You are on page 1of 16

Direct Protective and Buffering Protective

Factors in the Development of Youth Violence


Friedrich Lsel, PhD, David P. Farrington, PhD
Abstract: This article discusses conceptual issues and reviews knowledge about direct and buffering
protective factors in the development of youth violence. Direct protective factors predict a low
probability of violence, whereas buffering protective factors predict a low probability of violence in
the presence of risk (and often interact with risk factors). Individual, family, school, peer, and
neighborhood factors are reviewed. Heterogeneity of variables, measurement, contexts, study design,
sample, and other characteristics limit generalizations. However, there were various evidence-based
candidates for having a direct protective or buffering protective effect such as above-average intelligence, low impulsivity/easy temperament, enhanced anxiety, prosocial attitudes, high heart rate,
close relationship to at least one parent, intensive parental supervision, medium SES of the family,
sound academic achievement, strong school bonding, a positive school/class climate, nondeviant
peers, and living in a nondeprived and nonviolent neighborhood. The probability of violence
decreases as the number of protective factors increases (a doseresponse relationship). Implications
for future research and practice concern adequate research designs to detect nonlinear relationships;
conceptually and methodologically homogeneous studies; differentiated analyses with regard to age,
gender, and other characteristics; and greater integration of longitudinal correlational research with
(quasi-)experimental intervention studies.
(Am J Prev Med 2012;43(2S1):S8 S23) 2012 Published by Elsevier Inc. on behalf of American Journal of
Preventive Medicine

Introduction

chool shootings, knife murder, and other forms of


serious violence among young people have raised
much public concern in Western countries. In its
less spectacular forms as well, youth violence is a serious
problem for victims, parents, teachers, peers, neighbors,
and in the longer run, for the youth themselves. Although
there is no generally accepted defnition of violence, it is
widely seen as a subcategory of aggression (intended
harm of others) with a focus on physical impairment.
Violent behaviors are key symptoms of conduct disorders
that are particularly frequent syndromes among children
and youth.1 Early violence and delinquency are risk
markers for serious criminality and social and mental
health problems in youth and adulthood.2,3 Such children
are very diffcult to place, manage, and treat.4 Further, a
persistent serious young offender entails costs of approximately $1$5 million for society on average.5,6

From the Institute of Criminology (Lsel, Farrington), Cambridge University, Cambridge, United Kingdom; and the Institute of Psychology (Lsel),
University of Erlangen-Nuremberg, Erlangen, Bavaria, Germany
Address correspondence to: Friedrich Lsel, PhD, Cambridge University, Institute of Criminology, Sidgwick Avenue, Cambridge CB3 9DA,
United Kingdom. E-mail: fal23@cam.ac.uk.
0749-3797/$36.00
http://dx.doi.org/10.1016/j.amepre.2012.04.029

S8 Am J Prev Med 2012;43(2S1):S8 S23

For these and other reasons, prevention of youth


violence is highly important.7,8 Effective prevention
measures require a sound knowledge on the origins
and development of youth violence. Prospective longitudinal studies have led to a comprehensive understanding of the characteristics of youth who are at risk
of becoming serious and violent offenders.9 However,
most research concentrates on risk factors and defcits.
Much less attention has been paid to those cases that
are not in the focus of risk models. In particular, these
are individuals who abstain from violence in spite of a
high-risk background or desist from an antisocial
pathway.7,10,11 Such phenomena are not unusual. Even
in high-risk groups, more than half of the children
develop relatively well and a majority of serious offenders desist over time.12,13
The questions of why individuals do not set out on an
antisocial pathway, or why they leave it, led to the discussion of direct and buffering protective factors. Although
this is often seen as a paradigm shift from a focus on
defcits and pathology to strengths, well-being, and a
positive psychology,14 both perspectives are necessary for
an adequate understanding of human behavior problems.11 Risk models always account for a limited amount
of outcome variance that can be enhanced by investigating protective factors.

2012 Published by Elsevier Inc. on behalf of American Journal of Preventive Medicine

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

Conceptual and Methodologic Issues


Basic Concepts and Terminology
A large fraction of research on protective factors is related
to the topic of resilience in developmental psychopathology.19,20 This concept is used to describe both the process
and outcome of an elastic biopsychosocial adaptation to
stressors and diffcult circumstances in life.21 Resilience
refers to phenomena such as healthy development despite
a high-risk status (e.g., growing up in a multiproblem
milieu); maintaining competence under specifc stressors
(e.g., coping with parental divorce); or recuperating from
severe trauma (e.g., child abuse). These phenomena correspond to the elementary biological mechanisms of protection, regeneration, and repair.
Although resilience research has a broader focus, studies on youth violence can beneft from its theoretic and
methodologic discussions.11,19 This applies to the basic
concepts of risk and protective factors. There is not
yet a consistent terminology in this feld, in particular
with respect to protective factors. Risk factors are personal or social characteristics of an individual that predict
a high probability of a future behavior problem such as
the onset, persistence, or aggravation of youth violence.
In a complementary sense, protective factors should predict a low probability of such negative outcomes.
However, it is necessary to differentiate direct protective and buffering protective factors.22 Direct protective
factors predict a low probability of future problem behavior without taking other factors into account. They refer
to the main effect of a variable. In contrast, the term
buffering protective factors should be restricted to variables that predict a low probability of a negative outcome
August 2012

% violent

Research on the integration of both perspectives is


theoretically and methodologically more complicated
than risk research alone. Therefore, this article provides a
conceptual framework for the study of protective factors
in the feld of youth violence and contains a brief narrative review of research fndings. First, terminological,
conceptual, and methodologic issues are addressed. Second, a review is presented of the research on a number of
factors that may be candidates for having protective effects in relation to the onset, continuation, or aggravation
of youth violence. In part, reference is made to research
on closely related outcomes such as aggressive and criminal behavior because more systematic work on protective factors against youth violence is necessary (which is
the reason for the empirical studies in this supplement to
the American Journal of Preventive Medicine).1518 The
last part of the article draws conclusions and outlines
perspectives for future research as well as potential applications in violence prevention.

S9

Risk level
Group with the protective factor
Group without the protective factor

Figure 1. Interaction between a risk and a buffering protective factor in predicting youth violence

in the presence of risk factors. This terminology refers to


the moderating or interaction effects of the factors. In
other words, the buffering protective factor attenuates the
impact of a risk.
Risk and protective factors may not be different variables. The same variable may function simultaneously as
both a risk and protective factor, depending on what side
of the coin is viewed. For example, whereas the presence
of a stable emotional bond to a caregiver may be directly
protective, its absence is an important risk factor.23 Obviously, it can be confusing when two constructs are used
to describe different manifestations of one and the same
variable.
This is not just a problem in dichotomous variables,
but also in continuous ones. For example, whereas low
intelligence is a well-replicated risk factor for antisocial
behavior,24 high intelligence seems to have a direct protective or a buffering protective function.25 Accordingly,
the exact meaning of a risk or protective factor refers not
to the particular variable in general but to different poles
or different degrees of a continuous variable. A variable
can have direct protective effects, but not risk effects if it is
nonlinearly related to an outcome such as youth violence.
Loeber and Farrington26 have explained the meaning of
direct protective factors in detail.
Figure 1 gives an example of a buffering protective
variable. It shows the effect of this variable (e.g., strong
emotional bonding to a noncriminal reference person)
on the relationship between a risk factor (e.g., growing up
in more- or less-deprived neighborhoods) and a violent
outcome (e.g., probability of violent offending). There is a
linear increase for the group without the buffering protective factor but no increase for the group who possess it
(or a buffering protective degree of a quantitative variable). In Figure 1, the buffering protective factor nullifes

S10

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

the impact of a risk factor. In other cases, a buffering


protective variable could have a more graduated influence and lead to a reduction in the impact of the risk
factor. This would be indicated by a slope that lies between the two lines in Figure 1.

Research Designs
Although studies have used cross-sectional or retrospective longitudinal designs to investigate direct and buffering protective factors,27,28 prospective longitudinal designs are preferable.23,29 For example, they can address
longer-lasting effects in development and are less biased
by confounded information on predictors and outcomes.
In any case, the search for direct protective effects requires a design that is sensitive to nonlinear relations
between the predictors and outcomes.
Such a method was developed in the Pittsburgh Youth
Study.22,28 Age cohorts were split approximately at the
25th and 75th percentile of their score distribution on
potential risk and protective factors. This trichotomization resulted in one quarter of boys at low risk, a middle
half, and one quarter of boys at high risk. The degree of
delinquency was trichotomized in the same way. A direct
protective effect was inferred when the desirable quarter
of scores in a variable promoted nondelinquency or suppressed serious delinquency in comparison with the middle half. A risk effect was indicated when the undesirable
quarter fostered serious delinquency or suppressed
nondelinquency.
Prospective longitudinal designs can be used not only
to study direct and buffering protective factors for a specifc outcome at one time but also to investigate change
processes of, for example, desistance or aggravation of
violent behavior. Of course, it is arguable whether the risk
and protective poles of a variable should be defned according to an empirical frequency distribution or on the
basis of a theoretic distinction. In both cases, the setting of
a cut-off point is arbitrary in principle but may be guided
by specifcity and sensitivity coeffcients in prediction
models.30
The investigation of buffering protective factors requires the analysis of moderator effects of the protective
variable on the relationship between a risk factor and
violence. The most widely used design compares two
groups. Both are characterized by the same risk factors or
degree of risk. One group shows the predicted undesirable outcome whereas the other (resilient) group does not
become violent.
Typical examples of such a design are studies on children who experienced child neglect, abuse, or other
trauma and developed positive versus negative behavioral outcomes.31 Variables that differentiate between the
two groups can be interpreted as having a buffering pro-

tective effect because they nullifed or reduced the risk


effect in the group with the desirable outcome. There is an
interaction effect if the buffering protective variable predicts the outcome in the risk category differently from
that in the nonrisk category. Similar to the investigation
of direct protective (main) effects, buffering protective
(interaction) factors also can be studied in multivariate
designs in which other variables are controlled in regression analyses or structural equation models.
The study of Cicchetti et al.31 on relatively positive
behavioral outcomes of maltreated children who showed
ego-resilience or overcontrol (versus undercontrol) in
their personality is an example of the investigation of
buffering protective factors in high-risk samples, whereas
the above-mentioned Pittsburgh study addressed large
population samples. Findings from high-risk studies often have more clinical relevance, but population samples
are normally larger, have more statistical power, and permit analyses of complex interaction effects.

