! ‘QJ

ymgméiklzig'
~vwz‘ 23111;}, '9‘“
E 4' 5

,—

 

 

 

\
e.‘
'-s
1

 

 

”Mitt?!

£1
{70> \
S’iibO-UEL

This is to certify that the
dissertation entitled

Integrating Models of Risk and Protection for the Prevention of
Adolescent Delinquency

presented by
Jennifer L. Juras

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Psychologyi

41.4.; was;

Major‘F‘rofessors Signature
9"- ,13 - 07’
Date

MSU is an Amrmative Action/Equal Opportunity Institution

. "I I" .1- [1'

 

LIBRARY

Michigan State
University

*-

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE ' DATE DUE DATE DUE

 

mwlzzmg

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDue.p65-p.15

 

INTEGRATING MODELS OF RISK AND PROTECTION FOR THE
PREVENTION OF ADOLESCENT DELINQUENCY

By

Jennifer L. Juras

A DISSERTATION

Submitted to
Michigan State University
in partial fiilfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology, College of Social Sciences

Interdepartmental Graduate Specialization
in Ecological/Community Psychology

2004

ABSTRACT

INTEGRATING MODELS OF RISK AND PROTECTION FOR THE PREVENTION
OF ADOLESCENT DELINQUENCY

By

Jennifer L. Juras

A risk and resilience approach has been increasingly utilized to develop
interventions to prevent youth problems such as adolescent delinquency. This approach
to prevention uses ecological theory as a framework for understanding adolescent
development and behavior and adds information about delinquency from research on risk
and protective factors.

This study addressed gaps in research and practice by combining two risk and
resilience models of delinquency prevention - risk reduction and protection enhancement
- in a comprehensive manner. An additional goal was to examine the utility of
considering domains of risk and protection, rather than overall numbers of risk and
protective factors, for delinquency prevention efforts. Finally, gender, urban/rural, and
racial/ethnic differences in risk and protection were explored, as well as whether
individual, family, peer, school, and community protection moderated risk for youth
within these sociodemographic groups.

A survey measuring risk and protective factors in the individual, family, peer,
school, and community domains was administered to 452 students from an urban and a
rural middle school in a medium-sized Midwestern city. Of these, 57% were from the

urban school and 43% were from the rural school, and slightly over half were female.

Within the urban site, 48% were Caucasian and 52% were youth of color, and 92% were
Caucasian in the rural site.

The results revealed differences in levels and types of risk and protection within
sociodemographic groups, and confirmed that information about domains of risk and
protection significantly predicted delinquency beyond the amount explained by overall
levels of risk and protection. The findings also showed a main effect of family and
school protection on delinquency. In addition, school protection had a main effect on
delinquency and family protection buffered risk for youth with the highest levels of risk.
Finally, community protection was found to buffer the effects of risk for youth of color,
and individual protection had an exacerbating effect on risk, particularly for youth from
the urban area.

A number of implications for current practice emerged fi'om this study. One,
prevention efforts must attend to risk, as it is unlikely that protection by itself can
mitigate the effects of risk. Two, type of protection matters for preventing delinquency.
Using a protective factor enhancement model to build any type of protection will not
necessarily be beneficial and may potentially do more harm than good. Three,
consideration of youth sociodemographic characteristics may be useful in delinquency
prevention efforts. Four, information about protective factors youth have may be used to
direct strength-based interventions to prevent delinquency.

Future studies should include a greater number of protective factors and also re-
examine the role of individual protections such as self-esteem and independence. In
addition, future research should utilize longitudinal designs and ecological approaches to

assessing risk and protection for real world views of how they influence youth outcomes.

Cepyright by
JENNIFER L. JURAS
2004

ACKNOWLEDGEMENTS

My deepest appreciation goes to Cris Sullivan, my committee chair, for her time,
talent, and patience in guiding me to completion of this work. Her support made it
possible for me to both begin and finish this research. I would also like to thank my other
committee members: Deb Bybee, for her statistical expertise, support, and assistance;
Bob Caldwell, for his encouragement and help, especially at the earliest stages of this
work; and Tom Luster, for his time and helpful suggestions.

I am also grateful to a number of other people whose assistance contributed to the
completion of this work. Thanks goes to the Youth Violence Prevention Coalition Co-
chairs and members and the Michigan Council on Crime and Delinquency for their
assistance, time, and financial investment in developing and completing this research.
Thank you also to the staff and students at the middle schools that participated in the
YVPC Youth Survey; this research would not have been possible without them,

My deepest gratitude also goes to my other friends and family members who
supported me in the process of completing this project. Thank you to my parents and
grandparents for their love and support. Special thanks also go to: Lorraine McKelvey,
for managing to be there for me even while she was going through the same process;
Dan Cantillon, for nagging me incessantly and for finally reading my draft and evening
up the social capital; and Dave Loveland, for nagging me, harassing me, for always

remembering to bring the tequila, and listening to me whine until I was finally done.

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................... viii
LIST OF FIGURES ....................................................................................................... x
LIST OF APPENDICES ................................................................................................ xi
CHAPTER 1
INTRODUCTION
Overview of the Problem ......................................................................................... 1
Prevalence and Consequences of Adolescent Delinquency ............................... 2
Prevention of Adolescent Delinquency ............................................................. 6
Risk, Resilience, and Protection ........................................................................ 7
Ecological Framework ....................................................................................... 9
Concepts of Prevention ...................................................................................... 11
Risk and Protective Factor Research ....................................................................... 13
Risk Factor Research ......................................................................................... 13
Common or cross-cutting risk factors .......................................................... 14
Cumulative number of risks versus problem-specific risk pathways .......... 15
Problem-specific risk factors ....................................................................... 16
Communities That Care Model .................................................................... 17
Protective Factor Research ................................................................................ 20
Relationship between risk and protection .................................................... 21
Common or cross-cutting protective factors ................................................ 22
Problem-specific protective factors ............................................................. 22
Search Institute’s asset-building model ....................................................... 23
Risk and protection models .......................................................................... 26
Research incorporating both risk and protection ......................................... 27
Summary and conclusions about gaps in the literature ................................ 29
Potential Moderators of Risk, Protection, and Youth Outcomes ....................... 31
Gender .......................................................................................................... 31
Urban/Rural .................................................................................................. 35
Race/Ethnicity .............................................................................................. 37
Current Study ........................................................................................................... 39
Research Questions and Hypotheses ........................................................... 40
CHAPTER 2
METHODS
Development and Administration of Surveys .......................................................... 42
Sites ...................................................................................................................... 43
Participants ............................................................................................................... 43
Independent Variables ............................................................................................. 47
Risk Factors ....................................................................................................... 47

TABLE OF CONTENTS (Cont)

Individual ..................................................................................................... 47
Family .......................................................................................................... 51
Peer .............................................................................................................. 52
School .......................................................................................................... 53
Community .................................................................................................. 54
Protective Factors ............................................................................................... 55
Individual ..................................................................................................... 55
Family .......................................................................................................... 55
Peer .............................................................................................................. 56
School .......................................................................................................... 56
Community .................................................................................................. 57
Dependent Variables ................................................................................................ 58
CHAPTER 3
RESULTS
Procedures of Analysis ............................................................................................ 59
Dependent Variable: Youth Involvement in Delinquency ....................................... 62
Independent Variables: Youth Experience of Risk and Protection ......................... 62
Main effects of specific domains of protection on delinquency ........................ 73
Findings .................................................................................................................... 75
Research Question 1 .......................................................................................... 75
Research Question 2 .......................................................................................... 82
Research Question 3 .......................................................................................... 83
Research Question 4 .......................................................................................... 85
Research Question 5 .......................................................................................... 97
Final Analyses Examining Students with the Highest Risk .................................... 115
CHAPTER 4
DISCUSSION
Domain-Specific versus Cumulative Approach to Risk and Protection .................. 119
Demographic Differences in Risk and Protection ................................................... 122
Effects of Domains of Protection for Students with the Highest Risk ..... ’. .............. 128
Implications for Delinquency Prevention ................................................................ 129 '
Limitations of the Study ........................................................................................... 131
Strengths of the Study .............................................................................................. 133
Conclusions and Next Steps in Research ................................................................. 134
REFERENCES .............................................................................................................. 165

vii

Table 1.
Table 2.
Table 3.
Table 4.
Table 5.
Table 6.
Table 7.
Table 8.
Table 9.
Table 10.
Table 11.
Table 12.
Table 13.
Table 14.
Table 15.
Table 16.
Table 17.
Table 18.
Table 19.
Table 20.
Table 21.

Table 22.
Table 23.
Table 24.
Table 25.
Table 26.
Table 27.
Table 28.
Table 29.
Table 30.
Table 31.

Table 32.
Table 33.

LIST OF TABLES

Survey Participant Demographics .................................................................. 44
Comparison of Survey Participants in Urban and Rural Schools ................... 46
Risk and Protective Factors Measured by the YVPC Youth Survey .............. 48
Risk Factor Psychometrics .............................................................................. 49
Protective Factor Psychometrics ..................................................................... 50
Risk Factor Descriptive Statistics ................................................................... 64
Protective Factor Descriptive Statistics .......................................................... 65
Individual Risk and Protection Scale Correlations with Delinquency ........... 67
Family Risk and Protection Scale Correlations with Delinquency ................. 68
Peer Risk and Protection Scale Correlations with Delinquency .................... 69
School Risk and Protection Scale Correlations with Delinquency ............... 70
Community Risk and Protection Scale Correlations with Delinquency ....... 71
Risk and Protective Factor Domain Correlations ......................................... 72
Main Effects of Protection on Delinquency ................................................. 74
Risk Domains by Gender .............................................................................. 76
Protection Domains by Gender ..................................................................... 77
Risk by Urban/Rural ..................................................................................... 78
Protection by Urban/Rural ............................................................................ 79
Risk by Race/Ethnicity ................................................................................. 80
Protection by Race/Ethnicity ........................................................................ 81
Comparison of Ability of Overall Risk and Protection versus Domains of Risk
and Protection to Predict Delinquency ........................................................ 84
Do Individual, Family, Peer, School, and Community Protection Buffer the
Effect of Overall Risk on Delinquency? ...................................................... 86
Do Individual, Family, Peer, School, and Community Protection Buffer the
Effect of Individual Risk on Delinquency? ................................................. 88
Do Individual, Family, Peer, School, and Community Protection Buffer the
Effect of F arnily Risk on Delinquency? ....................................................... 91
Do Individual, Family, Peer, School, and Community Protection Buffer the
Effect of Peer Risk on Delinquency? ........................................................... 93
Do Individual, Family, Peer, School, and Community Protection Buffer the
Effect of School Risk on Delinquency? ....................................................... 94
Do Individual, Family, Peer, School, and Community Protection Buffer the
Effect of Community Risk on Delinquency? ............................................... 96
Gender by Individual Protection by Overall Risk ........................................ 99
Gender by Family Protection by Overall Risk .............................................. 100
Gender by Peer Protection by Overall Risk .................................................. 101
Gender by School Protection by Overall Risk .............................................. 102
Gender by Community Protection by Overall Risk ...................................... 103
Urban/Rural by Individual Protection by Overall Risk ................................ 104

viii

Table 34.
Table 35.
Table 36.
Table 37.
Table 38.
Table 39.
Table 40.
Table 41.
Table 42.
Table 43.

LIST OF TABLES (Cont)

Urban/Rural by Family Protection by Overall Risk ..................................... 105
Urban/Rural by Peer Protection by Overall Risk .......................................... 106
Urban/Rural by School Protection by Overall Risk ...................................... 107
Urban/Rural by Community Protection by Overall Risk .............................. 108
Race/Ethnicity by Individual Protection by Overall Risk ............................ 110
Race/Ethnicity by Family Protection by Overall Risk .................................. 111
Race/Ethnicity by Peer Protection by Overall Risk ...................................... 112
Race/Ethnicity by School Protection by Overall Risk .................................. 113
Race/Ethnicity by Community Protection by Overall Risk .......................... 114
Do Domains of Protection Buffer the Effect of Risk for Students with the

Highest Risk? ............................................................................................... 117

ix

LIST or FIGURES

Figure 1. Interaction of overall risk and individual protection in predicting
delinquency .................................................................................................................... 87

Figure 2. Interaction of individual risk and individual protection in predicting
delinquency .................................................................................................................... 90

Figure 3. Interaction of family risk and individual protection in predicting
delinquency .................................................................................................................... 92

Figure 4. Three-way interaction of individual protection, overall risk, and urban/rural in
predicting delinquency ................................................................................................... 109

Figure 5. Three-way interaction of community protection, overall risk, and race/ethnicity
in predicting delinquency ............................................................................................... 115

Figure 6. Interaction of family protection and overall risk in predicting delinquency for
youth with the highest risk ............................................................................................. 118

LIST OF APPENDICES

Appendix A. YVPC Youth Survey ............................................................................... 137

Appendix B. Family-Wise Bonferroni Corrections ...................................................... 155

xi

 

 

 

lhi

del

Chapter 1
INTRODUCTION ‘
Overview

Concerns about adolescent delinquency have created a demand for effective
strategies to prevent youth problem behaviors (Fraser, 1997). Prevention approaches
have long focused on identifying and reducing risk factors for adolescent problem
behaviors. More recently the prevention field has expanded to examine why some
adolescents whose exposure to risk might presage negative outcomes do not experience
serious problems. A number of researchers have suggested that both risk and protective
factors are promising targets for preventive intervention based on a risk and resilience
fiamework (Coie, Watt, West, Hawkins, Asarnow, Markman, Ramey, Shure, & Long,
1993; Fraser & Galinsky, 1997; Hawkins, Arthur, & Catalano, 1995; Luthar & Zigler,
1991; Mrazek & Haggerty, 1994; Wasserman & Miller, 1997). This perspective suggests
that preventive interventions focus on both risk reduction and protective factor
enhancement to prevent future crime, violence, and other problems.

This literature review examines research relevant to preventing adolescent
delinquency. The first section provides information about its prevalence and also
describes some of the major issues affecting public policy and research. Second,
concepts and concerns related to the prevention of this adolescent problem are identified.

In particular, the concepts of risk, resilience, and protection, models of prevention, and
the ecological framework underlying successful prevention efforts are delineated. The
third section provides a summary of what is known about risk factors for adolescent

delinquency and also describes a popular approach to risk reduction, the Communities

 

(f:

v:

That Care Model, that is widely used in schools and communities. Fourth, existing
knowledge about the factors that protect youth either by directly affecting negative
outcomes or by moderating the relationship among risk factors and problems is
considered. This section includes a description of the Search Institute Model, an
increasingly utilized approach to enhance protective factors. Finally, the review
highlights gaps in the literature, as well as current practice, and introduces a study that
addressed some of those gaps.
Prevalence and Consequences of Adolescent Delinquency

Policy makers, researchers, and the media have increasingly identified adolescent
delinquency, particularly violence, as an urgent problem in the United States throughout
the 19903 and early21st century (Glassner, 1999). The term delinquency suggests a wide
range of behaviors, from acting out or misbehaving during early childhood to violent and
destructive illegal behaviors. Violence, substance use and abuse, stealing, vandalism, and
truancy are some of the problem behaviors that fall under the rubric of delinquency for
adolescents. Legally, delinquent behavior is prohibited by law and is carried out by
youths approximately up to the age of eighteen, although the lower and upper age limits
vary by state. State laws legally prohibit two types of behavior for juveniles. The first
includes behavior that is criminal for adults, such as murder, rape, fraud, burglary, and
robbery, as well as less serious criminal offenses such as trespassing and possession of
drugs. The second type of delinquent behavior is status offenses that are not prohibited
for adults, such as running away from home, drinking alcohol, being out of parental
control (i.e., “unruly” or “ungovernable”), and being truant from school (Trojanowicz &

Morash, 1992).

While overall levels of adolescent delinquency have substantially decreased since
their peak in 1993 (Blumstein, 2000), public perceptions of juvenile offending have been
influenced by attention focused on high-profile incidents in recent years, most notably
school shootings occurring in Littleton, Colorado; Paducah, Kentucky; and Jonesboro,
Arkansas. Although fatal assaults in and around schools remain rare events, juvenile
offending has remained of utmost public concern, and citizens, parents, school officials,
and policy makers have called upon communities and schools to prevent these problems
as well as to adopt policies of zero tolerance for juvenile violence (Blurnstein &
Wallrnan, 2000; Snyder & Sickmund, 1999).

Rates of adolescent delinquency vary according to how they are measured,
primarily by either self-reports or official arrest rates. Compared with official studies,
self-report studies find a much higher proportion of the juvenile population involved in
delinquent behavior. Self-reported delinquency rates are typically higher than official
arrest rates because many juvenile crimes are never reported. A review of the literature on
self-report surveys concluded that no more than 15 percent of all delinquent acts result in
a police contact (Dryfoos, 1990). Furthermore, official arrest rates are influenced by bias
in the types of crimes and offenders that enter the juvenile justice system. Arrest rates
vary according to the current political and social climate and law enforcement policies
(Snyder & Sickmund, 1999). For example, juvenile violent crime arrest rates were higher
in 1997 than in 1980 even though victims’ reports of juvenile violent crime did not
increase and, after years of consistency, juvenile arrests for curfew law violations doubled
from 1993 to 1996 (Snyder & Sickmund, 1999). It is likely that these increases reflect

changes in public attitude and resulting law enforcement responses rather than changes in

juvenile behavior. In addition, racial and ethnic minorities are over-represented in arrest
statistics, possibly due to greater police surveillance and a tendency to arrest rather than
warn or otherwise informally sanction minority adolescents (Hawkins, Laub, & Lamitsen,
1998)

Although official crime statistics do not provide information about how many
youth are arrested because they are incident-based rather than person-based, the 1997
National Longitudinal Survey of Youth asked a nationally representative sample of 9,000
youth aged 12 through 16 about their behaviors as well as how often they had been
arrested. While only 8% had ever been arrested, 39% reported that they drank alcohol,
28% purposely destroyed property, 21% used marijuana, 18% committed assault, 11%
ran away from home, 10% carried a handgun, 8% stole something worth over $50, and
7% sold drugs. In addition, the 1997 Youth Risk Behavior Surveillance System found
that 9% of high school students reported that in the past 30 days they had carried a
weapon (e,g, gun, knife, or club) on school property, which was half the proportion of
students (18%) who said they had carried a weapon anywhere in the past month.

It should be noted that it is considered developmentally normal to commit some
delinquent acts during adolescence. Almost every child at one time or another defies
parents or teachers, commits minor acts of vandalism, or breaks the law. The decision to
label someone as “delinquent” is somewhat arbitrary, because being delinquent is a
matter of degree. If every child who had ever broken the law were labeled as delinquent,
then the majority of juveniles would be classified as such (1‘ rojanowicz & Morash, 1992).

Clearly most adolescents who commit delinquent and violent acts are not current or

future serious criminal and/or violent offenders. In fact, 54% of males and 73% of

females who initially enter the juvenile justice system never return on a new referral
(OJJDP, 1998), and only 8% of juveniles who enter the juvenile justice system later
become serious, violent, and chronic offenders (OJJDP, 2001).

However, many prevention and early intervention programs are designed to
prevent and reduce early delinquent behaviors both to reduce the rate of future offending
and to prevent the high cost of the behaviors themselves. For example, truancy is a status
offense and would not be labeled as serious delinquency, yet it is one of the strongest
predictors of future chronic and serious delinquency and itself carries serious negative
consequences, such as affecting youths’ educations and costing school districts hundreds
of thousands of dollars each year (Garry, 1996). Other early conduct problems have also
been found to be related to later delinquency and having an adult criminal career, and
may cause serious immediate consequences (Dryfoos, 1990). For example, our current
adherence to zero tolerance policies means that kids are suspended, expelled, or even land
in juvenile court for increasingly minor infractions, which has implications for their
education, their likelihood of participating in further delinquent acts, and their futures.

Prevention and early intervention programs also aim to prevent the multitude of
problems that adolescent delinquent behavior poses for victims, families of victims and
offenders, communities, and schools. For example, national surveys have consistently
indicated that the people at greatest risk for victimization of violent crime are youth
(Snyder & Sickmund, 1999), and studies that focus specifically on youth violence find
that for adolescents the majority of interpersonal violence occurs between same race and
age peers and commonly involves individuals who are acquaintances or friends

(Hausman, Spivak, Prothrow-Stith, & Roeber, 1992). Besides the risk of physical injury,

 

 

d:

exposure to violence among children and adolescents has been shown to increase mental
health disorders such as depression, anxiety, and post-traumatic stress disorder
(Aneshensel & Sucoff, 1996; Bell & Jenkins, 1993; Martinez & Richters, 1993).

Adolescent delinquency affects families by providing opportunities for siblings to
model delinquent or violent behavior and by causing emotional stress for family members
of both victims and perpetrators (F arrington, 1987). Delinquency may lead to diminished
quality of life for victims and people who reside in areas that are high in crime, and
communities pay the cost of vandalism and other property damage.

Finally, adolescent delinquency may have profound and far-reaching effects on
schools. According to the Centers for Disease Control and Prevention’s 1997 Youth Risk
Behavior Surveillance System, 37% of high school students reported they had been in one
or more physical fights during the past 12 months, and one-third of high school students
had property stolen or vandalized at school (Snyder & Sickmund). These problems may
impact the climate of the school, thereby affecting the learning environment, by causing
students to feel fearful of going to or being in school and reducing the amount of time
teachers spend on education as opposed to discipline.

Prevention science attempts to prevent or moderate human problems such as
adolescent delinquency. An important corollary of this goal is to eliminate or mitigate the
causes of disorder, thus prevention research focuses on examining antecedents of
dysfunction or health, called risk factors and protective factors respectively. The
following sections describe concepts and issues related to preventing adolescent

delinquency.

Prevention of Adolescent Delinquency

Public health professionals pioneered risk-focused approaches to prevention that
have been successfirlly applied to problems as diverse as cardiovascular disease and
traffic-related injuries. Prevention approaches attempt to interrupt the processes that
produce problem behaviors (Coie et al., 1993). During the past 30 years, research has
identified precursors of adolescent delinquency, called risk factors, as well as protective
factors that buffer the effects of exposure to risks and inhibit the development of problem
behaviors even in the face of risk.
_Ri_sl_(, Resilience, and Protection

The concept of risk is rooted in epidemiology and refers to conditions or variables
that are associated with a higher likelihood of negative or undesirable outcomes, such as
behaviors that compromise health or safety (lessor, Van Den Bos, Vanderryn, Costa, &
Turbin, 1995). Researchers have compiled an extensive list of predictors of adolescent
delinquency. For example, the US. Department of Justice’s Office of Juvenile Justice
and Delinquency Prevention (OJJDP) initiated a Program of Research on Causes and
Correlates of Juvenile Delinquency in 1986, consisting of three longitudinal studies based
in Denver, Rochester, and Pittsburgh. These and other studies have produced a unifying
framework focused on risk factors for delinquent behavior (i.e., a risk-focused approach)
that has dominated the field for more than a decade (Coie et al., 1993; Howell, Krisberg,
Hawkins, & Wilson, 1995; Huizinga, Loeber, & Thomberry, 1994).

The concept of resilience emerged from the search for risk factors, as researchers
consistently found that some children who faced stressful, high-risk situations fared well

in life (Garrnezy, 1985; Rutter, 1987; Werner, 1984; Werner & Smith, 1992). Data

 

 

 

ha

PH

suggest that only about one-third of any population of “at-ris ” children experience a
negative outcome and two-thirds appear to survive risk experiences without major
developmental disruptions (Wolin & Wolin, 1995). The term resilience has come to be
used to describe children who achieve positive outcomes in the face of risk (Kirby &
Fraser, 1997). For example, a child who is exposed to multiple risk factors for
delinquency (e. g., exposure to violence, poor family management practices, etc.) and does
not later exhibit serious or chronic delinquent behavior may be described in this manner.
Resilience is thus best defined not as the absence of risk but as successful adaptation
despite adversity (Masten, 1994; Werner & Smith, 1992).

In the face of growing dissatisfaction with pathology-focused intervention
strategies, professionals from mental health and other fields have joined public health
practitioners in the search for factors that might promote resilience in children (Kirby &
Fraser, 1997). Resilience research is concerned with the identification of protective
factors and mechanisms that operate to buffer youth facing multiple risks from problem
outcomes such as delinquency and violence (Smith, Lizotte, Thomberry, & Krohn, 1995).

Protective factors are the internal and external forces that help children resist or
ameliorate risk (Rutter, 1985). Although risk factor research is well-developed,
developmental research in criminology has only recently begun to focus on protective
factors related to resilience among youth at risk for delinquency (Smith et al., 1995).

The practice of reducing risk and increasing protective factors has been termed a
“risk and resilience” approach by some researchers (e.g., Fraser & Galinsky, 1997) and
has been increasingly utilized to understand and design interventions to prevent youth

problems. This approach to prevention uses ecological theory as a framework for

understanding both child development and human behavior and adds specific information
about adolescent delinquency from the growing body of research on risk and protective
factors because resilience is believed to be affected by these factors (Fraser & Galinsky,

1 997).

The risk and resilience perspective is comprised of two components. First, it
consists of a growing body of knowledge of individual and environmental risk and
protective factors that appear to underlie many adolescent social and health problems. In
aggregate, this body of knowledge provides the basis for a common or cross-cutting
model of risk and protective factors for a variety of youth problems. Second, building on
these common factors, the risk and resilience perspective recognizes that some risk
factors contribute uniquely to particular problems and that some protective factors
provide safeguards against particular problems. Thus, to understand and prevent a
particular childhood problem, one needs to consider not only common risk and protective
factors, but also problem-specific factors. The following sections describe the ecological
fi'amework guiding this perspective, prevention terminology and concepts for
distinguishing among different prevention goals, population foci, and timing of service,
and, finally, empirical research supporting the risk and resilience approach to prevention.
Ecological Framework

Adolescent delinquency is a complex problem with multiple causes that span
multiple domains of youths’ lives and include influences outside of their direct life
experiences. Ecological theory provides a framework not only for understanding
complex social problems, such as delinquency, but also for developing and evaluating

interventions that are sensitive to this complexity.

Ecological theorists emphasize that human behavior does not occur in a social
vacuum, but exists as part of a complex system. People operate in many settings and are
influenced by factors at many levels. A person exists in an environment, and it is the
interplay of individual characteristics with contextual influences that ultimately yields
human behavior (Germain, 1991).

Central to ecological theory is an emphasis on the interdependence of systems
(Kelly, 1966). Kelly’s principle of interdependence states that changes in one setting or
level of a system impact other settings and levels in that system. Because components
within a social system are interdependent, changes to any one component have the
potential to create radiating effects. Thus, it is important to attend to the complex
interconnections across system parts when planning interventions.

Ecological theory also provides an important frame of reference for understanding
the experience of childhood because it reminds us that child development and behavior
are shaped by both individual and environmental factors. As an early proponent of
ecological theory, Bronfenbrenner (1979, 1986) argued that children’s development is
strongly influenced by the family, school, peer, neighborhood, and community contexts in
which they live. From this perspective, the social ecology of childhood can be
conceptualized as consisting of interdependent and often nested parts or systems.

Ecological theory gives order to the growing body of research on risk and
protective factors by providing a conceptual framework that incorporates both individual
and contextual conditions affecting the probability of a problem (Fraser, 1997). Risk and
protective factors for juvenile delinquency are known to exist in all domains of a youth’s

life, including family, school, community, and peer group, as well as within individuals

10

themselves. Ecological theory argues that delinquency is influenced by multiple risk and
protective factors in many settings and that interventions to address this problem must be
responsive to this complexity. Consequently, the risk and resilience perspective is
consistent with ecological theory and provides a conceptual framework for prevention,
intervention, and treatment based on the person-in—environment model (Williams, Ayers,
and Arthur, 1997).
Concepts of Prevention

Caplan (1964) delineated three models of prevention - primary, secondary, and
tertiary- that have become standard terminology for differentiating prevention goals,
population foci, and timing of service delivery. Many prevention experts consider
primary prevention as prevention in its purest sense (Linney & Wandersman, 1991).
Primary prevention entails intervening before an unwanted condition or outcome occurs
and is aimed at entire populations to reduce overall incidence (rate of occurrence) of a
problem. An example of a primary prevention program is developing a community-wide
media campaign to promote non-violent conflict resolution. Secondary prevention refers
to programs or strategies to reduce the overall prevalence (total number of cases) of a
problem and involves identifying and targeting individuals who are at risk for an
unwanted outcome. An example of this type of prevention would be having a mentoring
program for students who have been identified by teachers and school administrators as
being at risk for engaging in delinquent behavior because they have fiiends who get into
trouble. Tertiary prevention represents prevention in its narrowest form and is targeted at
individuals who already manifest symptoms of a problem. Tertiary prevention strategies

are designed to reduce adverse consequences of a problem or prevent further participation

11

in the problem behaviors. An example of tertiary prevention would be providing a social
skills development program for youth who are in detention awaiting disposition.

The next sections will present risk and protective factor research as well as an
example of a model of risk reduction, Communities that Care (CTC), and an example of a
model of protective factor enhancement, the Search Institute Asset Model. The CTC and
Search models of risk reduction and protective factor enhancement are not themselves
primary, secondary, or tertiary prevention, but may be used in a range of settings to guide
prevention programs and strategies. In other words, these approaches serve as needs
assessments to guide programming, and primary, secondary, or tertiary prevention efforts
may evolve from them depending on the setting, population, prevention goals, etc. For
example, CTC and Search surveys are widely used to guide programming in schools and
communities, where the focus is likely to be on primary and secondary prevention, but
they have also been used in juvenile detention centers and prisons to guide tertiary
prevention programming.

Although existing risk reduction and protective factor enhancement models have
not been validated for use as instruments to classify individuals as “at-risk,” to the extent
these approaches may be used for secondary prevention, a caveat concerning prediction
and labeling is warranted. Although consensus among researchers regarding risk factors
for delinquency and violence is considerable, an important aspect of prediction is the base
rate of the behavior being predicted. A relatively small proportion of the individuals in
any birth cohort will engage in violent or serious delinquency during late adolescence or
early adulthood. Lipsey and Derzon (1998) found in their meta-analysis that

approximately 8% of juveniles were typically classified as violent or seriously delinquent

12

or criminal on outcome measures administered between the ages of 15-25. Even selecting
from youth who have been arrested, studies find that about 8% become serious repeat
offenders. The outcome of serious and violent delinquency or crime then, has a rather low
base rate and is consequently difficult to predict (Lipsey & Derzon, 1998). Therefore, the
potentially harmful effects of incorrectly labeling someone as “at-risk” need to be
considered when choosing secondary prevention strategies. Research suggests that
labeling can have powerful, harmful consequences (Cowen, 1996). This issue will be
further addressed in the sections below.

mind Protective F actor Research

Risk Factor Research

 

Ecological theory highlights the fact that risk factors in the individual, family,
school, peer group, and community environment are interdependent. Child development
occurs and human behavior unfolds in the context of multiple systems of influence that
may have varying impacts at different stages in development. Risk is conceptualized as
dynamic and interactional; individual risks are nested in the context of family, school,
neighborhood, and broader societal influences that both affect and are affected by
individual factors (Fraser, Richman, & Galinsky, 1999). Accordingly, most recent
research has focused on identifying and understanding processes underlying risks in
multiple domains of children’s lives. For example, Hawkins and Catalano (1992), in
conjunction with the Office of Juvenile Justice and Delinquency Prevention, reviewed
longitudinal studies of predictors of adolescent delinquency, violence, substance abuse,
pregnancy, and school drop-out. They organized 30 years of risk factor research into

community, school, family, and individual/peer domains for their “Communities That

13

Care” model of risk-focused prevention. This model is based upon the recognition that
because risks for adolescent problems exist in multiple domains, interventions to address
these problems should target multiple systems as well. More than 400 communities have
used the Communities That Care Model to examine the risks youths are facing and to
plan interventions accordingly. The CTC model and predictors of delinquency are
presented in a later section of this review.

