability. The values of the stan-
dardized residual scores of the OLS model predicting
violent delinquency ranged from -1.83 to 3.57. The range
of standardized residuals generated from the OLS model
predicting nonviolent delinquency was -2.08 to 3.46. The
final OLS model predicting overall delinquency resulted in
standardized residual scores ranging from -2.41 to 3.09.
As shown in Table 1, males displayed greater vulnerability
on average, and females tended to be more resilient.
3
Unique environmental estimates also include measurement error
(Plomin et al. 1997).
4
It is also possible to test for qualitative sex differences to determine
whether the specific genetic and environmental factors that are
account for variation in a trait differ in males and females. Our
preliminary analyses, however, did not reveal significant qualitative
differences. The results of the qualitative sex limitation models are
available upon request.
J Youth Adolescence
123
Comparing cross-twin correlations across all five
zygosity groups, shown in Table 2, provided an initial
assessment of genetic and environmental influences and
potential sex differences. Across all three measures of
differential response to risk correlations were stronger
among MZ twins relative to DZ twin pairs, suggesting
genetic factors are influencing each phenotype. Moreover,
the difference between MZ
F
and DZ
F
cross-twin correla-
tions is smaller than the difference between MZ
M
and DZ
M
pairs, suggesting that the magnitude of genetic effects may
be stronger among males than females across all three
measures.
The model-fitting results are shown in Table 3. The first
four models test for quantitative sex differences in differ-
ential response to risk for violent delinquency. The first
model (i.e., heterogeneity) permitted the path estimates for
males and females to vary (i.e., m =f), and served as the
model to which all subsequent models were compared.
Constraining the additive genetic (a), common environ-
mental (c), and unique environmental (e) parameters to be
equal between males and females in the second model (i.e.,
m=f) significantly reduced the fit of the model (p=.01,
AIC =1151.67), indicating the heterogeneity model pro-
vided a better fit to the data. Examining the path estimates
in the heterogeneity model suggested that the effects of
cmay be nonsignificant among males (c
m
=.00,
SE =.29). This was tested by fixing the parameter to zero
in the third model (i.e., c
m
=0), which provided a slight
improvement in the fit of the model (p=1.00,
AIC =1142.87). Further inspection of the path estimates
also suggested that the acomponent of variance may be
nonsignificant for females. The final model tested the sig-
nificance of both c
m
and a
f
by fixing both of these
parameters to zero (i.e., c
m
=0, a
f
=0). Placing these
constraints on the model provided a slight improvement in
model fit (p=.38, AIC =1142.83). The first two columns
in Fig. 1shows the proportion of variance in differential
response to risk for violent delinquency for males and
females.
5
For males, additive genetic and unique environ-
mental influences explained 41 and 59 % of the variance,
respectively. Among females, common environmental and
unique environmental influences explained 44 and 56 % of
the variance, respectively.
The second series of models shown in Table 3test for
quantitative differences in differential response to risk for
nonviolent delinquency. The heterogeneity model permits
the a,c, and eparameters to vary between males and
females (i.e., m =f), and equating the paths in the
homogeneity model (i.e., m =f) significantly reduced the
fit of the model (p=.03, AIC =1145.46). The two
remaining models tested the significance of a
f
(i.e., a
f
=0)
and then c
m
and a
f
(i.e., c
m
=0, a
f
=0), respectively,
because estimates for these paths were small in the
heterogeneity model. Fixing a
f
to zero provided a better fit
to the data (p=1.00, AIC =1140.13); however, fixing
both c
m
and a
f
to zero significantly reduced the fit of the
model (p=.02, AIC =1145.90). As shown in Fig. 1,
estimates produced in the best-fitting model indicated that
all three components of variance significantly contributed
to variation in differential response to risk for nonviolent
delinquency among males (a
2
=29 %, c
2
=11 %,
e
2
=60 %). For females, additive genetic influences were
not significant (c
2
=42 %, e
2
=58 %).
The final series of models shown in Table 3tested for
quantitative differences in differential response to risk for
overall delinquency. Similar to the models estimated for
differential response to risk for violent and nonviolent
delinquency, equating the path estimates in the homo-
geneity model (i.e., m =f) provided a significantly worse
fit (p=.00, AIC =1142.07) relative to the heterogeneity
model (i.e., m =f; AIC =1132.91). The estimate for c
m
was zero in the heterogeneity model, and the significance
of the parameter was tested by fixing it to zero in the third
model (i.e., c
m
=0). Dropping c
m
improved the fit of the
model (p=1.00, AIC =1130.91). The estimate for a
f
was
also somewhat small in the heterogeneity model, and the
final model tested whether dropping both c
m
and a
f
would
provide a better fit to the data (i.e., c
m
=0, a
f
=0).
Placing both constraints on the model provided the best fit
(p=.61, AIC =1129.89). The proportion of variance in
differential response to risk for overall delinquency among
males and females is shown in the last two columns of
5
Variance in a phenotype =a
2
?c
2
?e
2
=100 %.
Table 2 Cross-twin correlation
coefficients by zygosity and sex Pair type N Measure of differential response to risk
Violent delinquency Nonviolent delinquency Overall delinquency
MZ
F
155 .46*** .41*** .45***
DZ
F
130 .31** .39*** .38***
MZ
M
152 .44*** .42*** .44***
DZ
M
140 .27** .33*** .29**
DZ
O
204 .14* .22** .14
*p\.05; ** p\.01; *** p\.001
J Youth Adolescence
123
Fig. 1. For males, additive genetic (43 %) and unique
environmental (57 %) were influential. Among females,
however, the variance was attributed entirely to environ-
mental influences (c
2
=45 %, e
2
=55 %).