Issues of Causality
Even the most sophisticated longitudinal designs and
data analyses are limited in detecting causal effects.32
Therefore, research and practice must be sensitive to
issues of causality. In principle, risk factors and direct
protective/buffering protective factors indicate only correlations between time-sequenced variables and do not
necessarily show causal relationships.11,33 For example,
early antisocial behavior is a sound correlate/predictor
but not a cause of later youth violence.34
Accordingly, Kraemer and colleagues35 distinguish
among (correlative) risk factors; risk markers; and (truly)
causal risk factors.35 Risk markers are stable characteristics that may indicate an influence on the outcome but do
not have a causal impact by themselves (e.g., male gender
and violence). A causal risk factor can be assumed when
the change of the respective variable in a controlled intervention leads to a change in the outcome (e.g., less violence in participants of a social skills training group than
in an equivalent control group). In addition, one must
differentiate between more proximal and distal risks and
between the empirical factors and the underlying protective mechanisms that may form a causal explanation.20
However, there is not only one concept of causality36
and in felds where experimental manipulation is not
possible, causal inferences are often a matter of plausibility depending on the explicitness of hypotheses and differences in methodologic quality. Therefore, studies on
protective effects should fulfll similar methodologic criteria to those used in program evaluation.37 Murray et
al.38 have systematized such quality criteria for nonexperimental studies (e.g., a clear time sequence, control
for third variables, independent measurement). Accordwww.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

ingly, research on protective factors should aim for wellcontrolled quasi-experimental designs and withinindividual analyses of change in prospective longitudinal
studies.39,40
Although criminology needs more RCTs,41 this
method is not the primary choice for the present topic.
Protective mechanisms and effects are part of the natural
life course of individuals. They often refer to serious life
risks (e.g., child neglect, growing up in poverty) that are
not subject to randomization and experimental manipulation. Some potential protective factors are also extremely diffcult to implement in experimental designs
(e.g., attachment to a caregiver).
For these and other reasons, prospective longitudinal
designs are currently the method of choice in studying
protective factors. If this strategy leads to well-replicated
fndings, the particular factors could be manipulated experimentally as far as appropriate. Such a stepwise research strategy would not only make the best use of the
research on protective effects for violence prevention but
would also reduce the risk of implementing programs
that may be harmful.42

Definition and Measurement of the Outcome


Behavior
Research on direct protective and buffering protective
effects of variables requires the defnition and measurement of desirable behavioral outcomes. In many studies,
these are defned as the absence or reduction of mental
disorders or behavioral problems.19 Some authors
pleaded for more positive indicators of social competence,43 but this is not the topic of the present article. A
desirable outcome can be defned not only for a specifc
type of deviant behavior (e.g., no violence) but also more
broadly (e.g., generally no mental health problems). Both
approaches have their strengths and weaknesses and it
has always to be asked risk of what? and protection
against what?
A broad defnition of a desirable outcome could mask
specifc relationships that may be important for understanding causal processes. On the other hand, too narrow
a defnition can mistakenly suggest healthy development despite serious problems in other areas of functioning. For example, a child who exhibits no violence
may well have less-obvious internalizing symptoms that
could even lead to suicide. Studies have shown that nonantisocial children from a family background of abuse,
neglect, or accumulated stressors experienced enhanced
emotional or relationship problems.44 46 Similarly,
youth who start relatively late in exhibiting antisocial
behavior may not have been truly healthy in earlier life
but may have exhibited internalizing problems47,48 or
August 2012

S11

overcontrolled hostility until anger arousal overcame


strong inhibitions and resulted in violence.49
In addition to the defnition of outcome, research on
protective factors in the feld of youth violence must
clarify issues of outcome assessment. For example, it
needs to be specifed which types of behaviors (i.e., offenses) should be included and how seriousness or intensity should be graded. Although there seem to be only
minor differences between the risk factors for homicide,
assault, and less-serious forms of youth violence,50 it
must not be the same for protective factors. For statistical
reasons, one can expect smaller correlations when the
base rate of the behavioral outcome is low.
Findings also may vary with the type of measurement
(e.g., self-report, parent report, teacher report, or offcial
offending44,51). In general, a multi-setting and multiinformant approach is most appropriate.52 However, different informants show only moderate agreement, particularly in the broad middle range of problem intensity.
Although averaging the different assessments is often
appropriate, it may not generally be the best approach
(e.g., the information from external experts may be more
valid than that from parents who have limited comparisons with other children).
A further relevant issue is the developmental time period or age of the youth. Protective factors may be different for the onset of, or desistance from, violence. Although an early onset is most relevant for serious and
persistent antisociality, there are other pathways such as
late starters, early desisters, or children with a substantial
variety of crimes over time.13,44,53 In defning desirable
outcomes, researchers have to ask whether it is suffcient
to desist from violence over a relatively short period of
time or whether nonviolence is required for many years
or the rest of ones life. Of course, the duration of longitudinal studies sets practical limits here. Similar questions arise when desistance or aggravation is addressed as
not only a dichotomous issue but also a matter of degree
(i.e., a change in frequency or intensity).

Research Findings on Direct Protective


and Buffering Protective Factors
The candidates for protective effects against youth violence cover a very broad range of variables (e.g., biological, personality, family, school, peer, and community
variables).11,34,54,55 Therefore, at this conceptual stage a
narrative review is more adequate than a meta-analysis.56
Because previous reviews11,54 suggested only a small
number of studies that specifcally addressed protective
factors against youth violence, the current literature
search was widened to include outcome criteria that often
contain violence as a subcategory or are correlated with

S12

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

violence (e.g., aggressive behavior, serious offending, delinquency, antisocial behavior, gang involvement). The
focus of the search was on longitudinal studies published
between 1990 and 2009. More than a dozen databases
were searched (e.g., SSCI, MEDLINE, PsycINFO,
C2-Spectr) and articles in more than 30 journals.
Overall 112 studies met the inclusion criteria; relevant
older studies were added from previous reviews.11,54 The
available research turned out to be extremely heterogeneous with regard to research design, risk factors, protective factors, outcome measures, samples, data analyses,
follow-up periods, and other characteristics. Therefore,
the following discussion concentrates on potential protective factors that have at least been found in two longitudinal studies. If there were many more studies on a
specifc variable, space limits did not allow reporting all of
them. The fndings are categorized into individual, family, school, peer group, and neighborhood factors.

Characteristics of the Individual


Intelligence/Cognitive Competencies
Low intelligence is a risk factor for violence and delinquency.57,58 Conversely, above-average or high intelligence has a direct protective function against these problems.25,28,44,47,59,60 62 Above-average intelligence also
has a buffering protective function in the presence of
other risks such as a criminal father, a high-risk milieu, or
previous antisocial behavior.25,59,62 64 Some studies only
partially confrmed this buffering protective function, for
example, only for men/boys60,65 or for specifc subfactors
of intelligence.28,44 Cognitive competencies also seem to
be less important for desistance from offending.55
The protective function of above-average intelligence
relates to better executive neuropsychological functioning,24,66 as manifested in self-control or social information processing. Various studies suggest that it is not so
much an abstract intellectual capacity but more practical
intelligence, social competence, and realistic planning
that protect against antisocial development in the presence of risk.45,67 69 In high-risk migrant populations,
language competence seems to have a similar protective
function.70

Social Cognitions
Theories of social learning, information processing, and
action suggest that social and self-related cognitions are
important risk factors for violence and other forms of
aggression.71,72 Some fndings also support protective effects of the desirable pole of such variables. In the Pittsburgh Youth Study, for example, attitudes toward family
and school had both risk and direct protective functions.73 The perceived risk of being caught had a similar

two-sided effect. This is in accordance with Mofftt and


colleagues,74 who found that abstainers and those who
recovered from offending perceived the risk of detection
for a crime as higher than boys who showed adolescencelimited or persistent antisociality. Other fndings suggest
that deterrence is important for only those children who
are high in criminal propensity.75
Modes of perceiving, interpreting, and evaluating situations and action alternatives seem to have a buffering
protective function in the presence of risk factors.76 The
importance of social information processing also is supported by the literature on the prevention and treatment
of antisociality.77 However, some studies on natural
processes of desistance do not confrm the importance of
these factors. Loeber et al.,73 for example, found that
negative attitudes toward delinquency and high selfaspiration did not predict desistance from serious and
violent offending.
Various studies23,43,45,46,59,80 have shown that resilient
youth possess beliefs in self-effcacy and a low feeling of
helplessness. However, the relationship between violence
and self-esteem is complicated. Some studies show that
low self-esteem is related to aggression.81 Others suggest
no relationship82 or even enhanced/narcissistic selfesteem in aggressive and violent individuals.83 There are
likely additional factors related to self-esteem (e.g., the
degree of realism and fragility) that influence whether it
has a desirable developmental effect or not.
Low feelings of helplessness and of confdence in being
able to master life problems may be part of a good sense of
coherence84 or spiritual beliefs that give ones life structure and meaning. Various studies found that religious
orientations may promote nonviolence and nondelinquency in physically abused, high-risk, and representative samples of children and youth.85,86