Many adolescent problem behaviors have been consistently associated with
increasing exposure to risk factors (J essor & J essor, 1977; Newcomb & Felix-Ortiz,
1992; Osgood, Johnston, O’Malley, & Bachman, 1988; Rutter, 1990), and exposure to
multiple risk factors has cumulative effects (Bry, McKeon, & Pandina, 1982; Coie et al.,
1993; Newcomb, 1995). This cumulative effect is seen for risk factors within and across
domains. For example, OJ JDP Causes and Correlates researchers found that exposure to
more than one type of family violence (e.g., intimate partner violence, hostile family
climate, and child maltreatment) greatly increased the likelihood of subsequent violent
youth behavior. Other analyses revealed that multiple risk factors across domains also
interact to produce higher levels of risk than simple additive models would suggest. For
example, juveniles who have both delinquent fiiends and parents who are involved in
problem behaviors such as crime or drug abuse exhibit the highest levels of involvement
in delinquency, and this effect far exceeds the individual effects of either peers or parents
by themselves. The roles of parents and peers interact in influencing levels of delinquency
and violence (Thomberry, 1994).

Common or cross-cutting risk factors. Because many risk factors appear to be

common antecedents of a number of negative outcomes for adolescents, some researchers

l4

have advocated for identifying a set of common or “cross-cutting” risk factors that could
be targeted in interventions to prevent a range of mental health and behavioral problems
for adolescents (e.g., Coie, Watt, West, Hawkins, Asarnow, Markman, Rarney, Shure, &
Long, 1993). Common risk factors encompass multiple domains and include: family
factors such as family conflict, child abuse, poor bonding to parents, unskilled parenting,
and family disorganization; school problems; ecological/contextual factors such as
neighborhood disorganization, racial injustice, sexism, unemployment, and extreme
poverty; peer group factors such as peer rejection; and individual level factors such as
below average intelligence, learning disabilities, and attention deficit hyperactivity
disorder (Coie et al., 1993; Kirby & Fraser, 1997).

Cumugtive number of risks versus mblem-specific risk pathways. Some
researchers have argued that the cumulative number of risk factors may be more
important than examining problem-specific risk pathways (Rutter, Sameroff, Baldwin,
Baldwin, & Seifer, 1999; Sameroff, Seifer, Zax, & Barocas, 1987; Sameroff, Seifer,
Barocas, Zax, & Greenspan, 1987). For example, OJJDP’s Causes and Correlates
researchers and others (e. g., Smith, Lizotte, Thomberry, & Krohn, 1995) defined “high-
risk you ” as those who have five or more risk factors. In these studies it was considered
more important to know that a child was exposed to five risk factors than to know to
which factors the child was exposed. This perspective has led some researchers to
suggest that focusing on common risk factors and overall numbers of risks may be more
efficient than studying specific risk pathways for particular problems because the number

of risks is more important than any particular risk.

15

However, a cumulative risk perspective does not imply that problems are
explained fiJlly by generic risks. It merely suggests that the effects of poverty, racism,
gender discrimination, child maltreatment, unskilled parenting, and other negative
conditions elevate the odds for many types of problems and disorders (Fraser, Richman,
& Galinsky, 1999). For example, Loeber, Stouthamer-Loeber, Farrington, and Van
Kammen (1998) found that lack of guilt was associated more strongly with conduct
problems, physical aggression, and delinquency than with depression, substance use, and
shy or withdrawn behavior. Miller and Maclntosh (1999), also demonstrating a specific
risk and protective factor perspective, found that having a strong positive ethnic identity
may reduce the effect of daily hassles and promote academic achievement among African
American youths. These findings support an integrated perspective of risk that considers
both cumulative numbers of risks as well as risks that are specific to particular outcomes.

Problem-specific fisfiactors. Extensive research has identified risks for
delinquency, and numerous comprehensive risk factor reviews have been published (see
American Psychological Association, 1993; Dryfoos, 1990; Hawkins, Herrenkohl,
Farrington, Brewer, Catalano, & Hawkins, 1998; Lipsey & Derzon, 1998; OJJDP, 1995;
Reiss & Roth, 1993; Tolan & Guerra, 1994; Yoshikawa, 1994). For one of the most
commonly used models of risk-focused prevention, the Communities that Care model
(CTC), the OJJDP has compiled 30 years of risk research for delinquency and updates
this information on a regular basis. Risks specific to delinquency in this model are
detailed below and include factors within community, school, family, peer, and individual

domains.

16

Communities That Care model. Hawkins and Catalano (1992) developed The
Communities That Care model to prevent adolescent substance use, delinquency,
violence, pregnancy, and other adolescent problems. Communities That Care enumerates
a process of community mobilization and assessment and then development,
implementation, and evaluation of an action plan. As part of the community assessment
process, communities collect and assess data regarding the nature and extent of 19 risk
factors. Some risk factor data may be collected via community-level indicators, but
communities are also encouraged to use the CTC Youth Survey as a risk and needs
assessment instrument for youth in grades 6-12. Surveys are typically administered in
schools and the Charming Bete Company, which owns the rights to Communities That
Care, tabulates the results and provides the schools with a profile of risks and needs to
guide programming.

Empirically identified risks for delinquency in the CTC model include: 1)
individual factors such as early delinquent and aggressive behavior (F arrington, 1991;
Hawkins, Catalano, & Miller, 1992; Maguin et al., 1995; Rutter, 1990; Werner & Smith,
1992) and attitudes favorable to these behaviors (Kandel, Kessler, & Maguiles, 1978;

' Huesmann & Eron, 1986); 2) family factors such as poor family management practices
(Capaldi & Patterson, 1996; Farrington, 1991; Kandel & Andrews, 1987; Maguin et al.,
1995; Patterson & Dishion, 1985; Peterson, Hawkins, Abbott, & Catalano, 1994;
Thomberry, 1994), high levels of family conflict (Loeber & Dishion, 1984; Rutter &
Giller, 1983; Yoshikawa, 1994), and favorable parental attitudes and involvement in
criminal or violent behavior (Brook, Brook, Gordon, Whiteman, & Cohen, 1990; Hansen,

Graham, Shelton, Flay, & Johnson, 1987; Hawkins & Weis, 1985; Kandel, Kessler, &

17

Maguiles, 1978; Maguin et al., 1995); 3) peer influences, such as having friends who use
drugs or engage in delinquent or violent behavior (F arrington, 1991; Loeber, 1990) or
who have favorable attitudes toward delinquent or violent behavior (Ageton, 1993;
Elliott, 1994; Maguin et al., 1995); 4) school setting factors of academic failure (Maguin
& Loeber, 1996) and lack of commitment to school (Maguin et al., 1995); and 5)
characteristics of the community and neighborhood environment that are mediated
through family characteristics or affect the likelihood of individual, family, and peer
influences leading to delinquency or violence. These community and neighborhood
influences include low neighborhood attachment and community organization (Maguin et
al., 1995; Sampson, 1986; Sampson, Raudenbush, & Earls, 1997), high transitions and
mobility (Gottfredson, 1988; Maguin et al., 1995), and extreme economic deprivation
(Sampson & Lauritsen, 1994). Higher rates of juvenile drug problems, delinquency, and
violence occur in neighborhoods where people have little attachment to the community
and rates of vandalism. are high (Murray, 1983; Wilson & Hemstein, 1985). Availability
of drugs (Gottfredson, 1988) and firearms (Alexander, Massey, Gibbs, & Altekruse,
1985) and community norms favorable toward drug use, firearms, and crime also predict
delinquency and violence (Maguin et al., 1995; Thomberry, Huizinga, & Loeber, 1995).
As stated earlier, the CTC and similar models are not primary, secondary, or
tertiary prevention themselves, but assess risks and needs to guide prevention
programming. The CTC Youth Survey does not collect information identifying
individual students and schools do not keep the surveys; surveys are sent to the Charming
Bete company, which compiles the information and sends schools a report that profiles

the risks and needs faced by their students. In other words, the survey does not provide

18

schools with information to label certain youth as high risk so they may be targeted for
intervention. However, it should be noted that some of the risk factors that appear in
CTC and other risk reduction models (e.g., Developmental Services Group), such as early
and persistent anti-social behavior and early initiation of problem behaviors (e.g.,
delinquency or violence), do imply that secondary and tertiary prevention approaches
could be utilized. For example, if school staff and administrators learn that these risk
factors are prevalent in their middle or high schools, they could respond by initiating
primary prevention programs in their feeder elementary schools to prevent early behavior
problems; alternatively, they could identify on their own the students with early problems
and target them for intervention. Although experiencing early behavior problems is one
of the strongest predictors of future serious and violent delinquency and criminality
(r=.36), given the low base rate there is a high probability of mislabeling youth as at risk
for serious and violent delinquency. It would be more prudent to devise primary
prevention programs for younger children and base programs for youth already engaging
in problem behaviors, such as fighting and truancy, on addressing those problems because
of their high cost to youth themselves, schools, families, and communities, and not on the
pretext of preventing youth from becoming future serious or violent criminals.

The next sections of this review will examine research on factors that help buffer
the impact of risk. Although it was developed as a model of risk-focused prevention, a
small number of protective factors have recently been added to the Communities that
Care framework. Their protective factors were chosen based on well-established
criminological theories, such as social control theory (Hirschi, 1969), that support the role

of social bonds in protecting against delinquency. A recently published longitudinal

l9

study has associated these factors with a reduction in future violent behavior (e.g., Huang,
Kosterrnan, Catalano, Hawkins, & Abbott, 2001). The protective factors added to the
model are: bonds to school, family, and conventional peers, opportunities for involvement
in prosocial activities, skills to participate in such activities, and rewards or recognition
for participation. Additional research on protection as well as a popular model for
protective factor enhancement, the Search Institute Model, are considered below.
Protective Fgctor Resear_cl_r

The search for factors that mitigate the impact of risk has recently been extended
beyond child psychopathology to adolescent delinquency (e.g., Brook, Whiteman, Cohen,
& Tanaka, 1992; Newcomb & Feliz-Ortiz, 1992; Hawkins, Catalano, & Miller, 1992;
Stacy, Newcomb, & Bentler, 1992; Stacy, Sussman, Dent, Burton, & Flay, 1992; Wills,
Baccara, & McNamara, 1992). Although the concept of risk is widely understood, there
has been far less consensus about the concept and operationalization of protective factors
(Jessor, Van Den Bos, Vanderryn, Costa, & Turbin, 1995). Protection has been defined
by some as simply the absence of risk or as the low end of a risk variable. Rutter (1987)
argued that protective factors should be treated as conceptually distinct rather than as
opposite ends of a single dimension, and that view is now coming to be shared by many
others (Newcomb & F elix-Ortiz, 1992; Hawkins et al., 1992; Jessor, Donovan, & Costa,
1991; Luthar & Zigler, 1991; Pellegrini, 1990). However, the means for defining and
categorizing protective factors is not well specified. Whether high social support, for
example, is defined as a protective factor or whether low social support is defined as a
risk factor depends on earlier research and the nature of the relationship between social

support and a specified outcome (Fraser, Richman, & Galinsky, 1999).

20

R_elajon$1ip between risl(__ and protection. Research on risk and protective factors
has often shown them to be negatively related, which may reflect a particular organization
of the social environment; for example, in contexts in which protection is high, risk is
usually low, and vice versa. However, it is possible to find high risk accompanied by
high protection. An adolescent may have antisocial friends and yet be committed to and
involved in school. Although risk and protection may be inversely related empirically,
the conceptual perspective is that they are independent (J essor, Van Den Bos, Vanderryn,
Costa, & Turbin, 1995).

As with risk research, there has been debate over specific protective factor models
versus cumulative numbers of protective factors. Some researchers (e.g., Garrnezy, 1985)
have rejected the notion of specific risk models as old-fashioned and have asserted that
researchers should focus on overall numbers of common risk factors that are associated
with an array of negative outcomes. This assertion has led other researchers (e.g., Jessor
et al., 1995) to hypothesize that protection, like risk, has a wide spread of overlap and that
a number of common protective factors buffer many types of risk and protect against
many outcomes. Researchers should thus focus on cumulative levels or overall numbers
of protective factors rather than enumerating specific models of risk and protection for
particular outcomes. Other researchers (e.g., Guerra, 1998) believe that it would be a
notable advance in the literature to identify protective factors for specific risk profiles.
For example, which protective factors reduce risk for children living in neighborhoods
with high levels of violence and other problems? Are different factors more effective at
reducing risk for children who experience family problems? As with earlier risk research,

studies examining protection incorporate both cumulative and specific approaches. Thus,

21

it is important to consider factors that have been shown to protect children facing a
variety of risks for many problems as well as those that seem particularly effective against
delinquency specifically.

Common or cross-cutting protective factors. Researchers have found support for
a number of common or cross-cutting factors that protect youth from many problem
outcomes, such as substance use, teen pregnancy, school drop-out, delinquency, and
violence. These protection variables are posited to operate at three levels (Garmezy,
1985; Gordon, 1995; Luthar & Zigler, 1991; Rutter, 1987, Werner, 1998): a) individual
attributes such as high efficacy, high intelligence, independence, reflective rather than
impulsive decision making, internal locus of control, good communication and problem
solving skills, and resilient temperament; b) family attributes, such as parental support
and affection, appropriate discipline, adult monitoring and supervision, and strong family
bonds; and c) extrafamilial or societal circumstances, such as support from other adults or
strong community integration. In a review of resilience research, Davis (1999) identified
caring relationships, high expectations and adequate support, and opportunities to
contribute as general protective factors in families, schools, and communities. Masten
(2001) also reviewed variable and person focused resilience research and found support
for global factors associated with resilience that include connections to competent and
caring adults in the family and community, cognitive and self-regulation skills, positive
views of self, and motivation to be effective in the environment.

Problem;s_pecific protective factors. Studies that focus specifically on delinquent
behaviors such as violence, substance use, teen pregnancy, and school drop-out find

evidence that similar factors play a role in protecting youth from these outcomes. The

22

protective factors that have been shown to reduce the chance that a child experiencing
risk will become involved in delinquent behavior include strong attachment to parents,
commitment to family, resilient or positive temperament or disposition, ability to adjust
or recover, supportive family environment, strong external support system that reinforces
children’s coping efl’orts, healthy beliefs, prosocial orientation, and social problem
solving skills (Farrington, 1994; Hawkins et al., 1992, Hawkins et al., 1995).

Search Institute’s asset-building model. Recently, some authors have advocated
for a paradigm shift in the prevention field to focus exclusively on building assets - the
“positive relationships, opportunities, competencies, values, and self-perceptions that
youth need to succeed” (Scales & Leffert, 1999, p. 1) - rather than trying to reduce risk
(e.g., Benard, 1993; Benson, 1997; Scales & Leffert, 1999). These researchers assert that
targeting risk factors emphasizes the deficits of young people. They suggest that focusing
on building children’s strengths will produce more positive outcomes than interventions
focusing on reducing risk factors. In recent years, there has been enormous support by
community leaders and school officials across the country for asset building as promoted
by the Search Institute in Minneapolis. In the late 1990s, more than 300 communities
lalmched community-wide interventions to enhance the assets of their youth based upon
reports from the Search Institute (Price & Drake, 1999), and the model continues to grow
in popularity.

Search Institute has correlational data from surveys of over 600,000 6-12th grade
students showing that youth with high numbers of assets tend to engage in few problem
behaviors while adolescents with few assets tend to engage in many such behaviors

(Scales & Leffert, 1999). Developing assets in youth is intended to generally support

23

positive outcomes and reduce risky behaviors and does not focus on preventing specific
problems or buffering specific risks. Search’s 4O developmental assets are grouped into 8
categories representing broad domains of influence in youths’ lives. External assets,
which are relationships and opportunities that adults provide, consist of support,
empowerment, boundaries and expectations, and constructive use of time. The internal
assets, competencies and values that youth develop internally to help them become self-
regulating adults, are commitment to learning, positive values, social competencies, and
positive identity (Scales & Leffert, 1999).

Researchers have voiced numerous concerns with Search Institute’s asset-building
model (e.g., Price & Drake, 1999). First, surveys were administered within self-selected
communities that were primarily small, non-Metropolitan areas, and survey participants
were 98% White. Thus, any results reported or any validation of the instrument has been
conducted on a non-diverse sample that is not representative of US. youth. Second,
although asset building is promoted as focusing on youths’ strengths, it is not
implemented as a strength-based strategy, that is, leveraging youths’ strengths to help
address their needs. Instead, the Institute asserts that if schools and communities can
measure assets, they can then work to enhance the deficient assets for youth, thereby
reducing their at-risk status. Third, of the 27 subscales with two or more items, six had
reliability coefi'rcients of .50 or less, indicating weak internal reliabilities. In addition,
test-retest reliabilities have not been reported despite the fact that communities are
currently using the instrument as a pre-post test. Fourth, all assets are weighted equally.
Search aggregates data from the thousands of survey instruments provided by

communities and schools. For example, they suggest that students with 0-11 assets are 3

24

times as likely as those with 21-25 assets to have early sex experiences. Simply adding
the assets implies that they have the same effects on all students, although it seems logical
that different assets would have varying effects according to gender, race/ethnicity, age of
the student, and the types of risks faced. In addition, all Search data are correlational, and
there have been no studies published in peer reviewed journals to test asset theory. Fifih,
Search researchers themselves concede that their conceptualizations of some assets, such
as empowerment, positive values, and social competencies, are not well-grounded in
research (Scales & Leffert, 1999). Finally, the Search Institute survey, the validity and

' effectiveness of which has not been confirmed, has hindered the national longitudinal
assessment of key adolescent risks by replacing well-established risk assessment surveys,
such as the Youth Risk Behavior Survey, in many school districts. Many schools and
communities do not have the time or resources to administer multiple surveys, and the
Search survey cannot be shortened and combined with risk assessment questions because
Search requires that schools and communities sign a contract that prohibits them from
analyzing their own data.

Moreover, some researchers (e.g., Price & Drake, 1999; Tolan, 1996) have argued
that a sole focus on the assets of young people emphasizes individual characteristics and
ignores important social and contextual risk factors that should be a focus for prevention
policies and interventions. In addition, the effect of ignoring risk and focusing solely on
enhancing protection or assets on the development of adolescent problem behaviors is
unknown. No published study has shown that a major intervention to change youths’
assets, as defined by the Search Institute, reduces their risk for any type of negative

outcome (Price & Drake, 1999). Given the strong relationship between exposure to

25

increasing numbers of risk factors and negative outcomes, the likely effect of
interventions focused exclusively on building protection depends on whether protective
factors can fully mitigate the negative effects of exposure to multiple risk factors during a
child’s development.

Research to date suggests that risk may be more potent than protection in
influencing human behavior. Studies that have examined this issue indicate that simply
focusing on strengthening assets or protective factors without attending to risk exposure
is incomplete as a strategy for reducing the prevalence of problem behaviors. At higher
levels of risk, protective factors either do not exist or weakly counteract the effects of
adversity. It may be difficult to create and sustain high levels of assets or protection in
the highest risk environments unless efforts also seek to reduce risk exposure (Pollard,
Hawkins, & Arthur, 1999). Thus, both risk and protective factors should be considered to
understand social problems and devise intervention plans (Fraser et al., 1999).

Md protection models. It is widely believed that resilience results in some
way from the interplay between risk and protective factors, but the nature of these
interactions is poorly understood. Two basic models of interaction are described in the
literature: additive and interactive models.

Protective factors are considered to be independent variables that can have their
own direct effects on behavior as well as moderate the relationship among risk factors
and behavior (Rutter, 1987). In additive models, compensatory protective effects do not
affect risk but rather act directly to reduce a problem or disorder. For example, in a study
of over 36,000 7-12m grade students in Minnesota, compensatory protective effects for

externalizing behavior problems were found for family, school, and religious

26

connectedness in students exposed to low levels of risk (Resnick, Harris, & Blurn, 1993).
In interactive models, protective factors are believed to serve as buffers by exerting an
even stronger positive effect on children who have been exposed to adversity. This
interactional perspective does not exclude the possibility that a factor exerts an influence
on people who are not at risk, but rather asserts that the more important influence is on
people whose risk levels are high (Dubow, Roecker, & D’Irnperio, 1997). The increase in
variance accounted for by interactions is typically small (e.g., Newcomb & F elix-Ortiz,
1992; Luthar & Zigler, 1991), but existing data indicate that protective factors do interact
with risk factors and, thus, the processes leading to resilience are at least partly nonlinear
(Kirby & Fraser, 1997).

Research incormrafiag both risk__and protection. Pollard et al. (1999) measured a
range of risk and protective factors to examine the relationship among increasing levels of
risk exposure, increasing levels of protective factors, and adolescent substance use,
delinquency, and academic achievement. They found that increased levels of risk
exposure were associated with increases in the prevalence of substance use and
delinquency and decreases in academic achievement. Although for each risk level the
lowest prevalence rate of problem behavior was for students with the highest level of
protection, generally variation in levels of risk exposure appeared more strongly related to
the behavior outcomes than did variation in levels of protection. With increasing levels
of risk, fewer individuals had high levels of protection, and conversely, more individuals
with high levels of protection were found as overall risk decreased. At the highest levels
of risk exposure, high levels of protection were not associated with elimination of

problem behaviors. Even among those with high protection, prevalence rates of all

27

problem behaviors increased with more risk exposure. Finally, an interactive relationship
was found between overall levels of risk and protection for the two problem behaviors but
not for academic achievement. For each of the problem behavior outcomes, the reduction
in prevalence associated with higher levels of protection was greatest at the highest risk
levels. These findings support a buffering hypothesis of the relationship among aggregate
risk and protection and these behavioral outcomes.

Pollard et al. (1999) highlighted two points for future research. One, they stated
that more research is needed on the relationships between risk and protective factor
exposure and positive outcomes in adolescence, especially for youth with both low levels
of risk and low levels of protection. Two, although the strength and stability of their
findings across different adolescent problem behaviors are consistent with assertions that
aggregated risk and protection levels similarly impact multiple adolescent problem
behaviors (Jessor & Jessor, 1977; Osgood et al. 1988), the authors believed that it was
likely that different factors contributed differentially to overall risk and protection. Thus,
they stated that more research is needed on the relative strengths of specific risk and
protective factors in predicting various outcomes.

Many studies that examine the role of specific risk and protective. factors examine
only one or two factors at a time. However, there is evidence that some protective factors
may have more weight in protecting youth experiencing specific risks fiorn problem
outcomes such as delinquency. Smith, Lizotte, Thomberry, and Krohn (1995) examined
the effect of 18 family, school, and peer protective factors on serious delinquency for
youth that were classified as high-risk because they had at least five of nine possible

family risk factors (i.e., the top fifth of the distribution). Smith et al. found that two-thirds

28

of youth in the high-risk category were resilient to serious negative outcomes. They also
found that resilient youth had higher levels of protective factors than did non-resilient
youth; in other words, cumulative protective factors increased the likelihood of resilience
to serious delinquency. In addition, they found that given these family risk factors, the
most salient protective factors for resilience to delinquency and drug use were school
protective factors (e.g., commitment to school and teachers and aspirations to attend
college) followed by family protective factors (e.g., parental supervision, parent-child
attachment, and parent involvement in child’s activities). These findings provide support
for the idea that certain protective factors may be especially important in protecting youth
facing family risks from delinquency. The implications of these findings are that schools
may play a particularly important role in interventions for youth facing family risks that
may decrease the likelihood of them participating in delinquent behavior.

Summa_ry and conclusions abo_utgaps in the literature. The risk and resilience
perspective is increasingly being applied to understanding and preventing complex youth
problems such as delinquency. This perspective uses ecological theory as a fiamework
for understanding human development and behavior and adds information about risk and
protective factors for adolescent problems that occur in multiple domains of youths’ lives.

Many researchers believe that resilience results from the interplay between risk and
protective factors and have devoted their efforts to determining which risk and protective
factors are especially important, and the mechanisms by which they impact adolescents.

Research to date has provided evidence that both specific risk factors and
cumulative numbers of risk factors predict negative outcomes for adolescents. Likewise,

specific protective factors as well as cumulative numbers of protective factors have been

29

found to reduce the likelihood that adolescents will experience negative outcomes. While
increasing protective factors is related to a decrease in the likelihood of negative
outcomes for youth with few risks, available evidence suggests that focusing solely on
protective factors may not be effective for youth with a high number of risks. Thus,
interventions for these youth should focus on decreasing risks in conjunction with
increasing protection.

However, there remain many unanswered questions about the relationships among
risk, protection, and adolescent outcomes. Although some of the most widely used
models of risk reduction and protective factor enhancement incorporate large numbers of
risk and protective factors, little information is available on the relationships among the
risk and protective factors in these models and adolescent outcomes; it is not known how
well risk reduction and protective factor enhancement models work together. Although
many schools and communities utilize these models, they tend to be used separately. For
example, groups that utilize Search’s asset model tend to ignore risk factors and groups
that use risk assessment approaches incorporate only a few, if any, protective factors.
Although there is evidence that both risk and protection are important (e.g., Pollard et al.,
1999), research to date is piecemeal, examining overall or cumulative numbers of risk
and protective factors or the specific pathways of only a few risk and protective factors.

It is not known whether it is more effective for schools and communities using these
models as needs assessment checklists to focus solely on increasing the overall number of
protective factors and decreasing overall numbers of risks without regard to which factors
they are, or whether there are factors that are particularly important. There is some

evidence that specific protective factors do have more weight in protecting youth facing

30

specific risks from delinquency (e.g., Smith et al., 1995). Additional information of this
nature may be helpful for designing more effective programs.

In addition, very little research has explored gender, urban/rural, or race/ethnicity
differences in levels of risk and protection and how these factors relate to adolescent
outcomes. Risk and protective factors may affect children differently, depending on such
factors as gender, whether they live in urban or rural areas, and race/ethnicity. Although
little information is available for gender, urban versus rural, and race/ethnicity differences
in numbers of risk and protective factors or their relationship with outcomes such as
delinquency, what is known will be described in the section below.

Potentifi Modegtors of Risk, Protection, and Youth Outcomes

Prevalence data suggest that risk factors differ by a variety of sociodemographic
factors. However, researchers are just beginning to develop an understanding of the ways
in which risk and protective factors vary according to gender and race/ethnicity, and little
information is available regarding differences in risk and protective factors among youth
living in urban versus rural areas (Fraser, Richman, & Galinsky, 1999).

93m The US. Department of Justice’s Office of Juvenile Justice and
Delinquency Prevention (OJJDP) recently issued national reports calling for researchers
and practitioners to focus on the needs of females who engage in delinquent or violent
behavior, as girls in trouble have thus far been the afterthought of a juvenile justice
system designed to deal with boys (Bergsmann, 1989; Miller, Trapani, F ejes-Mendoza,
Eggleston, & Dwiggins, 1995). Research about female delinquents has been scarce
(OJJDP, 1998; Yoshikawa, 1994), and a number of researchers have called attention to

the need for more information (e.g., OJJDP, 1998; Snyder & Sickmund, 1999).

31

Adolescent female offenders have been called “the forgotten few” (Bergsmann,
1989), but their increasing numbers are now garnering attention from researchers, policy
and lawmakers, and service providers. In many cases, girls were victims themselves
before they became offenders (Davis, Schoen, Greenburg, Desroches, & Abrams, 1997;
Prescott, 1997). For example, girls are three times as likely as boys to have experienced
sexual abuse, which is often an underlying factor in high-risk behaviors (OJJDP, 1998).
Among female delinquents, an estimated 70% have a history of sex abuse (Calhoun,
Jurgens, & Chen, 1993). Because girls in crisis are more likely to strike inward and
threaten their own well-being, abusing drugs, prostituting their bodies, starving, or
mutilating themselves (Belknap, 1996), they seem less dangerous to society than boys,
resulting in their needs being overlooked and undertreated (Chesney-Lind, 1988).

Adolescent females often enter the juvenile justice system because they commit
status offenses such as running away from home. While status offenses continue to
account for the bulk of cases involving girls, females are now more likely to be arrested
for robbery, assault, drug trafficking, and gang activity-juvenile crimes only recently
considered the exclusive domain of young males (Poe-Yamagata & Butts, 1996).
Although girls still commit far fewer crimes than boys-they constitute about two out of
eight juvenile offenders- their numbers are increasing at a faster rate than that of boys and
violent crime has increased nearly four times as much among girls (16.5%) than among
boys (4.5 percent) during the past decade (Snyder & Sickmund; Richie, 1999). Between
1993 and 1997, increases in arrests were greater for girls than for boys in almost every
offense category (OJJDP, 1998).

Researchers have not only found different rates and patterns of offending for girls

32

and boys, but also differences in risk and protective factors that are similar to those
delineated by the Communities That Care and Search models. For example, Werner
(1990) noted that resilience in girls was promoted by parenting styles that placed
emphasis on risk taking, independence, and stable emotional support, while for boys,
resilience was fostered by parenting styles that provided higher degrees of supervision
and structure, the presence of a male role model, and support for expressing emotions.

Ensminger (1990) examined parental monitoring and boundary setting and found
that permissive family rules about curfews on school nights were associated with females
engaging in multiple problem behaviors (e.g., substance use and assaulting others).
Contrary to these findings, Seydlitz (1991) found that parental controls were more likely
to inhibit delinquency in males than females; moreover, these controls were more
effective in midadolescence for males and in later adolescence for females.

Researchers have also found that females tend to have larger possible networks of
support, including more adults and more connections across family, school, and
neighborhood (Coates, 1987; Svedhem, 1994) and that they seem to be more positively
affected by support and connectedness in general, especially by parent support ( Bailey &
Hubbard, 1990; Clark-Lempers, Lempers, & Ho, 1991; Cotterell, 1992; Eccles, Early,
Fraser, Belansky, & McCarthy, 1997; Leon, Fulkerson, Perry, & Dube, 1994; Margalit &
Eysenck, 1990; Windle, 1992). Some research, however, suggests that males may be even
more sensitive to and helped by support offered outside the family, in the neighborhood
and wider community (Entwisle, Alexander, & Olson, 1994; Werner, 1992).

Other research supports the idea that neighborhoods may influence males more

than they do females. Ensminger, Lamkin, & Jacobson (1996) examined the impact of

33

neighborhood indicators (percentage of residents below poverty level, percentage in
white-collar occupations, median income, median education) on the likelihood of school
drop-out and found no neighborhood effects for females as compared to males.