Discussion
A longstanding finding in criminological research is that
males tend to be involved in more delinquent behaviors
than females, and this is particularly evident with regard to
violent delinquency (Moffitt et al. 2001). More recently,
however, there has been a greater recognition that indi-
viduals may differ in the ways in which they respond to
risks (Ellis et al. 2011). Some youth display resilience and
others a greater vulnerability; however, researchers have
only recently begun to investigate the sources of these
differences. Further exploration into the factors that con-
tribute to differential response to risks may provide insights
to the observed differences in violent and nonviolent
offending generally, as well as by gender.
The current study investigated whether there are sex
differences in the genetic and environmental influences on
differential response to risk for violent, nonviolent, and
overall delinquency. Several interesting findings emerged
from the analyses. As shown in Table 1, there were sig-
nificant sex differences in differential response to risk
across all three measures of delinquency. On average,
females were more resilient to cumulative risk for delin-
quency while males were more vulnerable. Results from
the behavioral genetic analyses revealed that genetic fac-
tors played an important role in accounting for variation in
Table 3 Model-fitting results and unstandardized path estimates
Model -2LL df AIC DLL Ddf p Males Females
aceace
Violent delinquency
Heterogeneity (m =f) 4088.87 1472 1144.87 – – – .66
(.05)
.00
(.29)
.79
(.04)
.34
(.37)
.56
(.18)
.70
(.05)
Homogeneity (m =f) 4101.67 1475 1151.67 12.8 3 0.01 .65
(.10)
.06
(.91)
.75
(.03)
.65
(.10)
.06
(.91)
.75
(.03)
Heterogeneity (c
m
=0) 4088.87 1473 1142.87 0.00 1 1.00 .66
(.05)
– .79
(.04)
.34
(.19)
.56
(.10)
.70
(.04)
Heterogeneity (c
m
50, a
f
50) 4090.83 1474 1142.83 1.95 2 0.38 .66
(.05)
– .79
(.04)
–.64
(.05)
.71
(.03)
Nonviolent delinquency
Heterogeneity (m =f) 4088.13 1473 1142.13 – – – .56
(.21)
.35
(.30)
.80
(.04)
.00
(.35)
.62
(.05)
.73
(.03)
Homogeneity (m =f) 4097.46 1476 1145.46 9.33 3 0.03 .46
(.13)
.43
(.11)
.77
(.03)
.46
(.13)
.43
(.11)
.77
(.03)
Heterogeneity (a
f
50) 4088.13 1474 1140.13 0.00 1 1.00 .56
(.10)
.35
(.11)
.80
(.04)
– .62
(.05)
.73
(.03)
Heterogeneity (c
m
=0, a
f
=0) 4095.9 1475 1145.90 7.77 2 0.02 .67
(.05)
– .79
(.04)
– .62
(.05)
.73
(.03)
Overall delinquency
Heterogeneity (m =f) 4070.91 1469 1132.91 – – – .67
(.05)
.00
(.58)
.78
(.04)
.22
(.84)
.61
(.25)
.70
(.06)
Homogeneity (m =f) 4086.07 1472 1142.07 15.17 3 0.00 .63
(.10)
.19
(.27)
.75
(.03)
.63
(.10)
.19
(.27)
.75
(.03)
Heterogeneity (c
m
=0) 4070.91 1470 1130.91 0.00 1 1.00 .67
(.05)
– .78
(.04)
.22
(.20)
.61
(.07)
.70
(.03)
Heterogeneity (c
m
=0,a
f
= 0) 4071.89 1471 1129.89 0.99 2 0.61 .67
(.05)
– .78
(.04)
– .64
(.05)
.71
(.03)
Standard errors for each of the unstandardized path estimates are shown in parentheses. The best-fitting models are shown in bold
-2LL =negative two log likelihood, df =degrees of freedom, AIC =Akaike’s information criterion, DLL =difference in -2LL,
Ddf =difference in df
J Youth Adolescence
123
the manner in which males responded to cumulative risk,
while environmental factors appeared to better explain
variation in female responses to risks. More specifically,
genetic factors accounted for 41, 29, and 43 % of the
variance in responses to risk in males for violent, nonvio-
lent, and overall delinquency, respectively. Shared envi-
ronmental factors were only significantly influential to
differential response to risk for nonviolent delinquency for
males (11 %). For females, on the other hand, shared
environmental factors accounted for 44, 42, and 45 % of
the variance in responses to risk for violent, nonviolent,
and overall delinquency, respectively. All remaining vari-
ance was explained by nonshared environmental factors
(and error) with no significant genetic influence across the
three measures of differential response to risk.
The findings presented in this study are similar to those
reported in previous investigations of differences in
response to risk (Kim-Cohen et al. 2004; Newsome and
Sullivan 2014), and studies of resilience (Amstadter et al.;
Boardman et al. 2008; Waaktaar and Torgersen 2012). All
of these studies have provided evidence suggesting that
both genetic and environmental factors may influence the
ways in which individuals respond to the risks they
encounter. Estimates of genetic influences reported in these
studies have ranged from 38 to 77 %, and two studies
reported significantly stronger genetic influences in males
relative to females (Boardman et al. 2008; Waaktaar and
Torgersen 2012). While the research in this area remains
limited, this study is the only one in which genetic influ-
ences for females were found to be nonsignificant.