Temperament Factors
Impulsivity, hyperactivity, negative emotionality, sensationseeking, and risk-taking are predictors of violence,7,30,34,87 but
the opposite pole of these variables seems to predict desirable
development. Loeber and colleagues22 have shown that low
attention-defcit hyperactivity disorder (ADHD) mostly operates as a direct protective factor reducing the development of
violence and theft. Early psychopathic temperament features and depressed mood had both risk effects and at
the opposite pole direct protective functions. Other
studies found buffering protective effects of an easy
temperament88 in the presence of various social risk
factors. For example, sociability, positive mood, low
irritability, and low impulsivity seem to have such a
buffering protective function.43,44,46,59,63,65
In the Dunedin study,74 children who abstained from
violence and other antisocial behavior exhibited the easwww.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

iest temperament at the preschool age (e.g., low in emotional instability, restlessness, attention problems, and
roughness). Those on a persistent antisocial pathway, and
particularly those who became violent, had a more diffcult temperament in childhood than those with
adolescence-limited antisociality. Although boys who recovered from antisocial behavior had similar temperament diffculties at the preschool age as the persistently
deviant youngsters, their temperament improved later.
This fnding points to the potential of direct protective
changes in the phenotype of temperament (a disposition
that has a genetic component).
The characteristics of the easy child are similar to
Blocks89 concept of ego resiliency (flexible adaptation of
control to environmental demands). Whereas ego undercontrol (e.g., impulsiveness) is a risk factor for violence
and antisocial behavior,30,90 ego-resilience showed
buffering protective effects in children who were maltreated or came from otherwise disadvantaged backgrounds.31,44,59,78 Similar results were found for ego overcontrol (inhibition of action). However, the personality
facets of anxiety, shyness, and internalizing problems
seem to have two effects. They can protect against the
development of antisocial behavior in childhood and
youth,22,91 but they also may be risk factors for late-onset
criminality48 and hinder desistance in already violent
youngsters.22

Biological Factors
Numerous biological risk factors for violent behavior are
discussed in the literature (e.g., genetic factors, autonomic arousal, hormonal influences, neurotransmitter
processes, prenatal intoxication or perinatal complications). However, the causal function of biological variables is complex, and they often do not have main effects
but interact with social factors.20,21,55
On the genetic level, a functional polymorphism in the
promoter region of the monoamine oxidase A (MAO-A)
gene that is relevant for normal neurotransmitter function seems to have a buffering protective function. Caspi
et al.92 found linear relationships between the intensity of
child maltreatment in the family and later antisocial behavior of children. A differentiated analysis showed that
there was a less negative outcome of severe maltreatment
in children with high genetic MAO-A activity as compared to low MAO-A activity. Replication studies showed
mixed results93 or found a buffering effect in neglected or
abused Caucasians only and not in African-American
subsamples.94 A meta-analysis supports the interaction
of MAO-A and maltreatment,95 but it is not yet clear
whether the genetic disposition is as important as the
factual MAO-A activity in the brain.96,97
August 2012

S13

A larger number of studies have shown not only that


low heart rate is a risk factor for violence98 but also that
there is a buffering effect of high arousal in the presence of family and other risk factors.44,99 101 In addition, a higher heart rate level, higher skin conductance
arousal, and better skin conductance conditioning predicted desistance from aggressive behavior.102 Enhanced arousal goes along with the above-mentioned
protective factors of anxiousness/shyness,103 and these
traits also seem to have a buffering protective function.
It must also be assumed that there are biosocial interactions; for example, as low arousal seems to be associated with social influences such as the experience of
violence or emotional deprivation in the family during
early childhood.104,105
Hormonal functioning is a further candidate for
protective effects. For example, a high level of the stress
hormone cortisol is associated with more anxiety and
withdrawal103 and also may reduce an individuals testosterone level.106 However, children who are both
aggressive and anxious seem to have particularly high
cortisol levels.107,108 In general, the complex bidirectional and interactional effects between hormones,
aggression, and context variables suggest no simple
and well-identifed buffering protective effects.109
Similarly, complex relationships are suggested for
other biological variables that should be investigated in
multilevel approaches to better understand protective
mechanisms.20,21
Overall, the research on individual direct and buffering
protective factors in the development of youth violence is
heterogeneous. However, this brief overview showed a
number of promising variables. In particular, aboveaverage intelligence, sound planning behavior, nonaggression-prone cognitive schemes, spiritual values,
positive attitudes toward family and school, an easy temperament, low ADHD, enhanced anxiety and shyness,
and high heart rate show replicated effects. High MAO-A
activity is also a serious candidate for a buffering protective effect, but further replication of biological factors
seems to be necessary.

Family Factors
Characteristics of the parent child relationship, parenting behavior, parent attitudes, and the overall family climate are often closely inter-related. Therefore, protective
functions in one of these areas may be mediated by factors
in others. Family factors also do not have a unidirectional
impact on the child but are partially influenced by the
childs behavior110 and may even reflect an underlying
genetic transmission.111

S14

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

ParentChild Relationship
An emotionally positive parent child relationship and
secure bonding is a basic direct protective factor in child
development and has buffering protective effects in preventing a broad range of behavior problems in the presence of risk factors.112116 A close relationship with parents also promotes nonviolence and protects against the
aggravation of problems in youth.28,117 A positive relationship with at least one parent encourages social learning in a structured social environment, whereas early
deprivation may disrupt the brain development and neuronal functioning that are relevant for the regulation of
aggressive behavior.118 However, the protective effects of
stable positive emotional relationships do not appear to
be limited to early childhood and the core family. Studies
of children from multiple deprived backgrounds have
shown that a good bond to other reference people can
buffer risks and lead to nonviolent development.23,46 A
close relationship to a nondeviant partner also encourages the desistance of offenders in adulthood.119

Parenting Behavior
Numerous studies have shown linear relationships between parenting behavior and child aggression and violence. For example, in the Pittsburgh Youth Study parental reinforcement had a risk effect at its negative pole and
a direct protective effect at its positive pole. Intensive
supervision, high persistence of discipline, low physical
punishment, and strong involvement of the child in the
familys activities had mainly direct protective effects.22,28
Similar factors also seem to have a buffering protective
function in the presence of risks23 and in educational
contexts outside the family.120,121 Herrenkohl and colleagues85 showed that parental disapproval of antisocial
behavior predicted a low rate of violence in both physically abused and non-abused children.
Intensive supervision or frm parenting also improved
the behavior of children with conduct problems117 and
led to desistance from serious offending.22 An emotionally warm, attentive, accepting, norm-oriented, supervising, and structure-giving upbringing encourages the positive development of children.122 Such parenting
characteristics, together with parental stimulation, also
seem to have a buffering protective function in the presence of risk factors for antisocial development.115,116

Other Family Factors


Various other family characteristics may have protective
effects against violent behavior. For example, the SES of
the family had not only a risk effect at the lower pole but
also a direct protective effect at the upper pole.22 A similar
relationship was found for parental stress, and low paren-

tal stress had a buffering protective function that encouraged desistance from serious offending.22 The parents
own constructive coping with problems can be a model
for the child that contributes to the childs non-aggressive
development in a high-risk context.46,117 Interest and
positive parental attitudes toward the childs education
also had a direct protective effect against antisociality.22,44 Overall, early family factors seem to be more
relevant for promoting nonviolence and protecting
against violence in at-risk cases than for later desistance
from serious offending.22
In summary, research shows relatively consistent candidates for direct protective/buffering protective family
effects in the feld of youth violence. It suggests positive
influences of a close relationship with at least one parent,
parental acceptance, intensive supervision, high persistence of discipline, parental disapproval of antisocial behavior, low physical punishment, strong involvement in
family activities, above-average SES, low parental stress,
family models of constructive coping, and positive parental attitudes toward the childs education.

School Factors
Whereas negative school relationships and experiences
are important risk factors for serious and violent youth
offending,34 the positive side of the coin has a direct
protective/buffering protective function.85,121,123 The
main desirable influences of school factors derive from
the following areas: academic achievement, motivation,
school bonding, and class/school climate. Sound school
achievement or reaching higher education not only has a
direct protective effect22,44 but also has a buffering protective effect in the presence of other risks46 or against the
aggravation of criminality.28,124
Similar relationships were found for strong school motivation, commitment, and bonding to school.44,46 The
study of Herrenkohl et al.85 showed that good school
achievement had a buffering protective function against
violence and delinquency in a group that had been abused
as children. In the Dunedin study, the percentage of boys
leaving high school early was lowest for abstainers and
those who recovered from childhood antisociality.74 Although the desirable effects of good school achievement
and school bonding may be related to the buffering protective function of intelligence, this is not the main factor.46,61,65,79 In children from high-risk backgrounds,
sound school achievement and bonding to the school was
even more relevant for successful adjustment than was
high intelligence.67,68,125
The causal chains underlying the desirable effects of
school achievement and bonding are not fully clear. For
example, school success can be a source of self-effcacy
www.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

that compensates for negative experiences in other contexts.64,126 A close relationship to the school and a successful school career may indicate strong bonding to society127 and open the door to noncriminal opportunities
in life.128 Similarly, academic and professional motivation, desirable work behavior, and job stability are important factors influencing desistance from criminality in
late adolescence and early adulthood.119,129 However,
various studies also question the independent causal role
of school factors in positive versus negative social
development.130 133
Longitudinal intervention studies on anti-bullying
programs have shown that intensive supervision, clear
behavior rules, consistent negative reinforcement of aggression, engagement of parents and teachers, and other
characteristics of a positive school/class climate are relevant for reducing violence at school.134 Most of these
studies address the prevalence of aggression in the school
context. However, as there are long-term predictive relationships between school bullying and later violence in
other contexts,135137 it can be assumed that factors that
reduce school bullying may have a direct protective and
buffering protective effect in regard to later violence.
A summary of protective school factors against youth
violence has to take into account that these variables are a
mix of individual reactions to school and more- or lessobjective school characteristics. Overall, there is evidence
for desirable effects of good school achievement, bonding
to school, strong work motivation, reaching higher education, support and supervision by teachers, clear rules,
and other positive features of the school and class climate.
More research that disentangles individual and social
school-related factors is needed.