Studies have also found gender differences in achievement motivation. A study of
a diverse sample of more than 1000 7th graders followed through 8th grade (Roeser &
Eccles, 1998) found that females generally and African Americans of both genders valued
education significantly more than did males generally and Caucasian adolescents of both
genders. Females read for pleasure more than do males (Moffitt & Wartella, 1992) and
also spend more time on homework in middle school, although by the 9'” grade they
appear to spend less (Leone & Richards, 1989).

Achievement motivation has less of a relationship to self-image for females as
they make the transition from 6‘“ to 7”h grade, while for males, just the opposite occurs
(Elmen, 1991). Juvonen and Murdock (1995) found that “smart and diligent” students
are the most popular among peers in the 4th grade, but the least popular students by the 8“I
grade. If females tend to be engaged, do more homework, and get better grades overall
than males, their popularity may decrease as they move into early adolescence. The
researchers hypothesized that because group membership and belonging are important to
females, they may lose interest in doing well, particularly in those subjects where success
seems more connected to loss of popularity.

Although these studies provide information about gender differences in overall
levels of a few risk and protective factors and their relationship to a variety of outcomes,
information is needed on gender differences for a greater number of risk and protective

factors as well as their relevance for female delinquency. The Office of Juvenile Justice

34

and Delinquency Prevention (1998) has begun this process by examining the needs,
characteristics, and backgrounds of females in the juvenile justice system. The risk
factors they deem as especially salient for girls include early sexual experimentation,
academic failure, history of sexual abuse, low-self-esteem, dysfunctional family system,
racism, sexism, and substance abuse. Protective factors specified as important for girls
by the OJJDP (1998) include delay of sexual experimentation, academic
success/progress, positive sexual development, positive self-esteem, positive family
environment, positive minority identity, positive gender identity, and prosocial skills and
competence. However, more work is necessary to further understand the role of these
factors in promoting or inhibiting female delinquency and violence, as well as the role of
gender in moderating the relationship among risk, protection, and delinquency.

Urban versus ring Researchers have noted that although youth in urban settings
are at-risk for multiple behavior problems, few empirical investigations have been
undertaken regarding resiliency within this group (Luthar, Doernberger, & Zigler, 1993;
Miller, 1999). Likewise, researchers and practitioners have observed that youth in rural
settings face unique challenges, such as higher rates of tobacco and alcohol use (Cronk &
Sarvela, 1997), as well as barriers to accessing programs and resources. A study
conducted by a county health department examined urban versus rural differences in
levels of risk and protective factors measured by the Michigan Alcohol, Tobacco, and
Other Drug (ATOD) Survey and their relationship to substance use (Cheaturn, 1998).
The study examined local survey data gathered from 8‘”, 10’“, and 12th graders between
1992 and 1997. The outcome variable was the frequency with which survey participants

used alcohol, tobacco, or other drugs, and the measured risk factors were whether or not

35

respondents had been victims of violence and availability of alcohol, tobacco, and other
drugs, as measured by how often respondents went out at night. Measured protective
factors included family support, emotional well-being, enjoyment of school, and religious
service attendance.

The researchers found that patterns of substance use varied by region, with
cigarette and alcohol use higher in rural areas and marijuana use higher in urban areas.
However, marijuana use increased ten percent in the rural areas between 1992-93 and
1996-97, while there was only a two percent increase in urban areas during that time
period.

The results also indicated that number of protective factors varied by region.
Rural youth reported fewer protective factors than youth in urban communities. In
comparison with rural youth, urban youth reported higher levels of family support and
were more likely to report that they enjoyed school. However, urban youth were less
likely than rural youth to report that they were happy.

The researchers found that each of the measured risk and protective factors had a
direct relationship with youth alcohol, tobacco, or other drug use. They also reported
regional differences in the relationship of protective factors with substance use. While all
of the protective factors showed a significant negative relationship with substance use for
urban youth, enjoyment of school and family support did not predict substance use for
rural teens.

These results provide some support for the assertion that urban versus
rural differences may be important to consider in risk-focused prevention. This analysis

found that the outcome of interest - substance use - varied by region. Urban versus rural ‘

36

differences were also found in the number of protective factors youth had and the
relationship of these factors with substance use. However, more information is needed on
other outcomes, such as delinquency. In addition, while these findings support the
relevance of urban versus rural differences in risk focused prevention, it should be noted
that only a limited number of risk and protective factors were examined in this analysis
and clearly more information is needed to fully understand urban versus rural differences
in levels and salience of risk and protective factors.

Race/Ethnicig. Similarly, researchers have recognized the need for more studies
on racial and ethnic minorities. The exploration of additional protective factors within
populations that have unique stressors and histories is paramount for further
understanding resiliency in general and minority groups in particular (Miller, 1999). For
example, Afiican American children in urban settings often have numerous obstacles to
overcome, such as poverty, substandard housing, and inferior schools (Peters, 1985;
safyer, 1994). In addition, socialization of Afiican American children frequently occurs
in the context of racial discrimination and oppression (McCreary et al., 1996), an
environment that is not conducive to mental health (Thorton, Chatters, Taylor, & Allen,
1990). Research thus far suggests that risk and protective factors vary as a function of
race and ethnicity. For example, among African American families, strong social ties, a
deep sense of spirituality, racial identity, and flexibly configured families that include kin
and non-kin have been reported to contribute to resilience (McAdoo, 1998). Studies of
Afi'ican American adolescents have found that high racial identity is a significant factor in
successfully coping with stress, academic achievement, and lower participation in

problem behaviors (Bowman & Howard, 1985; McCreary, Slavin, & Berry, 1996).

37

A number of studies address the variation of family boundaries and expectations
as a fimction of racial and ethnic group differences. For example, Steinberg, Mounts,
Lambom, and Dombusch, (1991) reported that the relationship between authoritative
parenting and school performance is greater among Caucasian and Hispanic adolescents
than among their Afiican American or Asian American counterparts. They found no
racial or ethnic differences related to other adolescent outcomes such as behavior
problems.

Adolescents’ experience of family boundaries may vary considerably across
different contexts. Bulcroft, Carmody, and Bulcroft (1996) used a nationally
representative sample of households with a child between ages 12 and 18 to show that the
race, age, and gender of an adolescent interact to affect the kinds of boundaries and
independence parents grant. Hispanic parents maintain earlier curfews for both genders
at all ages than do Caucasian or African American parents. Afiican American males of
all ages are given more independence outside the home than are males of other races, but
African American females’ curfews stay relatively early even as they grow older. The
researchers concluded that Hispanic females in middle and late adolescence were among
the most restricted youth.

Roeser and Eccles (1998) formd that Afiican Americans of both genders valued
education significantly more than did Caucasian adolescents of both genders, which may
have implications for the protective factor of achievement motivation. Landrine et al.,
(1994) found some evidence that the impact of peer involvement in problem behaviors, a
risk factor specified by CTC and other risk reduction models, may vary by race/ethnicity.

They found smoking by peers to be the best predictor of smoking among Caucasian

38

adolescents, while peer smoking was a somewhat weaker predictor of smoking among
Hispanic and Asian youth and was not found to be a good predictor of smoking among
Afiican American youth (Landrine et al., 1994).

Although these studies suggest that race/ethnicity is an important factor to
consider, as with gender and urban/rural, additional information is needed on a greater
number of factors as well as the relationship of the factors to delinquency for youth from
different racial/ethnic backgrounds. Further work is necessary to further understand the
role of race/ethnicity in moderating the relationship among risk, protection, and
delinquency.

Current Study

The current study attempts to address several gaps in the literature with a survey
of middle school students. First, two popular approaches for targeting delinquency are
combined in a comprehensive manner. The survey incorporates many of the risk and
protective factors that appear in widely used risk reduction and protective factor
enhancement models, and this study examines the relationships of the risk and protective
factors with each other and with the outcome of adolescent delinquency. Second, gender,
race/ethnicity, and urban versus rural differences in risk and protective factors within the
individual, family, peer, school, and community domains are explored. Third, the utility
of considering the domains of the risk and protective factors rather than merely the
overall number of risk and protective factors is examined. Finally, to inform practice,
protective factors are examined to ascertain whether they buffer individual, family, peer,
school, and community risks for both genders, within different race/ethnicity groups, and

for students from rural and urban areas. Specific research questions and hypotheses for

39

the current study are listed below.

Research Questions and Hypotheses
1) Are there gender, urban/rural, or race/ethnicity differences in overall (cumulative)
levels of risk and protection?
Specific hypotheses:
a. Males will have higher levels of overall risk than females, and females will have
higher levels of overall protection than males.
b. Students from the urban school will have higher levels of overall risk and higher
levels of overall protection than students from the rural school.
c. Students of color will have higher levels of overall risk and higher levels of

overall protection than Caucasian students.

2) Are there gender, urban/rural, or race/ethnicity differences in levels of risk and

protection within individual, family, peer, school, and community domains?

3) Does a cumulative or a domain-specific model of risk and protection better predict
delinquency?
a. It is hypothesized that a domain-specific model of risk and protection will better

predict delinquency than a cumulative model of risk and protection.

40

4) Which domains of protective factors buffer the effects of overall, individual, family,
peer, school, and community risks on delinquency? Which protective factors buffer
overall risk for each gender, for students from urban versus rural areas, and for students

of color versus Caucasian students?

41

Chapter 2
METHOD
Development and Administration of Surveys

The Youth Survey was developed for the purpose of a risk and needs assessment
conducted in an urban and a rural site in mid-Michigan by the Youth Violence Prevention
Coalition (YVPC). The YVPC is a collaborative body charged with the responsibility of
advising the city and county on policies and programs concerning juvenile justice. The
YVPC consists of representatives from city and county government, schools, police, the
prosecutor’s office, the Family Independence Agency, and other child and youth welfare
organizations, as well as youths themselves.

Although copyright considerations prohibited utilizing existing surveys and
scales, YVPC members reviewed popular risk reduction and protective factor
enhancement models as well as risk and protective factor research to distill factors that
are recurrent in both the research literature and in these commonly used models. Sin-vey
questions were designed to measure risk and protective factors in the following five
domains: individual, family, peer, school, and community. First, other published or
locally available surveys were examined to find relevant items for each of the risk and
protection scales. When the number of items found in already existing surveys was
insufficient to measure a domain, new items were generated by YVPC members and
representatives from the two participating sites.

The survey was then tested with a sample of 20 students from the rural site to
ensure that the items were easy to understand and that the survey could be completed in

less than 45 minutes. Finally, a larger pilot study was conducted during Spring 1997 with

42

108 students from the urban and the rural sites. Letters and permission slips were sent to
all 6'h grade students and their parents to recruit participants for the pilot survey. A
YVPC representative visited the 6th grade lunch hour and made announcements over the
PA. system at both sites to remind students to turn in their permission slips. Students
completed their smveys during school hours and were paid $5 for their time.

After the pilot survey was administered, survey items were combined into scales
measuring risk and protective factors, and reliability statistics (e.g., item-scale
correlations) were used to eliminate unnecessary items fi'om the survey. The final survey,
administered in 1998 and 1999 to 6‘“, 7‘”, and 8‘“ graders in the urban and rural schools,
consisted of approximately 220 items and is contained in Appendix A.

SEES.

The two sites participated in the Youth Survey as part of a larger community
assessment project of the Youth Violence Prevention Coalition. The urban site consisted
of a large middle school with a diverse population. This middle school had gained a local
reputation for being unsafe and having a large number of students engaging in delinquent
and violent acts. School administrators and staff were interested in working with the
YVPC to assess the extent and nature of any problems they might have. The rural site
had a homogeneous population, and was the only middle school located in a rural
township within the same county as the urban site.

Participants
Participant demographics for the 1998 and 1999 years of survey administration, as

well as for the combined sample analyzed in this study, are shown in Table 1.

43

Table 1

Survey Participant Demographics

 

 

Participant 1998 1999 Combined Sample
Information (N=277) (N=245) (N=452)
Urban Site N=161 (58%) N=134 (55%) N=258 (57%)
Female 91 (57%) 74 (56%) 143 (55%)
Grade Level 6th — 55 (35%) 6th — 50 (3 7%) 6‘h — 92 (36%)
7"‘—57 (36%) 7'“—36 (27%) 7‘“-74 (29%)
8th — 44 (28%) 8‘” — 48 (36%) 8‘“ - 89 (34%)

Race/Ethnicity Afr. Amer— 43 (28%) Afi. Amen-28 (21%)
Caucasian- 70 (45%) Caucasian- 75 (56%)
Latino(a)-15 (10%) Latino(a)-10 (7%)
Asian Pac.-9 (6%) Asian Pac.-10 (8%)
Nat. Amen-2 (1%) Nat. Amen-1 (1%)
Multiracial-15 (9%) Multiracial-10 (7%)

Afi. Amen-65 (25%)
Caucasian- 124 (48%)

‘ Latino(a)—25 (10%)

Asian Pac.-17 (7%)
Nat. Amen-3 (1%)
Multiracial-21 (8%)

Rural Site N=115 (42%) N=111 (45%) N=194 (43%)
Female 59 (51%) 56 (51%) 97 (50%)
Grade Level 6th — 29 (25%) 6‘“ — 39 (35%) 6‘“ — 59 (30%)

7th — 37 (32%) 7‘h — 32 (29%) 7th — 51 (26%)
8m—42 (37%) 8th —40 (36%) 8‘“— 81 (42%)
Race/Ethnicity 93% Caucasian‘ 92% Caucasian] 92% Caucasian'

 

' Percentages of non-Caucasian students are not provided for the rural school survey

participants to protect the identity of the students.

In 1998, 277 students took the survey; 161 (58%) were from the urban site and
115 (42%) were from the rural site. In the urban sample, 57% were female, and 28%
were Afiican American, 45% Caucasian, 10% Latino or Hispanic, 6% Asian Pacific, 1%
Native American, and 10% Multiracial. Thirty-five percent of the urban sample were in
6th grade, 36% were in 7‘h grade, and 28% were in 8‘” grade. In the rural sample, 51%
were female, and 25% were in 6th grade, 32% were in 7‘“ grade, and 37% were in 8til
grade. Most (93%) of the rural sample were Caucasian.

In 1999, 245 students completed the survey; 134 (56%) were from the urban site
and 111 (45%) were fi'om the rural site. In the urban sample, 56% were female, and 21%
were Afiican American, 56% Caucasian, 7% Latino or Hispanic, 8% Asian Pacific, 1%
Native American, and 7% Multiracial. Thirty-seven percent of the urban sample were in
6til grade, 27% were in 7til grade, and 36% were in 8‘” grade. In the rural sample, 51%
were female, and 35% were in 6th grade, 29% were in 7th grade, and 36% were in 8th
grade. Again, most (92%) of the rural sample were Caucasian.

The 1998 and 1999 survey administrations were combined into one sample for
this study. The most recent (1999) survey was retained for students who participated in
the survey during both years. The combined sample consisted of 452 students; 258 (57%)
were from the urban site and 194 (43%) were from the rural site. In the urban sample,
57% were female, and 25% were African American, 48% Caucasian, 10% Latino(a) or
Hispanic, 7% Asian Pacific, 1% Native American, and 8% Multiracial. Thirty-six
percent of the urban sample was in 6th grade, 29% were in 7th grade, and 34% were in 8th
grade. In the rural sample, 50% were female, and 30% were in 6th grade, 26% were in 7th

grade, and 42% were in 8th grade. Most (92%) of the rural sample were Caucasian.

45

Table 2 compares demographic information for the survey participants with the

demographic characteristics of the entire urban and rural schools. In terms of gender,

Table 2

Comparison of Survey Participants with all Students in Urban and Rural Schools

 

 

Urban School All Urban Rural All Rural
Survey School School School
Participants Students Survey Students
Participants
Grade 6th — 36% 6‘” - 32% 6* — 30% 6‘” — 32%
Level 7‘“ — 29% 7th — 36% 7'“ — 26% 7‘“ — 37%
8'“—34% 8‘“—31% 8“'—42% 8‘“—31%
Gender 55% Female 50% Female 50% Female 47% Female
Race! 25% Afr. Amer 28% Afr. Amer. 92% were .9% Afr. Amer.
Ethnicity 48% Caucasian 46% Caucasian Caucasian’ 96% Caucasian
10% Latino(a) 15% Latino(a) 2% Latino(a)
7% Asian Pacific 9% Asian Pac. .2% Asian Pac.
1% Nat. Amer. 2% Nat. Amer. .3% Nat. Amer.
8% Multiracial

 

' Percentages of non-Caucasian students are not provided for the rural school survey participants to protect

the identity of the students.

grade, and race/ethnicity, the students sampled for the survey closely resembled the

populations of their schools. Females and 8m-graders were slightly over-sampled and 7th

graders were under-sampled in both the urban and rural sites (see Table 2), but the survey

participants are otherwise representative of the school populations.

46

Independent VaLabks;

The risk and protective factor scales measured by the YVPC Youth Survey
encompassed multiple domains; they are listed in Table 3 and described in the paragraphs
below. Indicators of internal consistency are based on the combined sample of 1998 and
1999 survey administrations. Psychometric information is listed in Table 4 for the risk
factor scales and Table 5 for the protective factor scales. All survey questions are shown
in Appendix A.

Risk Factors

Individual risk factors. Within the individual level domain, the risk factor of
Early Initiation of Problem Behaviors was assessed with 8 items that were scored on a
scale of 1 (never) to 5 (younger than 10). Youths were asked how old they were when
they first engaged in delinquent or violent behaviors such as fighting, skipping school or
carrying a weapon. Item-total correlations ranged from .24 to .71, and Cronbach’s alpha
was .80.

Poor Conflict Resolution Tactics were assessed with 3 items that were scored on a
4-point Likert-type scale, with higher scores indicating frequent use of poor conflict
resolution tactics. Youths were asked how often they used tactics such as yelling,
refusing to talk, or throwing things in response to conflicts with their friends. Item-total

correlations ranged from .49 to .57, and Cronbach’s alpha was .71.

47

Table 3

Risk and Protective Factors Measured by the YVPC Youth Survey

 

 

Domain Risk Protection
Early Initiation of Problem Independence
Individual Behavior
Self-Esteem
Poor Conflict Resolution Tactics
Family Management Problems
Role Models in the Home
Poor Family Supervision
Positive Parenting Behaviors
Family Family Conflict and Violence
Family Care and Social Support
Parent Attitudes toward Child
Problem Behaviors
Parent Participation in Violence
Peer Attitudes toward Prob. Behav. Peer Support for Non-Violence
Peer Peers Involved in Problem Behav. Conventional Friends
Academic Failure School Role Models/Mentors
Lack of Commitment to School Perceived Care and Social
School Support at School
Early and Persistent Anti-Social
Behavior at School Teacher Expectations of Success
and Achievement
Opportunities for Participation &
School Involvement
Availability of Firearms Sense of Community
Availability of Drugs Positive Community Orientation
Community toward Youth/Adult Role Models
Community Disorganization
Positive Police Presence in the
Experience of Racism Community

 

48

Table 4

 

 

Risk Factor Psychometrics

Risk Factor Number Item-Total Alpha
of Items Correlations

Individual
Early Initiation of Problem Behaviors 8 .24 - .71 .80
Poor Conflict Resolution Tactics 3 .49 - .57 .71
Family
Family Management Problems 6 .27 - .56 .67
Poor Family Supervision 8 .40 - .69 .83
Family Conflict and Violence 5 .35 - .65 .77
Parent Attitudes toward Child 11 .59 - .68 .88
Problem Behaviors
Parent Participation in Violence 9 .38 - .75 .84
Peer
Peer Attitudes Toward Problem 7 .54 - .76 .88
Behaviors
Peer Involvement in Problem 8 .55 - .77 .88
Behaviors
School
Academic Failure 3 N/A N/A
Lack of Commitment to School 8 .38 - .60 .77
Early and Persistent Anti-Social 11 .38 - .74 .82
Behavior at School
Perceived Lack of Safety in School 4 .59 - .66 .80
Community
Availability of Firearms 3 .50 - .71 .77
Availability of Drugs 6 .46 - .85 .89
Community Disorganization 6 .54 - .65 .82
Experience of Racism 4 .33 - .74 .76

 

49

Table 5

Protective Factor Psychometrics

 

 

Protective Factor Number of Item-Total Alpha
Items Correlations

Individual

Independence 7 .38 - .63 .80

Self-Esteem 3 .34 - .45 .57

Family

Family Role Models 3 .40 - .44 .62

Positive Parenting Behaviors 3 .56 - .63 .75

Family Care and Social Support 6 .58 - .71 .86

Peer

Peer Support for Non-Violence 4 .41 - .61 .72

Conventional Friends 5 .63 - .75 .86

School

Role Models or Mentors at School 3 .56 - .65 .77

Perceived Teacher Care and Social Support 3 .52 - .58 .73

Teacher Expectations of Success and 5 .24 - .51 .67

Achievement

Opportunities for Participation 3 .28 - .42 .55

School Involvement 4 N/A N/A

Community

Sense of Community 4 .62 - .77 .85

Positive Orientation Toward Youth and
Adult Role Models

Positive Police Presence in Community

Composite Measure

2 positive orientation items correlated

.55; 1 yes/no item

4

.53 - .64

 

50

Family risk factors. The family domain risks chosen for this study include Family
Management Problems, Poor Family Supervision, Family Conflict and Violence, Parent
Attitudes toward Child Problem Behaviors, and Parent Participation in Violence. Family
Management Problems were measured with 6 items that assessed the extent to which
youths experienced transitions or instability due to problems within their families. For
example, youths were asked how often they stayed in shelters or stayed with relatives
because their parents were having problems. Because this scale demonstrated low
internal consistency (Cronbach’s alpha was .60) in the 1997 pilot study, the items were
revised for subsequent survey administrations in 1998 and 1999. Item-total correlations
ranged from .27 to .56 and Cronbach’s alpha was .67 for the scale administered in the
1998 and 1999 surveys.

Poor F amily Supervision consisted of 8 items assessing level of family
supervision and consistent discipline. Responses were measured on a scale of l to 4, with
higher scores indicating low levels of supervision. This scale included questions about
whether or not parents consistently enforce clear rules for behavior and whether or not
they ask youth questions about schoolwork, their friends, and where they are going. Item-
total correlations ranged from .40 to .69 and Cronbach's alpha was .83.

Family Conflict and Violence were assessed with 5 questions measured on a 4-
point Likert-type scale (higher scores indicated higher conflict and violence). The scale
centered on verbal conflicts and threats as well as actual physical violence between

youths’ parents, and a question focusing on sibling violence was added to surveys

51

administered in 1998 and 1999. Item-total correlations ranged from .35 to .65 and
Cronbach’s alpha was .77.

Parental Attitudes toward Child Problem Behaviors were measured with 11 items
that assessed youths’ perceptions of whether their parents would accept them
participating in delinquent or aggressive behaviors. These items were scored on a 4-point
Likert-type scale, with higher scores indicating higher acceptance of delinquency and
aggression. Item-total correlations ranged from .59 to .68 and Cronbach’s alpha was .88.

Parent/Family Participation in Violence consisted of 9 items that assessed how
often youths witnessed their parents or other family members get into physical
confiontations or fights. The items were scored on a 4-point Likert-type scale, item-total
correlations ranged from .38 to .75, and Cronbach’s alpha was .84.

Peer risk factors. Risks assessed within the peer domain included Peer Attitudes
toward Problem Behaviors and Peer Involvement in Problem Behaviors. Peer Attitudes
toward Problem Behaviors were measured by 7 items scored with a 4-point Likert-type
response format. Youths were asked if their friends would think “it is alright” to engage
in behaviors such as skipping school, getting into a fight if challenged, or carrying a
weapon. Item-total correlations ranged from .54 to .76 and Cronbach’s alpha was .88.

The Peer Involvement in Problem Behaviors scale consisted of 8 items assessing
how often youths’ friends participated in behaviors such as skipping school, getting into
fights, carrying weapons, and getting into trouble at school. Items were scored with a 4-
point Likert-type scale, item-total correlations ranged from .55 to .77, and Cronbach’s

alpha was .88.

52

School risk factors. The YVPC Youth Survey measured school risks that
included: Academic Failure, Lack of Commitment to School, Early and Persistent Anti-
Social Behavior at School, and Perceived Lack of Safety in School.

Academic failure was measured with three items. Youths were asked about their
overall grades at schools. This item was scored on a 9-point scale, with l=mostly As,
2=mostly As and Bs, and 9=Mostly Es. Youths were also asked how many classes they
have failed and how many times they have had to repeat a grade at school. These items
were scored on a 7-point scale, with O=zero to 6=six or more.

Lack of Commitment to School was measured by 8 items in which youths were
asked how important doing well at school is to them, how important they think school is,
whether they intend to finish high school, how important education is for getting the job
or career they want, and whether they think they will go to college. These items were
scored on a 4-point Likert-type scale, item-total correlations ranged from .38 to .60, and
Cronbach’s alpha was .77.

Early and Persistent Anti-Social Behavior at School was measured with 11 items
asking youth how often they skip classes, get into fights, and get disciplined at school
(e. g., office referrals, detention, suspension, etc.). Items were scored on a 4-point Likert-
type scale, item-total correlations ranged from .38 to .74 and Cronbach’s alpha was .82.

Perceived Lack of Safety in School was measured with 4 items asking youth
whether they fear for their safety in or around school. Items were scored with a 4-point
Likert-type response format, item-total correlations ranged from .59 to .66, and

Cronbach’s alpha was .80.

53

Commu_nifl risk_f_a_<_:tgr_s_. Finally, the survey assessed the community risks of
Availability of Firearms, Availability of Drugs, Community Disorganization, and
Experience of Racism.

To assess Availability of Firearms, youths were asked 3 questions about whether
they could easily obtain a gun in their neighborhood and how often people in their
neighborhood carry weapons. All items were scored on a 4-point Likert-type scale, item-
total correlations ranged from .50 to .71, and Cronbach’s alpha was .77.

Availability of Drugs was measured by 6 items in which youth were asked how
often people use, buy or sell drugs in their neighborhood. These items were scored on a 4-
point Likert-type scale, item-total correlations ranged from .46 to .85 and Cronbach’s
alpha was .89.

Community Disorganization was measured by 6 items, such as “There are many
abandoned houses or other buildings in my neighborhood or community” and “There is a
lot of writing and other graffiti on the houses and other buildings in my neighborhood or
community.” These items were scored on a 4-point Likert-type scale and were drawn
from a Sense of Community Questionnaire developed by John Schweitzer and used in
research at Michigan State University. Item-total correlations ranged from .54 to .65 and
Cronbach’s alpha was .82.

Experience of Racism was measured by 4 items that were adapted from McCord
and Ensminger’s (1995) longitudinal research on pathways from childhood aggression to
adult violence. Youths were asked how often they had experienced problems With

teachers or had police officers or other people bothering them in their neighborhood

54

because of their race or ethnicity. Items were scored with a 4-point Likert-type scale,
item-total correlations ranged from .33 to .74, and Cronbach’s alpha was .76.
Protective farctors

Individuflotective factors. Individual-level protective factors measured on the
YVPC Youth Survey were Independence and Self-Esteem. The Independence scale
consisted of 7 items asking youth whether they feel comfortable voicing their opinions
and whether they think it is okay to be different from other kids. Responses were scored
on a 4-point Likert-type scale, item-total correlations ranged from .38 to .63, and
Cronbach’s alpha was .80.

Self-Esteem was measured with 3 items asking youth whether they are “able to do
things as well as others” or whether they are generally able to do whatever they set their
minds to do. Responses were scored on a 4-point Likert—type scale, item-total
correlations ranged from .34 to .45, and Cronbach’s alpha was .57.

F_amilv protective factors. Family protective factors measured on the survey
included Family Role Models, Positive Parenting Behaviors, and Family Care and Social
Support. Family Role Models was measured by 3 questions measured on a 4-point Likert-
type scale in which youth were asked if they have parents, siblings, or other relatives that
they admire and look up to. Item-total correlations ranged fi'om .40 to .44, and
Cronbach’s alpha was .62.

Positive Parenting Behaviors were measured by a 3-item scale asking youth how
ofien their parents reward or praise their good behavior or calmly discuss their
misbehavior with them. Items were scored on a 4-point Likert-type scale, item-total

correlations ranged from .56 to .63, and Cronbach’s alpha was .75.

55

The Family Care and Social Support scale consisted of 6 questions asking youth if
they are able to talk with their parents about problems and if their parents listen to them
and spend time with them. The items were scored on a 4-point Likert-type scale, with
higher scores indicating higher care and social support. Item-total correlations ranged
from .58 to .71 and Cronbach’s alpha was .86.

Peer protective fagors. Peer protective factors measured by the youth survey
included Peer Support for Non-Violence and Conventional Friends.

Peer Support for Non-Violence was assessed by 4 items in which youth were
asked whether their friends would support them or think they were afraid if they walked
away from a fight. The items were scored on a 4-point Likert-type scale, item-total
correlations ranged fi'om .41 to .61, and Cronbach’s alpha was .72.

Conventional Friends consisted of a 5-item scale that measured the extent to
which parents and teachers approve of youths’ friends. Youths were asked questions
about whether or not their parents and teachers think their fiiends are “good kids,” and if
their friends are interested in school and get good grades. The items were scored on a 4-
point Likert-type scale, item-total correlations ranged from .63 to .75, and Cronbach’s
alpha was .86.

School protective factors. School protective factors included Role Models or
Mentors at School, Perceived Care and Social Support at School, Teacher Expectations of
Success and Achievement, Opportunities for Participation, and School Involvement.

The Role Models or Mentors at School scale consisted of 3 questions that

assessed whether youths have teachers that they trust, admire, and respect. Items were

56

scored on a 4-point Likert-type scale, item-total correlations ranged from .56 to .65, and
Cronbach’s alpha was .77.

Perceived Care and Social Support at School were measured by a 3-item scale that
included questions about whether teachers seem to care about students and help them
when they have problems. The items were measured on a 4-point Likert-type scale, item-
total correlations ranged from .52 to .58, and Cronbach’s alpha was .73.

Teacher Expectations of Success and Achievement were measured by a 5-item
scale asking youths how many of their teachers expect that they will do well at school.
These items were scored on a 4-point Likert—type scale (1=none and 4=all), item-total
correlations ranged fiom .24 to .51, and Cronbach’s alpha was .67.

Opportunities for Participation were measured by 3 questions about how many
opportunities there are for youths to participate in school activities. The items were
scored on a 4-point Likert-type scale, item-total correlations ranged from .28 to .42, and
Cronbach’s alpha was .55.

School Involvement was measured by 4 items asking youth about the number of
school activities in which they were involved and about their level of participation in
school. This scale was scored as the total number of activities.

Community pmtecfivcflom. Community-level protective factors assessed on
the YVPC Youth Survey included Sense of Community, Positive Community Orientation
toward Youth and Adult Role Models, and Positive Police Presence in Community.

A 4-item scale measured Sense of Community. Examples of items included “It is
easy for me to tell a stranger in my neighborhood or community from somebody who

lives there” and “Neighbors take care of each others’ plants, pets, or children if needed.”

57

These items were scored on a 4-point Likert-type scale and were drawn from a Sense of
Community Questionnaire developed by John Schweitzer and used in research at
Michigan State University. Item-total correlations ranged from .62 to .77, and
Cronbach’s alpha was .85.

