USF Libraries
USF Digital Collections

A group-based approach to examining the association among risky sexual behavior, drug use, and criminal involvement in a...

MISSING IMAGE

Material Information

Title:
A group-based approach to examining the association among risky sexual behavior, drug use, and criminal involvement in a sample of newly arrested juvenile offenders
Physical Description:
Book
Language:
English
Creator:
Childs, Kristina K
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Substance use
Juvenile delinquency
Sexual practices
Problem behavior syndrome
Structural equation modeling
Dissertations, Academic -- Criminology -- Doctoral -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: This study focuses on the interrelationships between risky sexual practices, substance use, and arrest history. The sample consists of 948 newly arrested juvenile offenders processed at a centralized intake facility in 2006. A series of confirmatory factor analysis and structural equation modeling techniques are used to 1) determine if risky sexual behavior, marijuana and cocaine use, and arrest history form a unidimensional latent factor, 2) examine the direct effect of age on the latent factor, and 3) compare the factor structure, as well as the effect of age on the latent factor, across four demographic subgroups based on race and gender. Results provide moderate support for all three research objectives. Important similarities, as well as differences, in the factor structure across the four groups were found. The prevention and intervention implications of the findings, limitations of the current study, and directions for future research are discussed.
Thesis:
Dissertation (Ph.D.)--University of South Florida, 2008.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Kristina K. Childs.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 222 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002001274
oclc - 319699664
usfldc doi - E14-SFE0002626
usfldc handle - e14.2626
System ID:
SFS0026943:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 002001274
003 fts
005 20090430111217.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 090430s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002626
035
(OCoLC)319699664
040
FHM
c FHM
049
FHMM
090
HV6025 (Online)
1 100
Childs, Kristina K.
2 245
A group-based approach to examining the association among risky sexual behavior, drug use, and criminal involvement in a sample of newly arrested juvenile offenders
h [electronic resource] /
by Kristina K. Childs.
260
[Tampa, Fla] :
b University of South Florida,
2008.
500
Title from PDF of title page.
Document formatted into pages; contains 222 pages.
Includes vita.
502
Dissertation (Ph.D.)--University of South Florida, 2008.
504
Includes bibliographical references.
516
Text (Electronic dissertation) in PDF format.
3 520
ABSTRACT: This study focuses on the interrelationships between risky sexual practices, substance use, and arrest history. The sample consists of 948 newly arrested juvenile offenders processed at a centralized intake facility in 2006. A series of confirmatory factor analysis and structural equation modeling techniques are used to 1) determine if risky sexual behavior, marijuana and cocaine use, and arrest history form a unidimensional latent factor, 2) examine the direct effect of age on the latent factor, and 3) compare the factor structure, as well as the effect of age on the latent factor, across four demographic subgroups based on race and gender. Results provide moderate support for all three research objectives. Important similarities, as well as differences, in the factor structure across the four groups were found. The prevention and intervention implications of the findings, limitations of the current study, and directions for future research are discussed.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Co-advisor: Richard Dembo, Ph.D.
Co-advisor: John K. Cochran, Ph.D.
653
Substance use
Juvenile delinquency
Sexual practices
Problem behavior syndrome
Structural equation modeling
0 690
Dissertations, Academic
z USF
x Criminology
Doctoral.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.2626



PAGE 1

A Group-Based Approach to Examining the A ssociation among Risky Sexual Behavior, Drug Use, and Criminal Involvement in a Samp le of Newly Arrested Juvenile Offenders by Kristina K. Childs A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Criminology College of Behavioral and Community Sciences University of South Florida Co-Major Professor: Richard Dembo, Ph.D. Co-Major Professor: John K. Cochran, Ph.D. Christopher Sullivan, Ph.D. Steven Belenko, Ph.D. Date of Approval: November 17, 2008 Keywords: substance use, juvenile deli nquency, sexual practices, problem behavior syndrome, structural equation modeling, group-based modeling Copyright 2008, Kristina K. Childs

PAGE 2

Dedication This project is dedicated to two extraord inary people. To Gary, who walked into my life ten years ago and supported me ev er since. Thank you for your unwavering support, love, and belief in me. You have taught me so much. And, to my amazing mother, Cheryl, who is my best friend, great est cheerleader, helpfu l copy editor, and constant inspiration. The stre ngth and perseverance that you have exemplified over the years has made this journey possible for me. Witnessing the joy and passion that both of you have experienced in your car eers has inspired me to find this within my career. I could not have asked for better role models.

PAGE 3

Acknowledgements This project could not have been acc omplished without the assistance of many talented people. First, my sincerest th ank you to my mentor, Dr. Richard Dembo. You have extended me tremendous opportunities fo r which word cannot express my gratitude. Your patience, kindness, and wisdom are a gift to any student who is fortunate to work with you. Most importantly, it is your dedica tion to the field and sincere desire to improve the lives of adolescents that is an inspiration. Also, my si ncerest thank you to Dr. John Cochran. No matter how busy you were, you always took the time to listen and provide your honest advice – whether it was ab out which classes to take as a Masters student, my panic attacks when the word “comp” was mentioned, or when to go on the market as a doctoral candidate. Over the past five years, you have taught me so much. Also, thanks to Dr. Steven Belenko for allowing me to be a part of the STI/HIV research project, as well as your insightful comment ary and meaningful s uggestions throughout this project. I am also deeply indebted to Dr. Christopher Sullivan for his expertise and friendship over the past three years. The tim e and effort you have put into making this project a success is truly a ppreciated. You are an amazing role model and good friend. I would also like to thank Dr Paul Greenbaum. Your wisdom and willingness to help is greatly appreciated. Additionally, I would like to thank Dr. Michael Lynch, Dr. Jennifer Wareham, and Dr. Shane Jones for their cri tical commentary, as well their overwhelming support, throughout my graduate school experi ence. Whenever I needed to talk out an idea, you opened your door, listened, and talk ed it out with me. Fi nally, I would like to thank Laura Gulledge, Melissa Brownrigg, Dani elle Roda, Holly Haenel, Sarah Pasant, Melissa Cogswell, Raleigh Blasdell, and Gena Givens for their friendship and support.

PAGE 4

i Table of Contents List of Tables iv List of Figures vi Abstract vii Chapter 1 Introduction 1 Chapter 2 Substance Use and Adolescent Offending 8 Drug Use among the Adolescent Population 10 Drug Use and Adolescent Offending 13 Drug Use among Juvenile Offenders 17 Variations in the Substance Use-Delinquency Link 19 Gender 19 Race 22 Age 23 Possible Explanations for Substance Use-Delinquenc y Link 25 Chapter 3 Risky Sexual Behavior and A dolescent Offending 27 Risky Sexual Behaviors and STDs among the Adolescent Population 29 Risky Sexual Behaviors 29 Sexually Transmitted Diseases 31 Risky Sexual Behaviors and Adolescent Offending 32 Risky Sexual Behaviors and STDs among Incar cerated Adolescents 35 Risky Sexual Behaviors 35 Sexually Transmitted Diseases 37 Variations in the Risky Sexual Behavior-Delinquency Link 39 Gender 39 Race 40 Age 41 Chapter 4 Substance Use, Risky Sexual Behavior, and Adolescent Offending 44 Substance Use and Risky Sexual Behavior among Adolescents 46 Substance Users and Risky Sexual Behaviors 46 Using Substances Before or During Sexual Activity 50

PAGE 5

ii The Substance Use-Risky Sexual Behavi or Link across Delinquents and Nondelinquents 52 Substance Users and Risky Sexual Behaviors 53 Using Substances Before or During Sexual Activity 53 Substance Use and Risky Sexual Behaviors among Incarcerated Adolescents 54 Substance Users and Risky Sexual Behaviors 54 Using Substances Before or During Sexual Activity 56 Variations in the Risky Sexual Behavio r-Substance Use Link 58 Gender 58 Race 59 Age 60 Chapter 5 Explanations of Risk-Taking Behaviors in Adolescence 63 Social/Environmental Factors 64 Cognitive Developmental Factors 66 Personality Characteristics 68 Situational Factors 69 Problem Behavior Syndrome 70 Chapter 6 Problem Behavior Syndrome in Adolescence 72 Problem Behaviors Syndrome 73 Problem Behavior Theory 74 Disposition toward Deviance 76 Learning Theories 79 Evidence in Support of Problem Behavior Syndrome 84 Evidence against Problem Behavior Syndrome 88 Limitations of the Current Body of Research 90 Lack of Studies Involving Adolescent Offenders 91 Inconclusive Evidence on Variations in Problem Behavior Syndrome 93 Statistical Methods Used 98 Current Study 101 Chapter 7 Methods 102 Sample 102 Individual-Level Measures 106 UA Drug Test Results 106 Risky Sexual Behaviors 106 Criminal History 107 Demographic Characteristics 107 Analytic Steps 108 Multiple-Group Analyses 116

PAGE 6

iii Chapter 8 Results 125 Preliminary Analyses 125 Bivariate Analyses 125 Confirmatory Factor Analysis 130 Basic Structural Equation Model 132 Bivariate Analyses with Age 133 Structural Equation Model 134 Group-Based Confirmatory Factor Analysis 136 Bivariate Analyses 137 Separate CFA analyses 142 Unconstrained Group-Based CFA 144 Constrained Group-Based CFA 144 Final Group-Based CFA 145 Group-Based Structural Equati on Model 148 Chapter 9 Discussion and Conclusion 154 Implications of the Results 158 Limitations 165 Contributions to the Literature 169 Directions for Future Research 171 Conclusion 181 References 184 About the Author End Page

PAGE 7

iv List of Tables Table 1 STD Prevalence Rates of Incarcerated Adolescents in the US 37 Table 2 Sample Characteristics of the Weighted Sample 109 Table 3 Polychoric Correlations between the Observed Indicators 126 Table 4 Bivariate Relati onships between Marijuana Test Result and Risky Sexual Behavior, Arrest History, and Cocaine Test Result 127 Table 5 Bivariate Relationships betw een Cocaine Test Result and Risky Sexual Behavior, Arrest Histor y, and Marijuana Test Result 128 Table 6 Bivariate Relati onships between Risky Sexual Behavior and Arrest History, Marijuana Test Result, and Cocaine Test Result 129 Table 7 Bivariate Relationships betw een Arrest History and Risky Sexual Behavior, Marijuana Test Result, and Cocaine Test Result 130 Table 8 Confirmatory Factor Analysis 132 Table 9 Risky Sexual Behavior, Arre st History, Marijuana and Cocaine Test Result by Age 133 Table 10 Structural Equation Model 134 Table 11 Observed Indicators by Gender 138 Table 12 Observed Indicators by Race 139 Table 13 Observed Indicators by De mographic Subgroup 140 Table 14 Separate CFA Models across the Four Demographic Subgroups 143 Table 15 Final Group-Based CFA 147

PAGE 8

v Table 16 Bivariate Relationship be tween the Observed Indicators and Age across the Four Demographic Subgroups 148 Table 17 Group-Based St ructural Equation Model 153

PAGE 9

vi List of Figures Figure 1 Data Collection Protocol 105 Figure 2 General Structural Equa tion Model 114

PAGE 10

vii A Group-Based Approach to Examining the Association among Risky Sexual Behavior, Drug Use, and Criminal Involvement in a Samp le of Newly Arrested Juvenile Offenders Kristina K. Childs ABSTRACT This study focuses on the interrelation ships between risky sexual practices, substance use, and arrest hist ory. The sample consists of 948 newly arrested juvenile offenders processed at a centralized intake facility in 2006. A series of confirmatory factor analysis and structural equation modeling techniques ar e used to 1) determine if risky sexual behavior, marijuana and co caine use, and arrest history form a unidimensional latent factor, 2) examine the di rect effect of age on the latent factor, and 3) compare the factor structure, as well as th e effect of age on the latent factor, across four demographic subgroups based on race and gender. Results provide moderate support for all three research objectives. Important similarities, as well as differences, in the factor structure across the four groups were found. The prevention and intervention implications of the findings, limitations of th e current study, and directions for future research are discussed.