Peer Factors
Juvenile violence is often a group phenomenon and having deviant peers is strongly correlated with delinquent
behavior.138 140 However, Farrington et al.141 found that
delinquent peers did not predict a boys delinquency
within individuals, although this variable did predict between individuals. They concluded that peer delinquency
was not a cause of delinquency but a marker reflecting
group offending, unlike family factors, which did predict
within individuals.
Deviant peers can both model and reinforce violence.
Peer antisociality is an important predictor of serious and
violent juvenile offending for youth gangs.22,34,142 However, peers also can have a protective function.143145
Having a nondeviant good friend or being a member of a
peer group who disapprove of antisocial behavior has a
direct protective effect against youth violence and delinquency.22,44,74,85 Close relationships to nondeviant peers
August 2012

S15

or involvement in religious groups can have a buffering


protective effect against violence in the presence of
risks.46,85 These factors also seem to promote desistance
when antisocial development has already started.44,74,124
Although the correlations of violence with peer factors
are robust, the causal effects are less clear. On the one
hand, having deviant friends may lead to antisocial behavior. On the other hand, youth may join peer groups on
the basis of similar interests and attitudes.146,147 Most
fndings suggest that both selection and facilitation processes play a role.142
The positive or negative impact of peer factors depends
on moderators such as the age of the youth, personality
features, the type of antisocial pathway, and gender. For
example, the influence of deviant peers on violence and
delinquency is reduced by strong school bonding148 and
academic achievement.149 Early-starting antisocial children are often rejected by others and as a consequence
may join deviant peer groups.150,151 Antisocial adolescents are more accepted by deviant peers152 and may
therefore be reinforced in their deviant behavior.128,144
Complex constellations of risk and direct protective peer
group influences must also be assumed with regard to social
isolation. For example, having few or no friends is an indicator of social isolation that temporarily protects children
from antisocial behavior.153 Its negative mental health
implications, however, can lead to later antisociality143 or
noncriminal but otherwise unsuccessful lives.154 A similar
double effect of a variable is suggested for heterosexual
relationships. Having no girlfriend seems to be a direct
protective factor in younger boys45 whereas a nondeviant
girlfriend may become a buffering protective factor for desistance at later ages.119 The aggressive behavior of girls
depends even more on their intimate relationships.150,155
In summary, there are relatively consistent fndings on
peer factors. Having a nondeviant good friend, belonging
to a peer group who disapprove of antisocial behavior,
involvement in religious groups, or being socially isolated
seem to have direct protective or buffering protective
effects against youth violence.

Neighborhood and Community Factors


Juvenile violence is related to neighborhood and community characteristics such as high rates of families living in
economic deprivation, in public housing or on social
welfare, high area crime and violence rates, availability of
drugs, large proportions of non-integrated immigrants,
or racial discrimination.60,156 159 The causal impact of
these variables, or more integrative constructs such as
social disorganization, low social cohesion, and informal
control, is often confounded with individual and family
characteristics and may be both direct and indirect. The

S16

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

complex relationships and discrepancies among correlations on the aggregate and individual levels (the ecological fallacy) make it diffcult to disentangle and replicate
pure desirable neighborhood effects when other factors
are well controlled. Stouthamer-Loeber and colleagues,28
for example, found only risk effects and no protective
influence of community variables.
More-recent results from the Pittsburgh Youth Study
revealed that living in a good neighborhood mainly had a
direct protective effect and housing quality had both risk
and direct protective functions.22 Other analyses of the
Pittsburgh data support the potential effects of desirable
neighborhood characteristics on specifc subgroups. Lynam et al.160 found that living in a nondeprived neighborhood buffered to some degree the undesirable impact of
impulsivity on juvenile offending. According to Wikstrm and Loeber,159 advantaged neighborhoods protected against late-onset offending in relatively welladjusted adolescents.
In a study of younger children, Kupersmidt and colleagues161 found a buffering protective effect of living in
middle-class neighborhoods on the aggressive behavior
of black children from low-income, single-parent homes.
However, in these middle-SES neighborhoods, children
from low-income, white, single-parent families were rejected more often by their peers, and this rejection could
become a risk factor for later violence. These fndings
suggest that the same neighborhood may be buffering
protective for one family type but not for another. Studies
on the interaction of child maltreatment, violence in the
neighborhood, and child behavior point in the same direction.162,163 Other research reported main risk versus
direct protective effects of community violence on the
aggressive behavior of juveniles.156
Although social disadvantage is partially reproduced
when people move to better neighborhoods, some experiments that improved housing or supported at-risk families in moving to better neighborhoods revealed direct
protective/buffering protective effects.164 As there are interactions of these effects with desirable individual and
family characteristics, it is not a single neighborhood
characteristic but the accumulation of desirable compared to undesirable features and subjective experiences
that are primarily relevant for a childs resilience.165
Neighborhood poverty, for example, is not a risk per se
but can be associated with good cohesion and positive
experiences that may buffer the detrimental effects of
socioeconomic deprivation.158,166,167 Informal social
control and social trust/cohesion seem to be key neighborhood characteristics that promote nonviolence in various cultures.157 Because of the complex multilevel relationships, the protective effects of community and
neighborhood factors are most diffcult to prove. How-

ever, the brief discussion provided here shows that living


in a nondeprived, nonviolent, and cohesive neighborhood seems to have such positive effects.

DoseResponse Relationships
In predictions of serious and violent youth offending,
most single risks show relatively low correlations with the
outcome.30,34 The probability and intensity of antisocial
behavior increase strongly as a function of accumulated
risks.34,76,168 The single factors may be less or more independent from each other or causally related in such a way
that one factor may trigger another as in a cascade.11,54,165
Their cumulative effects are often additive but they also
can be multiplicative or exponential.54,169
Similar but inverse doseresponse relationships are
also relevant for direct and buffering protective factors.11,54,165 There is not much research on this issue, but
all available results point in the same direction. In a study
of groups of physically abused children, non-abused children, and children at risk of abuse, Herrenkohl et al.85
found that an increased number of protective factors
predicted low rates of violence and delinquency. With
regard to delinquency as an outcome, Smith and colleagues170 found that the percentage of non-offenders
increased with the number of desirable (direct protective)
factors to which the youngsters had been exposed.
Wikstrm and Loeber159 formed an overall risk vs
direct protective score by summing the undesirable versus desirable poles of variables such as high vs low impulsivity, weak vs strong parental supervision, high vs low
school motivation, many vs few peers, positive vs negative perception of antisocial behavior, and low vs high
levels of guilt. Youth with a high score at the direct protective end had the lowest rate of serious juvenile offending and were also least vulnerable to the additional effects
of living in disadvantaged neighborhoods.
Stattin et al.63 revealed doseresponse buffering protective effects in the presence of accumulated risks. The
authors followed up a large sample of Swedish conscripts
in offcial registers from ages 18 through 36 years. Risk
variables were recorded for the home background (e.g.,
divorce or fathers alcohol problems) and behavioral development (e.g., contact with the police, substance
abuse). Personal resources such as intellectual capacity,
emotional control, and social maturity were measured as
potential buffering protective factors. Approximately one
third of the young men received criminal convictions
during the follow-up period. Both behavioral risks and
home-background risks correlated positively with convictions and also (negatively) with personal resources.
However, among people with high behavioral risk, those
with no pronounced personal resources were convicted
www.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

S17

risk factors or very few


items on the positive side
of the coin.87 Adding
more items on sound direct protective/buffering
protective factors would
probably increase effect
sizes in prediction and
could lead to moredifferentiated patterns of
risk and more guidance
on appropriate prevenRisk fa
Risk
tion and treatment.
ctors
facto
rs
However, such a perspective
requires sensiFigure 2. A model of the relationship of accumulated risk and protective factors, with
behavior problem intensity (left); empirical test results (right)
tive analyses of the dose
response relationship. If
variables have both risk
and direct protective effects, one can essentially arbimuch more often in later life than their peers with many
trarily emphasize the undesirable or desirable influence
resources. Hardly any differences in conviction rates
of variables by accumulating risk versus direct protective
were found between low-risk men with some versus
factors. Doseresponse relationships are also important
many personal resources.
for the detection of buffering protective effects. Studies
A similar doseresponse relationship was found in the
vary widely in the degree of risk at which the moderating
Bielefeld Resilience Study that investigated youth from
or buffering effect of protective variables is studied. For
high-risk family backgrounds who lived in residential
example, Kandel and colleagues25 defned risk by only
care institutions.51,59 One group had developed stable
one factor (parental criminality); Werner and Smith64
behavior problems and others turned out to be resilient
used at least four factors; and Lsel and Bliesener59 used a
(e.g., no aggression). The results showed a doseresponse
risk index that contained 70 items.
relationship not only between risk factors and behavioral
The more risk factors enter into a multiple prediction,
outcome but also in the impact of buffering protective
the less variance is left over for protective effects of the
factors. Figure 2 displays a three-dimensional theoretic
desirable poles of variables. Therefore, researchers can
model of such relationships (left side) and the empirical
decide whether they want to give more emphasis to the
fndings of the study that used aggregated risk factors
risk or protective side of their investigation. Studying
and protective factors as predictors and the intensity of
11
effects of single variables at a very high risk level will
the problem behavior as outcome.
clearly decrease the possibility of detecting a buffering
Overall, the data ftted relatively well the model of cumuprotective effect. On the other hand, fnding a buffering
lative effects of both risk and buffering protective factors.
protective factor in children with many accumulated risk
However, as the right edge of the empirical Figure 2 shows,
factors would be particularly important because it would
buffering protective factors also seem to have a positive
reveal the potential of desirable development even in the
effect at a relatively low level of risk. This difference from the
most diffcult cases.
results of Stattin and colleagues63 may be attributable to the
54
There are also theoretic problems of the doseresponse
overall higher risk of the sample. Loeber et al. developed a
relationship. Although accumulations of both risk and
more comprehensive three-dimensional model of the reladirect protective/buffering protective factors may entionship between risk and direct protective/buffering prohance predictive validity, heterogeneous indices make
tective factors and outcomes. It also addresses the timing of
the meaning of the respective constructs unclear. As a
variables but is more diffcult to test empirically.
consequence, causal inferences become more diffcult to
As a rule of thumb, one may generalize the fndings of
171
:
the
higher
the
number
of
risk
factors
draw than in studies that only address single variables or
Sameroff et al.
the worse the behavioral outcome and, conversely, the
homogeneous constructs.
higher the number of protective factors the better the
Obviously the doseresponse influence of protective
outcome. This is a plausible assumption that has imporfactors is a key issue. Although research on this topic is
tant consequences for practice. Most structured riskchallenged by the above-mentioned and other problems,
assessment instruments for youth violence contain only
the empirical papers1518 in this supplement show a way
Behavior
problems

August 2012

Behavior
problems

S18

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

forward. On the one hand, they carefully disentangle the


independent direct protective effects of single variables.
On the other hand, they estimate the overall direct protective impact in multivariate analyses. Doing this in
multiple comparative studies will inform how relevant
dosages of direct protective patterns compensate for key
risk factors. As a practical consequence, prevention
should not focus too much on isolated single variables but
should target multiple desirable factors whenever possible. Further research needs to address theoretically more
explicit and homogeneous concepts of positive dosages
and contain more person-oriented analyses to complement the variable-oriented studies.