Positive Community Orientation toward Youth and Adult Role Models consisted
of a yes/no item asking youths if there are adults in their neighborhood that they admire
or would like to be like, and 2 questions asking youth whether adults in their
neighborhood know the names of kids who live there and seem to like them. The 2 items
were measured on a 4-point Likert-type scale and were correlated .55.

Positive Police Presence in the Community was measured with 4 items asking
youth whether they and the adults in their community trusted the police they saw in their
neighborhood. Item-total correlations ranged fi'orn .53 to .64, and Cronbach’s alpha was
.76.

Dependent Variables

Delinquency was measured by a modified version of the Self-Reported Delinquency
Index (SRD) that was developed for the National Youth Survey (Elliott, Huizinga, &
Ageton, 1985) and has been used extensively in delinquency research (e.g., Smith &
Thomberry, 1995; Smith et al., 1995; Thomberry, Lizotte, Krohn, Famworth, & Jang,
1991). The original scale consists of 21 self-report items assessing participation in a
range of delinquent behaviors, such as stealing, vandalism, arson, and using drugs.
Youths were asked to respond with the exact number of times they participated in these
behaviors during the last year. The index was shortened to 16 items for this study; item-

total correlations ranged from .29 to .70 and Cronbach’s alpha was .82.

58

Chapter 3
RESULTS

This chapter presents the findings of the current study. First, procedures of
analysis are delineated. Second, information is provided concerning the dependent
variable, youth participation in delinquency. Next, information about the independent
variables, risk and protection, is detailed. Finally, the results of the study are presented in
order of the four research questions.

Procedures of Analysis

Each student’s standardized scores were averaged across scales to compute risk
and protective factor scores for each domain (e.g, individual, family, peer, school, and
community), and then risk and protective scores were averaged across domains to
compute an overall or aggregate risk score and an overall protective score.

Data for research questions 1, 2, and 4 were analyzed using SPSS for windows. T-
tests were used to compare overall and domain-specific (e.g., individual, family, peer,
school, and community) risk and protection by gender, urban/rural, and race/ethnicity. T-
tests by race/ethnicity were computed using only students from the urban site to avoid
confounding the effects of urban/rural and race/ethnicity. For purpose of analysis,
race/ethnicity was recoded into a dichotomous variable (student of color or Caucasian).
On the basis of research showing that pairwise rather than listwise deletion may be
advantageous in terms of bias and efficiency (Arbuckle, 1996), analyses were computed
with pairwise deletion to account for missing data Graf and Alf’s (1999) method and
software for comparing R—squares with non-overlapping predictors were used to answer

research question 3.

59

To answer research question 4, multiple regression analysis was first used to
examine the ability of protective factors within each domain (protective factors were
grouped by domain and entered in blocks) to moderate the effects of overall risk
(standardized scores averaged across all risk scales) on delinquency, with gender, race,
and urban/rural entered as controls. Gender, race/ethnicity, and urban/rural were entered
in block one, followed by individual, family, peer, school, and community protection in
block two. Block three contained overall risk, and block four was comprised of the
interactions between overall risk and each of the protective factors.

Second, five separate regressions were computed to examine which domains of
protective factors moderate the effects of individual, family, peer, school, and community
risks on delinquency. In other words, a regression equation was computed to examine the
ability of the protective factors to buffer individual risks; a separate analysis examined the
protective factors that buffer family risks, then peer risks, etc. Gender, race/ethnicity, and
urban/rural were entered in block one as control variables, followed by individual, family,
peer, school, and community protection in block two. Block three contained individual,
family, peer, school, g community risk, and block four was comprised of the interactions
between individual, family, peer, school, or community risk and each of the protective
factors.

Finally, regression analysis was used to examine which domains of protective
factors buffer overall risk for females versus males (controlling for urban/rural and
race/ethnicity), urban versus rural students (controlling for gender and race/ethnicity), and
for students of color versus Caucasian students (controlling for gender and urban/rural).

Control variables were entered in block one, followed by gender, urban/rural, or

60

race/ethnicity; individual, family, peer, school, or community protection; and overall risk
in block two. Block three contained the two-way interactions of gender, urban/rural, or
race/ethnicity by individual, family, peer, school, or community protection; gender,
urban/rural, or race/ethnicity by overall risk; and individual, family, peer, school, or
community protection by overall risk. The fourth block contained the three-way
interaction of gender, urban/rural, or race/ethnicity by individual, family, peer, school, or
community protection, by overall risk. In total, fifteen multiple regression analyses were
computed to answer the final component of research question four.

In all multiple regression analyses, independent variables were centered to reduce
multicollinearity and increase interpretability of the predictor variables. As
recommended by Aiken and West (1991), interaction terms were formed by computing
the products of standardized variables and interpreted by examining the unstandardized
solution and plotting effects at one standard deviation above and below the mean.

Although most of the regression analyses build on each other in sets that repeat
the same main effects in the initial blocks, a large number of interactions were tested in
this study. Thus, modified Bonferroni corrections were applied to t-tests and regression
analyses to limit the inflation of alpha due to repeated testing while preserving a
reasonable level of power (Jaccard & Wan, 1996). The modified Bonferroni procedure
still retains a family-wise type 1 error rate of 5%. For this procedure, analyses may be
grouped into “families” of effects that have something in common for adjustment of
alpha. In this study, effects were grouped into families according to type of analysis and
domain of risk and protection (see Appendix B). Significance values from the multiple

tests were rank ordered within each family from smallest to largest. Within each family,

61

the significance of the first test was evaluated at alpha/number of tests. If the test statistic
was statistically significant after this adjustment was performed, the next test was
evaluated at alpha/(number of tests-l). Ifthis test was significant, the third test was
evaluated at alpha/(number of tests-2). Each test was evaluated in this fashion until a
non-significant test result was obtained. All tests after the first non-significant test were
deemed not significant. Tables showing the application of the modified Bonferroni
procedure to the analyses in this study are contained in Appendix B.
Dependent Variable: Youth Involvement in Delinquency

Students’ scores on the modified Self-Reported Delinquency Index ranged flour 0
to 72, and the mean score was 3.62. To correct for the highly skewed distribution
(skewness=4.6; kurtosis=25.62) and better meet the distributional assumptions of
multiple regression, a linear log transformation was applied to the scale. The resulting
distribution was less skewed (skewness=1 .20; kurtosis=.555).

Independent Variables: Youth Experience of Risk_and Protection

Tables 6 and 7 contain the means, standard deviations, skewness, and kurtosis
statistics for each of the risk and protective factor scales. Risk and protection scales were
scored so that lower scores indicate lower risk or protection and higher scores indicate
higher risk or protection (e.g., most scales were scored 1=not at all, 4=a lot). Skewed
distributions among the independent variables were transformed to better approximate
normality because they were used to form interaction terms; researchers recommend
transformation as a strategy in this instance because interactions magnify the effect of
extreme scores (e.g., Aiken & West, 1991). Accordingly, a linear log transformation was

applied to the family management and family violence scales to correct for skewed

62

distributions.

Tables 6 and 7 show that mean protective factor scores are generally higher than
mean risk factor scores, and the highest levels of protection are found in the individual
domain and the lowest levels of protection are found in the community domain. All of
the risk factor scales had mean scores ranging from 1 (not at all) to 2 (a little). Some of
the highest risks for the student survey- participants were the individual-level domain risk
factors of Early Initiation of Problem Behaviors and Poor Conflict Resolution Tactics, the
school domain risk of Lack of Comrrritrnent to School, and the community domain risk of
Availability of Drugs. Some of the lowest reported risks were the family domain risk of
Parent Participation in Violence and the school domain risk of Early and Persistent Anti-
Social Behavior in School.

Tables 8- 12 show the correlations of the dependent variable, delinquency, with the
risk and protective factor scales that were used to form the risk and protective factor
domain scores, and Table 13 contains correlations between overall risk and protection
and among risk and protective factor domain scores. Correlations among the risk and
protection scales, between overall risk and protection, and among risk and protective
factor domains were examined to determine whether the risk and protective factor scales
and domain scores measured distinct constructs or if they measured a single factor at
opposite ends of a continuum. Cohen’s (1988) standards were used to interpret small

(.10), medium (.30), and large (.50) effect sizes.

63

Table 6

Risk Factor Descriptive Statistics

 

 

Risk Factor N Range Mean SD Skew Kurt
Individual

Early Initiation of Problem 450 1.00-4.75 1.92 .904 .813 -.264
Behaviors

Poor Conflict Resolution 422 1.00-4.00 1.73 .708 1.23 1.25
Tactics

Family

Family Management 446 1.00-3.40 1.31 .375 2.42 7.80
Problems " 446‘ 0.00-1.22‘ll .241“ .236“ 1.40ll 2.33a
Poor Family Supervision 442 1.00-3.88 1.59 .586 1.32 1.76
Family Conflict and Viol. 426 1.00-4.00 1.55 .553 1.79 3.75
Parent Attitudes toward 435 1.00-3.73 1.36 .463 1.89 4.64
Child Problem Behaviors

Parent Participation in 432 1.00-3.56 1.17 .367 3.42 13.43
Violence ‘ 432al 0.00—1.27a .128a .229‘ 2.47‘I 6.48a
Peer

Peer Attitudes Toward 420 1.00-4.00 1.68 .653 .982 1.03
Problem Behaviors

Peer Involvement in 417 1.00-4.00 1.58 .632 1.51 2.25
Problem Behaviors

School

Academic Failure

Lack of Commitment to 450 1.00-3.75 1.80 .501 .920 .999
School

Early and Persistent Anti- 447 0.00-7.60 1.09 1.40 1.80 3.23
Social Behavior at School

Perceived Lack of Safety in 450 1.00-4.00 1.73 .665 1.00 .825
School

Community

Availability of Firearms 404 1.00-4.00 1.60 .703 1.35 1.35
Availability of Drugs 408 1.00-4.00 1.86 .804 .879 -.098
Community Disorganization 414 1.00-4.00 1.65 .621 1.06 .882
Experience of Racism 359 1.00-4.00 1.28 .508 2.37 6.25

 

‘ Scale descriptives after linear log transformation

Table 7

Protective Factor Scale Descriptive Information

 

 

Protective Factor N Range Mean SD Skew. Kurt.
Individual
Independence 450 1.43-4.00 3.33 .545 -1 .04 .810
Self-Esteem 446 2.00-4.00 3.49 .483 -.789 .044
Family
Role Models in the Home 434 1.00-4.00 2.78 .759 -.300 -.375
Positive Parenting Behaviors 440 1.00-4.00 3.14 .745 -.814 .147
Family Care and Social 447 1.00-4.00 3.30 .674 -1.11 .717
Support
Peer
Peer Support for Non- 451 1.00-4.00 3.32 .688 -1.09 .712
Violence
Conventional Friends 423 1.00-4.00 3.10 .717 -.667 -.061
School
Role Models or Mentors at 450 1.00—4.00 2.85 .790 -.307 -.764
School
Perceived Teacher Care and 451 1.00—4.00 2.98 .706 -.513 -.264
Social Support
Teacher Expectations of 451 1.20-4.00 3.1 1 .498 -.352 -.023
Success and Achievement
Opportunities for Participation 450 1.00—4.00 3.25 .686 -.877 .208
School Involvement 45 1 -1 .67- -.002 .823 -.028 -.332
(z-score) 1.97

, Community
Sense of Community 410 1.00-4.00 2.87 .852 -.442 -.685
Positive Orientation Toward 41 1 1.00-4.00 2.81 .884 -.271 -.823
Youth and Adult Role Model
Positive Police Presence 408 0.00-1.00 .790 .331 -1.32 .339

 

65

Table 8 contains the correlations between delinquency and individual risk and
protection scales. Within the individual domain, risk and protection were negatively
related and had small, non-significant correlations. Risk and protection scales within the
family domain (see Table 9) were negatively related and had mostly low to medium
significant correlations. However, Poor Family Supervision was a notable exception with
a moderate negative correlation with Family Role Models (-.377) and high negative
correlations with Positive Parenting (-.614) and Family Care and Social Support (-.761).
Within the peer domain (see Table 10), risk and protection scales had mostly medium-
sized significant negative correlations. However, Conventional Friends had medium to
high correlations with Peer Attitudes toward Problem Behaviors (-.450) and Peer
Involvement in Problem Behaviors (-.448). Within the school domain (see Table 11),
risk and protection scales had mostly low to moderate negative correlations. The highest
correlations were between Lack of Commitment to School and School Role
Models/Mentors (-.583) and Teacher Care and Social Support (-.427). Risk and
protection scales within the community domain had low to moderate significant negative
correlations (see Table 12. The highest correlations were in the medium range and
involved the relationship of Positive Police Presence in the Community with Availability

of Firearms (-.397) and Availability of Drugs (-.352).

66

Table 8

Individual Risk and Protection Scale Correlations with Delinquency

 

 

Protection Risk
Independence Self- Early Poor Delinquency
Esteem Initiation Conflict
of Resolution
Problem Tactics
Behaviors
Protection .
Independence .481 * "‘ -.059 -.094 -.1 15
Self-Esteem -.025 -.068 -.136**
Risk
Early
Initiation of .153" .609"
Problem
Behaviors
Poor Conflict .230”
Resolution
Tactics

 

” Correlation is significant at the 0.01 level (2-tailed). "' Correlation is significant at the 0.05 level (2

tailed).

67

Table 9

Family Risk and Protection Scale Correlations with Delinquency

 

Protection

Risk

 

F am.
Pos. Care F am.
Par. & Soc. Man.
Supp. Probs.

Poor

Super.

Family
Family Conflict
& Vio.

Parent
Attitude-
Child
Beh.

Parent
Partic.
in Vio.

DelincL

 

Protect.
Family .419* .459" -.069
a:

Role
Models

Positive .670** -.O65
Parent.

Pam Care -.060
& Social

Support

Risk
Family
Manag. -
Probs

Poor
Family
Superv.

Family
Conf. &
Violence

Parent
Attitudes
-Child
Beh.

Parent
Vio.

a377**

n614**

a76l**

.085

a189**

a180**

«296**

.222**

.222**

a140**

a240‘*

n292**

.227**

.380"

.367**

a156**

a226‘*

m265**

.361**

.297**

.549**

.484**

al7l**

-.201u

a342**

.094

.340**

.221*‘

.405“I

.273**

 

” Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2

unkd)

68

Table 10

Peer Risk and Protection Scale Correlations with Delinquency

 

 

 

Protection Risk
Peer Conventional Peer Peer Delinquency
Support for Friends Attitudes Involvement
Non— Toward in Problem
Violence Prob. Beh. Behaviors

Protection
Peer Support .335M -.301** -.246** -.162**
for Non-Viol.
Conventional -.450** -.448** -.338**
Friends
Risk
Peer Attitudes .726* * .451 * ’*
Toward
Problem
Behaviors
Peer .510"
Involvement
in Problem
Behaviors

 

** Correlation is significant at the 0.01 level (2-tailed). "‘ Correlation is significant at the 0.05 level (2
tailed).

69

Table 11

School Risk and Protection Scale Correlations with Delinquency

 

Protection Risk

 

Tch Tch Opp. Sch. Acad. Lack Early Lack Delinq
Care Exp. for Invol Fail. Sch. Anti- Sch.
Part. Com. Soc Safe
Beh.

 

Protect
School .692* .503“ .153* .094* -.139** -.583** -.330** -.035 -.329**
a:

Mentor * *

Teach 559* .l52"‘ .057 —.02l -.427** -.247’MI -.054 -.308**
Care * “

Teach .192* .145* -.ll9* -.344** —.180"‘* -.056 -.230**
Expt

Oppfor 581* -.245** —.333** -.067 -.102"‘ -.042
Partic. "‘

School
Involv, -.242"‘* -.203** -.046 -.074 .009

Risk
Acad. .251 "
pan. .267** .323** .183*

Lack

Com. to
Sch. .360** .204* .359**

Soc .088 .7oo*r

Sch.

 

" Correlation is significant at the 0.01 level (2-tailed). " Correlation is significant at the 0.05 level (2
tailed).

7O

Table 12

Community Risk and Protection Scale Correlations with Delinquency

 

Protection

Risk

 

Positive Positive
Comm. Police
Orient. Presence
to in
Youth Comm.

Comm.
Disorg.

Avail. of Avail.
Firearms of
Drugs

Exper.
of
Racism

Delinq

 

Protection
Sense of
Com.

Positive
Comm.
Orient.
Toward
Youth

Positive
Police
Presence
in Comm.

Risk
Avail. of
Firearms

Avail. of
Drugs

Comm.
Disorg.

Exper. of
Racism

.789" .268"

.283"

-.304** -.256** -.280"”"

-.273** -.259"‘* -.233"‘*

-.397** -.352"‘* -.281 **

558“ .496"

.621"

-.119*

-.104

-.299**

.280“

.204”

.132“

-.068

-.078

-.413”

.410”

.369”

.232"

.341”

 

" Correlation is significant at the 0.01 level (2-tailed). " Correlation is significant at the 0.05 level (2

tailed).

71

Table 13

Risk and Protective Factor Domain Correlations

 

Overall Individual Family Peer School Community

 

Risk Risk Risk Risk Risk Risk
Overall -.562** -.266** -.445** -.409" -.564** -.432**
Protection
Individual -.307** -.10 -.253** -.102* -.412** -.235**
Protection
Family -.454* * -.175 * * -.475* * -.340** -.374** -.314* *
Protection
Peer -.525* * -.319** -.400* * -.475"‘* -.486** -.344"‘ *
Protection
School -.372** -.181** -.251** -.315*"' -.438** -.242**
Protection
Community -.394** -.247** -.256** -.218** -.330** -.439**
Protection

 

" Correlation is significant at the 0.01 level (2-tailed). " Correlation is significant at the 0.05 level (2-
tailed).

72

For the most part, risk and protection scales had low to moderate negative
correlations, indicating that although the scales within each domain are related they
measure distinct constructs. One exception is that the very high correlation between Poor
Family Supervision and Family Care and Social Support signifies a high level of overlap
between these constructs. However, once the family risk and protection scales were
combined into farme risk and protection domain scores (see Table 13) the correlation
between family risk and protection was less extreme, which suggests that the overall
family risk and protection domain scores measured distinct constructs.

The correlation between overall risk and overall protection was -.562 (p<.01), and
the correlations between risk and protection within each domain are lower (individual= -
.10, p>.05; family=-.475, p<.01; peer=-.475, p<.01; school=-.438, p<.01; community=—
.439, p<.01). Most of the correlations between risk and protection are moderate to high
and none are very high, supporting the idea that although risk and protection are related,
the risk and protective factor domain scores measured distinct constructs.

Main effects of specific dom_ai_n_s of protection on delmcmency.

Table 14 shows the intercept, unstandardized coefficients, standardized
coefficients, zero-order correlations, and R2 and adjusted R2 after entry of demographic
control variables and all protective factor domains. Gender had a significant effect on
delinquency, and R2 was significant after entry of the demographic control variables in
block 1. R2 was also significant after all protective factor domains were entered in block
2, and family and school protection had significant main effects on delinquency. Being
male, having lower levels of family and school protection, and having higher levels of

risk were related to higher levels of delinquency.

73

Table 14

Main Eflects of Protection on Delinquency

 

 

 

Unstandardized Standardized
Coefficients Coefficients
Model 2 containing B Std. Beta Zero- R2 Adj.
demographics and Error t Order R2
protective factors Correl.
Constant .186* .084 2.22*
.085" .078"
Block 1.
Demographic
Controls
Gender -.488** .098 -.244 -4.97** -.285
Urban/Rural .107 .106 .053 1.01 .042
Race .033 .113 .016 .298 .024
195" 178"
Block 2. Protective
Factors
Individual
Protection .014 .066 .013 .222 -.148
Family -.195** .066 -.161 -2.95** -.289
Protection
Peer -.O73 .069 -.062 -1.06 -.285
Protection
School Protection -.232** .086 -.151 -2.68** -.278
Community -.101 .067 -.080 -1.52 -.232
Protection

 

"Denotes significant gender differences at 0.01 level (2-tai1ed). ‘Denotes significant gender differences at

0.05 level (2-tailed).

74

Ewing
The results of the study are presented in order of the four research questions.
Are there gender, race/ethnicity, or urban/rural differences in overall levels of riskan_d
protection?

Tables 15 and 16 show the means and standard deviations for overall risk and
protection for males and females. Gender differences were found in overall levels of risk
and overall levels of protection, and the hypothesis that males would have higher levels of
overall risk and females would have higher levels of overall protection was confirmed.

Tables 17 and 18 show the means and standard deviations for overall risk and
protection for students fi'om the urban and the rural school. Urban versus rural differences
were found in levels of overall risk, and the hypothesis that youth hour the urban school
would have higher levels of overall risk than students from the and school was
confirmed. However, support was not found for the hypothesis that urban students would
have higher levels of protection; urban versus rural differences were not found in levels
of overall protection.

Tables 19 and 20 show the means and standard deviations for overall risk and
protection for students of color and Caucasian students. Support was not found for the
hypotheses that students of color would have higher levels of overall risk and protection;
analyses including only students from the urban school district revealed no race/ethnicity

differences for levels of overall risk or levels of overall protection

75

Table 15

 

 

Risk Domains by Gender

Risk N M SD t 4"

Overall
Female 237 -0.102 0.520 4.08‘ 436
Male 201 0.1 18 0.605

Individual
Female 240 ~0.217 0.652 " 6.34* 389
Male 209 0.239 0.844

Family
Female 236 -0.062 0.613 " 2.01t 388
Male 199 0.069 0.729

Peer
Female 227 -0. 151 0.849 3.64‘ 418
Male 193 0.178 0.998

School
Female 240 -0.1 19 0.636 4.18‘ 447
Male 209 0.136 0.653

Community
Female 226 -0.048 0.705 1.55 41 1
Male 1 87 0.067 0.804

 

‘ Bonferroni-adjusted p<.05 (2-tailed).
tBonferroni-adjusted (p<. 10).
'Denotes significant difference in variances.

76

Table 16

Protection Domains by Gender

 

 

Protection N M SD t df

Overall
Female 240 0.05 1 0.541 -2.44* 447
Male 209 -0.077 0.562

Individual
Female 240 -0.030 0.828 .75 446
Male 208 0.031 0.893

Family
Female 239 0.056 0.809 -1.80 440
Male 203 -0.085 0.847

Peer
Female 240 0.205 0.732 ' -6.05 * 401
Male 209 -0.269 0.901

School
Female 240 0.029 0.630 -1 . 14 447
Male 209 -0.041 0.678

Community
Female 222 0.049 0.760 -1.47 405
Male 185 -0.067 0.828

 

‘Bonferroni-adjusted p<.05 (2-tailed).

'Denotes significant difference in variances.

Table 17

 

 

Risk by Urban/Rural

Risk N M SD t df

Overall
Rural 183 -0.066 0.521 -2.11** 439
Urban 258 0.050 0.602

Individual
Rural 194 -0. 164 0.679 " -3.92* 446
Urban 258 0.1 14 0.827

Family
Rural 183 -0.063 0.589 " -1.76 430
Urban 255 0.049 0.734

Peer
Rural 172 -0.086 0.850 " -1.66 399
Urban 251 0.063 0.984

School
Rural 194 -0.058 0.618 -1.69 450
Urban 258 0.047 0.679

Community
Rural 166 0.023 0.722 .368 414
Urban 250 -0.004 0.776

 

“ Bonferroni-adjusted p<.05 (2-tailed). “ Bonferroni-adjusted p<.05 (l-tailed).

"Denotes significant difference in variances.

78

Table 18

Protection by Urban/Rural

 

 

Protection N M SD t df

Overall
Rural 194 -0.010 0.524 -.108 450
Urban 258 -0.004 0.580

Individual
Rural 194 -0.069 0.829 -1 .45 449
Urban 257 0.050 0.885

Family
Rural 188 -0.103 0.824 -2.15t 443
Urban 257 0.067 0.826

Peer
Rural 194 0.046 0.769 1.38 450
Urban 258 -0.065 0.901

School
Rural 194 -0.009 0.630 -.232 450
Urban 258 0.005 0.669

Community
Rural 165 0.1 14 0.692 V 2.50' 408
Urban 245 -0.077 0.846

 

" Bonferroni-adjusted p<.05 (2-tailed).
‘Bonferroni—adjusted (p<. 10). ”Denotes significant difference in variances.

79

Table 19

 

 

Risk by Race/Ethnicity

Risk N M SD t (if

Overall
Caucasian 124 -0.024 0.580 -1 .76 253
Student of Color 131 0.106 0.598

Individual ,
Caucasian 124 -0.013 0.809 -2.29* 253
Student of Color 131 0.223 0.831

Family
Caucasian 123 -0.031 0.698 -1.51 251
Student of Color 130 0.104 0.720

Peer
Caucasian 122 0.042 0.954 -.341 246
Student of Color 126 0.084 1.006

School
Caucasian 124 -0.003 0.685 -.922 253
Student of Color 131 0.075 0.670

Community
Caucasian 122 -0.090 0.750 -1 .67 245
Student of Color 125 0.072 0.773

 

"‘ Bonferroni-adjusted p<.05 (2-tailed).

80

Table 20

Protection by Race/Ethnicity

 

 

Protection N M SD t df

Overall
Caucasian 124 -0.034 0.616" -.841 253
Student of Color 131 0.026 0.531

Individual“
Caucasian 123 -0.1 14 0.955 -2.94* 236
Student of Color 131 0.210 0.783

Family
Caucasian 123 0.038 0.849 -.567 252
Student of Color 131 0.096 0.802

Peer
Caucasian 124 -0.039 0.871 .449 253
Student of Color 131 -0.090 0.935

School
Caucasian 124 -0.049 0.755 " -1.22 229
Student of Color 131 0.053 0.571

Community
Caucasian 119 0.015 0.815 1.60 240
Student of Color 123 -0.158 0.862

 

r Bonferroni-adjusted p<.05 (2-tailed).

vDenotes significant difference in variances.

Are there gender. race/ethnicity. or urban versua rural differences in levels of mm
protection within individua_l,Lamily. school, peer. and community domains?

andea'. Tables 15 and 16 show the means and standard deviations for each of the
risk and protective factor domains for males and females. Significant gender differences
were found for individual, peer, and school risks, as well as peer protection. Males had
higher levels of individual, peer, and school risks. No gender diflemnces were found in
levels of family or community risk.

Females had higher levels of peer protection than did males. No gender
differences were found in levels of individual, family, school, or community protection.

Urban/Rural. Tables 17 and 18 show the means and standard deviations for each
of the risk and protective factor domains for students from the urban and the rural
schools. Urban/rural differences were found in individual risk, with students fi'om the
urban school experiencing higher levels of individual risk than students from the rural
school. Urban/rural differences were also found in community protection, with students
from the rural school having higher levels of community protection than students from
the urban school.

No urban/rural differences were found in levels of family, peer, school, or
community risk, or individual, peer, or school protection.

Race/Ethnicity. Tables 19 and 20 show the means and standard deviations for
each of the risk and protective factor domains for students of color and Caucasian
students. Analyses including only students fi'om the urban school showed race/ethnicity
differences for individual risk and individual protection. Students of color had higher

levels of both individual risk and individual protection.

82

No race/ethnicity differences were found in levels of family, peer, school, or
community risk, or in family, peer, school, or community protection.

Does a cumulative or a domain-specific model of risk_and protection better predict
delinquency?

Table 21 contains the results of two regression equations: one using two
predictors (overall risk and overall protection) for delinquency, and one using ten
predictors (risk and protection scores for individual, family, school, peer, and community
domains). Graf and Ali’s (1999) method and software for comparing squared multiple
correlations with non-overlapping predictors were used to compute R-squares, the
standard error, and confidence limits. Support was found for the hypothesis that a
domain-specific model of risk and protection would better predict delinquency than a
cumulative model of risk and protection. A significantly larger squared multiple
correlation was found for the ten domain-specific set of predictors than for the two
predictor set of overall risk and protection (R-square=0.452 versus 0.364;
difference=.08 8). The confidence interval for the difference was
0.049=<Difference<=0. 127 (SE=0.020).

One concern with the use of Graf and Ali’s method is that it does not adjust for
differences in the number of predictors (i.e., different levels of shrinkage) in the two
equations being compared. However comparison of the difference between the adjusted
R-square and the R-square statistic in an SPSS run of the two and ten predictor models
(see Table 15) reveals that the amount of shrinkage is small (2-predictor model=.004; 10-
predictor model=.014). Thus, we would expect that even if the adjusted R-squares were

used in this analysis, the 95% Confidence Limits would not encompass zero.

83

Azure E2 8... a segue beacons. agenda ”seen: 83$ es. 3... a sesame $035082 6855" 8.88.

 

 

5.. u can. t «a a3.
5." Sen“:
3.. 8.7 $4 .53- an. :35 :26 2.8a of. zone HS. EN; :32 “3.- .
e8. «8.. as. 02.- ”8. at. SN. 3. so: an. 2a.- 0%. so:
use
65am
«.3. 2.0. «.8. as. 98. N8. :5. m8. :3. 8o. ”8. as. as. as. mm
as. a8; a8. 5.- 95. RN. 9%. es. «2.- am. so. may a: 89- a
£80
425
.2.— ..E can can .2.— ..5. sea in ie. .3”—

EeU dam noon Sam >85 :30 .aom nook Bah >85 4250 8.5 via .350
.3388...— Q vim oEoonméma—uen 5:388.— Q «EH Enos

3...: 3.32:-..— 552 Susan."

 

DemawnmuQ 639$ S nomauucenm use 3.35% EmuEeQ 3.2m.» zehuuaeum 36 «Max 298.5% bmSVRe zeagfieb

flu 033.

84

Which protective factors buffer overall risk,_and which protectivefifactors buffer
individual, family, get, and communig risks?

Overall risk. Table 22 displays the unstandardized regressions coefficients (B),
intercept, zero-order correlations, and R2 and adjusted R2 after entry of all independent
variables. The change in R2 was significantly different from zero at the end of each
block. In the final model, gender, overall risk, and the interaction between overall risk
and individual protection significantly predicted delinquency. Being male and having
higher levels of overall risk were related to higher levels of delinquency. The nature of
the interaction between individual protection and overall risk in predicting levels of
delinquency is shown in Figure 1. Individual protection was found to have an
exacerbating effect on overall risk. When overall risk is low, higher levels of individual
protection are associated with lower levels of delinquency. However, when overall risk is

high, higher individual protection is associated with higher delinquency.

85

Table 22

Do Individual, Family, Peer, School, and Community Protection Bufler the Eflect of
Overall Risk on Delinquency?