PAGE 11

1 Chapter 1 Introduction For decades, researchers and practitione rs have attempted to identify which adolescents are most at-risk for engaging in de viant behavior. This endeavor is fueled by the need to develop effective prevention and intervention strategies that are able to address the needs of these adolescents. Base d on this research, a variety of problem behaviors have been found to be quite fr equent among at-risk adolescents. These behaviors include substance use, skipping school, vandalism, fighting, risky sexual behavior, reckless driving, and delinquent activity. This st udy focuses on three of the more common problem behaviors found among adolescents: delinque ncy, drug use, and risky sexual behavior. It has been consistently documented that the occurrence of these three behaviors is considerably high during th e adolescent years. Therefore, the main objective of this study is to examine the in terrelationship between these risk behaviors among newly arrested juvenile offenders and to determine if thes e relationships are consistent across individual-level factors. It is well established that juvenile o ffenders display a wide range of antisocial problems including poor academic performance (Dembo, Williams, Schmeidler, & Howitt, 1991), low self-control (Gottfredson & Hirschi, 1990), negative peer relations (Warr, 2002), poor decision-making (Farringt on, 1993; see Piquero & Tibbets, 2002), and strained family relations (Dembo & Schm eidler, 2002). Furthermore, research

PAGE 12

2 consistently documents that juvenile offende rs are also disproportio nately involved in a variety of adolescent problem behaviors. Compared to adolescent nonoffenders, studies document substantially higher levels of tr uancy (Gottfredson, 2001; Hallfors, Cho, & Brodish, 2006), reckless drivi ng (Junger et al., 1995; Trem blay et al., 1995), gambling (Welte, Barnes, & Hoffman, 2004), running away from home (Chapple, Johnson, & Whitbeck, 2004), breaking curfew (Farrington et al., 1988), unsupervised time spent with peers (Osgood & Anderson, 2004) and, most notably, substance use (Elliott et al., 1989) among adolescent offenders. In regard to drug use, the higher levels found among juvenile offenders has been considered a critical problem for well over 25 years (Huizinga, Loeber, Thornberry, & Cothern, 2000). Public health research has also suggeste d that juvenile offenders are engaging in risky sexual behaviors at a substantially higher rate than nonoffenders (Barthlow, Horan, DiClemente, & Lanier, 1995; Teplin, Mericl e, McClelland, & Abram, 2003), which has resulted in disproportionately higher rates of sexually transmitted di seases (STDs) among this population (Joesoef, Kahn, & Weinstock, 2006; Kahn et al., 2005). For example, Kingree, Braithwaite, and Woodring (2000) estimate that 15% of male and 30% of female juvenile detainees are infected with an STD at any given time. This critical public health issue has been relatively overlooked and understudied in both the criminological literature and juvenile justice prevention and treatment services. However, risky sexual behaviors, most notably STD infection, are serious public health problems that require the attention of both disciplines. Not only do these behaviors pose significant risks to the health and well-being of juvenile offenders but they also pose harmful risk to the community by allowing the spread of these diseases.

PAGE 13

3 Studies also show that drug use and risky sexual behavior are interrelated phenomena. Youth who report substance use ar e substantially more likely to report risky sexual behavior (Bryan & Stallings, 2002; Harwell, Trino, Rudy, Yorkman, & Gollub, 1999) and adolescents are more likely to repor t engaging in risky sexual behavior while they are high on drugs or alcohol (Castruc ci & Martin, 2002; To lou-Shams, Brown, Gordon, & Fernandez, 2007). These findings, c oupled with the higher rates of both drug use and risky sexual behaviors found among j uvenile offenders, suggest that delinquent behavior, drug use, and risky sexual practices may be part of a general syndrome of deviance. Several researchers have suggested that the tendency to engage in any one form of deviant behavior is part of a ge neral syndrome towards deviance. That is, engaging in a particular form of deviant beha vior is actually one sy mptom of the larger “general syndrome” of deviance, commonly referred to as problem behavior syndrome (Jessor & Jessor, 1977). Overa ll, research has provided supp ort for this argument (for a review, see LeBlanc & Bouthill ier, 2003). Adolescents who re port engaging in a specific form of deviant behavior (i.e., offending, drug us e) are significantly more likely to report engaging in other deviant behaviors. Drawing on this work, it is critical th at juvenile justice system prevention and treatment services are able to target all three of these behaviors in an integrated program. The first step toward the development of such services is identifying the characteristics associated with this general syndrome. Re search suggests that individual factors are important predictors of all th ree forms of deviant behavior s, and therefore, should be considered when examining problem behavior syndrome. In particular, a wealth of research has documented marked variations in the sexual practices and substance use of

PAGE 14

4 juvenile offenders across race, gender, and ag e. Therefore, identifying variations in the interrelationships among these three forms of be havior is crucial to obtaining an accurate understanding of the nature of problem behavior syndrome. Accordingly, examining the interrelati onship between drug use, risky sexual behavior, and criminal involvement is the focu s of this study. In particular, a series of confirmatory factor analysis and structur al equation models (Muthn & Muthn, 2007) involving data on 948 recently arrested juvenile offenders are used to: 1) examine the covariation in drug use, criminal involvement, and risky sexual behaviors, 2) assess the effect of age on this relationship, and 3) determine whether th e strength of the relationship among these behaviors, as well as the effect of age, varies across demographic subgroups of offenders. This study will add to the existing body of research in several important ways. First, the findings are expected to have important criminological and public health implications. The identification of differe nt demographic subgroups of offenders based on their criminal involvement, drug use, a nd risky sexual behavior, will highlight the need for these disciplines to come together in an effort to improve the well-being of adolescent offenders. These three forms of problem behaviors have long been considered both criminological and public health priorities. Second, the large majority of research that has been conducted on juvenile offenders is based on either general adolesce nt samples consisting of a relatively low number of serious juvenile offenders, or, on incarcerated adolescent samples characterized by the most high-risk adolesce nts. Although these samples have provided very important information regarding risk behaviors, they are unable to provide an

PAGE 15

5 adequate representation of the extent of th ese behaviors across the juvenile offender population, as a whole. This study addresses this limitation by examining risk behavior among recently arrested juvenile offenders which includes adolescents ranging from first-time offenders to more serious, chronic offenders. Most importantly, the findings of this study will aid in the development of effective juvenile justice system prevention and intervention services in two important ways. First, the examination of the covariation in risk behaviors will provide information regarding which areas of treatment (e.g., s ubstance abuse services) may benefit from including services re lated to additional risk behaviors (e.g., sex education). Secondly, examining the variation in the strength of the association among risky sexual behavior, criminal involvement, and substance use ac ross demographic characteristics will provide valuable information regarding the similarities and differences in the service needs of atrisk adolescents. Toward this end, Chapter 2 begins with a descriptive summary of the research that has examined drug use among community sample s of adolescents, and more specifically, juvenile offenders. Then, differences in substance use among juvenile offenders are reported. Specifically, variations in substance use by age, gende r, and race are discussed. This discussion provides support for examining the variation in these behaviors across demographic categories, rather than examining these behaviors across the entire sample. Chapter 3 involves a summary of the litera ture examining the prevalence of STDs and risky sexual behaviors among the general adolescent population, as well as juvenile offenders. A discussion of the differences in these behaviors across age, gender, and race

PAGE 16

6 is also provided. Similar to Chapter 2, this discussion further supports the use of groupbased modeling. Chapter 4 explores the interrelations hip between drug use and risky sexual practices. The co-occurrence of substan ce use and risky sexual behaviors among adolescent offenders across demographic characteristics is also addressed. Chapter 5 discusses common explanations for the strong association among risky sexual practices, substance use, and delinquenc y. This discussion draws from research regarding the causes of adolescent risk-taki ng behavior, in general. Then, possible explanations are supported w ith studies that have been conducted on risky sexual behavior, delinquency, and drug use. The goa l of this chapter is to highlight the commonality in risk factors for all three fo rms of problem behavior which supports the basic notion that engaging in multiple form s of risk behaviors represents a general syndrome of problem behavior. Chapter 6 presents a discussion and literat ure review of the concept of problem behavior syndrome. This review summarizes the evidence in support of generality, as well as the evidence against this concept. The major limitations of this body of research follow. Based on these limitations, it is argued that the examination of the differences in the covariation in problem behaviors acr oss demographic subgroups of adolescent offenders is needed. The research questions guiding this study conc lude this chapter. Chapter 7 consists of an overview of th e methods used for the current study. The sample consists of 948 newly arrested juvenile offenders processed at an intake screening facility in Hillsborough County, FL. The data collection procedure, sample characteristics, description of the vari ables, and analytic plan are presented.

PAGE 17

7 Chapter 8 summarizes the results of the current study. The analyses discussed in this chapter include a number of hypothesis tests examini ng the bivariate relationships among each of the problem behaviors and the bivariate relationships among each of the problem behaviors and individua l-level factors. In addition, a series of confirmatory factor analysis and structural equations models are reported. Chapter 9 includes a summary of the main findings of the study. Then, a detailed discussion of the treatment and theoretical implications of the results follows. The contributions, as well as the limitations, of th e current study are also highlighted. Finally, suggestions for future resear ch conclude this chapter.

PAGE 18

8 Chapter 2 Substance Use and Adolescent Offending The period of adolescence appears to be a crucial time frame for the emergence of several problem behaviors, most notably de linquency and drug use (Loeber, StouthamerLoeber, & White, 1999). Studies suggest that adolescence is the period when initiation into alcohol, tobacco, and marijuana use is most likely to occur (Johnston, O’Malley, Bachman, & Schulenberg, 2007; Kosterman, Hawkins, Guo, Catalano, Abbott, 2000; Tubman et al., 2005) and when regular use of such substances is solidified (Kandel & Logan, 1984; Office of the National Drug Cont rol Policy, 2004). Furthermore, regular use of substances in adolescence is an importa nt step toward the escalation to substance abuse and substance dependence later in life (Ellickson, Hays, & Bell, 1992). Turning to delinquency, adolescence is also the most critical time period for the onset of criminal behavior (Loeber & Farr ington, 1998; Moffitt, 1993). Age of onset and the level of offending during adolescence are bo th strong and consistent predictors of adult criminality (Nagin & Farrington, 1992; Tracy & Kempf, 1996). Furthermore, the co-occurrence of these problem behaviors duri ng adolescence is quite prevalent and the negative consequences of engaging in both of these behaviors, compared to only one or none, are substantially greater (Barnes, Welte, & Hoffman, 2002; Dembo, Pacheco, Schmeidler, Fisher, & Cooper, 1997; Ellicks on & McGuigan, 2000; Huizinga, Loeber, &

PAGE 19

9 Thornberry, 1994; OAS, 2004; OAS, 2003; Menard, Mihalic, & Huizinga, 2001; Tubman, Gill, & Wagner, 2004). Therefore, over the past two and a half decades, a wealth of research examining has been conducted on adolescent drug use and delinquency. This research consists of surveys on general adolescent samples (CDC, 2006; Huizinga, Loeber, & Thornberry, 1993; Johnson, O’Malley, Bachman, & Schulenbu rg, 2007; Prinz & Kerns, 2003), justice involved or incarcerated youth (Abrantes, Hoffman, & Anton, 2005; Lebeau-Craven et al., 2003; Dembo et al., 1991; Dembo, Wareha m, & Schmeidler, 2006; Helstrom, Bryan, Hutchinson, Riggs, & Blechman, 2004; National Institute of Justice, 1999; Robertson, Dill, Husain, & Undesser, 2004; Teplin, Ab ram, McClelland, & Dulcan, 2002; TimmonsMitchell et al., 1997; Vaughn, Wallace, Davis, & Fernandes, 2007; Winters, Weller, & Meland, 1993), youth in drug treatment faciliti es (Steven, Estrada, Murphy, McKnight, & Tim, 2004), and official statis tics on drug abuse violations (Snyder, 2006). The bulk of these empirical studies focus on: 1) the prevalence of drug use among criminal and noncriminal adolescents, 2) variations in drug use and delinquency among different demographic and offending subgroups, 3) iden tifying and comparing the risk factors for these behaviors, and 4) examining the causal linkages and/or temporal ordering of these behaviors. Across these studies, a rather robust and enduring relationship between delinquency and substance use has emerged. Th is chapter provides a detailed review of this body of research including community sample s of adolescents, as well as adolescent offenders.