Table 1. Variables that are strong candidates for having


protective effects against youth violence
Individual factors
Above-average intelligence
Positive attitudes toward family and school
Nonaggression-prone social cognitions and beliefs
Low impulsivity and an easy temperament
Low ADHD
Enhanced anxiety and shyness
High heart rate
High MAO-A activity

Discussion
The frst part of this article provided a conceptual framework for the research on direct protective and buffering
protective factors against the onset, persistence, or aggravation of youth violence. It showed that there are conceptual and methodologic diffculties that make work on this
topic more complicated than the prevailing risk-oriented
research. In principle, direct and buffering protective factors belong to the same pool of variables as risk factors.
However, they have effects at one of their poles that not
only reflect the absence of risk but also actively increase
the probability of a desirable development (no onset,
continuation, or aggravation of violence). Direct protective factors require the analysis of linear and nonlinear
main effects on the behavioral outcome. Buffering protective factors are moderators that reduce or nullify the
negative impact of risk factors on the outcome and therefore often require the analysis of interaction effects. The
introduction to the four empirical studies in this supplement describes the methods for the analysis of direct
protective effects in more detail.26
The second part of this article contained a brief narrative
review of research that demonstrated the necessity of the
systematic work in the four studies. The current discussion
revealed a number of factors that are candidates for having
direct protective/buffering protective effects against the onset, continuation, or aggravation of youth violence. The
most-promising variables are summarized in Table 1.
Many of these variables have risk effects at their undesirable poles, but these were not the topic of this article.
Some factors seem to have both direct protective effects
and buffering effects; however, there are too few studies
that directly analyze both types of effects using the same
sample, variables, and defnitions of the outcome. This is
also the case for differential effects on the onset of or
desistance from violence. Some variables may have opposite effects in different phases or patterns of development.
Anxiousneurotic personality characteristics and social

Family factors
Close relationship to at least one parent
Intensive parental supervision
Parental disapproval of aggressive behavior
Low physical punishment
Intensive involvement in family activities
Above-average SES of the family
Family models of constructive coping
Positive parental attitudes toward the childs education
School factors
Good school achievement
Bonding to school
Strong work motivation
Reaching higher education
Support and supervision by teachers
Clear classroom rules
Positive school climate
Peer factors
Nondeviant good friends
Peer groups who disapprove of aggression
Involvement in religious groups
Being socially isolated
Neighborhood factors
Nondeprived neighborhood
Nonviolent neighborhood
Cohesion and informal social control
ADHD, attention-deficit hyperactivity disorder; MAO-A, monoamine
oxidase A

isolation, for example, seem to protect against the early


onset of violence but may lead to a later onset or hinder
desistance in already violent youth.
www.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

Some direct protective or buffering protective effects


seem to be limited to specifc subgroups and may have no
desirable effect or even a risk effect for others (e.g., advantaged neighborhoods). Although such complex patterns
and double effects of variables are not uncommon in
research on protective factors, there is encouraging consistency in the fndings from different samples, study
methods, and outcomes. It needs to be noted that the
reviewed fndings focus mainly on men/boys. As with risk
factors, the patterns of protective factors in women/girls
seem to be partially different60,125,172174 (e.g., greater
relevance of relationship issues).
This article suggests a doseresponse relationship,
similar to that found for risk factors, between the number
of direct protective/buffering protective factors and
youth violence. Accumulated protective factors have a
much stronger effect on nonviolence or other forms of
antisocial behavior than do single factors. Such dose
response relationships are very important for prediction
and intervention. More direct protective/buffering protective factors should be included in structured riskassessment instruments,87,175 and interventions should
address multimodal patterns of these positive factors in
natural development.7,8
Although there is promising convergence of fndings
in this article, one must be aware of its limitations. Especially important are variations in construct defnitions,
measurement standardization, developmental differences in measurement of protective factors and violence,
gender and race differences, and differences in results
generated from high-risk or general population samples.
As a consequence, conceptually and methodologically
homogeneous studies are needed on the specifc protective effects of variables with regard to youth violence (as
represented in this supplement).
Also needed are RCTs, high-quality quasi-experimental designs and within-individual analyses39 to differentiate between protective effects as correlations versus truly causal
effects. This would require more integration of the
mainly prospective longitudinal research on risk and protective factors and the experimental research on prevention and treatment of youth violence. Similarities in results obtained for dynamic (changeable) protective
factors should lead to more-systematic comparisons of
observational and intervention data. Although Farrington et al.176 proposed more combined prospective
longitudinal and experimental research designs in criminology, only a few studies followed this recommendation.41 Accordingly, most meta-analyses on developmental prevention of violence reflect the separation of the two
areas of research.7,177 From a practical perspective, it also
would be important to integrate protective factors more
strongly in the design of intervention programs.
August 2012

S19
1518

This article and the four empirical studies


show
that a great deal can be learned from prospective longitudinal research about protective factors in the development of youth violence. These concepts are now better
understood than they were before. The time is ripe to
mount a new program of research on these topics, which
should advance knowledge about not only the explanation and prediction of youth violence but also its prevention and treatment.
The present analyses were supported by the CDC. The content
is solely the responsibility of the authors and does not necessarily represent the offcial views of the funding agencies.
Publication of this article was supported by the Centers for
Disease Control and Prevention through a Cooperative Agreement with the Association for Prevention Teaching and Research award # CIP-08-001.
No fnancial disclosures were reported by the authors of this
paper.

References
1. Lahey BB, Miller TL, Gordon RA, Riley AW. Developmental epidemiology of the disruptive disorders. In: Quay HC, Hogan AE, eds.
Handbook of disruptive behavior disorders. New York NY: Kluwer
Academic/Plenum, 1999:23 48.
2. Farrington DP. Antisocial personality from childhood to adulthood.
Psychologist 1991;4:389 94.
3. Robins LN, Price RK. Adult disorders predicted by childhood conduct problems: results from the NIMH epidemiologic catchment area
project. Psychiatry 1991;54:116 32.
4. Bender D, Lsel F. Working with violent children in German youth
services: results of a survey. In: Hagell A, Dent RJ, eds. Children who
commit acts of serious interpersonal violence. London, UK: Jessica
Kingsley, 2006:167 85.
5. Cohen M, Piquero A. New evidence on the monetary value of saving
a high risk youth. J Quant Criminol 2009;25(1):25 49.
6. Muntz R, Hutchings J, Edwards RT, Hounsome B, OCilleachair A.
Economic evaluation of treatments for children with severe behavioural problems. J Ment Health Policy Econ 2004;7(4):177 89.
7. Farrington DP, Welsh B. Saving children from a life of crime: early
risk factors and effective interventions. Oxford: Oxford University
Press, 2007.
8. Lsel F, Beelmann A. Effects of child skills training in preventing
antisocial behavior: a systematic review of randomized evaluations.
Ann Am Acad Polit Soc Sci 2003;587(1):84 109.
9. Loeber R, Farrington DP, eds. Serious & violent juvenile offenders:
risk factors and successful interventions. Thousand Oaks CA: Sage,
1998.
10. Loeber R, Stouthamer-Loeber M. Development of juvenile aggression
and violence. Some common misconceptions and controversies. Am
Psychol 1998;53(2):24259.
11. Lsel F, Bender D. Protective factors and resilience. In: Farrington
DP, Coid JW, eds. Early prevention of adult antisocial behaviour.
Cambridge, UK: Cambridge University Press, 2003:130 204.
12. Mofftt TE, Caspi A, Harrington H, Milne BJ. Males on the lifecourse-persistent and adolescence-limited antisocial pathways:
follow-up at age 26 years. Dev Psychopathol 2002;14(1):179 207.