There are at least two possible explanations for the sex
differences observed in this study. First, in relationship to
the differences between the findings in this study relative to
previous research, all of the prior studies have examined
different samples, risks, and outcomes. As a result, the
findings across studies may represent aspects of develop-
ment that are influenced by distinct processes. Given that
the findings from the current study suggest that boys are
more vulnerable and girls more resilient, a second possi-
bility is that these particular responses to risk may have
different origins. Newsome and Sullivan (2014) explored
this possibility and found that behavioral differences
between the extremely vulnerable group and the overall
population were largely attributed to genetic factors (group
heritability =55 %). In contrast, in moving across the
continuum of differential response to risk, heritability
estimates declined and differences between extremely
resilient youth and the overall population were entirely due
to environmental influences. Investigating the sources of
vulnerability and resilience to risk for delinquency by
gender could provide additional insights to these develop-
mental processes in males and females. Unfortunately, the
small number of twin pairs prohibits the investigation of
this possibility within the Add Health sample. Future
research should further explore these possibilities with
larger samples of kinship pairs.
Explanations as to why genetic influences may differ
across the sexes have been proposed including differences
in genetic expression, differences in sex hormones, and the
number of ‘‘active’’ X chromosomes (Amstadter et al.
2014; Ostrer 2001; Zahn-Waxler et al. 2006). Although
some research has investigated biological differences
between males and females that may contribute to variation
in antisocial outcomes specifically (Moffitt et al. 2001), this
is a highly under-researched area. Identifying the biological
sources of such differences could be used to further
develop the understanding of the development of antisocial
behaviors and inform prevention and intervention
strategies.
Another notable finding is that the genetic and envi-
ronmental influences on responses to risk for males vary
by type of delinquency. Variation in violent delinquency
in response to risk was moderately influenced by additive
genetic factors. These findings add further support to prior
research that has suggested violent behaviors may have a
biological basis (Eley et al. 1999,2003; Moffitt 1993).
That nonviolent offending appears to be partially influ-
enced by common environmental factors in males is
consistent with prior research suggesting that nonviolent
offenses may be influenced by social and situational fac-
tors, such as the presence of co-offenders (Reiss and
Farrington 1991; van der Laan et al. 2009). These findings
further highlight the need for continued investigation into
the factors that influence different types of delinquent
behaviors.
The environment appears to be responsible for the ways
that girls respond to risks, which can help to explain why
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Males Females Males Females Males Females
Violent Nonviolent Overall
Unique Environment
Common Environment
Addive Genec
Fig. 1 Proportion of the variance in differential response to risk for
delinquency attributed to genetic and environmental factors by sex
J Youth Adolescence
123
some girls who are at risk for delinquency are resilient and
other girls who are at a low risk for delinquency display
greater vulnerability. Many scholars advocate for interven-
tions that address unhealthy relationships that may con-
tribute to offending and emphasize strengthening prosocial
relationships that may act as protective factors (Blanchette
and Brown 2006; Bloom et al. 2003; Morash et al. 1998;
Prendergast et al. 1995; Van Wormer 2001). For example,
receiving an abundance of attention from caretakers, positive
relations with mentors, having prosocial peers, and being
connected to prosocial organizations and quality schools
have been shown to have protective effects among high-risk
youth (Masten and Powell 2003; Stouthamer-Loeber et al.
1993,2002; Werner and Smith 1982,1992). Continued
efforts to better understand how these, and other social
aspects of girls’ lives, influence their behavior may be the
most promising approaches to reducing vulnerability for
delinquency and fostering resilient development.
Limitations and Assumptions
The results presented in the current study should be con-
sidered with the following limitations and assumptions in
mind. While the measure of cumulative risk included
several known risk factors across multiple domains, it was
not possible to include all potential risk factors for ado-
lescent delinquency. Reasons for variable omission include
too few cases exposed to a given risk factor (e.g., neglect
and abuse), no direct measure of the risk factor (e.g., self-
control),
6
or the variables were not included in the Add
Health data (e.g., official criminal records). As a result, it is
possible that respondents’ calculated measure of risk was
underestimated, which in turn could have overestimated the
prevalence of vulnerability while also potentially under-
estimating the prevalence of resiliency.
There are also important considerations that should be
noted with regard to the use of a cumulative risk measure to
predict delinquency in this study. Specifically, it assumes
that each of the risk factors is equally influential, that the
same measure of cumulative risk is predictive of all three
measures of delinquency, and that it is equally applicable
for males and females. Some studies have found differ-
ences in the predictors of different types of delinquency, as
well as differences in the predictors of delinquent outcomes
across gender (Alarid et al. 2000; Booth et al. 2008; Daigle
et al. 2007; Mazerolle 1998). Although the measure of
cumulative risk used in this study was a significant pre-
dictor of all three types of delinquency, it is possible that
prediction could have been enhanced by creating separate
risk scores for each gender and each type of delinquency.
As more evidence with regard to differences in the expo-
sure to and impact of risks for males and females emerges,
future research should investigate whether the sources of
those differences can be attributed to genetic and envi-
ronmental factors.
It is also important to note that a cross-sectional design
was employed in the current study. Examining the
responses to risks longitudinally could further the under-
standing of development in light of the types of risks one
encounters across different developmental stages, and the
impact of the accumulation of risks throughout the life
course. Another possibility is that the duration of exposure
to risks could influence responses to risks. Investigating
these possibilities, particularly with respect to different
types of offending and gender, may uncover new insights
to the mechanisms underlying the relationships between
risk and development.
A brief discussion of the assumptions underlying ACE
modeling techniques and the use of twin samples is also
warranted. Central to these assumptions are the equal
environment assumption (EEA), the assumption of no gene-
environment interactions, and the assumption of no assor-
tative mating (Neale and Cardon 1992). First, critics of twin
studies argue that MZ twins experience more similar envi-
ronments compared to DZ twins and that is why they cor-
relate more strongly on behavioral measures (and not
because of genetic factors). It is important to note, however,
that the EEA can only be violated if two things occur: (1)
MZ twins experience more similar environments compared
to DZ twins, and (2) these similar experiences lead to more
behavioral similarities for MZ twins compared to DZ twins.