PAGE 20

10 Drug Use among the Adolescent Population Recent estimates suggest that drug use among the general adolescent population is declining (NIDA, 2007). According to th e Monitoring the Future (MTF) study, which consists of a nationally representative sa mple of 48,025 middle and high school students, past year use of any substance has declin ed 32% among eighth graders, 25% among tenth graders, and 13% among twelft h graders over the past five years (Johnson et al., 2007). In particular, since the peak in usage in 1996, marijuana rates have declined 18.3% among eighth graders, 25% among tenth grad ers, and 14% among twelfth graders. Significant declines in the reported use of cocaine, alcohol, methamphetamines, inhalants, and tobacco were also found in the 2007 MTF study. Despite recent declines, substance use among adolescents con tinues to remain relatively high (Dembo, Wareham, & Schmei dler, 2007a). According to the 2007 MTF lifetime prevalence rates, 19% of eighth graders, 36% of 10th graders, and 47% of 12th graders reported the use of any illicit s ubstance (Johnson et al., 2007). When broken down by specific drug category, lifetime preval ence rates for marijuana use were 14.2% of 8th graders, 31% of 10th graders, and 42% of 12th graders. Furthermore, 7.8% of 12th graders reported the use of cocaine, 6.5% reported lifetime use of MDMA, 3.0% reported lifetime use of methamphetamines, and 11.4% reported lifetime use of amphetamines. Alcohol was the most common substance used by adolescents in the study. Of the sample, 39% of 8th graders, 62% of 10th graders, and 72% of 12th graders reported using alcohol at least once in thei r lifetime, and; 18% of 8th graders, 41% of 10th graders, and 55% of 12th graders reported they had b een drunk in their lifetime.

PAGE 21

11 The National Youth Risk Behavior Surv ey (YRBSS) revealed similar results. This survey collects data from a nationally representative sample of 9th through 12th graders every two years (CDC, 2006). Drug us e trends obtained from the YRBSS also suggest that substance use among adolescents is declining. For example, from 2001 to 2005, the YRBSS reported declining rates of past 30 day use of alcohol, marijuana, cocaine, and tobacco. According to the 2005 YRBSS results, based on a sample of 13,953 high school students, 74.3% of the students reported li fetime use of alcohol, 43.3% reported drinking alcohol in the past month, and 25.5% had five or more drinks in one setting in the past month (CDC, 2006). Nationwide marijuana pr evalence rates were as follows: 38.4% reported lifetime use and 20.2% reported use in the past month. Cocaine prevalence rates were much lower, with 7.6% reporting life time use and 3.4% reporting past month use. Furthermore, 12.4% of the sample reported lif etime inhalant use, 4.0% reported lifetime steroid use, 8.5% reported lifetime hallucinoge n use, 2.4% reported lifetime heroin use, 6.2% reported lifetime methamphetamine use, and 6.3% reported lifetime ecstasy use. According to the 2003 National Survey on Drug Use and Health (NSDUH), nearly 8.6 million adolescents aged 12 to 17, or 34.3% of the adolescent population, had used alcohol in the past year (OAS, 2004), with roughly 2 million engaging in binge drinking.1 The same survey in 2002 estimated that 4 million youth (16%) aged 12 to 17 had used marijuana in the past year. Of these marijuana us ers, 38% used marijuana one to eleven days out of the year, 21% used 21 to 49 days, 23% used 100-299 days, and 9% used marijuana 300 or more days in the past year (OAS, 2003). 1 Based on the NSDUH estimates, ther e were approximately 25 million adolescents under the age of 17 in 2003 (OAS, 2004).

PAGE 22

12 Taken together, these nationally represen tative studies highlight the relatively high levels of drug and alcohol use occu rring among the adolescent population. In particular, all three of thes e studies indicate that roughly three-quarters of school aged children have used alcohol in their lifetime and 30 to 50% have used marijuana. These numbers are quite alarming, given the bulk of empirical evidence that suggests that adolescents who use alcohol a nd/or marijuana are substantia lly more likely to initiate other, more serious, forms of drug us e (Fergusson & Horwood, 1997; Fergusson & Horwood, 2000; Kandel & Logan, 1984; Kandel, Yamaguchi, & Chen, 1992; Kosterman, et al., 2000; Lessem et al., 2006; Menard et al., 2001) and to continue substance use into adulthood (Gfroerer, W u, & Penne, 2002; Office of the National Drug Council, 2004). Furthermore, it is well established th at substance use in adolescence places individuals at risk for a wi de range of physical and soci al problems, including negative family relations, poor academic performance, negative peer relations, mental health problems, and most notably, involvement in delinquent behavior. In particular, the prevalence of substance use is extrem ely high among adolescent offenders (Dembo, Wareham, Poythress, Cook, & Schmeidler 2006; Dembo et al., 2007a; Huizinga & Jakob-Chien, 1998; Huizinga, Loeber, Thor nberry, Cothern, 2000; Neff & Waite, 2007; Teplin, Mericle, McClelland, & Abram, 2003; Welte, Barnes, Hoffman, & Dintcheff, 1999). As such, an overview of the large body of research regarding the substance usedelinquency link is provided below.

PAGE 23

13 Drug Use and Adolescent Offending In 2005, over 2.1 million juvenile arrests were made. Of these arrests, 191,800 (9%) were for drug abuse violations. Over th e past ten years, alcohol and drug related arrests have steadily declined (Puzzanch era, Adams, Snyder, & Kang, 2007). For instance, in 1997, 715 per 100,000 youth aged 10-17 were arrested for a drug use violation and 578 per 100,000 youth were arrested for an alcohol relate d incident (i.e., driving under the influence, liquor law violation). In 2005, 570 per 100,000 youth aged 10-17 were arrested for a dr ug abuse violation and 476 per 100,000 youth were arrested for an alcohol related incident. However, as can be seen, even with these steady declines the prevalence of substance us e among juvenile offenders continues to remain extremely high. National statistics suggest that the prevalence of drug and alcohol use among adolescent offenders is a persistent and serious problem. For instance, the National Center on Addiction and Substance Abuse ( 2004) estimates that 78.4% of youth involved in the juvenile justice system are under the influence of alcohol or drugs while committing their crimes, test positive for drugs, are arrested for committing an alcohol or drug offense, admit to having substance abus e problems or share some combination of these characteristics. Such estimates highlig ht a critical social problem, considering the negative consequences of substance use am ong adolescent offenders (Stice, Myers, & Brown, 1998). Not only is involvement with drugs or alcohol associated with an increased likelihood of continued and serious contact with the juve nile justice system (National Center on Addiction and Substance Abuse, 2002), it is also related to higher rates of offending, more serious offending, a nd a longer duration of a criminal career

PAGE 24

14 (Greenwood, 1992; Huizinga et al., 1994; Hu izinga & Jacobs-Chien, 1998; Loeber, Green, Lahey, Frick, & McBurnett, 2000; Sealock, Gottfredson & Gallagher, 1997). Furthermore, longitudinal research has dem onstrated considerable continuity in the relationships between substa nce use and delinquency (Le Blanc & Loeber, 1998; Loeber et al., 1999). These studies suggest th at delinquency and substance use have contemporaneous relationships (Dembo et al., 2007a; Fagan, Weis, & Cheng, 1990; Horney, Osgood, & Marshall, 1995; Welte, Barnes, Hoffman, Wieczorek, & Zhang, 2005; White et al., 1999), even when controlling for important personality, family, and peer characteristics. To add to this, studi es also suggest that delinquency and drug use are two of the most treatment resistant form s of problem behavior found in adolescence (Huizinga et al., 1994; Mann, 2003). In general, an extensive body of resear ch, spanning several decades, consistently reveals higher levels of substance use among juvenile offenders, compared to nonoffenders (Beachy, Peterson, & Pearson, 1979; Elliott, Huizinga, & Menard, 1989; Huizinga et al., 1993; Johnson, Wish, Schm eidler, & Huizinga, 1991; OAS, 2003; OAS, 2004). Results from the 2002 NSDUH revealed that the percentage of youth reporting delinquent behavior was higher among mariju ana users than non-drug users, and the percentage of youth engaging in delinquent be havior rose with increasing frequency of past marijuana use (OAS, 2003). For exampl e, 2.9% of youth who reported no past year drug use stole something worth more than $50, whereas, 31.7% of youth reporting a high frequency of marijuana use (i.e., more than 300 times) admitted to the offense. Johnson et al. (1991) examined the relati onships between diffe rent typologies of substance users and delinquent offenders using cross-sectiona l data from the 1979

PAGE 25

15 National Youth Survey (NYS). This study invo lves a nationally repr esentative sample of 1,539 youth aged 14-20. On average, multiple index offenders who reported no substance use engaged in 5 delinquent acts, alcohol only users enga ged in 57 delinquent acts, marijuana users reported 89 delinque nt acts, and cocaine users reported 290 delinquent acts in one year. Among minor non-serious offenders, delinquency rates were over three times greater for cocaine users than nondrug users. Substance users comprised 30% of the NYS sample and accounted for over 85% of the delinquent acts reported. Elliot et al. (1989) examined the relationship between drug use and delinquency using three waves of data from the NYS (1976, 1980, and 1983). When the sample was broken down into drug use categories, results revealed a clear or dering of delinquency rates that increased from nonusers, to alcohol users, to marijuana users, and then to polydrug users (i.e., amphetamines, barbiturat es, cocaine, heroin, hallucinogens). For example, in 1983, 2% of nondrug users were in volved in index offenses, while 6% of alcohol users, 12.5% of marijuana users, a nd 23.5% of polydrug users engaged in index offenses. When the authors broke down the sample according to offending severity, similar results emerged; as se verity in offending increased, so did the prevalence rates for substance use. In 1980, 32.7% of nonoffenders were marijuana users compared to 72.2% of nonserious offenders and 85.1% of serious o ffenders. In regard to serious drugs, 8% of nonoffenders, 47.5% of nonserious offenders and 55.2% of serious offenders reported use. More recently, Huizinga & Jakob-Chien (1998) reviewed the findings from the 1990 Denver Youth Study which is comprised of 1,184 adolescents aged 11-17. Overall,

PAGE 26

16 serious juvenile offenders displayed the hi ghest prevalence rates, frequency, and abuse patterns for alcohol and marijuana and these rates declined as seriousness of offending declined. Similar results were obtained from the 1990 Pittsburgh Youth Study (n = 1,157) and the 1990 Rochester Youth Study (n = 1000) (Huizinga et al., 1993). These relationships were consistent across ag e, gender, and race (Huizinga, Loeber, & Thornberry, 1995; Thornberry, Hu izinga, & Loeber, 1995). A number of additional studies have found a strong relationship between violence and substance use during adolescence (D erzon & Lipsey, 1999; Farrington, 1998; Fergusson, Lynskey, & Horwood, 1996; Loeb er et al., 1999; Wagner, 1996; White, 1997). These studies highlight the notion th at, not only are drug use and delinquency related, but the extent of involvement in one be havior is strongly rela ted to the extent of involvement in the other. Additional general population studies on the relationship between substance use and delinquency have found that: 1) the earlie r the initiation of alcohol (Barnes et al., 2002; Lo, 2000; Newcomb & McGee, 1989) an d marijuana (Van Kammen, Loeber, & Stouthamer-Loeber, 1991) involvement, the grea ter chances of involvement in delinquent behavior, 2) being high on al cohol or drugs increases the odds of engaging in delinquent behavior, most notably viol ent behavior (Fergusson et al., 1996; Wagner, 1996; Zhang, Wieczorek, & Welte, 1997), 3) initiation of crime generally pro ceeds initiation of substance use (Bui, Ellickson, & Bell, 2000; Chaiken & Chaiken, 1990; Elliott et al., 1989; Huizinga & Jakob-Chien, 1998; Loeber et al., 1999; Menard et al., 2001), and 4) the substance use-delinquency association pe rsists throughout adolescence (Elliott et al., 1989; Loeber et al., 1999; Menard et al., 2001).