S20

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

13. Sampson RJ, Laub JH. Life-course desisters? Trajectories of crime


among delinquent boys followed to age 70. Criminology 2003;
41(3):55592.
14. Snyder CR, Lopez SJ. Handbook of positive psychology. Oxford:
Oxford University Press, 2005.
15. Pardini DA, Loeber R, Farrington DP, Stouthamer-Loeber M. Identifying direct protective factors for nonviolence. Am J Prev Med
2012;43(2S1):S28 S40.
16. Herrenkohl TI, Lee J, Hawkins JD. Risk versus direct protective factors and youth violence: Seattle Social Development Project. Am J
Prev Med 2012;43(2S1):S41S56.
17. Bernat D, Oakes JM, Pettingell SL, Resnick M. Risk and direct protective factors for youth violence: results from the National Longitudinal
Study of Adolescent Health. Am J Prev Med 2012;43(2S1):S57S66.
18. Henry DB, Tolan PH, Gorman-Smith D, Schoeny ME. Risk and direct
protective factors for youth violence: results from the Centers for
Disease Control and Preventions Multisite Violence Prevention
Project. Am J Prev Med 2012;43(2S1):S67S75.
19. Luthar SS, Cicchetti D, Becker B. The construct of resilience: a critical
evaluation and guidelines for future work. Child Dev 2000;71:543 62.
20. Rutter M. Resilience as a dynamic concept. Dev Psychopathol 2012;
24:335 44.
21. Cicchetti D. Resilience under conditions of extreme stress: a multilevel perspective. World Psychiatry 2010;9:14554.
22. Loeber R, Farrington DP, Stouthamer-Loeber M, White HR. Violence
and serious theft: development and prediction from childhood to
adulthood. New York NY: Routledge, 2008.
23. Werner EE, Smith RS. Journeys from childhood to midlife: risk,
resilience, and recovery. Ithaca NY: Cornell University Press, 2001.
24. Mofftt TE. The neuropsychology of conduct disorder. Dev Psychopathol 1993;5(12):13551.
25. Kandel E, Mednick SA, Kirkegaard-Sorensen L, et al. IQ as a protective factor for subjects at high risk for antisocial behavior. J Consult
Clin Psychol 1988;56:224 6.
26. Loeber R, Farrington DP. Advancing knowledge about direct protective factors that may reduce youth violence. Am J Prev Med
2012;43(2S1):S24 S27.
27. Cowen EL, Wyman PA, Work WC, Parker GR. The Rochester child
resilience project: overview and summary of frst year fndings. Dev
Psychopathol 1990;2(2):193212.
28. Stouthamer-Loeber M, Loeber R, Farrington DP, Zhang Q, Van Kammen W, Maguin E. The double edge of protective and risk factors for
delinquency: Interrelations and developmental patterns. Dev Psychopathol 1993;5(4):683701.
29. StouthamerLoeber M, Wei E, Loeber R, Masten A. Desistance from
persistent serious delinquency in the transition to adulthood. Dev
Psychopathol 2004;16(4):897918.
30. Lipsey MW, Derzon JH. Predictors of violent or serious delinquency
in adolescence and early adulthood: a synthesis of longitudinal research. In: Loeber R, Farrington DP, eds. Serious & violent juvenile
offenders: risk factors and successful interventions. Thousand Oaks
CA: Sage, 1998:86 105.
31. Cicchetti D, Rogosch FA, Lynch M, Holt KD. Resilience in maltreated
childrenprocesses leading to adaptive outcome. Dev Psychopathol
1993;5(4):629 47.
32. Shadish WR, Cook TD, Campbell DT. Experimental and quasiexperimental designs for generalized causal inference. Boston MA:
Houghton Mifflin, 2002.
33. Wikstrm P-OH. In search of causes and explanations of crime. In:
King RD, Wincup E, eds. Doing research on crime and justice. Oxford: Oxford University Press, 2007:11739.
34. Hawkins JD, Herrenkohl T, Farrington DP, Brewer D, Catalano RF,
Harachi TW. A review of predictors of youth violence. In: Loeber R,
Farrington DP, eds. Serious & violent juvenile offenders: risk factors
and successful interventions. Thousand Oaks CA: Sage, 1998:106 46.

35. Kraemer HC, Lowe KK, Kupfer DJ. To your health: how to understand what research tells us about risk. New York NY: Oxford University Press, 2005.
36. Bunge M. Causality. Cambridge MA: Harvard University Press, 1959.
37. Lsel F. Doing evaluation in criminology: balancing scientifc and
practical demands. In: RD King RD, Wincup E, eds. Doing research
on crime and justice. 2nd ed. Oxford, UK: Oxford University Press,
2007:14170.
38. Murray J, Farrington DP, Eisner MP. Drawing conclusions about
causes from systematic reviews of risk factors: the Cambridge Quality
Checklists. J Exp Criminol 2009;5(1):123.
39. Farrington DP. Studying changes within individuals: the causes of
offending. In: Rutter M, ed. Studies of psychosocial risk: the power of
longitudinal data. Cambridge, UK: Cambridge University Press,
1988:158 83.
40. Farrington DP. Criminological psychology in the twenty frst century.
Crim Behav Ment Health 2004;14(3):152 66.
41. Farrington DP, Welsh BC. A half-century of randomized experiments
on crime and justice. In: Tonry MH, ed. Crime and justice: a review of
research. Vol. 34. Chicago IL.: University of Chicago Press, 2006:
55132.
42. McCord J. Cures that harm: unanticipated outcomes of crime prevention programs. Ann Am Acad Polit Soc Sci 2003;587:16 30.
43. Garmezy N. Stress resistant children: the search for protective factors.
In: Stevenson JE, ed. Recent research in developmental psychopathology. Journal of Child Psychology and Psychiatry, Book Supplement
(vol. 4). Oxford, UK: Pergamon Press, 1985:21333.
44. Farrington DP. Protective factors in the development of juvenile
delinquency and adult crime. In: 6th Scientifc Meeting of the Society
for Research in Child and Adolescent Psychopathology. London,
England; 1994.
45. Luthar SS, Zigler E. Vulnerability and competencea review of research on resilience in childhood. Am J Orthopsychiatry 1991;
61(1):6 22.
46. Werner EE, Smith RS. Overcoming the odds: high risk children from
birth to adulthood. Ithaca: Cornell University Press, 1992.
47. Windle RC, Windle M. Longitudinal patterns of physical aggression:
associations with adult social, psychiatric, and personality functioning and testosterone levels. Dev Psychopathol 1995;7(3):563 85.
48. Zara G, Farrington DP. Childhood and adolescent predictors of late
onset criminal careers. J Youth Adolesc 2009;38:287300.
49. White A, Heilbrun K. The classifcation of overcontrolled hostility:
comparison of two diagnostic methods. Crim Behav Ment Health
1995;5:106 23.
50. Loeber R, Pardini D, Homish DL, et al. The prediction of violence and
homicide in young men. J Consult Clin Psychol 2005;73:1074 88.
51. Lsel F, Bliesener T. Resilience in adolescence: a study on the generalizability of protective factors. In: Hurrelmann K, Lsel F, eds. Health
hazards in adolescence. Berlin: W. de Gruyter, 1990:299 320.
52. Achenbach TM, McConaughy SH, Howell CT. Child/adolescent behavioral and emotional problems: implications of cross-informant
correlations for situational specifcity. Psychol Bull 1987;101(2):
21332.
53. Loeber R, Keenan K, Zhang Q. Boys experimentation and persistence
in developmental pathways toward serious delinquency. J Child Fam
Studies 1997;6(3):32157.
54. Loeber T, Slot W, Stouthamer-Loeber M. A three-dimensional cumulative developmental model of serious delinquency. In: Wikstrom
P-OH, Sampson RJ, eds. The explanation of crime: context, mechanisms and development. Leiden: Cambridge University Press,
2006:15394.
55. Loeber R, Pardini DA, Stouthamer-Loeber M, Raine A. Do cognitive,
physiological, and psychosocial risk and promotive factors predict
desistance from delinquency in males? Dev Psychopathol 2007;
19:867 87.

www.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23


56. Schmucker M, Lsel, F. Meta-analysis as a method of systematic
reviews. In: Gadd D, Karstedt S, Messner SF, eds. The Sage handbook
of criminological research methods. Thousand Oaks CA: Sage,
2011:425 43.
57. Lynam D, Mofftt T, Stouthamer-Loeber M. Explaining the relation
between IQ and delinquency class, race, test motivation, school
failure, or self-control. J Abnorm Psychol 1993;102(2):18796.
58. Ward DA, Tittle CR. IQ and delinquency: a test of two competing
explanations. J Quant Criminol 1994;10(3):189 212.
59. Lsel F, Bliesener T. Some high-risk adolescents do not develop
conduct problems: a study of protective factors. Int J Behav Dev
1994;17(4):75377.
60. McCord J, Ensminger ME. Multiple risks and comorbidity in an
African American population. Crim Behav Ment Health 1997;
7(4):339 52.
61. Radke-Yarrow M, Brown E. Resilience and vulnerability in children
of multiple-risk families. Dev Psychopathol 1993;5(4):58192.
62. White JL, Mofftt TE, Silva PA. A prospective replication of the
protective effects of IQ in subjects at high-risk for juveniledelinquency. J Consult Clin Psychol 1989;57(6):719 24.
63. Stattin H, Romelsjo A, Stenbacka M. Personal resources as modifers
of the risk for future criminalityan analysis of protective factors in
relation to 18-year-old boys. Br J Criminol 1997;37(2):198 223.
64. Werner EE, Smith RS. Vulnerable, but invincible: a longitudinal study
of resilient children and youth. New York NY: McGraw-Hill, 1982.
65. Elder G, Caspi A, Van Nguyen T. Resourceful and vulnerable children: family influence in hard times. In: Silbereisen RK, Eyferth K,
Rudinger G, eds. Development as action in context. Berlin: Springer,
1986:167 86.
66. Raine A. The psychopathology of crime: criminal behavior as a clinical disorder. San Diego CA: Academic Press, 1993.
67. Bender D, Bliesener T, Lsel F. Deviance or resilience? A longitudinal
study of adolescents in residential care. In: Davies G, Lloyd-Bostock S,
McMurran M, Wilson C, eds. Psychology, law, and criminal justice:
international developments in research and practice. Berlin: W. de
Gruyter, 1996:409 23.
68. Rutter M, Quinton D. Long-term follow-up of women institutionalized in childhoodfactors promoting good functioning in adult life.
Br J Dev Psychol 1984;2:191204.
69. Werner EE. Antecedents and consequences of deviant behavior. In:
Hurrelmann K, Lsel F, eds. Health hazards in adolescence. Berlin: W.
de Gruyter, 1990:219 31.
70. Schmitt-Rodermund E, Silbereisen RK. Well-adapted adolescent ethnic German immigrants in spite of adversity: the protective effects of
human, social, and fnancial capital. Eur J Dev Psychol 2008;
5(2):186 209.
71. Bandura A. Self-effcacy: the exercise of control. New York NY:
Worth Publishers, 1997.
72. Crick NR, Dodge KA. A review and reformulation of social information-processing mechanisms in childrens social adjustment. Psychol
Bull 1994;115(1):74 101.
73. Loeber R, Farrington DP, Stouthamer-Loeber M, Raskin-White H.
Violence and serious theft: development and prediction from childhood to adulthood. New YorkNY: Routledge, 2008.
74. Mofftt TE, Caspi A, Dickson N, Silva P, Stanton W. Childhood-onset
versus adolescent-onset antisocial conduct problems in males: natural
history from ages 3 to 18 years. Dev Psychopathol 1996;8(2):399 424.
75. Wikstrm P-OH. Deterrence and deterrence experiences. Preventing
crime through the threat of punishment. In: Shoham SG, Beck O, Kett
M, eds. International handbook of penology and criminal justice.
Boca Raton FL: CRC, Taylor & Francis, 2007:34578.
76. Lsel F, Bliesener T. Aggression and delinquency in adolescents:
studies on cognitive and social origins. Neuwied, Germany: Luchterhand, 2003.