Several lines of evidence suggest, however, that similar
treatment experienced by MZ twins does not significantly
affect their behavior, thereby not violating the EEA. For
example, research has shown that actual biologically
determined zygosity better predicts behavioral similarities
compared to perceived zygosity (Scarr 1968; Scarr and
Carter-Saltzman 1979). That is, DZ twins who are raised as
MZ twins tend to display behavioral similarities similar to
other DZ twins (despite being raised as MZ twins and
presumably experiencing more similar environmental con-
ditions). Research has also shown that MZ twins who are
raised by parents who purposefully treat them the same
(raised as a unit) are not any more similar behaviorally than
MZ twins who are raised as individuals (Kendler et al. 1993;
Loehlin and Nichols 1976). Again, this suggests that simi-
larities in experiences do not necessarily lead to behavioral
similarities. Given these findings, it appears that the con-
sequences of similar treatment experienced by MZ twins do
not significantly impact behavioral similarities, thereby not
violating the EEA.
6
While researchers have attempted to create measures of self-control
using various items in the Add Health (see Beaver et al. 2009; Perrone
et al. 2004), these items were included in other measures of risk in the
current study.
J Youth Adolescence
123
Second, with regard to gene-environment interactions,
the ACE model assumes that genes (A) do not interact with
either the shared (C) or nonshared (E) environment. It is
possible though that exposure to certain environments for
certain people (based on their genetic make-up) can be
particularly problematic or particularly advantageous
(Caspi et al. 2002; Schwartz and Beaver 2011; Simons
et al. 2011). It is currently unknown, however, whether
gene-environment interactions influence differential
responses to risk in adolescence. Since gene-environment
interactions were not specified in the current models, which
included twin pairs raised together, any effects of AxE
were captured in the estimates of E and any effects of AxC
were captured in the estimates of A (Heath et al. 2002).
Lastly, the assumption of no assortative mating refers to
individuals procreating at random and not based on similar
characteristics. Research has shown, however, that indi-
viduals tend to choose mates with similar behavioral and
personality traits as themselves (Caspi and Herbener 1990;
Quinton et al. 1993). Nonrandom mating would then vio-
late the assumption of a 0.50 genetic correlation for DZ
twins by making DZ twins more alike. From a statistical
standpoint, nonrandom mating could then lead to an
overestimation of shared environmental effects and an
underestimation of genetic effects (Neale and Cardon
1992). Maes et al. (1998), however, have shown that even
if this assumption is violated, the potential for biasing
results is minimal.
Practical Implications
With consideration to these precautions, it is nevertheless
possible to view the contributions of this study to pre-
vailing understandings of juvenile delinquency. The sex
differences discovered in the current study may be useful in
informing the current state of juvenile risk/need assessment
instruments. These assessment tools are intended to predict
adolescents’ risk for delinquency by quantifying and
combining static and dynamic risk factors. Examples of
these types of assessments include Youth Level of Service
Case Management Inventory (YLS/CMI) (Hoge and
Andrews 2002; Hoge et al. 2002) and the Ohio Youth
Assessment System (Latessa et al. 2009). The risk factors
included in these assessments are determined by reviews of
extant research regarding predictors of juvenile delin-
quency (Latessa et al. 2009), and are typically based in
traditional theories of criminal behavior. While evidence
suggests that these are valid and reliable assessment tools
(Schmidt et al. 2005), these tools give little to no attention
to gender. The findings in the current study that indicate
there are distinct sources of differences in response to risks
for males and females, which illustrates the importance of
investigating the potential for gender-specific factors to
improve prediction—both in terms of theory and in prac-
tical applications.
Given the significant influence of environmental factors,
interventions for girls would do well to continue identify-
ing and targeting gender-specific environmental factors
that promote resilience. Some scholars have argued that
environmental factors play an important role in the lives of
women offenders (Belknap 2014). As an example, Van
Voorhis et al. (2010) have identified several gender-re-
sponsive factors related to recidivism in female offenders
including: housing safety, abuse and trauma, family con-
flict, relationship difficulties, and parental stress. Their
research has also uncovered several protective factors
within the environment that may promote resilience such as
relationship support, parental involvement, family support,
and relationship satisfaction (Van Voorhis et al. 2010).
Prevention and intervention strategies that focus on envi-
ronmental factors such as these may prove to be the most
promising in fostering resilience and minimizing vulnera-
bility among females.
Conclusion
This study aimed to better understand the factors con-
tributing to differential response to risk for delinquency. By
assessing vulnerability and resilience in males and females
from a behavioral genetic perspective, our findings
underscore the necessity to examine differences in
responses to risk by sex. Our results revealed that genes
and the environment are influential in the way that males
respond to risks and only environmental factors affect
females’ responses to risks. While the environment is
important, incorporating genetic factors in the way that we
study responses to risks may be particularly relevant for
males. The finding that environmental factors were the only
source of influence on girls’ responses to risk for delin-
quency should be further examined with inclusion of
female-specific risk factors, which may show different
sources of influence on their behavior. Including specific
genotypes and female-specific risk factors should be the
next steps in research that strives to comprehensively
understand differences in responses to risk for delinquency.
Acknowledgments The authors wish to thank John Paul Wright and
Christopher J. Sullivan for their thoughtful comments on an earlier
draft of this manuscript. This research uses data from Add Health, a
program project directed by Kathleen Mullan Harris and designed by
J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the
University of North Carolina at Chapel Hill, and funded by grant P01-
HD31921 from the Eunice Kennedy Shriver National Institute of
Child Health and Human Development, with cooperative funding
from 23 other federal agencies and foundations. Special acknowl-
edgment is due Ronald R. Rindfuss and Barbara Entwisle for assis-
tance in the original design. Information on how to obtain the Add
J Youth Adolescence
123
Health data files is available on the Add Health website (http://www.
cpc.unc.edu/addhealth). No direct support was received from grant
P01-HD31921 for this analysis.