PAGE 27

17 As can be seen, these studies highlig ht a strong and consistent relationship between adolescent offending and drug use. However, a major limitation of research based on the general adolescent population is the small number of offenders, particularly serious offenders, included in the samples. As a result, the variability in adolescent offending, as well as additional problem be haviors that are co mmon among juvenile offenders, is somewhat low which inhibits the amount of detail on the behavior of juvenile offenders that can be obtained. As such, it is also important to consider the prevalence of substance use among samples co mprised of only juvenile offenders. To date, the majority of resear ch that examines substance use among juvenile offender populations is based on detained adolescents. These studies are also important because they provide information regarding the preval ence of substance use among juveniles at the back-end of the juvenile justice syst em, who tend to be more serious juvenile offenders. Drug Use among Juvenile Offenders Research involving justice-i nvolved adolescents also de monstrates a strong and consistent association between substance us e and delinquent behavi or (Belenko & Logan, 2003; Dembo et al., 1987; Dembo et al., 1991; Dembo, Williams, Fagan & Schmeidler, 1993; Dembo, Williams, Wothke, & Schmeidl er, 1994; Dembo, Wareham et al., 2006; Inciardi, Pottieger, Forney, Chitwood, & McBride, 1991; Potter & Jenson, 2003; Vaughn, Freedenthal, Jenson, & Howard, 2007; Winter et al., 1993). One of the most well-known trend studies of incarcerated offenders, the Arrestee Drug Abuse Monitoring Program (ADAM), provided self-reported data and dr ug test results on in carcerated juvenile offenders in nine cities across the United Stat es (National Institute of Justice, 2000).

PAGE 28

18 Using the ADAM data collected in 1997, Be lenko, Sprott, & Peterson (2004) reported that 75% of arrested adolescents reported a history of involvement with alcohol or other drugs. The 1999 results of this study indicated that the percen tage of juveniles, across the nine cities that tested positive for any one drug ranged from 38% to 69% (NIJ, 2000). Marijuana was the most commonly used drug and cocaine was the second most commonly used drug. Across the nine cities the median value for the percent of detainees who tested marijuana positive wa s 53% for males and 38% for females. Another well known study, the Northwestern Juvenile Project, has also provided information regarding the prevalence of subs tance use and substance use disorders (SUD) among incarcerated adolescents (Abram, Tepli n, McClleland, & Dulcan, 2003; Teplin et al., 2005; Teplin et al., 2003). This longit udinal study involved 1,829 juveniles detained in Cook County, IL between 1995 and 1998. McClelland, Teplin, & Abram (2004) reported that 77% of the youth included in the sample reported us ing alcohol or other drugs in the past 6 months. Nearly half of the detainees in this study met diagnostic criteria for one or more SUDs and over 20% met diagnostic criteria for two or more SUDs in the past six months (McClelland, Elkington, Teplin, & Abram, 2004). Based on a portion of the original sample, Teplin et al. (2003) reported substance use prevalence rates among 800 of the detain ees arrested between 1997 a nd 1998. Over 90% of the sample reported ever using marijuana; 60.7% of females and 68.9% of males reported using marijuana more than three times in the past three months, and; 21% of females and 14.6% of males reported ever using “other ” substances in their lifetime. In a study of 278 juvenile offenders pa rticipating in a l ongitudinal research project, Dembo and associates also found high prevalence rates of substance use. Of the

PAGE 29

19 sample, 61% of youth reported ever using marijuana and 21% reported using the drug more than 100 times (Dembo et al., 2007a). Furthermore, youth who reported higher frequencies of delinquent beha vior also reported higher frequencies of drug use. In another study, Dembo et al. (2007b) reported the prevalence of cocaine use. Based on both self-reported and/or hair test results, over 30 % of the sample were cocaine involved at some point during their lifetime. Taken together, the bulk of empirical research indicates that the association between substance use and juvenile offending is quite robust. In sum, this body of research suggests that 1) juvenile offenders are substantially more likely to engage in substance use compared to nonoffenders, 2) hi gher levels of involvement in delinquent behavior is associated with hi gher levels of substance use, an d 3) this association tends to persist throughout adolescence. Yet, variations in the st rength of these associations across important demographic factors includi ng age, gender, and race have also been documented (Barnes et al., 2002; Belenko et al ., 2004; Dembo et al., 2007a; Elliott et al., 1989; Huizinga & Jakob-Chien, 1998; Teplin et al., 2005). As such, it is important to consider the variation in drug use among j uvenile offenders across these demographic groups. Variations in the Substance Use-Delinquency Link Gender. On average, nationally representative studies indicate that male adolescents engage in substance use at a hi gher rate than female adolescents (CDC, 2006). However, research suggests that th e prevalence and pattern of substance use among juvenile offenders vary across gender groups (Helstrom et al., 2004; Huizinga & Jakob-Chien, 1998; Timmon-Mitchell et al., 1997) In regard to alcohol use, studies

PAGE 30

20 reveal similar prevalence rates across male and female juvenile offenders. Belenko et al. (2004) reported that 33% of females and 34% of males were alcohol involved; Teplin et al. (2003) reported 48% of ma les and 46.2% of females reported alcohol use more than three times in the past three months and Ne ff and Waite (2007) repor ted nearly identical rates of alcohol use across gender groups. Ho wever, these studies also indicate that females tend to initiate alcohol use earlier than their male counterparts. Gender differences in illicit drug use ar e somewhat complex. On average, the majority of studies suggest that male juvenile offenders report highe r levels of marijuana use (Barnes et al., 2002; Dembo et al. 2007a; Elliott et al., 1999; Johnston, et al., 2007; Steven et al., 2003). For example, Belenko et al. (2004) reported that 21% of arrested girls and 41% of arrested boys included in the ADAM project tested positive for marijuana and 73% of girls and 83% of boys re ported ever trying the drug. Wei, Makkai, & McGregor (2003) found higher marijuana positive rates in their sample of 493 juvenile detainees. In particular, 40% of female a nd 50% of male detainees tested marijuana positive. Teplin et al. (2003) found that 78% of males detainees, compared to 68% of female detainees, reported past month use of marijuana. However, Neff and Waite (2007) reported similar rates of marijuana use across gender groups and Helstrom et al. (2004) and Boyle et al. (1992) reported higher rates of marijuana use among female juvenile offenders. Typically, female juvenile offenders are found to report ear lier initiation and higher levels of serious dr ug involvement, such as cocaine and other polydrug use (Inciardi et al., 1991; Kim & Fendrich, 2002; Neff & Waite, 2007; Wei et al., 2003). For example, Belenko et al. (2004) reported that gi rls were significantly more likely to have

PAGE 31

21 tried cocaine, crack, amphetamines, and inhalant s. In the Northweste rn Juvenile Project, 21% of females and 14.6% of males reported li fetime use of “other drugs” (i.e., serious drugs other than marijuana) and 1.8% of fema les, compared to 0.8% of males, reported initiation of serious drug use prior to the age of 13 (Teplin et al., 2003). Stevens et al. (2003) found that males reported significantl y higher levels of marijuana use whereas females reported significantly higher levels of cocaine, heroin, and methamphetamine use. Females juvenile detainees also tend to display SUDs at a substantially higher rate than male detainees (Abrantes et al., 2005; Teplin et al., 2002; Timmon-Mitchell et al., 1997; Winters et al., 1993). The Office of Juvenile Justice Delinquency Prevention (OJJDP) also reported important gender differences in trends of a rrests for drug abuse viol ations (Snyder, 2006). From 1995 – 2004, the number of males arrested for drug abuse violations declined 8% but arrests for drug abuse viol ations increased 29% for females. However, females accounted for 27.2% of the overall cases heard in juvenile courts, but only 19.1% of drug-related cases (Puzzanchera & Kang, 2007) which means that female juvenile offenders continue to be underrepresen ted in drug abuse court cases. As can be seen, several discrepancies in substance use among juvenile offenders exist across gender groups. On one hand, male juvenile offenders typically reveal higher levels of marijuana use and greater numbers of drug abuse violations, however, females typically reveal higher levels of serious drug involvemen t and tend to initiate substance use at an earlier age. Thus, based on these in consistencies, it is important to account for gender effects when examining substance use among juvenile offenders. Failing to do so could mask important gender-specific differenc es in the association between drug use and

PAGE 32

22 delinquency, which in turn, could result in an inaccurate understandi ng of the nature of this association. Race. Interestingly, racial differences in subs tance use tend to be in contrast to racial differences found in criminal o ffending. On one hand, African American adolescents tend to report the highest rates of delinquent behavior (H uizinga et al., 1994; Vaughn et a., 2007), most notably serious delinquen t behavior (Elliott, 1994; Elliott et al., 1989; Hawkins, Laub, Lauritsen, & Cothern, 20 00; Loeber & Farrington, 1998). On the other hand, white adolescents, on average, are more likely to engage in substance use (Barnes et al., 2002; Elliott et al., 1989; Johnston et al., 2007). Studies examining racial differences in substance use among justice involved populations provide inconclusi ve results. According to the 1997 ADAM data, African American arrestees were more likely to test drug positive than white arrestees. Yet, whites self-reported substantially higher ra tes of lifetime prevalence of marijuana, cocaine, crack, amphetamines, and inhalants (Belenko et al., 2004). De mbo et al. (2007a) found that, compared to white offenders, Af rican American offenders reported lower levels of alcohol and marijuana use, however, they did not find any significant differences in cocaine use (see Dembo, et al ., 2007b). Stenmark, Wackwitz, Peffrey, and Dougherty (1974) found white juvenile drug offe nders to be heavie r substance users on all types of drugs including alc ohol, marijuana, and “other” il licit drugs; and Teplin et al. (2005) and Vaughn et al. (2007) found that white detainees were more likely to have substance-related problems, including SUDs blacking out, and earlier age of onset. In regards to official data, African American offenders make up a disproportionate amount of drug related arrest s. For example, in 2004 African American