August 2012

S21

77. McMurran M, McGuire J, eds. Social problem solving and offending:


evidence, evaluation, and evolution. Chichester, England: Wiley,
2005.
78. Cicchetti D, Rogosch FA. The role of self-organization in the promotion of resilience in maltreated children. Dev Psychopathol
1997;9(4):797 815.
79. Hetherington EM. Coping with family transitionswinners, losers,
and survivors. Child Dev 1989;60(1):114.
80. Wyman PA, Cowen EL, Work WC, Kerley JH.The role of childrens
future expectations in self-system functioning and adjustment to life
stress: a prospective study of urban at-risk children. Dev Psychopathol 1993;5:649 61.
81. Donnellan MB, Trzesniewski KH, Robins RW, Mofftt TE, Caspi A.
Low self-esteem is related to aggression, antisocial behavior, and
delinquency. Psychol Sci 2005;16(4):328 35.
82. Jang SJ, Thornberry TP. Self-esteem, delinquent peers, and delinquency: a test of the self-enhancement thesis. Am Sociol Rev
1998;63(4):586 98.
83. Baumeister RF, Smart L, Boden JM. Relation of threatened egotism to
violence and aggression: the dark side of high self-esteem. Psychol Rev
1996;103(1):533.
84. Antonovsky A. Unraveling the mystery of health: how people manage
stress and stay well. San Francisco CA: Jossey-Bass, 1987.
85. Herrenkohl TI, Tajima EA, Whitney SD, Huang B. Protection against
antisocial behavior in children exposed to physically abusive discipline. J Adolesc Health 2005;36(6):457 65.
86. Regnerus MD, Elder GH. Religion and vulnerability among low-risk
adolescents. Soc Sci Res 2003;32(4):63358.
87. Lsel F, Bender D. Risk factors for serious juvenile violence. In: Hagell
A, Jeyarajah-Dent R, eds. Children who commit acts of serious interpersonal violence: messages for best practice. London, UK: Jessica
Kingsley, 2006:4272.
88. Thomas A, Chess S. Temperament and development. New York NY:
Brunner/Mazel, 1977.
89. Block J. Lives through time. Berkeley CA: Bancroft Books, 1971.
90. Robins RW, John OP, Caspi A, Mofftt TE, Stouthamer-Loeber M.
Resilient, overcontrolled, and undercontrolled boys: three replicable
personality types. J Pers Soc Psychol 1996;70:15771.
91. Schwartz CE, Snidman N, Kagan J. Early childhood temperament as a
determinant of externalizing behavior in adolescence. Dev Psychopathol 1996;8(3):52737.
92. Caspi A, McClay J, Mofftt TE, et al. Role of genotype in the cycle of
violence in maltreated children. Science 2002;297:851 4.
93. Foley DL, Eaves LJ, Wormley B, et al. Childhood adversity, monoamine oxidase a genotype, and risk for conduct disorder. Arch Gen
Psychiatry 2004;61:738 44.
94. Widom CS, Brzustowicz LM. MAOA and the cycle of violence:
childhood abuse and neglect, MAOA genotype, and risk for violent
and antisocial behavior. Biol Psychiatry 2006;60(7):684 9.
95. Kim-Cohen J, Caspi A, Taylor A, et al. MAOA, maltreatment, and
gene-environment interaction predicting childrens mental health:
new evidence and a meta-analysis. Mol Psychiatry 2006;11:90313.
96. Alia-Klein N, Goldstein RZ, Kriplani A, et al. Brain monoamine
oxidase A activity predicts trait aggression. J Neurosci 2008;
28(19):5099 104.
97. Fowler JS, Alia-Klein N, Kriplani A, et al. Evidence that brain MAO A
activity does not correspond to MAOA genotype in healthy male
subjects. Biol Psychiatry 2007;62:355 8.
98. Ortiz J, Raine A. Heart rate level and antisocial behavior in children
and adolescents: a meta-analysis. J Am Acad Child Adolesc Psychiatry
2004;43(2):154 62.
99. Kindlon DJ, Tremblay RE, Mezzacappa E, Earls F, Laurent D, Schaal
B. Longitudinal patterns of heart rate and fghting behavior in 9through 12-year-old boys. J Am Acad Child Adolesc Psychiatry
1995;34:3717.

S22

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23

100. Lsel F, Bender D. Heart rate and psychosocial correlates of antisocial


behavior in high-risk adolescents. In: Raine A, Brennan PA, Farrington DP, Mednick SA, eds. Biosocial bases of violence. New York
NY: Plenum, 1997:321 4.
101. Raine A, Venables PH. Classical-conditioning and socializationa
biosocial interaction. Pers Individual Diff 1981;2(4):273 83.
102. Raine A, Reynolds C, Venables PH, Mednick SA. Biosocial bases of
aggressive behavior in childhood: resting heart rate, skin conductance
orienting, and physique. In: Raine A, Brennan PA, Farrington DP,
Mednick SA, eds. Biosocial bases of violence. New York NY: Plenum,
1997:10726.
103. Kagan J. Galens prophecy: temperament in human nature. New York
NY: Basic Books, 1994.
104. Gottman JM, Katz LF. Effects of marital discord on young childrens
peer interaction and health. Dev Psychol 1989;25(3):373 81.
105. Wadsworth MEJ. Delinquency, pulse rates and early emotional deprivation. Br J Criminol 1976;16(3):24556.
106. Tremblay RE, Schaal B, Boulerice B, Arsenault L, Soussignan R, Perusse D. Male physical aggression, social dominance, and testorone
levels at puberty: a developmental perspective. In: Raine A, Brennan
PA, Farrington DP, Mednick SA, eds. Biosocial bases of violence. New
York NY: Plenum, 1997:27191.
107. McBurnett K, Lahey BB, Rathouz PJ, Loeber R. Low salivary cortisol
and persistent aggression in boys referred for disruptive behavior.
Arch Gen Psychiatry 2000;57:38 43.
108. Lsel F, Beelmann A, Jaursch S, Stemmler M. Soziale Kompetenz fr
Kinder und Familien: Ergebnisse der Erlangen-Nrnberger Entwicklungs-und Prventionsstudie. Berlin: Federal Ministry for Family Affairs, Seniors, Women and Youth; 2004.
109. Susman EJ, Ponirakis A. Hormones-context interactions and antisocial behavior in youth. In: Raine A, Brennan PA, Farrington DP,
Mednick SA, eds. Biosocial bases of violence. New York NY: Plenum,
1997:251 69.
110. Boyce WT, Frank E, Jensen PS, Kessler RC, Nelson CA, Steinberg L.
Social context in developmental psychopathology: recommendations
for future research from the MacArthur Network on Psychopathology and Development. Dev Psychopathol 1998;10(2):143 64.
111. Plomin R. Genetics and experience: the interplay between nature and
nurture. Thousand Oaks CA: Sage, 1994.
112. Cummings EM, Davies P. Emotional security as a regulatory process
in normal development and the development of psychopathology.
Dev Psychopathol 1996;8(1):12339.
113. Farber EA, Egeland B. Invulnerability among abused and neglected
children. In: Anthony EJ, Cohler BJ, eds. The invulnerable child. New
York NY: Guilford, 1987:253 88.
114. Greenberg MT, Speltz ML, Deklyen M. The role of attachment in the
early development of disruptive behavior problems. Dev Psychopathol 1993;5(12):191213.
115. Osborn AF. Resilient children: a longitudinal study of high achieving
socially disadvantaged children. Early Child Dev Care 1990;62:23 47.
116. Wyman PA, Cowen EL, Work WC, Parker GR. Developmental and
family milieu correlates of resilience in urban children who have
experienced major life stress. Am J Community Psychol 1991;19:
40526.
117. Kolvin I. Continuities of deprivation? The Newcastle 1000 family
study. Aldershot, England: Avebury, Gower, 1990.
118. Kraemer GW. Social attachment, brain function, aggression, and
violence. In: Raine A, Brennan PA, Farrington DP, Mednick SA, eds.
Biosocial bases of violence. New York NY: Plenum, 1997:20729.
119. Sampson RJ, Laub JH. Crime in the making: pathways and turning
points through life. Cambridge MA: Harvard University Press, 1993.
120. Lsel F. Protective effects of social resources in adolescents at high risk
for antisocial behavior. In: Weitekamp EGM, Kerner H-J, eds. Crossnational longitudinal research on human development and criminal
behavior. Dordecht: Kluwer Academic, 1994:283301.