 

 

Final model B Std. Zero- R2 Adii.
containing all 4 (U nstand. Error t Order R
blocks Coeff.) Correl.
Constant .193 .075 258*
Block 1. .085" .078"
Demographic
Controls
Gender -.402 .084 —4.77** -.285
Urban/Rural .036 .091 .395 .042
Race/Ethnicity -.046 .097 -.471 .024
Block 2. Protective .195” .178"
Factors
Individual .01 1 .057 .202 -. 148
Protection
Family Protection -.000 .062 -.008 -.289
Peer .104 .063 1.65 -.285
Protection
School Protection -.1 18 .076 -1.54 -.278
Community Protection .012 .058 .205 —.232
Block 3. Overall Risk .408" .394"
Overall Risk 1.04 .094 l 1 . 12* * .603
Block 4. Risk x Prot. .427“ .406*
Interactions
Risk*Indiv. Protect. .258 .092 2.79* -.023
Risk‘Family Prot. -.101 .087 -1.16 -.l96
Risk*Peer Prot. .100 .081 1.11 -.186
Risk“ School Prot. -.168 .130 -1.29 -.l64
Risk‘Comm. Prot. -.186 .103 -1.81 -.104

 

Note. For the final model, F(14,375)=l9.98, p=.000.
"p<.01. * Bonferroni-adjusted p<.05 (2-tailed).

86

Figure 1. Interaction of overall risk and individual protection in predicting delinquency

 

 

 

 

0.8 ‘ """" i; /

0.6 — .____o_.n

 

 

0.4 A. -. --

 

 

0.2 -< & vLow Risk
f-O—High Risk

 

Delinquency

 

Low Prot High Prot

 

 

 

 

 

Individual Protection

 

Individual risg Table 23 displays the unstandardized regressions coefficients (B),

 

intercept, zero-order correlations, and R2 and adjusted R2 after entry of all independent
variables. The change in R2 was significantly different fi'om zero at the end of the fit
three steps, entry of demographic control variables, protective factors, and individual risk.
After the interactions between individual risk and each of the protective factors, the final
step, the change in R2 was not significantly different from zero.
In the final model, gender, family protection, school protection, individual risk,

and the interaction between individual risk and individual protection significantly

87

 

predicted delinquency. Being male, having lower levels of family and school protection,
and having higher levels of individual risk were related to higher levels of delinquency.
The nature of the interaction between individual protection and individual risk in
predicting levels of delinquency is shown in Figure 2. Individual protection was found to
have an exacerbating effect on individual risk. When individual risk is low, higher levels
of individual protection are associated with lower levels of delinquency. However, when

individual risk is high, higher individual protection is associated with higher delinquency.

Table 23
Do Individual, Family, Peer, School, and Community Protection Bufiizr the Efibct of

Individual Risk on Delinquency?

 

 

Final model B Std. Zero- R2 Adj
containing all 4 (U nstand Error t Order R2
blocks . Coefl'.) Correl.
Constant .159 .075 211*
Block 1. .085" .078”
Demographic
Controls
Gender -.296 .090 -3.29""" -.285
Urban/Rural -.003 .094 -.032 .042
Race/Ethnicity -.071 .100 -.713 .024
Block 2. Protective .195" .178“
Factors
Individual Prot. -.022 .058 -.376 -.l48
Family Protection -.161 .061 -2.65“ -.289
Peer Protection .021 .063 .330 —.285
School Protection -.171 .078 -2.20* -.278
Community Prot. -.041 .060 -.677 -.232
Block 3. Individual .370“ .355”
Risk
Individual Risk .589 .060 9.73" .550

88

Table 23 (Continued)

Do Individual, Family, Peer, School, and Community Protection Bufler the Eflect of

Individual Risk on Delinquency?

Final model B Std. Zero-
containing all 4 (U nstand Error t Order
blocks . Coeff.) Correl.

Block 4. Risk 1 Prot.
Interactions
Indiv. Risk * Indiv. .190 .071 2.66" -.001
Prot.
Indiv. Risk * Family -.045 .071 -.643 -.086
Prot.
Indiv. Risk * Peer -.066 .057 -1.16 -.l62
Prot.
Indiv. Risk * School -.194 .101 -1.92 -.062
Prot.
Indiv. Risk * Comm. -.006 .072 -.081 -.077
Prot.

R2 Adj
R2
.386 .363

 

Note. For the final model, F(14,375)=16.84, p=.000.
* Bonferroni-adjusted p<.05 (2-tailed).

89

Figure 2. Interaction of individual risk and protection in predicting delinquency.

 

Delinquency

 

Effect of Individual Protection on Delinquency by Individual Risk

 

0.8

 

0.6

0.4

 

 

 

_.________._‘
~ 0' LowRisk

 

0.2

 

-0.2 *

-o.4 --

 

 

 

 

 

 

 

-0.6

Individual Protection

 

 

 

"—___i“;¢h_lti§l‘-

 

Family risk. Table 24 displays the unstandardized regressions coefficients (B),

intercept, zero-order correlations, and R2 and adjusted R2 after entry of all independent

variables. The change in R2 was significantly different from zero at the end of the first

three steps in which demographic control variables, protective factors, and family risk

were entered. After the interactions between family risk and each of the protective

factors, the final step, the change in R2 was not significantly different from zero.

between family risk and individual protection significantly predicted delinquency. Being
male and having lower levels of school protection and higher levels of family risk were

related to higher levels of delinquency. The nature of the interaction between individual

In the final model, gender, school protection, family risk, and the interaction

protection and family risk in predicting levels of delinquency is shown in Figure 3.

90

Individual protection appears to have an exacerbating effect on family risk. When

individual protection is high, levels of delinquency are higher at high levels of family risk

and lower at low levels of family risk than when individual protection is low.

Table 24

Do Individual, Family, Peer, School, and Community Protection Bufier the Effect of

Family Risk on Delinquency?

 

 

Final model B Std. Zero- R2 Adj R2
containing all 4 (U nstand. Error t Order '
blocks Coeff.) Correl.
Constant .215 .083 260*
Block 1. Demog. .085" .078”
Controls
Gender -.503 .095 -5.31** -.285
Urban/Rural .080 .103 .784 .042
Race/Ethnicity -.022 .1 10 -.203 .024
Block 2. Prot Fact. .195“ .178"
Individual Prot. .008 .064 .120 -.148
Family Protection -.013 .071 -.184 -.289
Peer Protection -.005 .069 -.076 -.285
School Protection -.225 .084 -2.66** -.278
Community Prot. -.104 .064 -1.62 -.232
Block 3. Fam. Risk .254“ .236"
Family Risk .442 .083 5.30" .390
Block 4. Risk x .275 .248
Prot. Interactions
Fam. Risk * Indiv. .229 .086 2.66* -.007
Prot.
Fam. Risk * Fam. -.159 .084 -1.89 -.192
Prot.
Fam. Risk “ Peer .053 .082 .647 -.138
Prot.
Farn. Risk * School -.050 .122 -.406 -.085
Prot.
Fam. Risk " -.088 .099 -.895 -.083
Comm. Prot.

 

Note. For the final model, F(14,373)=10. 12, p=.000
" Bonferroni-adj usted p<.05 (2-tai1ed)

91

Figure 3. Interaction of individual protection and family risk in predicting delinquency.

 

Effect of Individual Protection on Delinquency by Family Risk

 

0.7

o 6 . /
0:5 /
/

 

 

 

i»

 

 

-- o —LowRisk
—o-—l-IighR'sk

 

 

 

 

9
.a
l

 

Delinquency
3 ° -

O

 

 

 

LowProt " ~ HighProt
-o.1 ~ . .

 

-o.2 _.,, ‘ ‘ ..

 

 

 

 

 

 

 

Individual Protection

 

Peer risk. Table 25 displays the unstandardized regressions coefficients (B),
intercept, zero-order correlations, and R2 and adjusted R2 after entry of all independent
variables. The change in R2 was significantly different fiorn zero at the end ofthe first
three steps. After the interactions between peer risk and each of the protective factors, the
final step, the change in R2 was not Significantly different from zero.

In the final model, gender and peer risk were significant predictors of
delinquency. Males and students with higher levels of peer risk reported engaging in
more delinquent behavior than did females and students with lower levels of peer risk.

Peer protection was related to delinquency on a trend level; students with higher levels of

92

peer protection reported fewer delinquent behaviors than did students with lower levels of
peer protection.
Table 25

Do Individual, Family, Peer, School, and Community Protection Bufler the Eflect of Peer

 

 

Risk on Delinquency?
Final model B Std. Zero- R2 Adj
containing all 4 (Unstand. Error t Order R2
blocks Coeff.) Correl.
Constant .191 .078 245"
Block 1. .085" .078"
Demographic
Controls
Gender -.463 .089 -5.21 "“" -.285
Urban/Rural .046 .096 .474 .042
Race/Ethnicity .074 .103 .723 .024
Block 2. Protective .195" .178”
Factors
Individual Prot. -.107 .061 -1.75 -.148
Family Protection -.081 .063 -1.29 -.289
Peer Protection .131 .067 1.94 -.285
School Protection -.083 .082 -1.01 -.278
Commrmity Prot. -.099 .061 -1.62 -.232
Block 3. Peer Risk .342" .326"
Peer Risk .492 .057 8.58" .518
Block 4.Risk x Prot .353 .329
Interactions
Peer Risk * Indiv. .103 .062 1.65 .000
Prot.
Peer Risk "' Family -.080 .054 -1.49 -.178
Prot.
Peer Risk * Peer .068 .051 1.34 -.152
Prot.
Peer Risk * School -.029 .078 -.366 -.149
Prot.
Peer Risk " Comm. -.079 .067 -1.17 -.106

Prot.

 

Note. For the final model, F(14,375)=l4.62, p=.000.
* Bonferroni-adjusted p<.05 (2-tailed).

93

School risk. Table 26 displays the unstandardized regressions coefficients (B),
intercept, zero-order correlations, and R2 and adjusted R2 after entry of all independent
variables. The change in R2 was significantly different fi'om zero at the end of the first
three steps. After the interactions between school risk and each of the protective factors,
the final step, R=.593=, R2=.351, adjusted R2=.327, Fchange (5, 375)=l.06, p>.05.

In the final model, gender and school risk were significant predictors of
delinquency. Males and students facing higher levels of school risk reported engaging in
more delinquent behavior than did females and students with lower school risk. Family
protection trended toward predicting delinquency, with students with higher levels of

family protection reporting fewer delinquent behaviors.

Table 26
Do Individual, Family, Peer, School, and Community Protection Bufler the Eflect of

School Risk on Delinquency?

 

 

Final model B Std. Zero- R2 Adj R’
containing all 4 (U nstand. Error t Order
blocks Coeff.) Correl.

Constant .171 .080 214*
Block 1. .085" .078W
Demographic
Controls
Gender -.369 .090 -4.1 1* * -.285
Urban/Rural .046 .096 .477 .042
Race/Ethnicity -.036 .103 -.345 .024
Block 2. Protective .195" .178"
Factors
Individual Prot. .092 .063 1.46 -.148
Family Protection -.116 .064 -1.81 -.289
Peer Protection .039 .067 .581 -.285
School Protection -.068 .083 -.811 -.278
Community Prot. -.075 .061 -1.22 -.232

94

Table 26 (Continued)
Do Individual, Family, Peer, School, and Community Protection Bufl'er the Efi'ect of

School Risk on Delinquency?

Final model B Std. Zero— R2 Adj R2
containing all 4 (U nstand. Error t Order
blocks Coeff'.) Correl.

Block 3. School .342” .327"
Risk
School Risk .712 .082 8.65" .536
Block 4.Risk x .351 .327
Prot. Interactions
School Risk *Indiv. .171 .082 2.10t -.021
Prot.
School Risk *Fam. -.032 .084 -.374 -.136
Prot.
School Risk *Peer -.018 .095 -. 190 -.1 10
Prot.
School Risk *Scho. -.141 .108 -1.30 -.179
Prot.
School Risk *Com. -.075 .102 -.743 -.028
Prot.

 

Note. For the final model, F(14,375)=14.51, p=.000.
" Bonferroni-adjusted p<.05 (2-tailed). "p<.01 (2-tai1ed).
tBonferroni-adjusted (p<. 10).

Communig risk. Table 27 displays the unstandardized regressions coefficients
(B), intercept, zero-order correlations, and R2 and adjusted R2 after entry of all
independent variables. The change in R2 was significantly different from zero at the end
of each step. In the final model, gender, family protection, school protection, and
community risk were significant predictors of delinquency. Students who were male, had
low levels of family and school protection, and who had high community risk reported

more delinquent behaviors than did students who were female, had high levels of family

95

and school protection, and lower levels of community risk.

Table 27

Do Individual, Family, Peer, School, and Community Protection Bufier the Effect of

Community Risk on Delinquency?

 

 

Final model B Std. Zero- R2 Adj
containing all 4 (U nstand. Error t Order R2
blocks Coeff.) Correl.
Constant .146 .079 1 .84
Block 1. .085" .078"
Demographic
Controls
Gender -.502 .090 -5.56*"‘ -.285
Urban/Rural .159 .098 1.62 .042
Race/Ethnicity .003 .104 .032 .024
Block 2. .195" .178"
Protective Factors .
Individual Prot. .018 .061 .298 -.148
Family Protection -.125 .063 -1.98* -.289
Peer Protection .010 .067 .157 -.285
School Protection -.255 .081 -3.16** -.278
Community Prot. .043 .067 .643 -.232
~ Block 3. .312" .295"
Community Risk
Community Risk .461 .070 6.56" .454
Block 4. Risk x .335* .310“
Prot Interactions
Com. Risk *Indiv. .175 .075 2.32“ -.051
Prot.
Com. Risk" Farrrily -.078 .072 -1.09 -.21_8
Prot.
Com. Risk“ Peer -.105 .071 -1.49 -.233
Prot.
Com. Risk" School -.178 .109 -1.64 -.l66
Prot.
Com. Risk“ Corn. .019 .079 .239 -.108
Prot. '

 

Note. For the final model, F(14,373)=13.44, p=.000.
‘ Bonferroni-adjusted p<.05 (2-tailed). "p<.01 (2-tailed).

96

Whicapromcfive factors buffer overall risl_< for each finder. for students of color versus
Caucasian students. and for students from urban versu_s rural M

Regression analyses for the final research question examined which domains of
protective factors most effectively buffer overall risk for girls versus boys (controlling for
race/ethnicity and urban/rural), students of color versus Caucasian students (controlling
for gender and urban/rural), and for urban versus rural students (controlling for gender
and race/ethnicity).

Gender. Tables 28-32 Show the unstandardized regression coefficients (B),

 

intercept, zero-order correlations, and R2 and adjusted R2 for each block of regression
analyses examining the ability of each of the protective factor domains to buffer overall
risk by gender. Individual, family, peer, and school protection did not moderate overall
risk for either female or male students.

Urban/rural. Regression analyses examining the ability of each of the protective
factor domains to moderate overall risk by urban/rural are shown in Tables 33-3 7.
Family, peer, school, and community protection did not buffer overall risk for students
from the urban or rural school. However, the three-way interaction of urban/rural, risk,
and individual protection was significant (see Table 33).

Figure 4 shows that among students fi'om the rural school at low risk, higher
individual protection is associated with higher levels of delinquency, while higher
individual protection for students at high risk is associated with little to no change in
delinquency. For urban students with low risk, high individual protection is related to

slightly lower delinquency, but for those with higher levels of risk, higher levels of

97

individual protection are related to higher delinquency (see Figure 4). Examination of the
slopes revealed that the exacerbating relationship of individual protection and overall risk
is statistically significant for the students from the urban school.

Race/ethnicig. Tables 38-42 contain the lmstandardized regression coefficients
(B), intercept, zero-order correlations, and R2 and adjusted R2 for each block of regression
analyses examining the ability of each of the protective factor domains to buffer overall
risk by race/ethnicity. Individual, family, peer, and school protection did not moderate
overall risk for students of color or Caucasian students. However, the three-way
interaction of race/ethnicity, risk, and community protection was bordering on
significance after Bonferroni corrections were applied (see Table 42). F iglue 5 shows
that community protection buffers the effect of overall risk for both students of color and
Caucasian students. Among students with low levels of overall risk, higher levels of
community protection are related to little to no change in levels of delinquency for both
students of color and Caucasian students. At higher levels of overall risk, higher
community protection is associated with lower levels of delinquency for both students of
color and Caucasian students. Examination of the Slopes revealed that the bufi‘ering
relationship of community protection and risk is statistically significant for students of

color.

98

Table 28

Gender by Individual Protection by Overall Risk

 

Unstand. Zero
Model gaggle: t Order R2 Adj.
B SE Correl R
l. Demog, Controls .002 -.003
Constant -.049 .076 -.651
Urban/Rural .079 .1 12 .705 .042
Race/Ethnicity .013 .118 .109 .024
2. Indiv. Prot. & Risk .395“ .388"
Constant .198 .070 2.81”
Urban/Rural .020 .088 .230 .042
Race/Ethnicity -.085 .094 -.905 .024
Individual Protection .035 .048 .719 -.148
Overall Risk 1.02 .074 13.84" .603
Gender -.342 .079 4.32” -.285
3. Two-way interaction terms .401 .389
Constant .215 .072 2.98"
Urban/Rural .029 .088 .324 .042
Race/Ethnicity -.101 .094 -1.07 .024
Individual Protection .008 .070 .109 -.148
Overall Risk 1.07 . 100 10.71 ” .603
Gender -.345 .079 —4.35** -.285
Gender‘lndividual Prot. .021 .096 .220 -.093
Gender‘Risk -.094 .147 -.639 .391
Risk'lndividual Protection .132 .076 1.74 -.023
4.Three-way interaction term .401 .387
Constant .215 .073 2.93"
Urban/Rural .029 .088 .324 .042
Race/Ethnicity -.101 .094 -1.07 .024
Individual Protection .008 .072 .104 -.148
Overall Risk 1.07 .100 10.69" .603
Gender -.345 .083 -4.14** -.285
Gender*1ndividual Protection .021 .098 .217 -.093
Gender‘Risk -.094 .148 -.636 .391
Risk‘lndividual Protection .133 .105 1.26 -.023
Gender*Risk*Individual Prot. -.001 .156 -.005 .018

 

Note. For the final model, F(9,406)=30.14, p=.000.

"p<.01 (2 tailed). * Bonferroni-adjusted p<.05 (2-tailed).

99

Table 29

Gender by Family Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
l. Demog. Controls .002 -.003
Constant -.049 .076 -.651
Urban/Rural .079 .1 12 .705 .042
Race/Ethnicity .013 .118 .109 .024
2. Faley' Prot. & Risk .394" .387"
Constant . .194 .071 2.74"
Urban/Rural .026 .088 .296 .042
Race/Ethnicity -.072 .093 -.773 .024
Family Protection -.020 .053 -.383 -.289
Overall Risk .986 .078 12.62" .603
Gender -.350 .079 -4.44** -.285
3. Two-way interaction terms .399 .388
Constant .173 .072 2.39*
Urban/Rural .025 .088 .279 .042
Race/Ethnicity -.061 .093 -.658 .024
F anrily Protection -.061 .076 -.809 -.289
Overall Risk 1.00 .102 9.82" .603
Gender -.355 .079 4.48" -.285
Gender‘Family Prot. .107 .107 .995 -.190
Gender‘Risk -.053 .158 -.335 .391
Risk’Family Protection -.071 .068 -1.05 -.196
4.Three-way interaction term .400 .386
Constant .176 .073 2.42*
Urban/Rural .024 .088 .274 .042
Race/Ethnicity -.058 .093 -.625 .024
Family Protection -.070 .079 -.886 -.289
Overall Risk 1.00 .102 9.81* * .603
Gender -.366 .084 4.34" -.285
Gender‘Family Protection .117 .111 1.06 -.190
Gender‘RiSk -.063 .160 -.394 .391
Risk‘Family Protection -.045 .094 -.482 -.196
Gender‘Risk‘Farnily Prot. -.054 .137 -.394 -.058

 

Note. For the final model, F(9,406)=30.02, p=.000.

”p<.01 (2-tailed). * Bonferroni-adjusted p<.05 (2-tailed).

100

Table 30

Gender by Peer Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
1. Demoga Controls .002 -.003
Constant -.049 .076 -.651
Urban/Rural .079 .1 12 .705 .042
Race/Ethnicity .013 .1 18 .109 .024
2. Peer Prot. & Risk .400" .393"
Constant .214 .071 3.02”
Urban/Rural .026 .087 .299 .042
Race/Ethnicity -.072 .092 -.779 .024
Peer Protection .107 .054 1.98“ -.285
Overall Risk 1.08 .079 13.61 ** .603
Gender -.383 .080 -4.77** -.285
3. Two-way interaction terms .406 .394
Constant .220 .072 3.04"
Urban/Rural .028 .087 .326 .042
Race/Ethnicity -.073 .092 -.791 .024
Peer Protection .053 .073 .730 -.285
Overall Risk 1.20 .118 10.15" .603
Gender -.376 .080 -4.68** -.285
Gender‘Peer Prot. .096 .11 1 .858 -.179
Gender‘Risk -.158 .168 -.940 .391
Risk"‘Peer Protection .125 .073 1 .70 -.1 86
4.Three-way interaction term .408 .395
Constant .227 .073 3.13"
Urban/Rural .026 .087 .298 .042
Race/Ethnicity -.062 .092 -.676 .024
Peer Protection .041 .074 .551 -.285
Overall Risk 1 .24 .123 10.09" .603
Gender -.413 .086 -4.79** -.285
Gender‘Peer Protection .105 .1 12 .942 -.179
Gender*Risk -.199 .171 -1.16 .391
Risk‘Peer Protection .184 .089 2.07* -.186
Gender*Risk*Peer Prot. -.189 .161 -1.18 .099

 

Note. For the final model, F(9,406)=31.07, p=.000.

”p<.01 (2-tailed). * Bonferroni-adjusted p<.05 (2-tailed).

101

Table 31

Gender by School Protection by Overall Risk

 

Unstand. Zero
Model Coerric. t Order R2 Adj.
B SE Correl R
l. Demag. Controls .002 -.003
Constant -.049 .076 -.651
Urban/Rural .079 .1 12 .705 .042
Race/Ethnicity .013 .1 18 .109 .024
2. School Prot. & Risk .398“ .390"
Constant .194 .070 2.75"
Urban/Rural .023 .087 .259 .042
Race/Ethnicity -.060 .093 -.651 .024
School Protection -.097 .064 -1.53 -.278
Overall Risk .957 .074 12.85" .603
Gender -.352 .079 4.48" -.285
3. Two-way interaction terms .403 .392
Constant .161 .072 2.23“
Urban/Rural .019 .087 .216 .042
Race/Ethnicity -.037 .093 -.401 .024
School Protection -.049 .090 -.548 -.278
Overall Risk 1.02 .099 10.29" .603
Gender -.361 .079 -4.59** -.285
Gender‘School Prot. -.087 .130 -.668 -.213 '
Gender‘Risk -.183 .151 -l.21 .391
Risk'School Protection -.168 .101 -1.66 -.l64
4.Three-way interaction term .405 .392
Constant .174 .073 2.37*
Urban/Rural .015 .087 .169 .042
Race/Ethnicity -.033 .093 -.354 .024
School Protection -.070 .092 -.762 -.278
Overall Risk 1.02 .099 10.32" .603
Gender -.393 .084 -4.67"'* -.285
Gender*School Protection -.074 .131 -.565 -.213
Gender‘Risk -.208 .153 -1.36 .391
Risk’ School Protection -.073 .136 -.537 -.164
Gender‘Risk‘School Prot. -.215 .204 -1.06 -.062

 

Note. For the final model, F(9,406)=30.70, p=.000.

”p<.01 (2-tailed). ‘ Bonferroni-adjusted p<.05 (2-tailed).

102

Table 32

Gender by Community Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
l. Demogg Controls .002 -.003
Constant -.049 .078 -.630
Urban/Rural .079 .115 .682 .042
Race/Ethnicity .013 .122 .106 .024
2. Comm. Prot. & Risk .394“ .386"
Constant .196 .073 2.69"
Urban/Rural .022 .091 .247 .042
Race/Ethnicity -.073 .096 -.766 .024
Community Protection .005 .055 .091 -.232
Overall Risk 1.00 .077 13.01 ** .603
Gender -.349 .081 -4.29** -.285
3. Two-way interaction terms .397 .384
Constant .172 .076 2.27“
Urban/Rural .020 .091 .215 .042
Race/Ethnicity -.065 .096 -.671 .024
Community Protection .016 .078 .201 -.232
Overall Risk 1.06 .104 10.16" .603
Gender -.350 .082 -4.30 -.285
Gender‘Community Prot. -.016 .114 -.142 -.161
Gender*Risk -.131 .157 -.836 .391
Risk‘Community Protection -.097 .092 -1.05 -.104
4.Three-way interaction term .403 .3 89
Constant .208 .078 2.68" »
Urban/Rural .008 .091 .090 .042
Race/Ethnicity -.056 .096 -.585 .024
Community Protection -.007 .078 -.095 -.232
Overall Risk 1.06 3.104 10.21 ** .603
Gender -.416 .088 -4.72** -.285
Gender‘Comm. Protection .003 .114 .027 -.161
Gender‘Risk -.152 .157 -.970 .391
Risk‘Community Protection .083 .131 -.632 -.104
Gender‘Risk*Comm. Prot. -.366 .189 -1.93 -.042

 

Note. For the final model, F(9,380)=28.50, p=.000.

”p<.01 (2-tailed). ‘ Bonferroni-adjusted p<.05 (2-tailed).

103

Table 33

Urban/Rural by Individual Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
l. Demog. Controls .083“ .078"
Constant .280 .075 3.72"
Gender -.575 .094 -6.09"”" -.285
Race/Ethnicity .082 .100 .821 .024
2. Indiv. Prot. & Risk .395" .388"
Constant .198 .070 2.81"
Gender -.342 .079 4.32” -.285
Race/Ethnicity -.085 .094 -.905 .024
Individual Protection .035 .048 .719 -.148
Overall Risk 1.02 .074 13 .84* * .603
Urban/Rural .020 .088 .230 .042
3. Two-way interaction terms .405 .393
Constant .232 .072 3.22"
Gender -.343 .079 4.34” -.285
Race/Ethnicity -.088 .095 -.928 .024
Individual Protection .035 .076 .467 -.148
Overall Risk 1.21 .124 9.76" .603
Urban/Rural .014 .088 .159 .042
Urban/Rural’Indiv. Prot. -.023 .098 -.232 -.091
Urban/Rural *Risk -.282 .153 -1.84 .445
Risk‘Indiv. Protection .140 .075 1.86 -.023
4.Three-way interaction term .415" .402"
Constant .176 .075 2.35* .
Gender -.339 .079 -4.31* * -.285
Race/Ethnicity -.086 .094 -.919 .024
Indiv. Protection .046 .075 .613 -.148
Overall Risk 1.15 .126 9.17" .603
Urban/Rural .086 .092 .934 .042
Urban/Rural *Indiv. Prot. -.055 .098 -.557 -.091
Urban/Rural *Risk -.216 .154 -1.40 .445
Risk*Indiv. Protection -.196 .149 -l .32 -.023
'Urban/Rural‘Risandiv Prot .454 .174 2.61 * .057

 

Note. For the final model, F(9,406)=31.64, p=.000.

"p<.01 (2-tai1ed). "‘ Bonferroni-adjusted p<.05 (2-tailed).

104

Table 34

Urban/Rural by Family Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
l. Demog= Controls .083" .078"
Constant .280 .075 3.72"
Gender -.575 .094 -6.08** -.285
Race/Ethnicity .082 .100 .821 .024
2. Fam‘ Prot. & Risk .394" .387"
Constant .194 .071 2.74”
Gender -.350 .079 -4.44*"‘ -.285
Race/Ethnicity -.072 .093 -.773 .024
Family Protection -.020 .053 -.383 -.289
Overall Risk .986 .078 12.62" .603
Urban/Rural .026 .088 .296 .042
3. Two-way interaction terms .409* .397*
Constant .161 .073 222*
Gender -.350 .079 -4.46** -.285
Race/Ethnicity -.067 .092 -.723 .024
Family Protection -. 148 .085 -l .75 -.289
Overall Risk .997 .138 7.24" .603
Urban/Rural .030 .088 .342 .042
Urban/Rural‘Family Prot. .257 .110 2.34“ -.144
Urban/Rural *Risk -.042 .167 -.249 .445
Risk*Farnily Protection -.092 .067 -1.37 -.196
4.Three-way interaction term .413 .400
Constant .133 .074 1.78
Gender -.353 .078 -4.51** -.285
Race/Ethnicity -.073 .092 -.791 .024
Family Protection -.137 .085 -1.62 -.289
Overall Risk .947 .141 6.73" .603
Urban/Rural .086 .094 .909 .042
Urban/Rural ‘Family Prot. .220 .112 1.97 -.144
Urban/Rural *Risk .013 .170 .079 .445
Risk’Family Protection -.223 .104 -2.15** -.196
Urban/Rural‘Risk*Fam Prot . .227 .137 1.66 -.075

 

Note. For the final model, F(9,406)=31.69, p=.000.
“p<.01 (2-tailed). ‘ Bonferroni-adjusted p<.05 (2-tailed).

105

Table 35

Urban/Rural by Peer Protection by Overall Risk

 

Unstand. Zero
Model Coeff'ic. t Order R2 Adj.
B SE Correl R
l. Demag. Controls .083" .078”
Constant .280 .075 3.72"
Gender -.575 .094 -6.08** -.285
Race/Ethnicity .082 .100 .821 .024
2. Peer Prot. & Risk .400M .393"
Constant .214 .071 3.02“
Gender -.383 .080 4.77" -.285
Race/Ethnicity -.072 .092 -.779 .024
Peer Protection .107 .054 1.98“ —.285
Overall Risk 1 .08 .079 13.61 "'* .603
Urban/Rural .026 .087 .299 .042
3. Two-way interaction terms .406 .395
Constant .231 .072 3.20"
Gender —.375 .080 -4.67** -.285
Race/Ethnicity -.065 .092 -.710 .024
Peer Protection .054 .092 .581 -.285
Overall Risk 1.21 .137 8.84" .603
Urban/Rural .018 .087 .21 1 .042
Urban/Rural*Peer Prot. .076 .1 13 .666 -.181
Urban/Rural *Risk -.184 .169 -1.09 .445
Risk“ Peer Protection .048 .068 .705 -.186
4.Three-way interaction term .407 .394
Constant .213 .078 2.71 **
Gender -.377 .080 -4.68** -.285
Race/Ethnicity -.065 .092 -.702 .024
Peer Protection .048 .093 .51 7 -.285
Overall Risk 1.19 .143 8.31" .603
Urban/Rural .042 .096 .439 .042
Urban/Rural *Peer Prot. .077 .114 .681 -.181
Urban/Rural *Risk -.152 .178 -.857 .445
Risk‘Peer Protection -.O34 .154 -.223 -.186
Urban/Rmal*Risk*Peer Prot. .103 .174 .593 -.153

 

Note. For the final model, F(9,406)=30.95, p=.000.

”p<.01 (2-tailed). ‘ Bonferroni-adjusted p<.05 (2-tailed).