PAGE 33

23 adolescents comprised 17% of the total popula tion of adolescents under the age of 18, but accounted for 27% of juvenile arrests for drug abuse violations (Snyder, 2006). Interestingly, examination of arrests for drug abuse violations from 1980 to 2004 reveal quite different trends. For white juvenile dr ug abuse violations, a rrests peaked in 1997 and have continuously declined, by 9%, sin ce that time. For Af rican American drug abuse violations, arrests peaked in 1995 and ha ve declined 44% since then. It is quite possible that the inconsistencies found between self-reported drug us e and official drug violation arrest statistics are due to racial bias in arrest procedures. However, other possible reasons may be related to 1) the wil lingness of African American adolescents to report drug use (Rosay, Najaka, & Herz, 2007), or 2) drug use and drug use violations represent different phenomena (e.g., drug use versus drug sales). Regardless of the nature and direction of the disparity in subs tance use among racial groups, it is apparent that important racial differences in the asso ciation among drug use a nd delinquency exist. Thus, it is imperative that racial charact eristics are accounted for when examining substance use patterns among juvenile offenders. Age. The bulk of empirical evidence suggest s that age influences the substance use-delinquency link in several ways. First, studies on both adoles cent populations and juvenile offenders suggest that, as an individual moves through adolescence, the likelihood of engaging in substance use and delinquency increases (E lliott et al., 1989; Kelley, Huizinga, Thornberry, & Loeber, 1997; LeBlanc & Loeber, 1998; Menard et al., 2001; White et al., 1999). For example, in a sample of 5,045 students, Tubman et al. (2004) found that the proportion of students reporting both deli nquency and substance use increased four fold from early to late a dolescence. Second, the earlier the age of

PAGE 34

24 initiation into substance use, the greater the likelihood of se vere and persistent drug use (Office of the National Drug Control Po licy, 2004; Prinz & Ke rns, 2003; Potter & Jenson, 2003). Third, youth who initiate deli nquent behavior at an early age are more likely to use substances throughout adolescen ce (Farrington, 1995; Menard et al., 2001; Van Kammen & Loeber, 1994). Research conducted on juvenile offende r populations highlight s a positive linear relationship between substance use and age (Loeber et al., 19 99; McClelland et al., 2004; Teplin et al., 2003). For inst ance, Dembo et al. (2007a) repo rted higher levels of heavy drinking and marijuana use among the older ju stice-involved youth in their study. In addition, arrests for drug abuse violations al so fit this linear re lationship. In 2004, youth aged 12 and under accounted for 2.7% of drug abuse violations, youth aged 15 -17 accounted for 39.8% and youth 16 and olde r accounted for 57.5% (Snyder, 2006). As can be seen, variations in the su bstance use-delinquency link are complex and somewhat inconsistent across studies. On av erage, female offenders (compared to male offenders), white offenders (compared to minority offenders), and older adolescent offenders tend to report more serious use of substances. Although these relationships are far from conclusive, it is quite clear that th ese socio-demographic characteristics need to be accounted for when considering the asso ciation among substan ce use and delinquent behavior. Failing to account for these charac teristics has the potenti al to lead to an inaccurate understanding of the co-occurrence of the behaviors among juvenile offenders, in turn, leading to the assumption that the n eeds of juvenile offenders are similar across demographic categories, when in fact, they may be substantially different. Indeed, a large body of research highlights racial and ge nder differences in cultural expectations,

PAGE 35

25 socialization practices, and cognitive development that could potentially lead to variations in problem behaviors across th ese demographic characteristics (Bennett, Farrington, & Huesmann, 2005; Kotchick et al., 2001; Morash & Chesney-Lind, 1991). Possible Explanations for the Substance Use-Delinquency Link Even with the extens ive body of research conduc ted on the substance usedelinquency link, the cause and direction of this association remains unanswered (Wagner, 1996). Currently, three main expl anations for the strong and consistent relationship found between drug use and deli nquency have been discussed in the literature: 1) substance use causes crime, 2) crime causes substance use, and 3) the relationship between substance us e and crime is spurious (Jes sor & Jessor, 1977; Jessor et al., 1991; Menard et al., 2001; White, 1990). Reviews of the research on drug use and delinquency conclude that there is little evidence that s ubstance use causes crime (White et al., 1999; White, 1997). In fact, most longitudinal studies find that delinquency developmentally precedes the initiation in to substance use (Cha iken & Chaiken, 1990; Elliot et al., 1989; Menard et al., 2001). Although this finding provides preliminary support for the second explanation, it does not prove that delinquency causes substance use. Most research indicates that the rela tionship does not prove to be this simplistic, given that all juvenile offenders do not use drugs (Dembo et al., 2007a; Huizinga & Jakob-Chien, 1998). Those who support a spurious model of the substance use-delinquency link suggest that both behaviors are predicted by the same set of common risk factors or that drug use and delinquent be havior cluster together as part of a general problem behavior syndrome (Dembo et al., 1992; Elliot et al., 1989; Jessor & Jessor, 1977; Neff & Waite,

PAGE 36

26 2007; Menard et al., 2001; Osgood, Johnst on, O’Malley & Bachman, 1988). This explanation is supported by a large amount of evidence th at highlights common risk factors for both drug use and delinquency. Furt hermore, the same factors that are able to predict age of onset of delinquency, serious delinquency, and persis tent delinquency have also been shown to predict initiation of substance use, serious substance use, and persistent substance use into adulthood (Elliott et al., 1989; Hawkins, Catalano, & Miller, 1992; Stice et al., 1998; White, 1997; Zhang et al., 1997). These risk factors include poor family relations, negative peer influe nces, educational difficulties, community disadvantage, and personality characterist ics (Dembo et al., 2007b; Dembo et al., 2006; Hawkins et al., 1992; Jessor, Donovan, & Co sta, 1991; Jessor & Jessor, 1977; Neff & Waite, 2007; White et al., 1999). Understanding the exact nature of th e substance use-delinquency link is beyond the scope of this study. However, based on the strong empirical ev idence highlighting shared risk factors for both drug use and de linquency, in addition to the high prevalence of substance use among juvenile offenders, it seems reasonable to examine the cooccurrence of these behaviors, as well as additional problem beha viors discussed in Chapter 3, among juvenile offenders. Furthermor e, the empirical evid ence that highlights marked variation in the substance use-de linquency link across race, gender, and age highlight the need to examine the co-occurrence of these behaviors across these demographic subgroups in an effort to obt ain the most accurate knowledge of this relationship.

PAGE 37

27 Chapter 3 Risky Sexual Behavior a nd Adolescent Offending Adolescence is also an important time period for the development of risky sexual behavior (e.g., high number of partners, inc onsistent condom use, sex while using drugs or alcohol).2 Specifically, several researchers have suggested that adolescence is the most critical time period for initiation into sexual behavior, the pr ogression into positive or negative sexual practices, and the developm ent of perceptions re garding the risks and consequences of sexual behavior (Mal ow, Devieux, Jennings, Lucenko, & Kalichman, 2001; St. Lawrence, 1993; Tapert, Aarons, Sedl ar, & Brown, 2001). As such, a wealth of studies indicates that the teen years tend to be the time pe riod when individuals are the most vulnerable to engaging in risky sexual behavior (E ricksen & Trocki, 1992). For example, national survey data suggest that 20 to 47% of sexually active fifteen to seventeen year old high-school students report having four or more partners in their lifetime (Santelli, Lindberg, Abma, McNeely & Resnick, 2000), and recent data indicate that nearly one-third of sexually active teen s use contraception inconsistently (National Campaign to Prevent Teen Pregnancy, 2000). Because of these relatively high rates of risky sexual practices, adolescents have been found to be at high risk for many nega tive health consequen ces related to unsafe 2 It should be noted that the use of substances prior to or during sex is also considered a risky sexual behavior. However, discussion of the link between substance use and risky sexual behavior is the topic of Chapter 4.

PAGE 38

28 sexual practices (Kotchick, Shaffer, Foreha nd, & Miller, 2001). Fo r example, in 2006, the Centers for Disease Control and Preventi on (CDC) indicated that youth aged 10-19 accounted for 35% of reported Chlamydia a nd 28% of reported gonorrhea infections (CDC, 2007). Compared to adults, this age range accounted for the highest proportion of new infections. To make matters worse, individuals infected with sexually transmitted diseases (STDs) are 3 to 5 times more likely to be infected with the Human Immunodeficiency Virus (HIV) (CDC, 1998) and other long-lasting se xually transmitted diseases (i.e., syphilis, herpes) (CDC, 2007). For females, infections can also cause reproductive problems including infertility, pe lvic inflammatory disease (PID), and unintended pregnancy (CDC, 2007; Chacko, Barnes, Weiman, & Smith, 2004). The CDC (2007) indicates that females under the age of 25 are the most likely age group to develop PID from an untreated STD. Given the significant negative health c onsequences associated with risky sexual behavior, a large body of public health resear ch has been conducted on sexual behaviors among adolescents. This body of research is based on community adolescent samples (Elliott & Morse, 1989; Evans et al., 2004; Malow et al., 2001; Morrison et al., 2003; Tolou-Shams, Brown, Gordon, & Fernandez, 2007; Warren et al., 1997), youth in substance abuse treatment programs (Bryan & Stallings, 2002; Deas-Nesmith, Brady, White, & Campbell, 1999; Gordon, Kinlock, & Battjes, 2004; Mezzich et al., 1997; Tapert, Aarons, Sedlar, & Brown, 2001), and incarcerated adolescent offenders (Barthlow, Horan, DiClemente, & Lanier, 1995; Canterbury et al ., 1995; Joesof, Kahn, & Weinstock, 2006; Kahn et al., 2005; Kingree, Braithwaite, & Woodring, 2000; Mertz, Voigt, Hutchins, & Levine, 2002; Morris, Bake r, Valentine, & Penni si, 1998; Oh et al.,

PAGE 39

29 1998; Pack, DiClemente, Hook, & Oh, 2000; R obertson, Thomas, St. Lawrence, & Pack, 2005; Teplin et al., 2005). Across these studie s, three important findings have emerged: 1) risky sexual behavior among the adolesce nt population is quite prevalent (CDC, 2007; Kotchick et al., 2001; Malow et al., 2001), 2) youth who engage in risky sexual behaviors are substantially more likely to be infected with a sexually transmitted disease (Castrucci & Martin, 2002; Pack et al., 2000; Robers ton et al., 2005), and 3) compared to nonoffenders, adolescent offenders are substantia lly more likely to engage in risky sexual behaviors and to test positive for a sexua lly transmitted disease (Barthlow, Horan, DiClemente, & Lanier, 1995; Canterbury et al., 1995; Joesof et al., 2006; Kahn et al., 2005; Pack et al. 2000, Teplin, Meri cle, McClelland, & Abram, 2003).3 This chapter explores the existing research on risky sexual practices a nd the prevalence of STDs among adolescents, most notably, juvenile offenders. Risky Sexual Behaviors and STDs among the Adolescent Population Risky sexual behaviors. Nationally representative studies suggest that adolescents are engaging in risky sexual behaviors at alarming rates (Santelli et al., 2000). For instance, nearly half of the 13,953 high school students incl uded in the 2005 National Youth Risk Behavior Survey (YRBSS) reporte d ever engaging in se xual intercourse. Of the respondents who reported a sexual expe rience, 34% were sexually active at the time of the survey (i.e., 2005), 37% reported se x without a condom, and 14% reported sexual intercourse with more than four partners (CDC, 2006). Furthermore, the examination of 3 Research also highlights a strong relationship between risky sexual behavior and substance use. However, a discussion regarding this association is not included in this chapter because Chapter 4 is devoted solely to this topic.