121. Mortimore P. The positive effects of schooling. In: Rutter M, ed.


Psychosocial disturbances in young people: challenges for prevention.
Cambridge: Cambridge University Press, 1995:333 63.
122. Baumrind D. Rearing competent children. In: Child development
today and tomorrow. San Francisco CA: Jossey-Bass, 1989:349 78.
123. Gottfredson DC. Schools and delinquency. Cambridge, UK: Cambridge University Press, 2000.
124. Hoge RD, Andrews DA, Leschied AW. An investigation of risk and
protective factors in a sample of youthful offenders. J Child Psychol
Psychiatry Allied Disciplines 1996;37(4):419 24.
125. Werner EE. Risk, resilience, and recovery: perspectives from the
Kauai longitudinal-study. Dev Psychopathol 1993;5(4):50315.
126. Rutter M. Psychosocial resilience and protective mechanisms. In: Rolf
JE, Garmezy N, eds. Risk and protective factors in the development of
psychopathology. Cambridge, UK: Cambridge University Press,
1990:181214.
127. Hirschi T.Causes of delinquency. Berkeley: University of California
Press, 1969.
128. Caspi A, Mofftt TE. The continuity of maladaptive behavior: from
description to understanding in the study of antisocial behavior. In:
Cicchetti D, Cohen DJ, eds. Developmental psychopathology (vol. 2):
risk, disorder, and adaptation. New York NY: Wiley, 1995:472511.
129. Farrington DP. Later adult life outcomes of offenders and nonoffenders. In: Brambring M, Lsel F, Skowronek H, eds. Children at risk:
assessment, longitudinal research, and intervention. Berlin: de
Gruyter, 1989:220 44.
130. Bergman LR, Magnusson D. A person-oriented approach in research
on developmental psychopathology. Dev Psychopathol 1997;9(2):
291319.
131. Elliott DS. Serious violent offenders onset, developmental course,
and termination. Criminology 1994;32(1):121.
132. Huesmann LR, Eron LD, Yarmel PW. Intellectual functioning and
aggression. J Pers Soc Psychol 1987;52(1):232 40.
133. Tremblay RE, Masse B, Perron D, Leblanc M, Schwartzman AE,
Ledingham JE. Early disruptive behavior, poor school-achievement,
delinquent-behavior, and delinquent personalitylongitudinal analyses. J Consult Clin Psychol 1992;60(1):64 72.
134. Ttof MM, Farrington DP. Effectiveness of school-based programs to
reduce bullying: a systematic and meta-analytic review. J Exp Criminol 2011;7:2756.
135. Ttof MM, Farrington DP, Lsel F, Loeber R. The predictive effciency
of school bullying versus later offending: a systematic/meta-analytic
review of longitudinal studies. Crim Behav Ment Health 2011;21(2):
80 9.
136. Bender D, Lsel F. Bullying at school as predictor of delinquency,
violence and other antisocial behaviour in adulthood. Crim Behav
Ment Health 2011;21(2):99 106.
137. Farrington DP, Ttof MM. Bullying as a predictor of offending, violence and later life outcomes. Crim Behav Ment Health 2011;21(2):
90 8.
138. Elliott DS, Huizinga D, Menard SW. Multiple problem youth: delinquency, substance use, and mental health problems. New York NY:
Springer-Verlag, 1989.
139. Reiss AJ, Farrington DP. Advancing knowledge about co-offending
results from a prospective longitudinal survey of London males. J
Crim Law Criminol 1991;82(2):360 95.
140. Van Mastrigt SB, Farrington DP. Co-offending, age, gender and crime
type: implications for criminal justice policy. Br J Criminol
2009;49(4):55273.
141. Farrington DP, Loeber R, Yin Y, Anderson SJ. Are within-individual
causes of delinquency the same as between-individual causes? Crim
Behav Ment Health 2002;12(1):53 68.
142. Thornberry TP. Membership in youth gangs and involvement in
serious and violent offending. In: Loeber R, Farrington DP, eds.
Serious & violent juvenile offenders: risk factors and successful interventions. Thousand Oaks CA: Sage, 1998:147 66.

www.ajpmonline.org

Lsel and Farrington / Am J Prev Med 2012;43(2S1):S8 S23


143. Bender D, Lsel F. Protective and risk effects of peer relations and
social support on antisocial behaviour in adolescents from multiproblem milieus. J Adolesc 1997;20(6):66178.
144. Jessor R, Donovan JE, Costa FM. Beyond adolescence: problem behavior and young adult development. Cambridge: Cambridge University Press, 1991.
145. Mofftt TE. Adolescence-limited and life-course-persistent antisocial
behavior: a developmental taxonomy. Psychol Rev 1993;100(4):
674 701.
146. Dishion TJ, Patterson GR, Griesler PC. Peer adaptations in the development of antisocial behavior: a confluence model. In: Huesmann LR,
editor. Aggressive behavior: current perspectives. New York NY: Plenum, 1994:6195.
147. Tremblay RE, Msse LC, Vitaro F, Dobkin PL. The impact of friends
deviant behavior on early onset of delinquency: Longitudinal data
from 6 to 13 years of age. Dev Psychopathol 1995;7:649 68.
148. Sprott JB, Jenkins JM, Doob AN. The importance of school. Youth
Violence Juvenile Justice 2005;3(1):59 77.
149. Fergusson DM, Vitaro F, Wanner B, Brendgen M. Protective and
compensatory factors mitigating the influence of deviant friends on
delinquent behaviours during early adolescence. J Adolesc 2007;
30(1):3350.
150. Cairns RB, Cairns BD. Social cognition and social networks: a developmental perspective. In: Pepler DJ, Rubin KH, eds. The development
and treatment of childhood aggression. Hillsdale NJ: Lawrence Erlbaum, 1991:249 78.
151. Patterson GR, Reid JB, Dishion TJ. Antisocial boys. Eugene OR:
Castalia, 1992.
152. Stattin H, Magnusson D. Antisocial development: a holistic approach.
Dev Psychopathol 1996;8(4):617 45.
153. Farrington DP, Gallagher B, Morley L, St Ledger RJ, West DJ. Are
there any successful men from criminogenic backgrounds? Psychiatry
1988;51(2):116 30.
154. Farrington DP, Gallagher B, Morley L, St. Ledger RJ, West DJ. A
24-year follow-up of men from vulnerable backgrounds. In: Jenkins
RL, Brown WK, eds. The abandonment of delinquent behavior: promoting the turnaround. New York: Praeger, 1988:15573.
155. Caspi A, Lynam D, Mofftt TE, Silva PA. Unraveling girls delinquency: Biological, dispositional, and contextual contributions to adolescent misbehavior. Dev Psychol 1993;29(1):19 30.
156. Gorman-Smith D, Tolan P. The role of exposure to community violence and developmental problems among inner-city youth. Dev Psychopathol 1998;10(1):10116.
157. Sampson RJ, Wikstrm P-O. The social order of violence in Chicago
and Stockholm neighborhoods: a comparative study. In: Kalyvas SN,
Shapiro I, Masoud TE, eds. Order, conflict, and violence. Cambridge:
Cambridge University Press, 2008:97119.
158. Seidman E, Yoshikawa H, Roberts A, et al. Structural and experiential
neighborhood contexts, developmental stage, and antisocial behavior
among urban adolescents in poverty. Dev Psychopathol 1998;
10(2):259 81.
159. Wikstrm P-OH, Loeber R. Do disadvantaged neighborhoods cause
well-adjusted children to become adolescent delinquents? A study of
male juvenile serious offending, individual risk and protective factors,
and neighborhood context. Criminology 2000;38(4):1109 42.

August 2012

S23

160. Lynam DR, Caspi A, Mofftt TE, Wikstrom PO, Loeber R, Novak S.
The interaction between impulsivity and neighborhood context on
offending: the effects of impulsivity are stronger in poorer neighborhoods. J Abnorm Psychol 2000;109(4):56374.
161. Kupersmidt JB, Griesler PC, DeRosier ME, Patterson CJ, Davis PW.
Childhood aggression and peer relations in the context of family and
neighborhood factors. Child Dev 1995;66(2):360 75.
162. Lynch M, Cicchetti D. Trauma, mental representation, and the organization of memory for mother-referent material. Dev Psychopathol
1998;10(4):739 59.
163. Richters JE, Martinez PE. Violent communities, family choices, and
childrens chances: an algorithm for improving the odds. Dev Psychopathol 1993;5:609 27.
164. Sampson RJ. Moving to inequality: neighborhood effects and experiments meet social structure. Am J Sociol 2008;114(1):189 231.
165. Jaffee SR, Caspi A, Mofftt TE, Polo-Tomas M, Taylor A. Individual,
family, and neighborhood factors distinguish resilient from nonresilient maltreated children: a cumulative stressors model. Child
Abuse Negl 2007;31(3):23153.
166. Garbarino J, Kostelny K. Child maltreatment as a community problem. Child Abuse Negl 1992;16(4):455 64.
167. Wilson WJ. When work disappears: the world of the new urban poor.
New York NY: Knopf, 1996.
168. Farrington DP. Early prediction of violent and non-violent youthful
offending. Eur J Crim Policy Res 1997;5(2):51 66.
169. Masten AS, Best KM, Garmezy N. Resilience and development: contributions from the study of children who overcome adversity. Dev
Psychopathol 1990;2(4):425 44.
170. Smith C, Lizotte AJ, Thornberry TP, Krohn MD. Resilient youth:
identifying factors that prevent high-risk youth from engaging in
delinquency and drug use. In: Hagan J, ed. Delinquency and disrepute
in the life course. Greenwich CT: JAI Press, 1995:217 47.
171. Sameroff AJ, Bartko WT, Baldwin C, Seifer R. Family and social
influences on the development of child competence. In: Lewis M,
Feiring C, eds. Families, risk, and competence. Mahwah NJ: Lawrence
Erlbaum, 1998:161 85.
172. Crosnoe R, Erickson KG, Dornbusch SM. Protective functions of
family relationships and school factors on the deviant behavior of
adolescent boys and girls. Youth Soc 2002;33(4):515 44.
173. Fagan AA, Van Horn ML, Hawkins JD, Arthur MW. Gender similarities and differences in the association between risk and protective
factors and self-reported serious delinquency. Prev Sci 2007;
8(2):11524.
174. Hart JL, OToole SK, Price-Sharps JL, Shaffer TW. The risk and
protective factors of violent juvenile offending. Youth Violence Juvenile Justice 2007;5(4):367 84.
175. Arthur MW, Hawkins JD, Pollard JA, Catalano RF, Baglioni AJ.
Measuring risk and protective factors for substance use, delinquency,
and other adolescent problem behaviorsthe Communities That
Care youth survey. Eval Rev 2002;26(6):575 601.
176. Farrington DP, Ohlin LE, Wilson JQ.Understanding and controlling crime:
toward a new research strategy. New York NY: Springer-Verlag, 1986.
177. Lsel F, Bender D. Child social skills training in prevention of antisocial development and crime. In: Farrington DP, Welsh BC, eds. Handbook of crime prevention. Oxford: Oxford University Press, 2012:
10229.

You might also like