Authors Contributions JN conceived of the study and its design,
performed the statistical analyses, and contributed to and coordinated
the draft of the manuscript. JV participated in the interpretation of the
results and in preparing the draft of the manuscript. KG participated
in the interpretation of the results and their implications and in
preparing the draft of the manuscript. DB participated in the design of
the study and preparing the draft of the manuscript. All authors read
and approved the final version of this manuscript.
Conflicts of interest The authors report no conflict of interests.
References
Alarid, L. F., Burton, V. S. J., & Cullen, F. T. (2000). Gender and
crime among felony offenders: Assessing the generality of social
control and differential association theories. Journal of Research
in Crime and Delinquency, 37, 171–199.
Amstadter, A. B., Myers, J. M., & Kendler, K. S. (2014). Psychiatric
resilience: Longitudinal twin study. The British Journal of
Psychiatry, 205, 275–280.
Appleyard, K., Egeland, B., van Dulmen, M. H., & Sroufe, L. A.
(2005). When more is not better: The role of cumulative risk in
child behavior outcomes. Journal of Child Psychology and
Psychiatry, 46, 235–245.
Beaver, K. M., Ratchford, M., & Ferguson, C. J. (2009). Evidence of
genetic and environmental effects on the development of self-
control. Criminal Justice and Behavior, 36, 1158–1172.
Belknap, J. (2014). The invisible woman: Gender, crime, and justice
(4th ed.). Belmont: Thompson Wadsworth.
Belknap, J., & Holsinger, K. (2006). The gendered nature of risk
factors for delinquency. Feminist Criminology, 1, 48–71.
Belsky, J. (1997). Variation in susceptibility to environmental
influence: An evolutionary argument. Psychological Inquiry, 8,
182–186.
Belsky, J. (2005). Differential susceptibility to rearing influence: An
evolutionary hypothesis and some evidence. In B. Ellis & D.
Bjorklund (Eds.), Origins of the social mind: Evolutionary
psychology and child development (pp. 139–163). New York:
Guilford Press.
Blanchette, K., & Brown, S. L. (2006). The assessment and treatment
of women offenders: An integrative perspective. Chichester:
Wiley.
Bloom, B., Owen, B., & Covington, S. (2003). Gender responsive
strategies: Research, practice, and guiding principles for women
offenders. Washington: U.S. Department of Justice, National
Institute of Corrections.
Blumstein, A., & Cohen, J. (1987). Characterizing criminal careers.
Science, 237, 985–991.
Boardman, J. D., Blalock, C. L., & Button, T. M. (2008). Sex
differences in the heritability of resilience. Twin Research and
Human Genetics, 11, 12–27.
Booth, J. A., Farrell, A., & Varano, S. P. (2008). Social control,
serious delinquency, and risky behavior: A gendered analysis.
Crime & Delinquency, 54, 423–456.
Boyce, W. T., & Ellis, B. J. (2005). Biological sensitivity to context:
An evolutionary-developmental theory of the origins and
functions of stress reactivity. Development and Psychopathol-
ogy, 17, 271–301.
Broidy, L., & Agnew, R. (1997). Gender and crime: A general strain
theory perspective. Journal of Research in Crime and Delin-
quency, 34, 275–306.
Bronfenbrenner, U. (1979). The ecology of human development:
Experiments by nature and design. Cambridge: Harvard Univer-
sity Press.
Buehler, C., & Gerard, J. M. (2013). Cumulative family risk predicts
increases in adjustment difficulties across early adolescence.
Journal of Youth and Adolescence, 42, 905–920.
Caspi, A., & Herbener, E. S. (1990). Continuity and change: Assortative
marriage and the consistency of personality in adulthood. Journal
of Personality and Social Psychology, 58, 250–258.
Caspi, A., McClay, J., Moffitt, T. E., Mill, J., Martin, J., Craig, I. W.,
et al. (2002). Role of genotype in the cycle of violence in
maltreated children. Science, 297, 851–854.
Cicchetti, D., & Rogosch, F. A. (1996). Equifinality and multifinality
in developmental psychopathology. Development and Psy-
chopathology, 8, 597–600.
Daigle, L. E., Cullen, F. T., & Wright, J. P. (2007). Gender
differences in the predictors of juvenile delinquency: Assessing
the generality-specificity debate. Youth Violence and Juvenile
Justice, 5, 254–286.
Dekovic
´, M. (1999). Risk and protective factors in the development of
problem behavior during adolescence. Journal of Youth and
Adolescence, 28, 667–685.
Eley, T. C., Lichtenstein, P., & Moffitt, T. E. (2003). A longitudinal
behavioral genetic analysis of the etiology of aggressive and
nonaggressive antisocial behavior. Development and Psy-
chopathology, 15, 383–402.
Eley, T. C., Lichtenstein, P., & Stevenson, J. (1999). Sex differences in
the etiology of aggressive and nonaggressive antisocial behavior:
Results from two twin studies. Child Development, 70, 155–168.
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J.,
& van Ijzendoorn, M. H. (2011). Differential susceptibility to the
environment: An evolutionary-neurodevelopmental theory.
Development and Psychopathology, 23, 7–28.
Evans, G. W. (2003). A multimethodological analysis of cumulative
risk and allostatic load among rural children. Developmental
Psychology, 39, 924–933.
Evans, G. W., Kim, P., Ting, A. H., Tesher, H. B., & Shannis, D. (2007).