106

Table 36

Urban/Rural by School Protection by Overall Risk

 

Unstand. Zero Adj.
Model Coeffic. t Order R2 R
B SE Correl
1. Demag. Controls .083" .078“
Constant .280 .075 3.72"
Gender -.575 .094 -6.08” -.285
Race/Ethnicity .082 .100 .821 .024
2. School Prot. & Risk .398“ .390"
Constant .194 .070 2.75”
Gender -.352 .079 -4.48** -.285
Race/Ethnicity -.060 .093 -.65 1 .024
School Protection -.097 .064 -1.53 -.278
Overall Risk .957 .074 12.85" .603
Urban/Rural .023 .087 .259 .042
3. Two-way interaction terms .404 .392
Constant .183 .072 253*
Gender -.357 .079 -4.54** -.285
Race/Ethnicity -.035 .093 -.377 .024
School Protection -.051 .106 -.481 -.278
Overall Risk 1.09 .135 8.09* "' .603
Urban/Rural .01 1 .088 .129 .042
Urban/Rural“ School Prot. -.042 .133 -.318 -.183
Urban/Rural *Risk -.223 .162 -1.37 .445
Risk‘School Protection -.125 .101 -1.25 -.164
4.Three-way interaction term .405 .392
Constant .206 .076 2.71 **
Gender -.356 .079 -4.53** -.285
Race/Ethnicity -.029 .094 -.305 .024
School Protection -.052 .106 -.486 -.278
Overall Risk 1.13 .141 8.04" .603
Urban/Rural -.025 .095 -.264 .042
Urban/Rural *SchOOl Prot. -.034 .134 -.255 -.183
Urban/Rural ‘Risk -.264 .167 -1.58 .445
Risk“ School Protection .006 .165 .034 -.164
Urban/Rural*Risk*Sch. Prot. -.210 .210 -.998 —.088

 

Note. For the final model, F(9,406)=30.77, p=.000.

”p<.01 (2-tailed). "' Bonferroni-adjusted p<.05 (2-tailed).

107

Table 37

Urban/Rural by Community Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
1. Demog= Controls 1 .083" .078”
Constant .280 .078 3.60”
Gender -.575 .098 -5.89” -.285
Race/Ethnicity .082 .104 .794 .024
2. Comm. Prot. & Risk .394” .386"
Constant .196 .073 2.69”
Gender -.349 .081 -4.29** -.285
Race/Ethnicity -.073 .096 -.766 .024
Community Protection . .005 .055 .091 -.232
Overall Risk 1.00 .077 13.01” .603
Urban/Rural .022 0.91 .247 .042
3. Two-way interaction terms .406 .393
Constant .174 .075 232*
Gender -.344 .081 -4.25** -.285
Race/Ethnicity -.057 ' .096 -.596 .024
Community Protection .169 .103 1.65 -.232
Overall Risk 1.27 .135 9.39“ .603
Urban/Rural .018 .091 .197 .042
Urban/Rural‘Comm. Prot. -.221 .126 -1.76 -.185
Urban/Rural *Risk -.397 .165 -2.40* .445
Risk’Community Protection -.103 .091 -1.12 -.104
4.Three-way interaction term .407 .393
Constant .205 .082 250*
Gender -.348 .081 -4.29** -.285
Race/Ethnicity -.055 .096 -.572 .024
Community Protection .164 .103 1.60 -.232
Overall Risk 1.27 .135 9.38" .603
Urban/Rural -.020 .100 -.203 .042
Urban/Rural *Comm. Prot. -.212 .126 -1.69 -.183
Urban/Rural *Risk -.399 .165 -2.42"‘ .445
Risk‘Community Protection .052 .191 .270 -.104
Urban/Rural'Risk*Com Prot. -.207 .225 -.918 -.123

 

Note. For the final model, F(9,380)=29.00, p=.000.

”p<.01 (2-tailed). " Bonferroni-adjusted p<.05 (2-tailed).

108

Figure 4. Three-way interaction of individual protection, overall risk, and urban/rural in

predicting delinquency.

Efl’ect of Individual Protection on Delinquency, by Overal R’nk and Site

 

Delinquency
o
l

 

 

“/4

 

 

 

 

-1

 

Lo Indiv Prof Hi Indiv Plot
‘- ~ ‘ ‘ ~
‘‘‘‘‘‘‘‘ .x
”‘rra
x
Inrivirhal Protection

109

—o—Higll R'sk/Urban ’
+High Risk/Rural
- 1— - Low R'sk/Urban
1  x~___ 10“ “fist/BR“!

 

Table 38

Race/Ethnicity by Individual Protection by Overall Risk .

 

Unstand. Zero
Model Coeffic. t Order it2 Adj.
B SE Correl R
l. Demog. Controls .085" .080"
Constant .241 .086 2.80" '
Gender -.578 .095 -6.1 1" -.285
Urban/Rural .120 .095 1.25 .042
2. Indiv. Prot. & Risk .395“ .388"
Constant .198 .071 2.80”
Gender -.342 .079 -4.31** -.285
Urban/Rural .020 .088 .230 .042
Individual Protection .035 .048 .718 -.148
Overall Risk 1.02 .074 13.82" .603
Race/Ethnicity -.085 .094 -.904 .024
3. Two—way interaction terms .404 .393
Constant .233 .072 3.25”
Gender -.349 .079 -4.41** -.285
Urban/Rural .036 .088 .416 .042
Individual Protection .060 .057 1.04 -.148
Overall Risk 1.13 .095 1 1.85“ * .603
Race/Ethnicity -.088 .094 -.93 1 .024
Race‘Indiv. Protection -.130 .106 -1.22 -.061
Race‘Risk -.247 .156 -1.58 .341
Risk*Indiv. Protection .172 .078 2.21 * -.023
4.Three-way interaction term .406 .393
Constant .244 .072 3.37"
Gender -.347 .079 -4.39** -.285
Urban/Rural .035 .088 .400 .042
Indiv. Protection . .057 .058 .996 -.148
Overall Risk 1.15 .097 1 1.85” .603
Race/Ethnicity -.1 15 .098 -1.18 .024
Race‘lndiv. Protection -.098 .110 -.886 -.061
Race’Risk -.234 .156 -1.50 .341
Risk‘lndiv. Protection .228 .094 2.42* -.023
Race‘Riskflndiv Prot. -.180 .171 -1.05 .074

 

Note. For the final model, F(9,405)=30.74, p=.000.

"p<.01 (2-tailed). ‘ Bonferroni-adjusted p<.05 (2-tailed).

110

Table 39

Race/Ethnicity by Family Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl - R
l. Demog. Controls .085“ .080“
Constant .241 .086 2.80”
Gender -.578 .095 -6.1 1" -.285
Urban/Rural . 120 .095 1 .25 .042
2. Faml_ly' Prot. & Risk .394" .387"
Constant .194 .071 2.73"
Gender -.350 .079 -4.43** -.285
Urban/Rural .026 .088 .295 .042
Family Protection -.020 .053 -.3 82 -.289
Overall Risk .986 .078 12.60" .603
Race/Ethnicity -.072 .093 -.772 .024
3. Two-way interaction terms .399 .387
Constant .192 .072 2.65"
Gender -.362 .079 4.57" -.285
Urban/Rural .018 .089 .206 .042
Family Protection .045 .068 .668 -.289
Overall Risk 1.04 .102 10.24" .603
Race/Ethnicity -.051 .093 -.547 .024
Race‘Family Protection -.131 .1 14 -1.15 -.152
Race*Risk -.165 .160 -1.03 .341
Risk‘Family Protection -.071 .068 -1.06 -.196
4.Three-way interaction term . .399 .386
Constant .193 .073 2.65"
Gender -.361 .080 4.54" -.285
Urban/Rural .018 .089 .208 .042
Family Protection .044 .068 .647 -.289
Overall Risk 1.04 .103 10.17“ .603
Race/Ethnicity -.057 .098 -.583 .024
Race‘Family Protection -.123 .121 -1.02 -.152
Race‘Risk -.168 .161 -1.04 .341
Risk’Family Protection -.062 .082 -.754 -.196
Race’Risk‘Family Prot. -.030 .147 -.203 -.106

 

Note. For the final model, F(9,405)=29.91, p=.000.

I”p<.01 (2-tailed). "‘ Bonferroni-adjusted p<.05 (2—tailed).

111

Table 40

Race by Peer Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl . R

1. Demog, Controls .085M .080"
Constant .241 .086 2.80"

Gender -.578 .095 -6.1 1** -.285

Urban/Rural .120 .095 1.25 .042

2. Peer Prot. & Risk .400" .393"
Constant .214 .071 3.02”

Gender -.383 .080 -4.77** -.285

Urban/Rural .026 .087 .299 .042

Peer Protection .107 .054 1.97* -.285

Overall Risk 1.08 .079 13.60" .603

Race/Ethnicity -.072 .092 -.778 .024
3. Two-way interaction terms .418** .407"
Constant .246 .072 3.44"

Gender -.393 .080 -4.93** -.285

Urban/Rural .042 .087 .487 .042

Peer Protection .241 .069 3.51 ** -.285

Overall Risk 1.25 . 100 12.45" .603

Race/Ethnicity -.083 .091 -.906 .024

Race*Peer Protection -.3 75 .1 12 -3.35** -.226

Race*Risk -.351 .166 -2.1 1* .341

Risk*Peer Protection .127 .069 1.83 -.186

4.Three-way interaction term .420 .407
Constant .263 .073 3.58"

Gender -.389 .080 -4.87** -.285

Urban/Rural .042 .086 .490 .042

Peer Protection .245 .069 3.57" -.285

Overall Risk 1.28 .104 12.30" .603

Race/Ethnicity -.117 .097 -1.20 .024
Race*Peer Protection -.356 .114 -3.13** -.226

Race*Risk -.403 .174 -2.31* .341

Risk*Peer Protection .204 . 103 1 .99* -.1 86

Race*Risk*Peer Prot. —.143 .141 -1.02 -.159

 

Note. For the final model, F(9,405)=32.53, p=.000.

“p<.01 (2-tailed). * Bonferroni-adjusted p<.05 (2-tailed).

112

Table 41

Race by School Protection by Overall Risk

 

Unstand. Zero
Model Coeffic. t Order R2 Adj.
B SE Correl R
1. Demog. Controls .085" .080"
Constant .241 .086 2.80"
Gender -.578 .095 -6.1 1” -.285
Urban/Rural .120 .095 1.25 .042
2. School Prot. & Risk .398" .390”
Constant .194 .070 2.75"
Gender -.352 .079 -4.48** -.285
Urban/Rural .023 .088 .259 .042
School Protection -.097 .064 -1.52 -.278
Overall Risk .957 .075 12.84" .603
Race/Ethnicity -.060 .093 -.650 .024
3. Two-way interaction terms .403 .392
Constant .174 .072 2.41 *
Gender ~.358 .079 -4.54** -.285
Urban/Rural .024 .087 .273 .042
School Protection -.034 .077 -.441 -.278
Overall Risk .977 .100 9.80" .603
Race/Ethnicity -.038 .093 -.404 .024
Race*School Protection -.178 .143 -1.24 -.136
Race*Risk -.068 .158 -.431 .341
Risk*School Protection -.138 .103 -1.35 -.164
4.Three-way interaction term .404 .391
Constant .182 .073 249*
Gender -.356 .079 -4.52** -.285
Urban/Rural .025 .088 .280 .042
School Protection -.034 .077 -.447 -.278
Overall Risk .989 .101 9.81“ .603
Race/Ethnicity -.059 .097 -.612 .024
Race*School Protection -.156 .146 -1.07 -.136
Race*Risk -.058 .158 -.365 .341
Risk“ School Protection -.093 .1 17 -.797 -.164
Race*Risk* School Prot. -.206 .251 -.821 .002

 

Note. For the final model, F(9,405)=30.55, p=.000.

”p<.01 (2-tailed). r Bonferroni-adjusted p<.05 (2-tailed).

113

Table 42

Race/Ethnicity by Community Protection by Overall Risk

 

Unstand. Zero R2 Adj.
Model Coeffic. t Order R2
B SE Correl
l. Demog. Controls .085" .080"
Constant .241 .089 2.71* *
Gender -.578 .098 -5.92** -.285
Urban/Rural .120 .098 1.22 .042
2. Comm. Prot. & Risk .394" .386"
Constant .196 .073 2.69"
Gender -.349 .081 -4.30** -.285
Urban/Rural .022 .091 .247 .042
Community Protection .005 .055 .091 -.232
Overall Risk 1 .00 .077 13.01" .603
Race/Ethnicity -.073 .096 -.766 .024
3. Two-way interaction terms .401 .388
Constant .178 .075 237*
Gender -.346 .081 -4.26** -.285
Urban/Rural .026 .091 .283 .042
Community Protection .093 .074 1.25 -.232
Overall Risk 1.09 .100 10.84" .603
Race/Ethnicity -.073 .096 -.757 .024
Race*Community Protection -.186 .117 -1.59 -.136
Race*Risk -.189 .160 -1.18 .341
Risk*Community Protection -.069 .093 -.742 -.104
4.Three-way interaction term .409* .395*
Constant .212 .076 2.79”
Gender -.341 .081 -4.21** -.285
Urban/Rural .034 .090 .381 .042
Community Protection .100 .074 1.35 -.232
Overall Risk 1.11 .100 11.10** .603
Race/Ethnicity -.148 .101 -1.46 .024
Race*Community Protection -.151 .1 17 —1.29 -.136
Race*Risk -.208 .159 -l.31 .341
Risk*Community Protection .126 .124 1.02 -.104
Race*Risk*Comm. Prot. -.463 .198 -2.34t -.1 17

 

Note. For the final model, F(9,3 80)=29.23, p=.000.
**p<.01 (2-tailed). * Bonferroni-adjusted p<.05 (2-tailed).
tBonferroni-adjusted (p<.10).

114

Figure 5. Three-way interaction Of community protection, overall risk, and race/ethnicity

 

 

 

 

 

 

 

 

 

 

in predicting delinquency.
Effect of Conlnunity Protection on Delinquency, by Overall Risk and
Race/Ethnicity
2
1 _____. ___
E- —o—High Risk/ SOC
8- +High Risk /Cauc
E + Low Risk/SOC
a 0 x Low Risk/Cane
7“” If _ ~,,_,_j —.—*’—°+HA— _- ___ /
low Prot High Pl'ot
ap__ _ - - A
X — r _,.._#7. ’ x
-1
Community Protection

 

_Fi_rhal Analyses Exarninin Students with the Hi est Risk

Finally, students with the highest amount of risk in the sample were examined to
see which protective factors had Significant main effects on delinquency and which
protective factors buffered overall risk for this sub-sample. Students in the highest
quartile, or an overall risk score Of .30 or above were selected for this analysis
(standardized overall risk; overall sample mean=0). This sub-sample included 110
students whose risk scores ranged from .30 to 2.43 and had a mean risk score of .68.
Sixty-six percent of the sub-sample were fi'om the urban school, 44% were female, and

39% were students of color. Their overall protection (standardized overall protection;

115

overall sample mean=0) ranged from -.157 to .90 and the mean overall protection was -
.43. Table 43 Shows the standardized coefficients, intercept, zero-order correlations, R2
and adjusted R2 after entry of demographic control variables, each domain of protection,
overall risk, and the interactions between each protective factor domain and overall risk.
Gender, school protection, and overall risk’had significant main effects on delinquency
for this sub-sample, and the interaction between overall risk and family protection was
significant. Figure 6 shows that for students with the highest amounts of risk, higher
levels Of family protection are related to lower levels of delinquency. However, higher
levels of protection for students in the sub-sample with comparatively lower risk are

related to higher amounts Of delinquency.

116

Table 43

Do Domains of Protection Bufler the Eflect of Risk for Students with the Highest Risk?

 

Unstand. Zero R2 Adj.
Model Coeffic. t Order R
B SE Correl
l. Demog= Controls - .060 .029
Constant 1.25 .205 6.13* *
Gender -.474 .219 -2.16* —.230
Urban/Rural -.1 17 .262 -.446 -.094
Race/Ethnicity -.100 .254 -.393 —.088
2. Domains of Protection & .203* .118*
Overall Risk
Constant .702 .269 2.60”
Gender -.453 .224 -2.02* -.230
Urban/Rural -.138 .258 -.534 -.094
Race/Ethnicity -.046 .255 -.179 -.088
Individual Protection .189 .135 1.40 .020
Family Protection -.085 .132 -.646 -.175
Peer Protection .099 .144 .690 -.036
School Protection -.401 .186 -2.15* -.229
Community Protection -.120 .156 -.765 -.072
Overall Risk .557 .230 2.42* .270
3. Interaction terms .294 .169
Constant 1.27 .335 3.79**
Gender -.531 .220 -2.41* -.230
Urban/Rural -.248 .259 -.957 -.094
Race/Ethnicity .1 12 .258 .435 -.088
Individual Protection .027 .266 .101 .020
Family Protection .461 .260 1.78 -.175
Peer Protection .352 .263 1.34 -.036
School Protection —.556 .394 -1.41 -.229
Community Protection .326 .330 .986 -.072
Overall Risk -. 133 .378 -.351 .270
Individual Prot * Risk .174 .288 .606 -.037
Family Prot "' Risk -.579 .249 -2.33* -.314
Peer Prot * Risk -.252 .258 -.977 -.192
School Prot * Risk .149 .432 .344 -.3 1 7
Community Prot * Risk —.603 .368 -1.64 -.l45

”Denotes significant race/ethnicity differences at 0.01 level (2-tailed). *Denotes significant race/ethnicity

differences at 0.05 level (2-tailed).

117

Figure 6. Interaction of family protection and overall risk in predicting delinquency for

students with the highest risk.

 

Delinquency
C O O —l A .5
lb '0: m -t iv in 3!)

.°
N

 

Effect of Family Protection on Delinquency by Overall Risk for

Students with Highest Risk

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Low Prot

118

 

Chapter 4

DISCUSSION

The purpose Of this study was to addreSs gaps in delinquency prevention research
by combining two popular approaches for preventing delinquency - risk reduction and
protective factor enhancement - in a comprehensive manner. Additional aims of the
study included exploring the utility Of considering domains, rather than merely overall
levels, of risk and protection, examining the relationship of protection with delinquency,
and examining sociodemographic differences in risk and protection. The findings
corroborate previous research showing that preventing and reducing delinquency requires
reducing risk in conjunction with increasing protection (e.g., Pollard et al., 1999). This
study also found evidence that type of risk and protection matters, supporting an
ecological perspective for delinquency prevention. The results highlight the relevance of
domain of protection both in directly affecting delinquency and in moderating the effects
of risk. In addition, this study found that contextual information about youth and the risks
they are facing has implications for which types Of protection Should be fostered. The
specific results, as well as the implications of the findings for prevention of delinquency,
are discussed below.

Domain-Specific versu_s Cumulative Approach to Risk and Protection

A hypothesis of this study was that a domain-specific model of risk and protection
would better predict delinquency than a cumulative model of risk and protection. This
hypothesis was confirmed; information about the domains of risk and protective factors

significantly predicted delinquency beyond the amount explained by overall or

119

cumulative levels of risk and protection. Although cumulative levels of risk and
protection were also found to significantly predict delinquency, the findings corroborate
the utility of a domain-specific approach to risk and protection.

In terms of delinquency prevention efforts, these results suggest that it is not
sufficient to concentrate solely on overall levels of risk and protection, because different
types Of risk and protection are not equal and interchangeable. The finding that the types
of risk and protection youth have are. important confirms and builds upon the work of a
number Of researchers (e.g., Smith et al.,1995; Guerra, 1998; Price and Drake, 1999;
Pollard et a1, 1999) and has implications for how models of risk reduction and protective
factor enhancement to prevent adolescent delinquency are currently utilized. In addition
to being used separately, these models are also currently used as checklists, where the
goal is to reduce risk and increase protection as much as possible without regard to type
of risk or protection. The finding that types of risk and protection are relevant suggests
that arbitrarily decreasing risk and increasing protection may be an inefficient and less
effective way of using these models, and that practitioners may need to attend to which
types of protection they build.

This finding substantiates the idea that although cumulative risk and protection
are valid for understanding and preventing delinquency, researchers and practitioners
should not discontinue seeking and utilizing more specific information about risk and
protection in the pursuit of efficiency. Continuing efforts to determine which risk and
protective factors are especially relevant for delinquency are necessary to enhance the
effectiveness, and eventually the efficiency, of initiatives to prevent delinquency and help

youth. This type of information may eventually help practitioners to build the types of

120

protection that are most needed for particular groups of youth facing specific types of
risks.

To this end, this study explored the role of different types of protection in directly
affecting delinquency and in moderating the effects of various types of risk. Using
hierarchical regression, family and school protection were found to exert direct effects on
delinquency. Interventions to build family protections, such as family care and support,
and school protections, such as teacher care and social support, high expectations of
students, and opportunities for students to be involved in and bond to school, may be
helpful in directly preventing and reducing delinquency regardless of risk. This finding
supports other research regarding the importance of family and school protection (e. g.,
Smith et al., 1995).

When moderating relationships among risk, protection, and delinquency were
examined, however, family and school protection did not buffer the effect of risk.
Instead, the most consistent finding was the positive relationship of individual protection
with risk. Surprisingly, individual protection was found to have an exacerbating, rather
than buffering, effect on overall, individual, and family risks. For youth with low levels
of these risks, higher individual protection is associated with lower levels of delinquency.

Conversely, when youth have higher amounts of these types of risks, higher individual
protection is related to higher delinquency.

Caution should be employed in interpreting this finding to suggest that increasing
self-esteem may cause youth facing higher risks to engage in more fiequent delinquent
behavior, however, as research regarding the relationship between delinquency and

individual protections such as self-esteem and independence is complex and has mixed

121

results (see Costello & Dunaway, 2003). Some research shows that inflated self-esteem
promotes a protective information processing style that hinders youths’ ability to benefit
from corrective feedback about their behavior (see Wyman, 2003). In other words,
inflated self-esteem may lead to youths ignoring teacher and family reactions to their
inappropriate behavior or acting in ways that protect their inflated self views. Other
research shoWs that increased participation in delinquency can result in improved self-
esteem (e.g., Jang & Thomberry, 1998). The cross-sectional design of this study hinders
determining whether higher self—esteem led to increased delinquency or if youth with
higher levels of risk participated in a greater amount of delinquent behavior, which
subsequently increased their self-esteem.

It is also possible that the measure of self-esteem used in this study may not have
captured the multi-faceted nature of self-esteem or distinguished self-esteem fi'om other
related constructs such as egotism or bravado. Further research to determine whether
building individual protections such as self-esteem is helpful or harmful for youth facing
higher levels of risk is greatly needed, as these factors are featured in Search and other
protective factor enhancement models.

Demographic Differences in Risk and Protection

This study also examined how different types of risk and protection vary by
gender, urban/rural, and race/ethnicity and whether different types of protection moderate
the effects of overall risk for youth within different sociodemographic groups.
Information about sociodemographic differences in types Of risk and protection could
help practitioners use models of risk reduction and protective factor enhancement in a

true strength-based manner. For example, practitioners could incorporate protective

122

factors generally found in rural youth within prevention programs to meet their risks and
needs. Data regarding which protective factors moderate risk for boys and girls, youth
from urban and rural areas, or youth of color and Caucasian youth may also facilitate the
design of interventions to prevent delinquency.

fiends:

The hypothesis that males would report significantly higher levels of overall risk
compared to females, and that females would report significantly higher levels of overall
protection was confirmed. This study also contributes to previous research by exploring
differences in types of risk and protective factors for delinquency by gender. Males
reported significantly higher levels of individual risks such as early initiation of problem
behaviors (e.g., delinquency and violence) and poor conflict resolution skills, peer risks
such as having friends who engage in delinquent acts, and school risks such as academic
failure, lack of commitment to school, and early and persistent behavior problems at
school. There was also some evidence that males experienced higher levels of family
risks such as poor supervision from their caregivers, conflict and violence at home, and
caregivers who engage in crime and violence; however, this finding became merely trend-
level once Bonferroni corrections were applied.

This study adds to scarce information about risk and protective factors for females
by showing that they have significantly higher levels of peer protection than males.
Females reported having fiiends that earned good grades and did not get into trouble in
school, and that their fiiends did not support use of violence to solve problems.

Researchers are beginning to focus on the needs of girls for prevention and

intervention, because gender differences have been found in risk and protective factors

123

(e.g., Ensminger, 1990; OJJDP, 1998; Werner, 1990) and the reasons why girls enter the
juvenile justice system are different from boys (e.g., OJJDP, 1998). The results of this
study indicate that gender is important to consider in prevention and intervention. Gender
continued to exert a significant direct effect on delinquency after domains of risk and
protection were entered in hierarchical regression analysis, while the other
sociodemographic factors examined in this study, urban/rural and race/ethnicity, did not.
In addition, a greater number of overall and domain-specific risk and protective factor
differences were found by gender than for other demographic variables. The information
about gender differences in risk and protective factors for delinquency found in this and
other studies could help direct preventive interventions. The results of this study suggest
that peer groups are strengths for girls and could thus form the basis of strength-based
preventive interventions to help address their risks and needs. Researchers have noted
that group membership and belonging are especially important to females (e. g., Juvonen
& Murdock, 1995), and a number of initiatives that aim to target the unique needs of
female juvenile offenders are beginning to recognize the importance of incorporating
girls’ peer groups within gender responsive programming (e.g., Acoca, 1999). The
results of this study suggest that utilizing girls’ peer groups may be successful for
prevention programs in the general population as well.

The findings of this study also indicate that boys have a number of risks to address
in order to prevent delinquency. Although it is important to address the unique needs of
girls because juvenile justice research and programming have typically been developed
for boys, the result of this study also call attention to the fact that boys generally have

more risks and appear in the juvenile justice system more Often than girls. Practitioners

124

should focus on developing family and school protections for boys while targeting their
risks.
Urban/Rural

The hypothesis that urban students would report higher levels of overall risk
compared to rural students was confirmed; however, the hypothesis that they would also
have higher levels of overall protection was not. The current study did not support the
findings of previous research that rural students have fewer protective factors than urban
youth (e.g. Cheaturn, 1998). When domains of risk and protection were explored, urban
students reported significantly greater levels of individual risks such as early initiation of
problem behaviors and poor conflict resolution skills and rural students reported higher
levels of community protection. Some support was found for the results of previous
research showing that urban students report higher levels Of family protection than do
rural students (Cheatum, 1998).

It is notable that of the community-level factors, it is community protection, not
risk, that differs for urban versus rural students. Contrary to media stereotypes, the
urban/rural community differences for students in this sample were not related to the
. urban area having greater problems with drugs, firearms, and general disorganization, but
were instead related to the rural students reporting a greater sense of community and
positive orientation toward youth in their neighborhoods. The finding that rural youth
report a greater sense of community fits with the observations of sociologists and
researchers that sense of community diminished as people moved fi'om smaller, tight-knit
communities where everyone knew and depended upon their neighbors to larger, more

anonymous cities where social isolation was greater. Perhaps the high sense of

125

community and positive orientation toward youth found in rural communities could be
used to address youths’ risks and needs. For example, these community protections could
be utilized in a deliberate manner to support and provide needed resources for families,
foster adult mentors, and recruit adults to supervise youth in the community and provide
tutoring and other skill building activities.

Examination of the moderating effects of protection within demographic groups
revealed that for rural students, higher levels of the individual protective factors of
independence and self-esteem are associated with higher delinquency for students with
low risk and no change in delinquency for students with higher risk. However, among
urban students, individual protection was found to significantly exacerbate the effects of
risk; higher levels of individual protection were associated with lower levels of
delinquency for students with low risk and higher levels of delinquency for students with
higher risk. As discussed previously, this effect is difficult to interpret due to the cross-
sectional design of this study, and existing research on the relationship between self-
esteem and delinquency is inconclusive (Costello & Dunaway, 2003). Perhaps inflated
self-esteem leads youth to ignore environmental responses to their inappropriate
behaviors or causes them to act in ways that protect their inflated self-views (W yman,
2003). It is possible that this effect might be more pronounced in an urban context, where
youths report higher levels of individual risks such as early initiation of delinquent
behaviors and poor conflict resolution Skills. For example, inflated self-esteem may
cause youth to ignore feedback about their poor conflict resolution skills. Further, youth
may respond to criticism from parents and teachers about poor conflict resolution skills

by acting out in an attempt to dispel the threat to their inflated self-esteem.

126

On the other hand, research has also shown that participation in delinquent
behavior may increase self-esteem (e.g., Jang & Thomberry, 1998). Longitudinal studies
are needed to further examine this relationship to determine whether self-esteem
programs may actually do more harm than good and should be avoided when working
with urban youth facing higher risk, or if the positive relationship between individual
protection and delinquency Simply indicates that delinquent behavior increases self-
esteem for urban youth.

Race/Ethnicity

The hypothesis that youth of color would report higher levels of overall risk and
overall protection than Caucasian youth was not confirmed. When domain-specific risk
and protection were explored, few differences were found in levels of risk and protection
by race/ethnicity. Contrary to previous research suggesting that risk and protective
factors, especially family factors, vary by race/ethnicity (e.g., Safyer, 1994; McAdoo,
1998), there were no differences in overall or family, peer, school, or community risk and
protection. Youth of color did have higher levels of both individual risks, such as early
initiation of problem behaviors and poor conflict resolution skills, and individual
protection, such as self—esteem and independence, than Caucasian students. However,
although individual risk was related to higher levels of delinquency and individual
protection appeared to exacerbate delinquency and both of these factors were higher in
youth of color, there was not a main effect of race/ethnicity on delinquency. Several
issues that may have limited finding racial/ethnic variations in risk and protection are
discussed later in this section.

Some evidence for a buffering effect of protection domains was also found when

127

moderating effects were examined within demographic groups. For students of color,
higher community protection was related to little to no change in delinquency at low
levels of risk. At higher levels of risk, however, high levels of community protection
were associated with a reduction in delinquent behavior. These results suggest that
building community protections such as sense Of community and positive community
orientation toward youth may be especially helpful in buffering the effects of risk for
youth of color.
Effects of Domg'naaf Protection for Students with th_e Higlaest Ris_k

Finally, the role of protection was examined for youth with the highest level of
overall risk. Gender, school protection, and overall risk had direct effects on delinquency
for this subgroup. Females and students with higher school protection and relatively
lower risk had lower levels of delinquency, and males and students with less school
protection and relatively higher risk engaged in greater amounts of delinquent behavior.
The interaction between overall risk and family protection was also significant for this
group. Within the sub-sample of students with the highest levels of risk, those with
relatively lower risk actually had higher levels of delinquency at high levels of family
protection, while for those with the absolute highest levels of risk, higher family
protection is related to comparatively lower levels of delinquency. Interpretation of this
finding is hindered by the study’s cross-sectional design. Perhaps there is a level of risk
at which youth rebel and engage in increased amounts of delinquent behavior in response
to family attempts to provide positive discipline and support, and a threshold of risk at
which family protection begins to exert a buffering effect. It is also possible that other

factors, such as the types of risks youths are facing, explain this finding.

128

Implication; for Deliaquencv Prevention

Similar to the findings of other research (e.g., Pollard et al., 1999), the results of
this study highlight the importance of increasing protection in conjunction with
decreasing risk; delinquency prevention programs cannot focus exclusively on increasing
protection without regard for risk. While the current research was concerned primarily
with the effects of protection, it is noteworthy that risk had a much stronger relationship
with delinquency than did protection in every analysis conducted in this study. These
findings confirm those of researchers (e.g., Pollard et al., 1999) who have found that
protection enhancement alone is incomplete as a delinquency prevention strategy. It is
unlikely that protection by itself can mitigate the effects of risk, thus prevention efforts
must also attend to risk.