PAGE 40

30 trends over time suggests that these estimates have remained stable over the past five years. The 1995 (Wave 1) data from The National Longitudinal Survey of Adolescent Health (Add Health) reported similar resu lts. Based on a sample of 14,151 students in grades 7-12, 39% reported that they had ev er had sex. Among those who reported ever having sex (n = 6,887), 35% did not use contra ception at first sexual experience and 32% did not use contraception at the most recen t sexual experience (Cubin, Santelli, Brindis, & Braveman, 2005). The National Survey of Youth Knowledge and Attitudes on Sexual Health Issues, funded by the Kaiser Family Foundation, wa s designed to examine the attitudes and behavior of sexually active a dolescents and young adults (Ho ff, Greene, & Davis, 2003). Of the 483 respondents aged 15 to 17, 29% report ed feeling pressure to engage in sexual activity, 10% believed “it is not that big of a deal to ha ve sex without a condom,” and 37% reported ever having sexual intercourse. Of the sexually activ e respondents, 38% reported having sex without a condom, 39% re ported more than two partners in their lifetime, 33% reported being in a sexual relationship where they felt things were moving too fast, and 14% felt that it was a possibility that they were STD positive. In a sample of 571 sexually active fema le adolescents, 33% reported their first sexual experience prior to the age of 14, 18% re ported two or more pa rtners in the past two months, 38% reported four or more part ners in their lifetime, 42% reported never using condoms with new partne rs, and 20% tested STD positive (Millstein & Moscicki, 1995). Likewise, using information collected from the daily diaries of 112 sexually experienced adolescents, Morrison et al. ( 2003) found that 38% of respondents engaged

PAGE 41

31 in anal sex, 22% reported a hist ory of an STD, and 37% reported rarely or never using a condom. Additional studies indicate that only a small proportion, approximately 10-20% of sexually active adolescents, report consiste nt use of condoms (DiClemente et al., 1992; Kahn et al., 1995; Seidman & Reider, 1994). Furthermore, youth who engage in on e form of risky behavior are also substantially more likely to enga ge in other forms of risky se xual behavior (Biglan et al., 1990; Gillmore, Butler, Lohr, & Gilchrist, 1992). For example, DiClemente et al. (1996) found that the frequency of sexual intercourse was inversely related to condom use and Richter, Valois, McKeown, and Vincent ( 1993) found a strong corre lation between the number of lifetime partners and the failure to use condoms. Research also highlights the strong association between risky sexual beha vior and sexually transmitted diseases. Put simply, the greater the level of risky se xual behavior, the grea ter the likelihood of contracting a sexually transmitted disease (C astrucci & Martin, 2002; Hoff et al., 2003; Millstein & Moscicki, 1997; Pack et al., 2000; Roberston et al., 2005). Sexually transmitted diseases. Data from the CDC (2007) indicates that adolescents and young adults represent the highest risk group for sexually transmitted diseases. In particular, it is estimated that young people, ages 15 to 24, account for nearly half of all new infections. According to the 2006 estimates, 65 per 100,000 youth aged 10-14 and 1,674 per 100,000 youth aged 15-19 test ed positive for Chlamydia and 20 per 100,000 youth aged 10-14 and 459 per 100,000 youth aged 15-19 tested positive for gonorrhea. In addition, the latest availabl e data (i.e., 2006 estimates) suggest that Chlamydia and gonorrhea rates are increasi ng for the adolescent population. Although these statistics seem alarming, it must be not ed that these figure s are based on reported

PAGE 42

32 cases, and therefore, represen t only a small proportion of the national STD estimates. The CDC estimates that a substantial number of cases of treatable STDs go undetected, untreated, and unreported (CDC, 2007), which means that the prevalence rates for these diseases are actually much highe r among the adoles cent population. Furthermore, the CDC (2000) recently estimated that nearly 50% of new HIV infections are under the age of 25. Overall, national estim ates suggest that the number of new HIV infections is decreasing (CDC, 2007) However, new HIV infections for 13-14 year olds seem to be stabilizing while new in fections for 15-19 year olds have steadily increased. To add to this, the World Health Organization (WHO) esti mates that half of the 14 million people infected with HIV worldwide were infected between the ages of 15 and 24 (Goldsmith, 1993). Statistics such as these underscore the fact that the consequences of sexual risk-taking during ad olescence can have long-lasting effects. Given that juvenile offenders tend to experience lower access to health care services, fewer institutional resources, and in adequate sexual educ ation services (Dembo & Schmeidler, 2003), the adolescent offending p opulation appears to be at heightened risk for engaging in risky sexual behaviors, and in turn, contra cting and spreading sexually transmitted diseases (Joesof et al ., 2006). As such, sexual behavior among juvenile offenders has recently begun to receiv e attention in the public health field. This research has provided overwhelming support fo r the link between juvenile offending and risky sexual behavior. Risky Sexual Behaviors and Adolescent Offending Research comparing the sexual behavi or of delinquents to non-delinquents suggests that criminally involved adolescents are substantially more likely to report risky

PAGE 43

33 sexual behavior (DiClemente, Lanier, Ho ran & Lodico, 1991; Gordon et al., 2004; Mezzich et al., 1997; Robertson & Levin, 1999; Rowe, Rodgers, Meseck-Bushey, & St. John, 1989; Rodgers & Rowe, 1990; Scarame lla, Conger, Simons, & Whitbeck, 1998; Stouthamer-Loeber and Wei, 1998; St Lawarence, Crosby, & O’Bannon, 1999; Timmermans, Van Lier, & Koot, 2007). For ex ample, compared to high school students, detained adolescents display much higher ra tes of sexual intercourse, nearly double the STD incidence rate, and are s ubstantially less likely to repor t the use of condoms (Morris et al. 1995; Schafer et al., 1993). One r ecent study conducted by Crosby, DiClemente, Wingood, Rose, & Levine (2003) found that adol escents with a history of adjudication were significantly more likely to engage in sexual risk behaviors, including early initiation, multiple partners, sex with an infect ed partner, and the us e of drugs or alcohol during sex, compared to those who did not have a history of adjudication. Based on a community sample of 1,400 at -risk adolescents, Tolou-Shams et al. (2007) compared self-reported drug use and risky sexual behavior among youth with a history of arrest to youth wit hout a history of arrest. Y outh with a criminal history reported significantly higher frequencies of unprotected sex, a history of an STD, poor attitudes toward safe sex pr actices, and substance use dur ing sex compared to youth without a criminal past. Similarly, De vine, Long, and Forehand (1993) found that general delinquency was related to a greater number of sexual partners. Using the Add Health data, Armour and Haynie (2007) compared each respondent’s age of sexual debut (only respondents who were virgins at Wave 1) to the mean age of sexual onset for the respondent’s school. They found that experien cing early sexual initiation, measured by comparing the age of sexual onset of the res pondent to the mean age of sexual onset of

PAGE 44

34 the respondent’s school, was associated w ith a 20% increase in later delinquency compared to initiating sex on time. On th e other hand, youth who reported late initiation of sexual intercourse were 50% less likely to engage in delinquency compared to youth who initiated on time. Using the NYS, Elliott and Morse (1989) compared the sexual behavior of all non-married respondents across different le vels of offending. In 1976, when the respondents were 11-17 years old, 43% of se rious offenders were sexually active; however, only 6% of the nonoffenders were se xually active. In 1980, when respondents ranged from 15-21 years of age, 83% of se rious offenders were sexually active while only 36% of nonoffenders were sexually active. Furthermore, approximately 7% of the “patterned offenders” aged 11-14 in their sample reported more than six partners, whereas 1-3% of “exploratory” offenders and less than one percent of nonoffenders reported more than six partners. Among res pondents aged 15-17, 1727% of “patterned” offenders, 8-11% of “exploratory” offenders and 3-7% of nonoffenders reported more than six lifetime partners. Research conducted on the general adol escent population provides important insight into the prevalence of risky sexua l behaviors, and STDs, among both delinquent and nondelinquent adolescents by allowing the direct comparison of these behaviors across delinquent categories. However, simila r to studies examining substance use, these survey techniques only provide information on a small number of offenders within the larger adolescent population. Therefore, it is al so important to consider the prevalence of risk behaviors found among adolescent o ffender populations. Since it has been established that juvenile offenders are dispr oportionately more likely to engage in risky

PAGE 45

35 sexual practices, the studies are able to pr ovide a more in-depth understanding of the prevalence and severity of risky sexual pr actices among adolescent offenders. To date, this body of research is based, in la rge part, on incarce rated adolescents. Two decades ago, Bell and colleagues sugge sted that “adolescent detainees may be disproportionately important as core-group transmitters of STDs” (1985: 33). A number of subsequent studies on incarcerated adolescents indicate this statement is true. Therefore, the next two secti ons are devoted to this body of research. In particular, a detailed discussion of the public health re search conducted on juvenile offenders and risky sexual behaviors, includi ng variations in this associ ation across socio-demographic characteristics is provided. Risky Sexual Behaviors and STDs among Incarcerated Adolescents Risky sexual behaviors. Research suggests that adol escent offenders tend to initiate sex earlier, report higher numbers of partners, a history of an STD, use condoms less often, and are substantially more likel y to test positive for Chlamydia and/or gonorrhea (Barthlow et al., 1995; Castrucc i & Martin, 2002; Kingree & Phan, 2001; Rickman et al., 1994). For instance, Canter bury et al. (1995) found that 76% of the detainees in their study reported having three or more sex partners in their lifetime, 20% reported never using a condom, a nd 22% reported a past history of an STD. In a sample of 6,581 juvenile detainees, Morris et al (1998) found that 96% of the sample was sexually active. Of this subgroup, 71% re ported never using a condom and only 5% reported using a condom all of the time. Teplin et al. (2003) studied STD risky behaviors in a sample of 800 adolescents incarcerated at the Cook County Juvenile Te mporary Detention Center. Ninety-one

PAGE 46

36 percent of the youth reported being sexually active: 35% re ported having unprotected sex in the past month, 90% reported engaging in at least three risky sexual behaviors, and 65% reported engaging in 10 or more. Pack et al. (2000) found similar results based on the behavior of 284 African American male detainees. Among their sample, the average age of sexual debut was 11.9 and the mean nu mber of sexual partners was 11. Of the sexually active youth (98%), only 37% repor ted using condoms consistently, 20% reported using condoms more than half of the time, and 14% reported never using condoms. In regard to sexually transmitted diseases, 19.7% reported a history of an STD and 18% tested positive for Chlamydia, gonorrhea, or both. Research also suggests that chronic j uvenile offenders are substantially more likely to engage in risky sexual behavior compared to nonserious juvenile offenders and/or nonoffenders (Timmermans, Van Lier & Koot, 2007; Tolou-Shams et al., 2007; Robertson & Levin, 1999). For instance, Ha rwell, Trino, Bret, Yorkman, and Gollub (1999) compared the sexual behavior of youth on their first admission to juvenile detention to youth will multiple admissions to juvenile detention. Multiple admissions youth were significantly more likely to initia te sex before age 13, report eight or more lifetime partners, exchanged sex for drugs or money, and to use condoms inconsistently. Taken together, these studies highlight the troubling rates of ris ky sexual behavior among adolescent offenders. Across the studie s reviewed above, over 90% of adolescent offenders were sexually active, while only 30-50% of the adolescents in the community samples were sexually active. Of the se xually active offenders, over three-fourths reported multiple partners, whereas 14-40% of the community samples reported sex with