Cumulative risk,maternal responsiveness, and allostaticload among
young adolescents. Developmental Psychology, 43, 341–351.
Farrington, D. P. (1986). Age and crime. In M. Tonry & N. Morris
(Eds.), Crime and justice: An annual review of research (Vol. 7,
pp. 189–250). Chicago: University of Chicago Press.
Farrington, D. P. (1995). The development of offending and antisocial
behaviour from childhood: Key findings from the Cambridge
Study in Delinquent Development. Journal of Child Psychology
and Psychiatry, 36, 929–964.
Farrington, D. P., & Welsh, B. C. (2007). Saving children from a life
of crime: Early risk factors and effective interventions. Oxford:
Oxford University Press.
Federal Bureau of Investigation. (2015). Crime in the United States,
2014. Washington: U. S. Government Printing Office.
Fergus, S., & Zimmerman, M. A. (2005). Adolescent resilience: A
framework for understanding healthy development in the face of
risk. Annual Review of Public Health, 26, 293–302.
Fergusson, D. M., Boden, J. M., Horwood, L. J., Miller, A., & Kennedy,
M. A. (2012). Moderating role of the MAOA genotype in antisocial
behaviour. The British Journal of Psychiatry, 200, 116–123.
Fergusson, D. M., & Woodward, L. J. (2000). Educational,
psychosocial, and sexual outcomes of girls with conduct
problems in early adolescence. Journal of Child Psychology
and Psychiatry and Allied Disciplines, 41, 779–792.
J Youth Adolescence
123
Gerard, J. M., & Buehler, C. (2004). Cumulative environmental risk
and youth problem behavior. Journal of Marriage and Family,
66, 702–720.
Giordano, P. C. (1989). Confronting control theory’s negative cases.
In S. F. Messner, M. D. Krohn, & A. E. Liska (Eds.), Theoretical
integration in the study of deviance and crime: Problems and
prospects (pp. 261–278). Albany: State of University of New
York Press.
Guo, G., Roettger, M. E., & Cai, T. (2008). The integration of genetic
propensities into social-control models of delinquency and
violence among male youths. American Sociological Review,
73, 543–568.
Harris, K. M., Halpern, C. T., Smolen, A., & Haberstick, B. C. (2006).
The national longitudinal study of adolescent health (add health)
twin data. Twin Research and Human Genetics, 9, 988–997.
Haynie, D. L., Steffensmeier, D., & Bell, K. E. (2007). Gender and
serious violence: Untangling the role of friendship sex compo-
sition and peer violence. Youth Violence and Juvenile Justice, 5,
235–253.
Heath, A. C., Todorov, A. A., Nelson, E. C., Madden, P. A. F.,
Bucholz, K. K., & Martin, N. G. (2002). Gene-environment
interaction effects on behavioral variation and risk of complex
disorders: The example of alcoholism and other psychiatric
disorders. Twin Research, 5, 30–37.
Herrenkohl, T. I., Maguin, E., Hill, K. G., Hawkins, J. D., Abbott, R.
D., & Catalano, R. F. (2000). Developmental risk factors for
youth violence. Journal of Adolescent Health, 26, 176–186.
Hoge, R., & Andrews, D. A. (2002). The youth level of service/case
management inventory manual and scoring key. Toronto: Multi-
Health Systems.
Hoge, R., Andrews, D. A., & Leschied, A. (2002). The youth level of
service/case management inventory. Toronto: Multi-Health Systems.
Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L.
J. (1993). A test of the equal-environment assumption in twin
studies of psychiatric illness. Behavior Genetics, 23(1), 21–27.
Kim-Cohen, J., Moffitt, T. E., Caspi, A., & Taylor, A. (2004). Genetic
and environmental processes in young children’s resilience and
vulnerability to socioeconomic deprivation. Child Development,
75, 651–668.
Latessa, E., Lovins, B., & Ostrowski, K. (2009). The Ohio Youth
Assessment System: Final report. Cincinnati: University of
Cincinnati Corrections Institute.
Lee, S. S. (2011). Deviant peer affiliation and antisocial behavior:
Interaction with monoamine oxidase A (MAOA) genotype.
Journal of Abnormal Child Psychology, 39, 321–332.
Loeber, R., & Dishion, T. (1983). Early predictors of male
delinquency: A review. Psychological Bulletin, 94, 68–99.
Loeber, R., Slot, W., & Stouthamer-Loeber, M. (2008). A cumulative
developmental model of risk and promotive factors. In R.
Loeber, N. Slot, & M. Hoeve (Eds.), Tomorrow’s criminals: The
development of child delinquency (pp. 153–161). Farnham:
Ashgate Publishing Limited.
Loehlin, J. C., & Nichols, R. C. (1976). Heredity, environment, &
personality: A study of 850 sets of twins. Austin: University of
Texas Press.
Maes, H. H. M., Neale, M. C., Kendler, K., Hewitt, J. K., Silberg, J.
L., Foley, D. L., et al. (1998). Assortative mating for major
psychiatric diagnosis in two population-based samples. Psycho-
logical Medicine, 28, 1389–1401.
Masten, A. S., Best, K. M., & Garmezy, N. (1990). Resilience and
development: Contributions from the study of children who
overcome adversity. Development and Psychopathology, 2, 425–444.
Masten, A. S., & Powell, J. L. (2003). A resilience framework for
research, policy, and practice. In S. S. Luthar (Ed.), Resilience
and vulnerability: Adaptation in the context of child adversities
(pp. 1–25). New York: Cambridge University Press.
Mazerolle, P. (1998). Gender, general strain, and delinquency: An
empirical examination. Justice Quarterly, 15, 65–91.
Miller, J. (1998). Up it up: Gender and the accomplishment of street
robbery. Criminology, 36, 37–66.
Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent
antisocial behavior: A developmental taxonomy. Psychological
Review, 100, 674–701.
Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P. A. (2001). Sex
differences in antisocial behavior: Conduct disorder, delin-
quency, and violence in the Dunedin Longitudinal Study.
Cambridge: Cambridge University Press.
Monroe, S. M., & Simons, A. D. (1991). Diathesis-stress theories in
the context of life stress research: Implications for depressive
disorders. Psychological Bulletin, 110, 406–425.
Morash, M., Bynum, T., & Koons, B. (1998). Women offenders:
Programming needs and promising approaches. Washington:
USDOJ, National Institute of Justice.
Neale, M. C., & Cardon, L. R. (1992). Methodology for genetic
studies of twins and families. Dordrecht: Kluwer Academic.
Newsome, J., & Sullivan, C. J. (2014). Resilience and vulnerability in
adolescents: Genetic influences on differential response to risk
for delinquency. Journal of Youth and Adolescence, 43, 1–16.
Ostrer, H. (2001). Invited review: Sex-based differences in gene
expression. Journal of Applied Physiology, 91, 2384–2388.
Perrone, D., Sullivan, C. J., Pratt, T. C., & Margaryan, S. (2004).
Parental efficacy, self-control, and delinquency: A test of a
general theory of crime on a nationally representative sample of
youth. International Journal of Offender Therapy and Compar-
ative Criminology, 48, 298–312.
Plomin, R., DeFries, J. C., McClearn, G. E., & Rutter, M. (1997).
Behavioral genetics (3rd ed.). New York: W. H. Freeman.
Prendergast, M., Wellisch, J., & Falkin, G. (1995). Assessment of
services for substance-abusing women offenders and correc-
tional settings. Prison Journal, 75, 240–256.
Quinton, D., Pickles, A., Maughan, B., & Rutter, M. (1993). Partners,
peers, and pathways: Assortative pairing and continuities in conduct
disorder. Development and Psychopathology, 5, 763–783.
Raine, A. (2002). Biosocial studies of antisocial and violent behavior
in children and adults: A review. Journal of Abnormal Child
Psychology, 30, 311–326.
Raine, A. E., Brennan, P. A., Farrington, D. P., & Mednick, S. A.
(1997). Biosocial bases of violence. In NATO advanced study
inst on the biosocial bases of violence, May, 1996, Rodos Palace
Hotel, Rhodes, Greece. Plenum Press.
Reiss, A. J., & Farrington, D. P. (1991). Advancing knowledge about co-
offending: Results from a prospective longitudinal survey of London
males. The Journal of Criminal Law and Criminology, 82, 360–395.
Ribeaud, D., & Eisner, M. (2010). Risk factors for aggression in pre-
adolescence: Risk domains, cumulative risk and gender differ-
ences—Results from a prospective longitudinal study in a multi-
ethnic urban sample. EuropeanJournal of Criminology, 7, 460–498.
Rowe, D. C., & Jacobson, K. C. (1998). National longitudinal study of
adolescent health: Pairs code book. Chapel Hill: Carolina
Population Center, University of North Carolina at Chapel Hill.
Rutter, M. (1985). Resilience in the face of adversity: Protective
factors and resistance to psychiatric disorder. British Journal of
Psychiatry, 147, 598–611.
Rutter, M. (1990). Psychosocial resilience and protective mecha-
nisms. In J. Rolf, A. S. Masten, D. Cicchetti, K. H. Neuchterlein,
& S. Weintraub (Eds.), Risk and protective factors in the
development of psychopathology (pp. 181–214). New York:
Cambridge University Press.
Sameroff, A. J., Bartko, W. T., Baldwin, C., & Siefer, R. (1998).
Family and social influences on the development of child
competence. In M. Lewis & C. Feiring (Eds.), Families, risk, and
competence (pp. 161–185). Mahwah: Erlbaum.
J Youth Adolescence
123
Scarr, S. (1968). Environmental bias in twin studies. Eugenics
Quarterly, 15(1), 34–40.
Scarr, S., & Carter-Saltzman, L. (1979). Twin method: Defense of a
critical assumption. Behavior Genetics, 9(6), 527–542.
Schmidt, F., Hoge, R., & Gomes, L. (2005). Reliability and validity
analysis of the Youth Level of Service/Case Management
Inventory. Criminal Justice and Behavior, 32, 329–344.
Schwartz, J. A., & Beaver, K. M. (2011). Evidence of a gene x
environment interaction between perceived prejudice and
MAOA genotype in the prediction of criminal arrests. Journal
of Criminal Justice, 39, 378–384.
Simons, R. L., Lei, M. K., Beach, S. R., Brody, G. H., Philibert, R. A.,
& Gibbons, F. X. (2011). Social adversity, genetic variation,
street code, and aggression: A genetically informed model of
violent behavior. Youth Violence and Juvenile Justice, 10, 3–24.
Sjo
¨berg, R. L., Ducci, F., Barr, C. S., Newman, T. K., Dell’Osso, L.,
Virkkunen, M., & Goldman, D. (2008). A non-additive interac-
tion of a functional MAO-A VNTR and testosterone predicts
antisocial behavior. Neuropsychopharmacology, 33, 425–430.
Skuse, D. H. (2006). Sexual dimorphism in cognition and behaviour:
The role of X-linked genes. European Journal of Endocrinology,
155, S99–S106.
Spear, L. P. (2009). Heightened stress responsivity and emotional
reactivity during pubertal maturation: Implications for psy-
chopathology. Development and Psychopathology, 21(01),
87–97.
Sprott, J. B., Jenkins, J. M., & Doob, A. N. (2005). The importance of
school: Protecting at-risk youth from early offending. Youth
Violence and Juvenile Justice, 3, 59–77.