In terms of protection, the results of this study suggest that type of protection
matters for preventing delinquency. Family and school protection had direct effects on
delinquency, while individual, peer, and community protection did not. When
moderating relationships were examined, individual protection had an exacerbating effect
on risk. At the very least, these findings suggest that people who utilize protective factor
enhancement models would be more efficient and effective by focusing their efforts on
family and school protective factors. Increasing family protections such as family care
and social support and school protections such as teacher care and social support, high
expectations of students, and opportunities for students to be involved in and bond to
school appears to have positive effects for all students.

In addition, the results of this study also have serious implications for how

protective factor enhancement models are currently used. Models such as the Search

129

Institute employ a checklist of protective factors, and the goal is to build as many
protective factors, regardless of type, as possible. The results of this study indicate that
using a protective factor enhancement model to build any type of protection will not
necessarily be beneficial. If the goal is strictly prevention of delinquency, efforts to build
protection types other than family and school may be a waste of valuable resources for
low risk youths. The only type of protection shown to buffer the effects of risk was
family protection, and this effect was only seen for students with the highest levels of
risk; again, delinquency initiatives focusing on other types of protection for this
population may be wasting resources. Furthermore, programs that aim to increase
individual-level protections may actually do more harm than good, as factors such as self-
esteem and independence may actually be related to an increase in delinquency.

Furthermore, the results suggest that consideration of not only risk but also youth
sociodemographic characteristics is helpful in delinquency prevention efforts. For
example, initiatives to prevent delinquency for youth of color may be especially efficient
and effective if they focus on building community protection.

Finally, the results Of this study have implications for utilizing a true strength-
based approach to preventing delinquency. Existing prOtective factor models typically
use “asset building” in a deficit-oriented manner. Youth are surveyed to ascertain which
protective factors they lack, and then those assets are targeted for development.
Information gained fi'om this study about protective factors youth have could also be used
to direct interventions for particular youth in a strength-based approach; that is, the
protective factors youth possess could be used to target their risks or needs. For example,

current initiatives utilize girls’ peer groups in interventions to meet their needs. Rural

130

youths’ community protective factors could be employed in a similar manner within an

intervention to alter or offset their risks.

Limitations of the Study

In addition to the limitations imposed by this study’s cross-sectional design and
measurement of self-esteem, there are methodological problems that may have hindered
accurate measurement of the independent variables and generalizability of the findings.
The results of this study need to be replicated with a new sample due to the multiple
families of analyses performed on this one dataset.

A limitation to understanding how risk and protection operate in the real world to
affect delinquency is the individual-level measurement of all variables. This method was
utilized in an attempt to simulate how people might combine existing risk reduction and
protective factor enhancement models. As these models currently tend to be used
separately, information about how the models work together would advance knowledge
regarding delinquency prevention. However, a limitation to applying the results of this
study to the use of popular risk reduction and protective factor enhancement models is
that the exact scales to measure risk and protective factors used in these models could not
be utilized in this study. Copyright regulations prohibit shortening the risk and protection
surveys in order to combine them, and the models cannot be used separately and then
combined in a meaningful way because the owners of the models do not allow consumers
to keep their data.

Additionally, there were problems that may have precluded uncovering gender or

racial/ethnic variation in risk and protective factors that may exist. One methodological

131

issue that may have obscured differences is the manner in which protective factor
domains, rather than single protective factors, were examined in this study. Differences
in particular protective factors may have diminished when they were combined into
domain scores. For instance, females may have experienced higher family support and
males may have experienced higher family supervision, but these differences may have
been masked once the factors were combined into the family domain score. As a further
example, research shows that Afiican Americans value education more (achievement
motivation) than Caucasian or Hispanic students (Roeser & Eccles, 1998), and peer
pressure may affect Caucasian students more than students fiern other racial/ethnic
groups (e.g., Landrine et al., 1994). Any variation in these particular factors within this
sample may have faded once they were averaged with other protective factors in the
school and peer domains.

Moreover, a number of gender and race/ethnicity differences in risk and protection
identified in previous research were not examinedin this study. Factors that may be
salient for females that were not examined in this study include risk factors such as
experience of sexism and history of sexual abuse, and protective factors such as positive
gender identity, positive minority identity, positive sexual development, and delay Of
sexual experimentation. Factors that are likely to be relevant for protecting youth of color
from delinquency that were not examined in this study include a deep sense of
Spirituality, racial identity, and flexibly configured families that include kin and non-kin
(e.g., McAdoo, 1998, Miller, 2002).

Finally, Hispanic, Asian, Native American and multi-racial/ethnic youth were

combined with African American youth into a “youth of color” category because the

132

number of survey participants within each of these categories was quite small.
Combining these racial/ethnic groups may have been problematic, because research
suggests they may vary in their experience of and reaction to risk and protective factors.
For example, Afiican American families have been found to grant youth more
independence, while Hispanic families have been found to provide the highest levels of
youth supervision (Bulcroft et al., 1996). Research also suggests that youth from
different racial/ethnic groups vary in their response to risk and protective factors. For
example, the relationship of family boundaries and expectations with school performance
is greater for Hispanic and Caucasian youth than for Afiican American and Asian youth
(Steinberg et al., 1991).
Strengtl_rs of the Study

A number of this study’s strengths contribute to the delinquency prevention
literature. Research to date has focused primarily on either risk or protective factors;
studies combining these approaches have examined cumulative risk and protection or
have taken a narrow methodological approach of measuring only a few specific risk and
protective factors (e.g., Florsheim, Tolan, & Gorman—Smith, 1998; Pollard et al., 1999).
This study explored a large number of risk and protective factors for delinquency in a
more comprehensive manner, moving beyond looking solely at cumulative levels of risk
and protection. This study also contributes to delinquency prevention practice by
attempting to combine two common prevention models, risk reduction and protection
enhancement.

The findings built upon studies looking at the relationships of particular types of

risk and protection with delinquency. For example, Smith et al. (1995) found that school

133

and family factors buffered the effect of family risks. This study included risks fiom
other domains and also explored individual, family, peer, school, and community
protection by gender, race/ethnicity, and urban/rural differences.

Conclusions and Next Steps in Research

Future studies should build on this research by using a longitudinal design tO
examine how risk and protective factors that appear in widely utilized risk reduction and
protective factor enhancement surveys affect delinquency as well as other outcomes,
including positive behaviors such as academic achievement. Research could also
examine a greater variety of protective factors, such as gender and racial identity, and
examine the effect of single protective factors rather than domains of protection.

Future studies could also utilize a more ecological approach to assessing domains
of risk and protection. Rather than surveying youth about their perceptions of multiple
settings in their lives, domains of risk and protection could be measured in other ways.
For example, discipline records, grades, standardized test scores, and attendance data
from a large number of schools could be examined to assess school risks and protection.
Crime mapping data and community asset mapping procedures could be utilized to assess
community risks and protection, and data on child abuse and neglect, runaway, domestic
violence, and delinquency rates could be employed to assess family and peer risks and
protective factors within a large sample of urban and rural neighborhoods.

This study represents a first step in understanding the role of specific types of risk
and protection in predicting delinquency. The results confirm the findings of other
researchers that protective factor enhancement alone is incomplete, and delinquency

prevention efforts must also target risk. The results of the study support the idea that

134

domains of risk and protection are important, and these findings have practical
implications for how we try to prevent adolescent delinquency.

This study also provided information about types and levels of risk and protection
by gender, race/ethnicity, and urban/rural. The findings suggest that having information
about youth sociodemographic characteristics as well as the risks they are facing is
important when using risk reduction and protection enhancement to prevent delinquency.

Practitioners should gather this information before using these models, as the findings of
this study suggest that increasing any type of protection without regard for this contextual
information is at best inefficient and at worst may do harm. Researchers Should continue
to seek specific information about risk and protection specifically for delinquency rather

than relying solely on cumulative levels Of generic risk and protection.

135

APPENDIX A

136

A. Demographic Questions:

1. Race/Ethnicity:

Black/African American
White/Caucasian American
Lalino/Chicano/Mexican American
Asian Pacific American

Native American

Other (indicate)

@@®@®®

 

2. What is your age?

10111213 141516

@®®©®@©

3. What grade are you in at school?

(Dem (2 7th
CD 8th

4. What is your gender:

CD Male @ Female

5. Does your mom or other adult female
help take care of you?

@Yes ®No

6. Does your dad or other adult male
help take care of you?

@Yes @No

7. Does someone else help take
care of you? Who?

CD brother sister

(2) uncle ® aunt

@ grandmother grandfather

G) godparent foster parent

(5) stepparent (9 other adult___

B. School Questions:

. What are your overall grades at school?

(Please fill in one of the circles)

@Mostly As @Mostly As & Bs
®Mostly 83 @Mostly Bs & Cs
®Mostly Cs ©Most|y Cs & 05
@Mostly 05 @Mostly Ds 8 Es
@Moslly Es

2. How many classes have you failed during

this school year?
01234 Sormore

©®®®®©©

3. How many times have you had to repeat

a grade at school?
1 2 3 4 5 ormore

0
©®®®®©©

4. What do you think you will do after

high school?

@Look for a job

®Enlist in the military

@Attend a four-year college
@Attend a two-year college
($60 to vocational/training school

©Don't know 3
a? «5
cs .3?
5. How much do you agree or f a, 0
. 8. o. ‘3
dlsagree with the 5 g? a? 39
following statements? ‘3 ‘9 6‘ “
a. I plan to improve my grades
in school next year Q) ® (9 (D
b. I would like to get better
grades (9 Q) Q 0
c. I think I need special help
with my schoolwork that
I’m not getting Q Q) Q CD
d. I don't think I will finish
high school 6) ® Q) Q

137

6. How important are your
grades to you?

7. How important is education
for getting the job you want?

8. How many hours do you

spend each week on homewor

I
I
4%"!
49%,

@ son, 0%!er

5 . 5

'30 r" e° f: a;

@ (93 c. N my SChool. it is not safe to ‘0 Y 'o‘ a}
go into the bathroom alone 6) Q) Q)

Q Q Q Q d. Because I lear for my physical
safety. I go out of my way to

avoid certain parts of my school 6) C3) C9 C9

k or studying?

® 0 Q ® ® ® Q Q G @ Q3 0, me“, 9. While I'm walking to or from

a. expect you to do well in
school

b. use creative activities and
other ways oI teaching class

c. have students work together
in groups during class time?

d. have group projects that you
work on with other students?

e. seem to care about students?

i. seem to understand students'
problems?

9. tell students when they do
good work?

h. help you when you're having
problems or are upset about
something? '

i. make you want to learn new
things?

i. are people you trust and
respect?

10. How much do you agree or
disagree with the following
statements:

a. In general. I leel safe in school
b. I am afraid that someone will

try to start a physical fight with
me or attack me at school

school (or the bus). I'm afraid

0 :-
~§ 096 ,f’ 2&9 that someone-will challenge me
6) @ C2) G to a physucal light or attack me (D GD @ CD
,5?
11. How true are the following ‘5‘? f is": o
6) @® CD statements for your school? is s S" be
A" ‘3 v?’ 2°
. a. I feel like Hit in” with other kids ®®®®
@@ (9 (D
b. I feel it is okay to be me (9 C9 (9 O

®@®® c. lfeellikelcan saywhatlthink (999
it ' ‘ diti t th
6) @® ® even it IS erent ram 0 ers 0

Id. It is okay to be different from .
other kids Q (9 (9 O

®@®®
' e.ladmireandlookuptomyteadiers 0930

G) G) G) C9 12. How true are the following

statements?
is";
®@®® . . i *3
a.l have friends who are there lot1 9 § §
me when I need them A
coco
(D @ ® (9
b. If I have a problem. I can talk. to my
friends ' O 0 O O
Q) @Q) (9 ' .
c. If I reluse to light, my friends will
.0 é? think I m afraid (D G G O
s g ‘
.9 ‘ :d. If I walked away from a light. my

0

3 ‘3 3 friends would stand by
5 {5965. ‘5’ me C) G G O
o G GO 9. If I’m challenged. i'm going 10 “QMDGGO

i. It Is easy to make new friends at

my school
0000 g.0verall.lliketheotherstudents .GGGO
thatgotomyschool 0930

IS, 15:3
h. I am able to do things as well A: V g
as most other people Q Q Q Q

i. Overall. I feel accepted by other
students who go to my school Q) Q @Q

i. It is important for me to do well at
activities other than sports. such
as music. drama. clubs. etc Q ® G) (D

k. I am interested in the activities
my school offers (sports. clubs,

etc.) ® G) (9 CD

I. I am interested in the social
activities my school offers

(dances 8. other activities) . Q Q Q (D

m.l participate in activities at my

school @960

I3. How many opportunities are
there for you to participate _In _' 5‘; 3 §
activities (sports,social,clubs, v v e
dances) at your school? 6) G) G) (9

14. How many of the following
activities are you Involved In? 0 i 2 a 4 s

a. Sports teams in the community @ d (2) Q Q Q G)

in 5

b. School activities (clubs. student 1, ., '
government. band .choir.sports) Q) Q ® ® Q (s) Q

c. Religious youth groUps @ Q Q Q (9 Q Q

35' 3._

Cor more

d. Community activities. such as

scouts .service clubs .hobby clubs Q) Q (2) Q Q Q Q
e. Other (what?) @ G) ® 5 (9 @ q?

15. How much do you agree or
disagree with the following
statements?

 

a. I don’t learn much in school

b.0veraII.IfeeIgoodaboutmy
school -

c. Ithinkthatthisschoolisasgood
orbetterthanotherschools-

 

139

(I. My classes are boring G) (3) C9 C9

e. I think that school is important Q) Q Q) Q

I. It is important for me to do well

at school @6930
g. It is important for me to do well
at sports @ Q) C2) (9
h. When I do a job. I do it well @@®®
k. I can do just about anything I set
my mind to Q) Q) @ (D
16. How much gang activity do you
, think there is at your school? 3;? j
0 C9 Q Q
C. Questions About Yourself: 0
1. About how old were you when cg
you first? £3 ,3 ,9
.5 5:5 .t‘

a. get suspended from school? Q‘Q ©,®®

iii %
b. got in trouble with the police 5;:
or courts? 3 O Q Q G)
c. carried a weapon? (9'0 9 G (D
d. challenged someone'to a physical i. 3:.
fight? Q0 Q}® G)
e. got into a physical fight? GO é®®
I. had to change schools because , , "7 ,
of your behavior? QG®®®
g. were challenged to a physical ? :5

140

0

fight? 3'
h. skipped classes without an excuse? ,0

' Too

3..

G)

About how many times In the last year
have YOU?(glve number): 0 l 2 3 4 5 6

a.

b.

c or more

709

run awaylrom home? QQQQQQQQQ®®

carried a hidden knife.

gun. or other weapon? @QQQQQ©®.Q®

. Damaged. destroyed or

marked up somebody
else' 5 property on purpose Q Q Q Q Q Q Q Q0 0 Q Q

. Set or tried to set fire on

purpose to a house.
building, car. or

someone'spropeny? ©®®©®©©®d©®

. Gone into or broken into

a building to steal or ' '
damage something? ©®®©®©@@.@®
Tried to steal or actually ‘

stolen money or things

worth:

SSorless? @®®@®©.@@@®

Between $5 and $100? QQQQQQHQQQ

mmmwmn ©®®®®QQ®0©®
. Tried to buy or sell things ' A . '

that were stolen? QQQQQ@®@.@®

. Taken a car or motorcycle

for a ride without the

owner's permission? Q Q Q Q Q Q Q Q @ CD

Stolen or tried to steal a

car or other motor vehicle?@ Q Q Q Q ($39 9 Q Q @ ®

Forged a check or used
take money to pay for

something? ©®®®®©O©®©®

. Usedl‘tried to use a credit

card. bank card. or auto-
matic teller card without

permission? @®®®®©O@®@®

Been challenged to a f -

physicalfight? @®®_©®©@@@0®
.Challenged someone to ‘ 4 r

a physical fight? @®®@®@@@@O®
. Gotten Into a physlcal .

mm @®®®®©@@0@®

141

OfO

‘
oOfm

0i23456789

0. get suspended from

school? ©®®©®©©®C©®

p. get in trouble with the

police or courts? @®®®®©..@@®

q. hit someone because

they made you angry?©®®©®©©®.@®

r. skip classes without

an excuse? ©®®©®©©®O®®

s. got in trouble because
of your behavior in
class, at school. or

on the bus? ©®®®®®©®®®®

t. Used a weapon with
the idea of seriously

hurting someone? Q) 0 Q Q) (9 ® © G (9 ® 0

u. Hit someone or thrown :
something at someone 'A
with the idea of hurting -‘
them (other than what

you wrote above)? ©0Q®®®®®©®®

v. Been involved in

wwmm ooooéeoooeo

w. Used a weapon or
physical lorce to
,makesomeone give

you money or things? @Q®@®@©®©®®
x. Used alcohol? ©0®®®®®®®®®

y. Used marijuana or

W? ooeeoooooeo

2. Used tobacco
(smoked cigarettes or ‘
cigars. or chewed

thW cocooeooeoe

142

3. How often have you experienced
any of the following problems
because of your race or
ethnicity?

a. someone bothering you because
of your race or ethnicity while g
you were out walking somewhere g g
(in your own neighborhood or in if; #3
other neighborhoods) Q Q Q Q

b. Someone causing problems for
you because of your race or
ethnicity when you were going
out somewhere for fun or

entertainment 6) @ C9 C9

c. A problem with your teachers
(any teachers you've had since

kindergarten) because of your g I
race or ethnicity Q Q Q Q
344
d. The police bothering you g; a:
because of your race or '43; 3';
ethnicity Q Q 'Q
4. Are you a peer mediator?
@Yes ' @No
5. If you are not a peer mediator would you like
to be one?
C?) Yes (D No
@ Not applicable

6. Have you used the peer mediation program
at your school to resolve a conflict that
you were having?

@ Yes (D No

7. if you haven’t used peer mediation to resolve
a conflict, why not?

Was not available
@Jidn’t have a conflict
@idn't want to use (why?)
@idn’t know enough about it
@Jke to solve my own conflicts
@ther

 

 

143

8. How often do you get in verbal (talking or yelling)
arguments or fights with other kids?

Q A lot Q Sometimes
Q A little (D Never

D. Family Questions:

1. How often have you and 5’.”
your family? ‘ s .
:ffg
3. moved in your lifetime? (9 C3) (9 0
b. moved in the past year? ®©®®

c. been without your own place Q , .
WM? ®@®®

d. stayed with relatives because you .
did not have a place of your own? ©©®®

e. stayed in a shelter? G) @ @‘0
a} f'

f. How often have you had to stay ,
with someone else because of the ‘ .

courts or because your parents .3”; "‘
were in trouble? (9 G G O
2. How true are these statements for A:
your parents or caregivers? 43 ~ 45’
3
Is
a. I share my thoughts and feelings .é‘oigg
with them (9 @ G 0
b. I enjoy spending time with my A .
family ® @ 9 CD
c. If I had a personal problem, i ‘ . l 7
could ask them for help (9 @ G 0

d. They would miss me if I didn't .
come home from school (9 @ G G

e. They want to know where i am or

where l'm going @ (9 G 0
r. They ask me about my day? @GG‘Q

144

2. How true are these statements

for your parents or ,5?
caregivers? (cont.) 5 ‘ ,5?
k 3
ff§ k
9. They ask me about the work I r a?

have completed in school G) (ID (‘9 CD

h. They meet the friends I hang

wmm C©Q®

i. They make sure I get to school
everyday. even if it means that
they have to take me themselves ®@ (9 (D

j. They spend time with me ®® @G
k. They tell me they love me G) G!) G) G)
I. They ask me to tell them what
IWM ©©®®
m. They listen to what I say 6) G) ® C9
n. They criticize me ©®®®
0. They enforce clear rules about
what I can and cannot do 9 @ C2) ®_.__

3. How are you usually punished when you
do something wrong?

Q My parents and I usually talk about it

Q l'm usually given a time-out or sent to my room

Q I’m usually given extra chores

Q I'm usually put on restriction or lose privileges
such as using the phone or watching t.v.

Q I’m usually grounded

Q I'm usually yelled at

Q) l'm usually hit or spanked

4. How often in the past year have

your parents or caregivers (3‘?
done any of the following? 5 ago 3
V a 4
. a. rewarded you ®©®®
b. praised you ®©®®

c. discussed an issue calmly with you @© (9 C9

d. sent you to your room @@ ®@

a. warned you not to do something

again @@@G

145

i. took possessions or privileges
away from you

Q. grounded you 7

h. scolded or yelled at you

i. spanked or slapped you

j. asked you to leave your home

k. taken you to church, temple or
other religious service

353;

©®®®
@@®®
®@®®
®®®®
®®®®

®®®®

5. What usually happens when your parents
or caregivers disagree with one another?

Q They usually talk it out

Q They usually walk away or stop talking to each

other for a while

Q They usually argue or yell at each other
Q They usually throw things at each other
Q They usually push or hit each other

6. What usually happens when you have a
disagreement with your parents or

caregivers?

(D We usually talk it out

Q I usually go to my room
Q We usually argue or yell
® Other:

 

7. How much do you agree with
these statements?

a. I admire, look up to, or want to be
like one of my parents or
caregivers

b. I admire, look up to. or want to be
like one of my siblings

c. I admire. look up to. or want to be

like one of my other relatives

146

©®®®

©©®®

8. How true are these statement for

your family?

a.

My dad or male caregiver thinks
it's okay if I get into a physical
fight at school

My dad or male caregiver thinks
it's okay for me to solve my
arguments by fighting

. My mom or female caregiver

thinks it's okay if I get into a
physical fight at school

. My mom or female caregiver

thinks it's okay for me to solve
my problems by fighting

9. My parents or caregivers feel its
okay for me to:

a.

b.

9.

drink beer, wine or hard alcohol
smoke cigarettes

smoke marijuana or take
other drugs

. steal

. disobey my teachers

get suspended from school

hit someone

10. How often do your parents or
caregivers do the following:

a.

b.

use drugs

challenge someone to a physical

fight

$8

©©®®

threaten to seriously hurt someone ®@ ® (9

. steal

. hit someone

147

©®®®
©©®®

11. How often do these things happen
in your family?

. Parents or caregivers argue?

Parents or caregivers yelling
and screaming at each other, or
one yelling or screaming at

the other?

Seen or heard your parents
or caregivers threaten to hurt
each other, or one threaten
to hurt the other?

. Parents or caregivers hurt
each other physically?

. How often are you afraid that
one of your brother or sisters
will hurt or try to hurt you?

12. In the past year, how often have you
seen your family members do
the following things?

a. Dad or male caregiver get
into a physical confrontation
or fight?

Mom or female caregiver get
into a physical confrontation
or fight?

. one or more of your male relatives
(like uncles. brothers, or
grandfathers) get into a physical
confrontation or fight?

get into a physical confrontation
orflght

148

. one or more of your female relatives
(like aunts. sisters. or grandmothers)

©®®®

oeoo

©®®®

@®®®

0’

$90 s
e s
w§45

®®®O

®®®O

@960

®@®®

E. Questions About Your
Peers & Friends:

1. How often do you do the following things when

you disagree with your peers

or friends? ‘5’
. . 55$?
a. get angry. but discuss the issue 'r v e
with them and work it out? G) 6) ® CD
b. yell or insult them? CD (:9 Q G)

c. suik and refuse to talk to them? @@ @G

d. push, grab. hit or throw things

at them? CD G G 0
e. other 7 Q @ ® 0

2. How true are the following

a.

statements for your friends? 3 “
Sgs .3
Teachers approve of most of my ’9 9‘ ea
friends 0 Q) G)
. My parents approve of most of my
friends Q 6) Q G)

. Adults (like parents and teachers)

think my friends are ‘good kids" 6) 6) Q G)

. My friends get good grades in

school 0990

. My friends are interested in school 63 @@ CD

149

§
0

3. My friends think it is alright to:

a?
.69 a? 49
a. get into a physical fight if another 0? ‘9 6 0,
students challenges them Q G ('9 O
b. take a weapon to school or carry
a weapon @000
c. challenge another student to a
physical fight G (3 O Q
d. attack someone with the idea of
seriously hurting them Q G G 0
e. skip school G G G O
i. use threats in order to have sex
with someone Q) 6) ® (9

9. take something they want from
someone even though it doesn't

belong to them. 9 Q) Q G

4. Think of your closest friends (Friends outside
of school too- anyone you spend a lot of time
with). in the past year, how often have any
of them:

a. gotten in trouble for misbehaving 5,595; f
in class. the lunchroom, ‘-

or hallway? @ Q) (9 G
b. been suspended from school for v

fighting? ‘ GDGDQQ
c. carried a weapon? 6) ® ® (9
d. been arrested? 6) ® ® 6)
e. tried to steal anything? 6) ® CD (D

l. screamed or swore at a teacher? 6) ® (19 G)

9. hit someone because they were

angry? @ @ Q G)
h. skipped classes without f a ‘
an excuse? 0 (5.) ® 6)

150

F. Community Questions:

1. How much do you agree with these

statements? 8 a:
s .3
'r c
5‘ 4" e
3. There are many abandoned é: g 35 08
houses or other buildings in my ‘5 V9 6 a?
neighborhood or community. G) (:9 ® (9

b. There is a lot of writing and other
graffiti on the houses and other
buildings in my neighborhood or

community. Q) @ C2) (9

c. There is a lot of litter on the
ground in my neighborhood

or community. 6) @ ® (9

d. There are a lot of alcoholics and/
or drug abusers in my
neighborhood or community 6) @ ® Q

e. l'm afraid that someone will
challenge me to a fight or attack
me while l’m out walking or
hanging out in my neighborhood
or community. Q) (a) Q Q

151

2. How well does each of these
statements describe your
neighborhood or community?

a.

3. How many gangs do you think

It is easy for me to tell a stranger
in my neighborhood or community
from somebody who lives there

The adults in my neighborhood
or community seem to like the
kids that live there

. People in my community or
neighborhood know each other

. It is fairly safe to walk in my
neighborhood or community

at night

. Neighbors take care of each others'
plants. pets, or children if needed
(if someone goes out of town)

Neighbors feel like a family

. Adults in my neighborhood know
the names of the kids who live

there

CD®®®

®©®®
®@®®

©©®®
®®®®

®@®®

353$

there are in your community ®@®C9

4. How often do people your age
who live in your neighborhood:

9.0.0:»

5. On average, how often do the
adults in your community do
any of the following?

a.
b.
c.

use tobacco?
use alcohol?
use drugs?

buy or sell drugs?

use drugs?
steal?
hit someone?

6. Do you think the adults in your community like

or trust the police?

0 Yes

7. Do you like or trust the police?

(2) Yes (D No

8. Do you think the police you see in your
community are there to help people?

(9 Yes G) No

9. Are there any adults in your community that
you admire or would like to be like when

you grow up?

®Yes (D No
5‘ 83" a
s5 3439’ «50’
10. It is easy to get a handgun ‘° *9 0 a?
in my community ®@®®
8*
is
o
9 S
11. How easy would it be for (in 5x 3“, s
you to get a handgun? g» «7 (005 6s
(96) (D
12. How often do these things
happen? o
s"

a. On the average. how often do the 3,.
adults in your community carry w éf

a weapon? ®® @C;

b. On average. how often do you
watch television shows that

contain Violence? 6 G G O

c. On average, how often do you
watch movies that contain

violence? 6) ® (9 O

 

 

Thank you for participating in this survey! Please
bring your survey to us when you're finished. We
will discuss the survey with you and answer any
questions you may have when everyone has
finished. We will also share the results of the
survey with you at a later date.

 

 

153

APPENDIX B

154

Significant Findings of Study
After Application of Family-Wise Bonferroni Adjustments

 

 

 

 

 

 

 

 

Overall Risk

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender — .000 .05 3 .02 Yes

Overall Risk

Urban/Rural — .031 .OS 2 .03 Yes-l

Overall Risk tailed

test

Race/Ethnicity .079 .05 1 .05 No

Overall Risk

Individual Risk

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender — .000 .05 3 .02 Yes

Individual

Risk

Urban/Rural — .000 .05 2 .03 Yes

Individual

Risk

Race/Ethnicity .023 .05 1 .05 Yes

Individual

Risk

Family Risk

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender — .042 .05 3 .02 No

Family Risk

Urban/Rural — .091 .05 2 .03 No

Family Risk

Race/Ethnicity .133 .05 l .05 No

Family Risk

 

155

Peer Risk

 

 

 

 

 

 

 

 

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender — Peer .000 .05 3 .02 Yes

Risk

Urban/Rural — .108 .05 2 .03 No

Peer Risk

Race/Ethnicity .733 .05 l .05 No

Peer Risk

School Risk

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender - .000 .05 3 .02 Yes

School Risk

Urban/Rural — .091 .05 2 .03 No

School Risk

Race/Ethnicity .358 .05 1 .05 No

— School Risk

Community Risk

t-test Obtained Original Divisor New Sig?
Significance Atha Alpha

Race/Ethnicity .096 .05 3 .02 No

Community

Risk

Gender — .122 .05 2 .03 No

Community

Risk

Urban/Rural -— .713 .05 1 .05 No

Community

Risk

 

156

Overall Protection

 

 

 

 

 

 

 

 

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender — .015 .05 3 .02 Yes

Overall

Protection

Race/Ethnicity .403 .05 2 .03 No

Overall

Protection

Urban/Rural— .914 .05 1 .05 No

Overall

Protection

Individual Protection

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Race/Ethnicity .004 .05 3 .02 Yes

— Individual

Protection

Urban/Rural— .149 .05 2 .03 No

Individual

Protection

Gender — .456 .05 1 .05 No

Individual

Protection

Family Protection

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Urban/Rural— .032 .05 3 .02 No

Family Prot.

Gender — .074 .05 2 .03 No

Family Prot.

Race/Ethnicity .572 .05 1 .05 No

Family Prot.

 

157

Peer Protection

 

 

 

 

 

 

 

 

t-tect Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender — .000 .05 3 .02 Yes

Peer Protect

Urban/Rural — .168 .05 2 .03 No

Peer Protect

Race/Ethnicity .653 .05 l .05 No

Peer Protect

School Protection

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Race/Ethnicity .221 .05 3 .02 No

School Protect

Gender — .254 .05 2 .03 No

School Protect

Urban/Rural — .817 .05 1 .05 No

School Protect

Community Protection

t-test Obtained Original Divisor New Sig?
Significance Alpha Alpha

Urban/Rural — .013 .05 3 .02 Yes

Community

Protection

Race/Ethnicity .1 10 .05 2 .03 No

Community

Protection

Gender — .142 .05 1 .05 No

Community

Protection

 

158

2-Way Intergctions

Which protective factors bufler overall risks?

 

 

 

 

 

 

 

 

 

 

2-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Overall risk by .005 .05 5 .01 Yes
individual protection

Overall risk by .070 .05 4 .01 No
community protection

Overall risk by school .197 .05 3 .02 No
protection

Overall risk by family .246 .05 2 .03 No
protection

Overall risk by peer .266 .05 1 .05 No
protection

Which protective factors bufler individual risks?