PAGE 47

37 multiple partners. Inconsistent condom use ranged from 60 to 95% among sexually active offenders and from 25 to 42% among sexually active community adolescents. Sexually transmitted diseases. Table 1 provides a summar y of previous studies that report the prevalence of STDs among in carcerated adolescents in the United States. As can be seen, a disproportionate number of juvenile offenders are infected with Chlamydia and/or gonorrhea. To add to this, Mertz et al. (2002) re ported data in five different detention centers across the United States. The median positivity rate for Chlamydia infections was 15.6% for fema les and 7.6% for males and the median positivity for gonorrhea infections was 5.2% for females and 0.9% for males. Table 1. STD Prevalence Rates from Studies of Incarcerated Adolescents in the US. Authors Sample + Chlamydia + Gonorrhea Broussard et al. (2002) Males ( n = 5029 ) Females ( n = 529 ) 12.9% 32.5% 4.3% 13.6% Canterbury et al. (1995) Males ( n = 1068 ) Females ( n = 147 ) 8.6% 9.5% 0.6% 5.4% Lofy et al. (2006) Females ( n = 3,593 ) 13.7% -Kahn et al. (2005) Males ( n = 98,296 ) Females ( n = 33,619 ) 5.9% 15.6% 1.3% 5.1% Katz et al. (2004) Females ( n = 101 ) 13.9% 5.9% Oh et al. (1998) Males ( n = 217 ) Females ( n = 46 ) 8.8% 28.3% 2.8% 13.1% Pack et al. (2000) Males ( n = 297 ) 14.4% 6.7% Risser et al. (2001) Males ( n = 450 ) Females ( n = 139 ) 9.6% 28.1% 6.7% 23.4% Robertson et al. (2005) Males ( n = 400 ) Females ( n = 218 ) 8.1% 24.7% 1.5% 7.3% Shafer et al. (1993) Males ( n = 269 ) 10.7% 6.6%

PAGE 48

38 To compare these estimates to the ge neral population, the CDC (2006b) recently reported a 6.3% median state STD positive rate for females aged 15 to 24 tested at family clinics, whereas the median percent positive rate for females tested in juvenile correctional facilities was 14.2%. Moreover, Chlamydia and gonorrhea rates among male adolescent detainees have been found to be 152 times greater than the general population in the same age range (CDC, 1996). These figures are quite disturbing given th at the large majority of Chlamydia and gonorrhea infections are asympt omatic (CDC, 2006b; Kahn et al., 2005). For example, in a sample of 263 juvenile detainees, 37 (14 %) tested STD positive and all but one were asymptomatic. Based on a sample of 284 incarcerated minority males, Pack et al. (2000) found that 84% of the STD positive youth ( 18% of the sample) self-reported no symptoms. As mentioned above, juvenile offenders ar e substantially less likely to have access to health care services (Dembo & Schmeidler 2002), which puts them at heightened risk for undetected and untreated disease. Thus, a large portion of risk-taking, sexually active juvenile offenders in our communities contin ue to unknowingly spread STDs. Therefore, reducing risky sexual behavior among this population, in addition to providing STD screening and treatment to justice involved a dolescents, is critical. The juvenile justice system has the potential to serve as an important avenue for providing much needed public health services to youth at-risk for experiencing the negative consequences of risky sexual behavior and reduc ing the spread of disease. As the research reviewed above indi cates, a very strong and consistent relationship exists between risky sexual behavior and adolescent offending. These

PAGE 49

39 studies concur that adolescent offenders are at heightened risk for engaging in risky sexual behavior and contracting STDs. Howe ver, several important socio-demographic differences in the risky sexual behavior-deli nquency link have also been revealed. In particular, important variations in the tendency to engage in risky sexual behavior and STD prevalence rates have been found across gender, race, and age. Therefore, it is important to consider these differences when examining the risky sexual behaviordelinquency link. Variations in the Risky Sexual Behavior-Delinquency Link Gender. Research on community samples of adolescents tends to suggest that males are more likely to report engaging in risky sexual behavi ors including early initiation and multiple partners, and females tend to report higher levels of inconsistent condom use (CDC, 2006; Luster & Small, 1994; Reitman et al., 1996; Shrier et al., 1996; Tubman et al., 1996; Warren et al., 1997) However, comparison of risky sexual behaviors among male and female detainees rev eals inconsistent patterns. Teplin et al. (2003) found that males were more likely to report being sexually active, multiple partners, and sex while drunk or high. Howeve r, female detainees reported higher levels of unprotected sex in the past month, sex w ith high risk partners, unprotected sex while drunk or high, and trading sex for money. Canterbury et al. (1995) failed to find significant gender differences in the number of sex partners or use of condoms, but a significantly higher percentage of females (44%), versus males (19%), reported a history of an STD. Kingree et al. (2000) found that females were significantly more likely to have sex without a condom than males, wh ereas, Morris et al. (1998) found that male detainees were more likely to report never using a condom, higher numbers of partners,

PAGE 50

40 and engaging in bisexual beha viors. Finally, Castrucci and Martin (2002) failed to find any significant gender differences in nu mber of partners or condom use. Despite the unresolved relationship between gender and risky sexual behaviors, empirical research has consistently documented that female juvenile offenders have substantially higher prevalence rates of STD infec tion than males (see Table 1). Furthermore, in a review of articles repor ting STD prevalence rates among incarcerated adolescents published between 2004 and 2005, Jo esof et al. (2006) concluded that Chlamydia rates range from 15.6 to 28.3% for females and from 5.2 to 14.4% for males; Gonorrhea rates ranged from 4.5 to 7.3% fo r females and 0.9 to 1.5% for males. As can be seen, significant gender differen ces exist in regard to risky sexual behavior and STD infection. In addition, Robertson et al. (2 005) found that predictors of STD positivity differed by gender. For male offenders, condom use was an important predictor of positivity, however, for female offenders, number of partners and sexual activity in conjunction with alcohol use pred icted positivity. Thus, it is important to consider gender differences in risky sexua l behavior and STD prevalence rates among juvenile offenders to accurately represent the sexual behavior-d elinquency link across male and female offenders. Given this body of research highlighting gender differences, ignoring such demographic differe nces could lead to inaccura te generalizations regarding the nature and strength of the association. Race. Similar to studies on the general adol escent population, studi es suggest that minority juvenile offenders engage in risky se xual behaviors at a hi gher rate than white juvenile offenders (Canter bury et al., 1995; Kahn et al., 2005; Lofy, Hofmann, Mosure, Fine, & Marrazzo, 2006; Morris et al., 1995; Ri sser, Risser, Gefter, Brandsetter, &

PAGE 51

41 Cromwell, 2001). For example, Canterbury et al. (1995) found that African American detainees reported higher numbers of sexual pa rtners, less use of condoms, a history of STD infections, and a current STD. DiClem ente (1991) found that non-Black juvenile detainees were significantly more likely to report consistent c ondom use. However, Morris et al. (1998) found that white detainees were more likely to use condoms with a steady partner, but were less likely to use c ondoms with a casual partne r. In regard to STDs, Lofy et al. (2006), Mert z et al. (2002), and Kahn et al. (2005) reported higher STD rates for minority detainees compared to white detainees, but, Risser et al. (2001) failed to find significant racial differe nces in STD test result. Thus, although the evidence tends to reveal higher risky behavior among mi nority offenders, results remain somewhat inconclusive. Age. Studies that focus on age differences in the sexual behaviors of adolescent offenders are relatively rare (Teplin et al. 2003). The handful of studies that do exist, in addition to studies based on community sample s, suggest that older adolescents report higher levels of risky sexual behaviors (Morris et al., 1998; Kingree et al., 2000; Shafer et al., 1993). That is, a positive linear relations hip between age and risky sexual practices is typically revealed. For instan ce, Teplin et al. (2003) found that, compared to detainees under the age of fourteen, older adolescent de tainees (14 and older) were more likely to report unprotected sex in the past 30 days, sex while drunk or high, unprotected sex while drunk or high, trading sex for money, and havi ng more than three partners in the past three months. Results for variation in STD infection acr oss age are mixed (see Lofy et al., 2006; Kingree & Phan, 2001). For example, Mertz et al. (2002) found that male detainees aged

PAGE 52

42 15-19 were more likely to test positive fo r Chlamydia and gonorrhea, whereas, female detainees aged 10-14 were more likely to test positive for gonorrhea bu t less likely to test positive for Chlamydia. Kahn et al. (2005) found that older male detainees were significantly more likely to test positive for Chlamydia. However, tests for Chlamydia among the female detainees did not reveal signi ficant age differences. Furthermore, there were no significant age differences in gonorrhea positivity, but younger adolescents were significantly more likely to test positive for both diseases (i.e., Chlamydia and gonorrhea). On the other hand, Risser et al. (2001) and Robertson et al. (2005) found higher prevalence rates of STD infection am ong older incarcerated adolescents. Taken together, it seems that older a dolescents are slightly more lik ely to be at risk for risky sexual practices and STD infection. Indeed, these studies unde rscore the importance of accounting for demographic factors when examining the occurrence of risky sexual practices among adolescent offenders. Taken together, African-American and older adolescents tend to report higher levels of risky sexual practices. Female offe nders are substantially more likely to test STD positive; however, variation in risky sexu al practices across gender is rather mixed. Similar to substance use, th e research above suggests that failing to account for such variation may lead to an imprecise understa nding of the strength and direction of the risky sexual behavior-delinquency associa tion across different offender groups, and in turn, lead to the assumption that the sexual health service needs are similar across race, gender, and/or age. Nevertheless, it is clear that risky sexual behavior is quite common among adolescent offenders; therefore, this behavior should be a top priority for prevention and

PAGE 53

43 intervention services. One addi tional factor that has been shown to play an important role in the relationship between risky se xual practices and delinquent behavior is substance use. Apart from the strong associations found among substance use and delinquent behavior (discussed in Chapter 2) and risky sexual be havior and delinquency (discussed throughout the current chapter), a strong and consistent association among all three forms of problem behaviors is also docum ented in the literature. As such, Chapter 4 explores the interrelationships between sexual risk-taking, substance use, and delinquent behavior and provide s further support for the need to examine these behaviors simultaneously.

PAGE 54

44 Chapter 4 Substance Use, Risky Sexual Behavior, and Adolescent Offending In addition to the well established relationships found among drug use and delinquency, and risky sexual behavior and delinquency, a strong co rrelation between risky sexual behavior and subs tance use is also consistently documented by public health research. In general, researchers have reli ed on two types of methods for assessing the link between risky sexual behavior and substance us e. The first method, global association, is routinely employed to determ ine whether substance users are more likely than non-substance users to engage in risky sexual practices over a specified period of time (Kingree et al., 2000). One example of a global association study would be to compare the reported number of sexual partners in the past three months of marijuana users to nonmarijuana users. These types of studies provide insight into which types of substances (i.e., alcohol, mariju ana, or other serious drugs) and which types of users (e.g., heavy users versus occasional users) are more likely to engage in risky sexual practices. The second method commonly used to asse ss the risky sexual behavior-substance use link is situational association. This me thod assesses the influence of substance use during sexual episodes. An example of a s ituational study would be one that examined whether or not individuals who reported the use of marijuana prior to or during sexual intercourse were more likely to report unprot ected sex. These type s of studies provide

PAGE 55

45 insight into the types of substa nces and level of use that in fluences risky sexual practices during sexual contexts (Kingree et al., 2000). Based on both types of observational met hods, it has been well es tablished that: 1) youth who report using alcohol, marijuana, and other illegal drugs (i .e., cocaine, crack) are substantially more likely to report enga ging in risky sexual behavior (i.e., higher number of partners, inconsistent condom use) compared to youth who do not use substances, 2) youth who report using alc ohol or drugs before or during sex are substantially more likely to engage in sexual risk-taki ng, 3) the more serious the substance use, the greater the likelihood of risky sexual beha vior, 4) the co-occurrence of risky sexual practices and substance use significantly increase s the likelihood of contracting an STD, compared to engaging in on ly one or none of these behaviors, and 5) compared to nonoffenders, juvenile offenders are substantially more likely to report the co-occurrence of sexual risk-t aking and substance use. Based on these findings, examining the interrelationships between risky sexual behavior, substance use, and delinquency during adolescence seems crucial to both researchers and practitioners. Not only can th is endeavor offer important insight into the nature of these relationships, it will also help to improve prevention and treatment services aimed at any one of these behaviors. Accordingly, this ch apter is devoted to exploring the interrelationship among these risk behaviors. First, an overview of the substance use-risky sexual behavior link among the general adol escent population is presented. Then, a detailed discussion of th e link between sexual risk-taking, delinquent behavior, and substance use is provided. Taken as a whole, the body of research