Stein, K. B., Vadum, A. C., & Sarbin, T. R. (1970). Socialization and
delinquency: A study of false negatives and false positives in
prediction. The Psychological Record, 20, 353–364.
Steketee, M., Junger, M., & Junger-Tas, J. (2013). Sex differences in
the predictors of juvenile delinquency: Females are more
susceptible to poor environments; males are influenced more
by low self-control. Journal of Contemporary Criminal Justice,
29, 88–105.
Stoddard, S. A., Zimmerman, M. A., & Bauermeister, J. A. (2012). A
longitudinal analysis of cumulative risks, cumulative promotive
factors, and adolescent violent behavior. Journal of Research on
Adolescence, 22, 542–555.
Stouthamer-Loeber, M., Loeber, R., Farrington, D. P., Wikstro
¨m, P.
H., & Wei, E. (2002). Risk and promotive effects in the
explanation of persistent serious delinquency in boys. Journal of
Consulting and Clinical Psychology, 70, 111–123.
Stouthamer-Loeber, M., Loeber, R., Farrington, D. P., Zhang, Q., van
Kammen, W., & Maguin, E. (1993). The double edge of
protective and risk factors for delinquency: Interrelations and
development patterns. Development and Psychopathology, 5,
683–701.
Sullivan, C. J. (2011). The utility of the deviant case in the
development of criminological theory. Criminology, 49,
905–920.
van der Laan, A. M., Blom, M., & Kleemans, E. R. (2009). Exploring
long-term and short-term risk factors for serious delinquency.
European Journal of Criminology, 6, 419–438.
van der Laan, A. M., Veenstra, R., Bogaerts, S., Verhulst, F. C., &
Ormel, J. (2010). Serious, minor, and non-delinquents in early
adolescence: The impact of cumulative risk and promotive
factors. The TRAILS study. Journal of Abnormal Child
Psychology, 38, 339–351.
Van Voorhis, P., Wright, E., Salisbury, E., & Bauman, A. (2010).
Women’s risk factors and their contributions to existing risk/
needs assessment: The current status of a gender-responsive
supplement. Criminal Justice & Behavior, 37, 261–288.
Van Wormer, K. (2001). Counseling female offenders and victims: A
strengths-restorative approach. New York: Springer.
Vaske, J. C., Boisvert, D. L., & Wright, J. P. (2014). A biosocial
perspective of female offending. In F. T. Cullen, P. Wilcox, J.
L. Lux, & C. L. Jonson (Eds.), Sisters in crime revisited. New
York: Oxford University Press.
Waaktaar, T., & Torgersen, S. (2012). Genetic and environmental
causes of variation in trait resilience in young people. Behavior
Genetics, 42, 366–377.
Wang, P., Niv, S., Tuvblad, C., Raine, A., & Baker, L. A. (2013). The
genetic and environmental overlap between aggressive and non-
aggressive antisocial behavior in children and adolescents using
the self-report delinquency interview (SR-DI). Journal of
Criminal Justice, 41, 277–284.
Welsh, J. L., Schmidt, F., McKinnon, L., Chattha, H. K., & Meyers, J.
R. (2008). A comparative study of adolescent risk assessment
instruments predictive and incremental validity. Assessment, 15,
104–115.
Werner, E. E., & Smith, R. S. (1982). Vulnerable but invincible: A
longitudinal study of resilient children and youth. New York:
McGraw Hill.
Werner,E.E.,&Smith,R.S.(1992).Overcoming the odds: High risk
children from birth to adulthood. Ithaca: Cornell University Press.
Wikstro
¨m, P. H. (2008). In search of causes and explanations of
crime. In R. King & E. Wincup (Eds.), Doing research on crime
and justice (2nd ed., pp. 117–139). New York: Oxford Univer-
sity Press.
Zahn-Waxler, C., Crick, N. R., Shirtcliff, E. A., & Woods, K. E.
(2006). The origins and development of psychopathology in
females and males. In D. Cicchetti & D. J. Cohen (Eds.),
Developmental psychopathology (pp. 76–138). Hoboken: Wiley.
Zheng, Y., & Cleveland, H. H. (2013). Identifying gender-specific
developmental trajectories of nonviolent and violent delinquency
from adolescence to young adulthood. Journal of Adolescence,
36, 371–381.
Jamie Newsome is an Assistant Professor in the Department of
Criminal Justice at the University of Texas at San Antonio. She
received her doctorate in Criminal Justice from the University of
Cincinnati in 2013. Her research interests include biosocial crimi-
nology; developmental, life-course criminology; and the policy
implications of biosocial research on crime and delinquency.
Jamie C. Vaske is an Associate Professor in the Department of
Criminology and Criminal Justice at Western Carolina University.
She received her doctorate in Criminal Justice from the University of
Cincinnati in 2009. Her research interests include quantitative
methods, gender and crime, and biosocial and life-course
criminology.
Krista S. Gehring is an Assistant Professor in the Department of
Criminal Justice at the University of Houston–Downtown. Her
primary research interests include correctional risk/needs assessment
and treatment intervention strategies, with a particular focus on
female offenders. Her research has appeared in such journals as
Criminal Justice and Behavior,Feminist Criminology, and Women,
Girls, & Criminal Justice.
Danielle L. Boisvert (Ph.D., University of Cincinnati) is an
Associate Professor and Graduate Director in the Department of
Criminal Justice and Criminology at Sam Houston State University.
Her research focuses on the influence of genes and the environment
on antisocial behaviors throughout the life course.
J Youth Adolescence
123
A preview of this full-text is provided by Springer Nature.
Content available from Journal of Youth and Adolescence
This content is subject to copyright. Terms and conditions apply.