2—way Obtained Original Divisor New Sig?

Significance Alpha Alpha

Indiv risk by indiv .008 .05 5 .01 Yes
protection

Individual risk by .056 .05 4 .01 No
school protection

Individual risk by peer .245 .05 3 .02 No
protection

Indiv risk by family .520 .05 2 .03 No
protection

Indiv risk by com .936 .05 l .05 No

tection

 

 

 

 

 

 

159

 

Which protective factors bufler family risks?

 

 

 

 

 

2-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Family risk by _ .008 .05 5 .01 Yes

individual protection

Family risk by family .059 .05 4 .01 No

protection

Family risk by .371 .05 3 .02 No

community protection

Family risk by peer .518 .05 2 .03 No

protection

Family risk by school .685 .05 1 .05 No

protection

Which protective factors bufihr peer risks?

2-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Peer risk by indiv . 100 .05 5 .01 No

protection

Peer risk by family .136 .05 4 .01 No

protection

Peer risk by peer .181 .05 3 .02 No

protection

Peer risk by comm. .241 .05 2 .03 No

Protection

Peer risk by school .714 .05 l .05 No

protection

 

160

Which protective factors bufikr school risks?

 

 

 

 

 

2-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

School risk by .037 .05 5 .01 No

individual protection

School risk by school .193 .05 4 .01 No

protection

School risk by .458 .05 3 .02 No

community protection

School risk by family .708 .05 2 .03 No

protection

School risk by peer .850 .05 1 .05 No

protection

Which protective factors bufikr community risks?

2-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Community risk by .021 .05 5 .01 No

individual protection

Community risk by .102 .05 4 .01 No

school protection

Community risk by .137 .05 3 .02 No

peer protection

Community risk by .275 .05 2 .03 No

family protection

Community risk by .811 .05 1 .05 No

community protection

 

161

Three-way interactions: Individual Protection

 

 

 

 

 

 

 

 

3-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Urban/rural by risk by .009 .05 3 .02 Yes

individual protection

Race/Ethnicity by .292 .05 2 .03 No

Risk by Individual

Protection

Gender by Risk by .996 .05 1 .05 No

Individual Protection

T hree-way interactions: Community Protection

3-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Race/Ethnicity by .020 .05 3 .02 trend

Risk by Community

Protection

Gender by Risk by .050 .05 2 .03 No

Community Protection

Urban/Rural by Risk .359 .05 3 .05 No

by Community

Protection

Three-way interactions: Family Protection

3-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Urban/Rural by Risk .098 .05 3 .02 No

by Family Protection

Gender by Risk by .694 .05 2 .03 No

Family Protection

Race/Ethnicity by .840 .05 1 .05 No

Risk by Family

Protection

 

162

Three-way interactions: Peer Protection

 

 

 

 

 

3-way Obtained Original Divisor New Sig?
Significance Alpha Atha

Gender by Risk by .239 .05 3 .02 No

Peer Protection

Race/Ethnicity by .309 .05 2 .03 No

Risk by Peer

Protection

Urban/Rural by Risk .553 .05 1 .05 No

by Peer Protection

Three-way interactions: School Protection

3-way Obtained Original Divisor New Sig?
Significance Alpha Alpha

Gender by Risk by .291 .05 3 .02 No

School Protection

Urban/Rural by Risk .319 .05 2 .03 No

by School Protection

Race/Ethnicity by .412 .05 1 .05 No

Risk by School

Protection

 

163

LIST OF REFERENCES

164

LIST OF REFERENCES

Aiken, LS. and West, S.G. (1991). Multiple regression: Testing and interpreting
interactions. Newbury Park, CA: Sage.

Alexander, G.R., Massey, R.M., Gibbs, T., and Altekruse, J .M. (1985). F irearm-related
fatalities: An epidemiological assessment of violent death. American Journal of
Public Health, 75, 165-168.

Almeida, D., Wethington, E., & Chandler, A. (1999). Daily transmission of
tensions between marital dyads and mother-child dyads. Journal of Marriage and the
Family, 61, 49-57.

American Psychological Association (1993). Violence and youth: Psychology ’s response.
Washington DC.

Aneshensel, CS, and Sucoff, C.A. (1996). The neighborhood context of adolescent
mental health. Journal of Health and Social Behavior, 3 7, 293 -310.

Arbuckle, J .L. (1996). Full information estimation in the presence of incomplete data. In
G.A. Marcoulides, & R.E. Schumacker (Eds), Advanced structural equation
modeling: Issues and techniques (pp. 243-278). Mahweh, NJ: Lawrence Erlbaum.

Bailey, S.L. and Hubbard, RL. (1990). Developmental variation in the context of

marijuana initiation among adolescents. Journal of Health and Social Behavior, 31,
58-70.

Belknap, J. (1996). The invisible woman: Gender, crime, and justice. Belmont, CA:
Wadsworth Publishing Company.

Bell, CC. and Jenkins, El. (1993). Community violence and children on Chicago’s
Southside. Psychiatry, 56, 46-54.

Benard, B. (1993). Turning the corner: From risk to resiliency. Portland, OR:
Northwest Regional Educational Laboratory.

Benson, PL. (1997). All kids are our kids: What communities must do to raise caring
and responsible children and adolescents. San Francisco, CA: Jossey-Bass.

Bergsmann, LR. (1989). The forgotten few: Juvenile female offenders. Federal
Probation, 53(1), 73-78.

Blurnstein, A. & Wallrnan, J. (2000). The crime drop in America Cambridge University
Press: Cambridge; New York.

165

Bowman, P.I., & Howard, C. (1985). Race-related socialization, motivation, and
academic achievement: A study of black youths in three-generation families. Journal
of the American Academy of Child Psychiatry, 24(2), 134-141.

Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature
and design. Cambridge, MA: Harvard University Press.

Bronfenbrenner, U. (1986). Ecology of the family as a context to human development:
Research perspectives. Development Psychology, 22, 723-742.

Brook, J.S., Brook, D.W., Gordon, A.S., Whiteman, M., and Cohen, P. (1990). The
psychosocial etiology of adolescent drug use: A family interactional approach.
Genetic, Social, and General Psycholog Monographs, I 16.

Bry, B.H., McKeon, P., and Pandina, RJ. (1982). Extent of drug use as a function of
number of risk factors. Journal of A bnormal Psychology. 91, 273-279.

Bulcroft, R.A., Carmody, DC, and Bulcrofi, K.A. (1996). Patterns of parental
independence giving to adolescents: Variations by race, age, and gender of child.
Journal of Marriage and the Family, 58, 866-883.

Calhoun, G., Jurgens, J ., and Chen, F. (1993). The neophyte female delinquent: A review
of the literature. Adolescence, 28(110), 461-471.

Capaldi, D.M. and Patterson, GR. (1996). Can violent offenders be distinguished from
frequent offenders? Prediction from childhood to adolescence. Journal of Research
in Crime and Delinquency, 33, 206-231.

Caplan, G. (1964). Principles of preventive psychiatry. New York: Basic Books.

Chesney-Lind, M. (1988). Girls and status offenses: Is juvenile justice still sexist?
Criminal Justice Abstracts, 20(1), 144-165.

Clark-Lempers, D.S., Lempers, J .D., and Ho, C. (1991). Early, middle, and late
adolescents’ perceptions of their relationships with significant others. Journal of
Adolescent Research, 6, 296-315.

Coates, D.L. (1987). Gender differences in the structure and support characteristics of
Black adolescents’ social networks, Sex Roles, I 7, 667-687.

Coie, J.D., Watt, N.F., West, S.G., Hawkins, J .D., Asarnow, J.R., Markman, H.J., Ramey,
S.L., Shure, M.B., and Long, B. (1993). The science of prevention: A conceptual

framework and some directions for a national research program. American
Psychologist, 48(10), 1013-1022.

166

Costello, BI. and Dunaway, R.G. (2003). Egotism and delinquent behavior. Journal of
Interpersonal Violence, 18(5), 572-590.

Cotterell, IL. (1992). The relation of attachments and support to adolescent well-being
and school adjustment. Journal of Adolescent Research, 7, 28-42.

Cowen, EL. (1996). The ontogenesis of primary prevention: Lengthy strides and stubbed
toes. American Journal of Community Psychology, 24, 235-249.

Cronk, C., and Sarvela, PD. (1997). Alcohol, tobacco, and other drug use among
rural/small town and urban youth: A secondary analysis of the Monitoring the Future
data set. American Journal of Public Health, 87(5), 760-764.

Davis, NJ. (1999). Resilience: Status of the research and research-based programs —
Working Paper. Substance Abuse and Mental Health Services Administration,
Center for Mental Health Services.

Davis, K., Schoen, C., Greenburg, L., Desroches, C., and Abrams, M. (1997). The
Commonwealth Fund survey of the health of adolescent girls. New York:
Commonwealth Fund.

Dryfoos, J .G. (1990). Adolescents at risk: Prevalence and prevention. New York:
Oxford University Press.

Dubow, E.F., Roecker, CE, and D’Irnperio, R. (1997). Mental health. In R.T.
Ammerman & M. Hersen (Eds), Handbook of prevention and treatment with
children and adolescents: Intervention in the real world context (pp. 259-286). New
York: John Wiley & Sons. '

Eccles, J .S., Early. D., Fraser, K., Belansky, E., & McCarthy, K. (1997). The relation of
connection, regulation, and support for autonomy to adolescents’ functioning.
Journal of Adolescent Research, 12, 263-286.

Elliott, D.S. (1994). Serious violent offenders: Onset, developmental course, and
termination - The American Society of Criminology 1993 address. Criminology, 32,
1-21.

Ensminger, M.E. (1990). Sexual activity and problem behaviors among Black, urban
adolescents. Child Development, 61, 2032-2046.

Ensminger, M.B., Lamkin, RP, and Jacobson, N. (1996). School leaving: A longitudinal
perspective including neighborhood effects. Child Development, 67, 2400-2416.

Entwhisle, D.R., Alexander, KL, and Olson, LS. (1994). The gender gap in math: Its

possible origins in neighborhood effects. American Sociological Review, 59, 822-
838.

167

 

Farrington, DP. (1987). Early precursors of frequent offending. In J .Q. Wilson & G.C.
Loury (Eds), From children to citizens: Vol. 3. Families, schools, and delinquency
prevention (pp. 27-50). New York: Springer-Verlag.

F arrington, DP (1991). Childhood aggression and adult violence: Early precursors and
later-life outcomes. In DJ. Pepler & K.H. Rubin (Eds), The development and
treatment of childhood aggression (pp. 5-29). Hillsdale, NJ: Lawrence Erlbaum.

F lorsheirn, P., Tolan, P., and Gorrnan-Smith, D. (1998). Family relationships, parenting
practices, the availability of male family members, and the behavior of inner-city
boys in single-mother and two parent families. Child Development, 69 (5), 143 7-
1447.

Fraser, M.W. (1997). The ecology of childhood: A multisystems perspective. In M.W.
Fraser (Ed.), Risk and resilience in childhood: An ecological perspective (pp. 1-9).
Washington, DC: NASW Press.

Fraser, M.W. and Galinsky, M.J. (1997). Toward a resilience-based model of practice.
In M.W. Fraser (Ed.), Risk and resilience in childhood: An ecological perspective
(pp. 265-276). Washington, DC: NASW Press.

Fraser, M.W., Richman, J .M., and Galinsky, M.J. (1999). Risk, protection, and resilience:
Toward a conceptual framework for social work practice. Social Work Research, 23
(3), 131-143.

Garmezy, N. (1985). Stress-resistant children: The search for protective factors. In J .E.
Stevenson (Ed.), Recent research in developmental psychopathology (pp. 213-233).
New York: Pergamon Press.

Garry, EM. (1996). Truancy: First step to a lifetime of problems. Office of Juvenile
Justice and Delinquency Prevention, US. Department of Justice.

Germain, CB. (1991). Human behavior in the social environment: A n ecological view.
New York: Colombia University Press.

Glassner, B. (1999). The culture of fear: Why Americans are afiaid of the wrong things —
Crime, drugs, minorities, teen moms, killer kids, mutant microbes, plane crashes,
road rage, and so much more.

Gordon, K.A. (1995). Self-concept and motivational patterns of resilient African-
American high school students. Journal of Black Psychology, 21(3), 239-255.

Graf, R.G. and Alf, RF. (1999). Correlations redux: Asymptotic confidence limits for

partial and squared multiple correlations. Applied Psychological Measurement,
23(2), 116-119.

168

Guerra, N. (1998). Serious and violent juvenile offenders. In R. Loeber & D.P.
Farrington (Eds), Serious and violent juvenile oflenders: Risk factors and successful
interventions. Thousand Oaks: Sage.

Hansen, W.B., Graham, J .W., Shelton, D.R., Flay, BR, and Johnson, C.A. (1987). The
consistency of peer and parent influences on tobacco, alcohol, and marijuana use
among young adolescents. Journal of Behavioral Medicine, I 7, 135-154.

Hausman, A.J., Spivak, H., Prothrow-Stith, D., and Roeber, J. (1992). Patterns of teen
exposure to a community-based violence prevention project. Journal of Adolescent
Health, 13, 668-675.

Hawkins, J .D., Arthur, M.W., and Catalano, RF. (1995). Preventing substance abuse. In
M. Tonry & D.P. Farrington (Eds), Building a safer society: Strategic approaches
to crime prevention: Vol. I 9. Crime and Justice: A review of research (pp. 343-
427). Chicago: University of Chicago Press.

Hawkins, J .D., and Catalano, RF. (1992). Communities that care. San Francisco:
Jossey-Bass.

Hawkins, J ., Catalano, RF, and Miller, J .Y. (1992). Risk and protective factors for
alcohol and other drug problems in adolescence and early adulthood: Implications
for substance abuse prevention. Psychological Bulletin, 112 (1), 64-105.

Hawkins, D.P., Laub, J .H., and Lauritsen, J .L. (1998). Race, ethnicity, and serious
juvenile offending. In R. Loeber & D.P. Farrington (Eds), Serious and violent
juvenile offenders: Risk factors and successfid interventions. Thousand Oaks: Sage.

Hawkins, JD. and Weis, J .G. (1985). The social development model: An integrated
approach to delinquency prevention. Journal of Primary Prevention, 6.

Hirschi, T. (1969). Causes of Delinquency. Berkley, CA: University of California Press.

Howell, J .C., Krisberg, B., Hawkins, J .D., and Wilson, JJ. (1995). Sourcebook on serious,
violent, and chronic juvenile oflenders (pp. 275-278). Thousand Oaks, CA: Sage.

Huang, B., Kosterrnan, R., Catalano, R.F., Hawkins, J .D., and Abbott, RD. (2001).
Modeling mediation in the etiology of violent behavior in adolescence: A test of the
social development model. Criminology, 39(1), 75-107.

Huesmann, LR. and Bron, L.D. (1986). The development of aggression in American
children as a consequence of television violence viewing. In L.R. Huesmann & L.D.
Eron (Eds), Television and the Aggressive Child: A Cross-National Comparison.
Hillsdale, NJ: Lawrence Erlbaum Associates.

169

Huizinga, D., Loeber, R. and Thomberry, T.P. (1994). Urban delinquency and substance
abuse: Initial findings. Washington, DC: Office of Juvenile Justice and Delinquency
Prevention.

Jang, SJ. and Thomberry, T.P. (1998). Self-esteem, delinquent peers, and delinquency:
A test of the self-enhancement thesis. American Sociological Review, 63 (4), 586-
598.

Jessor, R., Donovan, J .E., and Costa, RM. (1991). Beyond adolescence: Problem
behavior and young adult development. New York: Cambridge University Press.

Jessor, R. and Jessor, S.L. (1977). Problem behavior and psychosocial development. San
Diego, CA: Academic Press.

Jessor, R., Van Den Bos, J., Vanderryn, J., Costa, F.M., and Ttu'bin, MS. (1995).
Protective factors in adolescent problem behavior: Moderator effects and
developmental change. Developmental Psychology, 31(6), 923-933.

Jessor, R., Van Den Bos, J., Vanderryn, J., Costa, F.M., and Turbin, MS. (1997).
Protective factors in adolescent problem behavior: Moderator effects and
developmental change. In G.A. Marlatt & G.R. Van Den Bos (eds), Addictive
behaviors: Readings on etiology, prevention, and treatment (pp. 23 9-264).
Washington, DC: American Psychological Association.

Juvonen, J. and Murdock, T.B. (1995). Grade-level differences in the social value of
effort: Implications for self-preservation tactics of early adolescents. Child
Development, 66, 1694-1705.

Kandel, DB. and Andrews, K. (1987). Processes of adolescent socialization by parents
and peers. International Journal of the Addictions, 22, 319-342.

Kandel, D.B., Kessler, R.C., and Magulies, RS. (1978). Antecedents of adolescent
initiation into stages of drug use: A developmental analysis. Journal of Youth and
Adolescence, 7, 13-40.

Kelly, J. (1966). Ecological constraints on mental health services. American
Psychologist, 21, 535-539.

Kirby, K.D. and Fraser, M.W. (1997). Risk and resilience in childhood. In M.W. Fraser
(Ed.), Risk and resilience in childhood: An ecological perspective (pp. 10-33).
Washington, DC: NASW Press.

Landrine, H., Richardson, J .L., Klonofl‘, E.A., and Flay, B. (1994). Cultural diversity in

the predictors of adolescent cigarette smoking: The relative influence of peers.
Journal of Behavioral Medicine, I 7, 331-346.

170

Leon, G.R., Fulkerson, J .A., Perry, CL, and Dube, A. (1994). Family influences, school
behaviors, and risk for the later development of an eating disorder. Journal of Youth
and Adolescence, 23, 499-515.

Leone, CM. and Richards, M.H. (1989). Classwork and homework in early adolescence:
The ecology of achievement. Special issue: The changing life space of early
adolescence. Journal of Youth and Adolescence, 18, 531-548.

Linney, J .A. and Wandersman, A. (1991). Prevention Plus 111: Assessing alcohol and
other drug prevention programs at the school and community level: A four-step
guide to useful program assessment. Rockville, NH): US. Department of Health and
Human Services, Public Health service, Alcohol, Drug Abuse, and Mental Health
Administration, Office for Substance Abuse Prevention.

Lipsey, M.W. and Derzon, J .H. (1998). Predictors of violent or serious delinquency in
adolescence and early adulthood. In R Loeber & D.P. Farrington (Eds), Serious
and violent juvenile oflenders: Risk factors and successfill interventions. Thousand
Oaks: Sage.

Loeber, R. (1990). Development and risk factors of juvenile antisocial behavior and
delinquency. Clinical Psychology Review, 10, 1-41.

Loeber, R. and Dishion, T.J. (1984). Boys who fight at home and school: Family
conditions influencing cross-setting consistency. Journal of Consulting and Clinical
Psychology, 52 (5), 759-768.

Loeber, R., F arrington, D.P., Stouthamer-Loeber, M., and Van Kammen, W.B. (1998).
Antisocial behavior and mental health problems: Explanatory factors in childhood
and adolescence. Mahwah, NJ: Lawrence Erlbaum.

Luthar, S.S., Doerenberger, C.H.., & Zigler, E. (1993). Resilience is not a unidimensional
construct: Insights from a prospective study of inner-city adolescents. Development
and Psychopathology, 5, 703 -71 7.

Luthar, S.S., & Zigler, E. (1991). Vulnerability and competence: A review of research on
resilience in childhood. American Jom'nal of Orthopsychiatry, 61(1), 6-21.

Maguin, E. and Loeber, R. (1996). Academic performance and delinquency. In M.
Tonry (Ed.), Crime and justice: A review of research (V 01. 20, pp. 145-264).
Chicago: University of Chicago Press.

Margalit, M. and Eysenck, S. (1990). Prediction of coherence in adolescence: Gender

differences in social skills, personality, and family climate. Journal of Research in
Personality, 24, 510-521.

171

I.

Martinez, P. and Richters, J .E. (1993). The NIMH community violence project, 11:
Children’s distress symptoms associated with violence exposure. Psychiatry, 5 6, 22-
35.

Masten, A. (1994). Resilience in individual development: Successful adaptation despite
risk and adversity. In M.C. Wang & E.W. Gordon (Eds), Educational resilience in
inner-city America: Challenges and prospects (pp. 3-26). Hillsdale, NJ: Lawrence
Erlbaum.

Masten, AS. (2001). Ordinary magic: Resilience processes in development. American
Psychologist, 56(3), 227-238.

McCord, J. and Ensminger, M. (1995). Pathways fiom aggressive childhood to
criminality. Paper prepared for The American Society of Criminology, Boston, MA,
November 15-18, 1995.

McCreary, M.L., Slavin, L.A., & Berry, El. (1996). Predicting problem behavior and
self-esteem among African American adolescents. Journal of Adolescent Research,
I 1(2), 216-234.

Miller, DB. (1999). Racial socialization and racial identity: Can they promote resiliency
for African American adolescents? Adolescence, 34 (135), 493-501.

Miller and Maclntosh (1999). Social Work Research, 23 (3), 159-169.

Moffitt, MA. and Wartella, E. (1992). Youth and reading: A survey of leisure reading
pursuits of female and male adolescents. Reading Research and Instruction, 31, 1-
17.

Mrazek, PJ. and Haggerty, R.J. (Eds). (1994). Reducing risks for mental disorders:
Frontiers for preventive intervention research. Washington, DC: National Academy
Press.

Murray, C.A. (1983). The physical environment and community control of crime. In
J .Q. Wilson (Ed.), Crime and public policy (pp. 107-122). San Francisco, CA:
Institute for Contemporary Studies.

Newcomb, MD. (1995). Identifying high-risk youth: Prevalence and patterns of
adolescent drug abuse. In E. Rahdert, D. Czechowicz, & I. Amsel (Eds), Adolescent
drug abuse: Clinical assessment and therapeutic intervention (pp. 7-3 8). Rockville,
MD: National Institute on Drug Abuse.

Newcomb, MD. and F elix-Ortiz, M. (1992). Multiple protective and risk factors for

drug use and abuse: Cross-sectional and prospective findings. Journal of
Personality and Social Psychology, 63(2), 280-296.

172

Office of Juvenile Justice and Delinquency Prevention (1998). Guiding principles for
promising female programming: An inventory of best practices. Office of Juvenile
Justice and Delinquency Prevention, US. Department of Justice.

Office of Juvenile Justice and Delinquency Prevention (2001 ). The 8% solution OJJDP
Fact Sheet #39. Office of Juvenile Justice and Delinquency Prevention, US.
Department of Justice.

Osgood, D.W., Johnston, L.D., O’Malley, F.M., and Bachman, JG. (1988). The
generality of deviance in late adolescence and early adulthood. American
Sociological Review, 53, 81-93.

Patterson, GR. and Dishion, T.J. (1985). Contributions of families and peers to
delinquency. Criminology, 23, 63-77.

Peters, M.F. (1985). Racial socialization of young Black children. In H.P. McAdoo and
J .L. McAdoo (Eds), Black children: Social, educational, and parental environments
(pp. 159-173). Newbury Park, CA: Sage.

Peterson, P.L., Hawkins, J .D., Abbott, RD, and Catalano, RF. (1994). Disentangling the
effects of parental drinking, family management, and parental alcohol norms on
current drinking by Black and White adolescents. Journal of Research on
Adolescents, 4(2), 203-227.

Poe-Yamagata, E. and Butts, J .A. (1996). Female oflenders in the juvenile justice
system: Statistics summary. Washington DC: US. Department of Justice, Office of
Juvenile Justice and Delinquency Prevention.

Pollard, J .A., Hawkins, J .D., and Arthur, A.W. (1999). Risk and protection: Are both
necessary to understand diverse behavioral outcomes in adolescence? Social Work
Research, 23 (3), 145-158.

Prescott, L. (1997). Adolescent girls with co-occurring disorders in the juvenile justice
system. New York: Policy Research, Inc.

Price, J .H. and Drake, J .A. (1999). Asset building: Rhetoric versus reality — A cautionary
note. Journal of School Health, 69(6), 215-216.

Reiss, A.J. and Roth, J .A. (Eds) (1993). Understanding and preventing violence.
Washington, DC: National Academy Press.

Resnick, M.D., Harris, L.J., and Blum, R.W. (1993). The impact of caring and

connectedness on adolescent health and well-being. Journal of Pediatrics and Child
Health, 29 (Suppl. 1), S3-S9.

173

Roeser, R.W. and Eccles, IS. (1998). Adolescents’ perceptions of middle school:
Relation to longitudinal changes in academic and psychological adjustment. Journal
of Research on Adolescence, 8, 123-158.

Rutter, M. (1985). Resilience in the face of adversity: Protective factors and resistance to
psychiatric disorder. British Journal of Psychiatry, 14 7, 598-611.

Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American
Journal of Orthopsychiatry, 5 7, 316-331.

Rutter, M. (1990). Psychosocial resilience and protective mechanisms. In J. Rolf, A.S.
Masten, D. Cicchetti, K.H. Nuechterlein, & S. Weintraub (Eds), Risk and protective
factors in the development of psychopathology (pp. 181-214). New York: Cambridge
University Press.

Rutter, M. and Giller, H. (1983). Juvenile delinquency: Trends and perspectives. New
York, NY: Penguin.

Safyer, A.W. (1994). The impact of inner-city life on adolescent development:
Implications for social work. Smith College Studies in Social Work, 64(2), 153-167.

Sameroff, A., Seifer, R., Barocas, R., Zax, M., and Greenspan, S. (1987). Intelligence
quotient scores of 4-year old children: Social-environmental risk factors. Pediatrics,
79, 343-350.

Sameroff. A., Seifer, R., Zax, M., and Barocas, R. (1987). Early indicators of
development risk: The Rochester Longitudinal Study. Schizophrenia Bulletin, 13,
383-3 93.

Sampson, R.J. (1986). Crime in cities: The effects of formal and informal social control.
In A.J. Reiss and M. Tonry (Eds), Crime and justice, volume 8. Chicago, IL:
University of Chicago Press.

Sampson, R.J., Raudenbush, SW. and Earls, F. (1997). Neighborhoods and violent
crime: A multilevel study of collective efficacy. Science, 277, 918-924.

Scales, RC. and Leffert, N. (1999). Developmental assets: A synthesis of the scientific
research on adolescent development. Minneapolis, MN: Search Institute.

Seydlitz, R. (1991). The effects of age and gender on parental control and delinquency.
Youth and Society, 23, 175-201.

Smith, C., Lizotte, A.J., Thomberry, T.P., and Krohn, MD. (1995). Resilient youth:
Identifying factors that prevent high-risk youth fi'om engaging in delinquency and
drug use. In Smith-Blau, Z. (Ed.) Current perspectives: Delinquency and disrepute
in the life course. Vol. 4. Greenwich, CN: JAI Press.

174

 

Smith, C. and Thomberry, T.P. (1995). The relationship between child maltreatment and
adolescent involvement in delinquency. Criminology, 33, 451-481.

Snyder, H.N. and Sickmund, M. (1999). Juvenile oflenders and victims: A national
report. Washington DC: US. Department of Justice, Office of Juvenile Justice and
Delinquency Prevention.

Steinberg, L., Mounts, N.S., Lambom, SD, and Dombusch, SM. (1991). Authoritative
parenting and adolescent adjustment across varied ecological niches. Journal of
Research on Adolescence, I, 19-36.

Svedhem, L. (1994). Social network and behaviour problems among 11-13 year-old
schoolchildren: A theoretical and empirical basis for network therapy. Acta
Psychiatrica Scandinavica, 89, 4-84.

Thomberry, T.P. (1994). Violent families and youth violence. OJJDP Fact Sheet #21.
Washington, DC: US. Department of Justice, Office of Juvenile Justice and
Delinquency Prevention.

Thomberry, T.P., Huizinga, D., and Loeber, R. (1995). The prevention of serious
delinquency and violence: Implications from the program of research on the causes
and correlates of delinquency. In J .C. Howell, B. Krisberg, J .D. Hawkins, and J .J .
Wilson (Eds), A sourcebook: Serious, violent, and chronic juvenile oflenders.
Thousand Oaks, CA: Sage.

Thomberry, T.P., Lizotte, A.J., Krohn, M.D., Famworth, M., and Jang, S.J., (1991).
Testing interactional theory: An examination of reciprocal causal relationships
among family, school, and delinquency. The Journal of Criminal Law &
Criminology, 82(1), 3-35.

Thorton, M.C., Chatters, L.M., Taylor, R.J., & Allen, W.R. (1990). Sociodemographic
and environmental correlates of racial socialization by black parents. Child
Development, 61 , 401-409.

Tolan, PH. (1996). How resilient is the concept of resilience? Community Psychologist,
29, 12-15.

Tolan, PH. and Guerra, N.G. (1994). What works in reducing adolescent violence: An
empirical review of the field Monograph prepared for the Center for the Study and
Prevention of Youth Violence. Boulder: University of Colorado.

Trojanowicz, RC. and Morash, M. (1992). Juvenile delinquency: Concepts and control.
Fifth Edition. Engiewood Cliffs, New Jersey: Prentice-Hall Inc.

175

u
.«n.

m

I???

Wasserman, GA. and Miller, L. (1997). The prevention of serious and violent youth
offending. In R. Loeber & D.P. Farrington (Eds), Serious and violent juvenile
ofirenders: Risk factors and successfitl interventions (pp. 197-247). Report by the
OJJDP Study Group, prepared for the Office of Juvenile Justice and Delinquency
Prevention. Thousand Oaks, CA: Sage Publications.

Werner, EB. (1984, November). Resilient children. Young Children, pp. 68-72.

Werner, EB. and Smith, RS. (1992). Overcoming the odds: High risk children fiom
birth to adulthood. Ithaca, NY: Cornell University Press.

Williams, J .H., Ayers, CD, and Arthur, M.W. (1997). Risk and protective factors in the
development of delinquency and conduct disorder. In M.W. Fraser (Ed.), Risk and

resilience in childhood: An ecological perspective (pp. 140-170). Washington, DC:
NASW Press.

Wills, T.A., Baccara, D., and McNamara, G. (1992). The role of life events, family
support, and competence in adolescent substance use: A test of vulnerability and
protective factors. American Journal of Community Psychology, 20, 349-3 74.

Wilson, J.Q. and Hemstein, R. (1985). Crime and human nature. New York: Simon and
Schuster.

Windle, M. (1992). A longitudinal study of stress buffering for adolescent problem .
behaviors. Developmental Psychology, 28, 522-530.

Wolin, S. and Wolin, S. (1995).-Resilience among youth growing up in substance-
abusing families. Pediatric Clinics of North America, 42, 415-429.

Wyman, P. (2003). Emerging perspectives on context specificity of children’s adaptation
and resilience: Evidence from a decade of research with urban children in adversity.
In S.S. Luthar (Ed.), Resilience and Vulnerability. Cambridge, UK: Cambridge
University Press.

Yoshikawa, H. (1994). Prevention as cumulative protection: Effects of early family

support and education on chronic delinquency and its risks. Psychological Bulletin,
1 I 5, 28-54.

176

 

   

Iiyiiiyiiiiifiiii«(my