PAGE 56

46 summarized in this chapter s uggests that further investiga tion of the co-occurrence of these behaviors is warranted. Substance Use and Risky Sexual Behavior among the Adolescents Substance users and risky sexual behaviors. National data indicate that risky sexual practices are more prevalent among yout h who use substances, including alcohol, marijuana, cocaine, and stimulants (Lowry et al., 1994). For example, results from the 2003 National Youth Risk Behavior Survey indicated that 71% of adolescents who reported lifetime use of marijuana reported noncondom use at last sexual experience and 77% of marijuana users reported having multiple partners in the past three months. Of the respondents who did not use marijuana, 30% reported the failure to use condoms at last sexual experience and 23% reported having multiple partners (Yan, Yu-Wen, Stoesen, & Wang, 2007). The two most common forms of risky se xual behaviors examined in previous research that compares substance users to nonsubstance users are: 1) the failure to use condoms and 2) having multiple partners in a given time period. A wealth of studies has revealed a negative correlation between subs tance use and condom use. These studies suggest that adolescents who report the use of specific substances such as alcohol, marijuana, or cocaine, or score higher on a “substance use” index, are significantly more likely to report unprotected sexual inte rcourse (Bachanas et al., 2002; Brown, DiClemente, & Park 1992; Boyer, Tschann, & Shafer, 1999; Copper, Peirce, & Huselid, 1994; Fergusson & Lynskey, 1996; Fullilove et al., 1993; Hingson, Strunin, Berlin, & Heeren, 1990; Lowry et al., 1994; Luster & Small, 1994; Millstein & Moscicki, 1995; Strunin & Hingson, 1992). For instance, Bailey, Camlet, & Ennett (1998) found that

PAGE 57

47 marijuana users were 1.4 times more likely to report noncondom use. Likewise, Yan et al. (2007) reported that marijuana users were 1.4 and cocaine users were 1.8 times more likely to report unprotected sex. Baskin -Sommers & Sommers (2006) found that the odds of inconsistent condom use were 7.2 times higher for alcohol users and 15.9 times higher for methamphetamine users. Studies also consistently find that adolescent substance users are considerably more likely to report having multiple partne rs compared to their non-substance using counterparts (Bachanas et al., 2002; Devine et al., 1993; Duncan, Strycker, & Duncan, 1999; Fullilove et al., 1993; Fergusson & Lynskey, 1996; Koniak-Griffin & Brecht, 1995; Ramisetty-Mikler, Caetano, Goebert, & Ni shimura, 2004; Tubman, Windle, & Windle, 1996). Shrier, Emans, Woods, & DuRant (1997), for example, examined the association between drug use and number of lifetime partners in a sample of 1,078 students in Massachusetts. Respondents who reported a gr eater frequency and severity of lifetime drug use were significantly more likely to re port a greater number of lifetime partners compared to non-drug users. National data also indicate that adol escent substance users are more likely to report risky sexual behavior. Using data from the 2003 Nati onal Youth Risky Behavior Survey, Yan et al. (2007) found that marijuan a users were 1.8 and cocaine users were 2.5 times more likely to report having multiple partners in the past three months. The National Youth Survey found a positive linear relationship between seriousness of drug use and the percentage of youth reporting more than six partners. Of the youth aged 1114, less than 2% of non-drug users reported more than six partners, compared to more than 12% of illicit drug user s; among respondents aged 1517, less than 8% of the non-

PAGE 58

48 drug users reported more than six partners comp ared to over 20% of the illicit drug users (Elliott and Morse, 1989). In addition, early initiati on into sexual intercourse (Rosenbaum & Kandel, 1990; Smith, 1997; Rosenthal, Smith, & De Visse r, 1999), teenage pregnancy (Gillmore, Butler, Lohr, & Gilchrist, 1992; Yamaguc hi & Kandel, 1987), having unprotected sex with a casual partner (Bailey et al., 1998), an al intercourse without a condom (Tapert et al., 2001) and trading sex for m oney (Bailey et al., 1998) have also been shown to be significantly higher among adolescent substance users compared to non-substance users. At the same time, substance use is also a c onsistent predic tor of a history of an STD (Halpern et al., 2004) and current STD in fection (Boyer et al., 2000; De Genna, Cornelius, & Cook, 2007; Millstein & Moscoc ki, 1995). For example, Boyer et al. (1999) found that adolescents who reported us ing marijuana more than once per week were 2.3 times more likely to test positive for Chlamydia and/or gonorrhea than youth who did not use marijuana. Of the STD positive youth included in their study, 55% reported the use of marijuana in the past 30 days. One common approach to examining th e global association between risky sexual behavior and drug use has been to compare the sexual behavi or of youth with substance use disorders to community adolescents. Su ch studies also confir m a strong association among these risk behaviors (Malow et al., 2001 ; Mezzich et al., 1997). Adolescents with substance use disorders are substantially mo re likely to report risky sexual practices compared to adolescents without substance us e disorders. For example, Tapert et al. (2001) compared youth who were involved in su bstance abuse treatment to a sample of community adolescents. At the twoyear follow-up, the treatment group was

PAGE 59

49 significantly more likely to report multiple partners (72% vs. 50%), sex with casual partners in the past six mont hs (46% vs. 25%), and less lik ely to report condom use (20% vs. 46%). Furthermore, twice as many youth from the treatment group reported a history of STD infection (12% vs. 5%). Deas-Nesmith et al (1999) also found that youth diagnosed as chemically dependent were signi ficantly more likely to report engaging in sexual intercourse without a condom, anal intercourse without a condom, sex with a prostitute, and intercourse with a strange r, compared to youth diagnosed with nonsubstance abuse disorders a nd a community control group. The majority of research comparing the use of various substances suggests that severity of drug use also influences the seve rity of risky sexual pr actices. Lowry et al. (1994) reported a linear relationshi p between having multiple partners (> 4 partners) and substance use and condom use and substance use. For example, among the sexually active teens included in their sample, 4% of non-substance users reported multiple partners, whereas 10% of alcohol users, 30% of marijuana users, and 46% of cocaine and other drug users reported having sex with more than four partners. Adolescents who reported cocaine and other drug use were al so substantially more likely to report a combination of risky sexual behaviors. In re gard to engaging in both types of sexual risktaking behaviors, only 3% of non-substance us ers reported more than four partners and the failure to use a condom at last intercourse, whereas 18 % of marijuana-only users and 30% of cocaine and other drug users reported mo re than four partners and the failure to use a condom at last intercourse. As can be seen, a robust global asso ciation between risk sexual practices and substance use has been established. Clearly, substance using

PAGE 60

50 adolescents are at a heightened risk for e ngaging in sexual risk-tak ing, and in turn, STD infection. Using substances before or during sexual activity. National data also indicate that the concurrent use of substances a nd sexual intercourse is also common among adolescents. For example, of the sexually active teenagers include d in the 2005 National Youth Risk Behavior Survey, 23% admitted to using alcohol or drugs before their last sexual intercourse (CDC, 2006). National data also highlight the greater likelihood of sexual risk-taking when adolescents use substa nces before or during sexual activity. For example, data from the Kaiser Foundatio n’s Youth Knowledge and Attitudes on Sexual Health indicates that 29% of sexually active adolescents feel that alcohol or drugs have “influenced their decisions about sex.” In particular, 24% of sexually active 15 to 17 year olds admitted to “doing more” than pla nned while under the influence, 13% reported having unprotected sex “because they were drinking or using drugs” and 26% have “worried about STDs or pregnancy because of something they did while drinking or using drugs” (Hoff et al., 2003). The National Center on Addiction and Substa nce Abuse (1999) al so reports that one in five sexually active teens reported al cohol use during last sexual intercourse and up to 18% of adolescents reported drinking during their first sexua l experience. For adolescents who have never had vaginal interc ourse, use of alcohol or marijuana tripled the likelihood of engaging in unprotected or al sex. A study conducted in 2000 by the National Campaign to Prevent Teen Pregnancy revealed that more than half of the respondents who reported inconsis tent condom use said it was b ecause of the influence of alcohol or drugs.

PAGE 61

51 Similar studies regarding th e use of specific substances such alcohol, marijuana, and cocaine, in addition to studies employi ng an overall substance use index, have documented similar results regarding the use of substances prior to or during sexual encounters (Fullilove et al., 1993; Gillmor e et al., 1992; Robertson & Plant, 1988; Ramisetty-Mikler, 2004). Based on a sa mple of 522 sexually active adolescents, DiClemente et al. (2002) f ound that use of alcohol during intercourse significantly predicted a history of STD in fection. Millstein and Mosc icki (1995) found that youth who reported the use of substances (alcohol, marijuana, hallucinogens, or “other drugs) during any of their last four sexual experien ces reported higher numbers of partners and current STD. Jemmott and Jemmott (1993) found that African American male adolescents who reported having “sex while high” were considerably more likely to report unprotected intercourse, a greater number of sexual partners, and a greater number of “risky” sexual partners. As a whole, these studies suggest that a dolescent substance users are more likely to report a range of risky sexual behaviors a nd sexual risk-taking is more likely to occur when adolescents use substances prior to or during sexual activity. Given that juvenile offenders are disproportionately more likely to engage in both substance use and risky sexual practices, it is not su rprising that the co-occurre nce of these behaviors is substantially greater among th is population. For that reason, examining the substance use-risky sexual behavior link am ong juvenile offenders is cruc ial to 1) understanding the significance of this problem among juvenile o ffenders, 2) exploring the interrelationship between delinquency, drug use, and risky sexual behavior, 3) identify ing the risk factors for the co-occurrence of these behaviors, a nd 4) informing the development of juvenile

PAGE 62

52 justice programs. As such, the rest of this chapter is devoted to reviewing the existing research on the substance use-risky sexual be havior link among adol escent offenders. Knowledge regarding the co-occurrence of risky sexua l practices and substance use among juvenile offenders comes from studi es based on the self-reported behaviors of incarcerated adolescents or studies compar ing these behaviors across delinquent and nondelinquent adolescents. Overall, the large majority of thes e studies indicate that those who engage in both delinquent behavior and substance use are substantially more likely to report early initiati on of sexual activity (Weber et al ., 1989), inconsistent condom use (DiClemente et al., 1991), a highe r number of partne rs (Teplin et al., 2005), a history or current STD (Morris et al., 1995; Oh et al., 1998; Robertson et al., 2005; Shafer et al., 1993), trading sex for drugs or money (Wood & Shoroye, 1993), and anal sex without a condom (Bryan & Stallings, 2002; Teplin et al., 2005), than adolescents who engage in only one or none of th ese behaviors. The Substance Use-Risky Sexual Behavior Link across Delinquents and Nondelinquents Only a small number of studies have compared the co-occurrence of substance use and risky sexual practic es across delinquent and nondelinquent adolescents. However, the few studies that have been c onducted underscore the heightened risk for the co-occurrence of these behavior s, and in turn, STD infection among juvenile offenders. These studies are valuable because they provi de a direct comparison of the risky sexual behavior-substance use link across subgroups, a nd therefore, provide a vivid picture of the disproportionate number of offenders that are at risk for enga ging in these problem behaviors simultaneously.

PAGE 63

53 Substance users and risky sexual behaviors. The handful of studi es that compare the substance use-risky sexual behavior a ssociation for juvenile delinquents to nondelinquents indicate that adolescent offenders display substantially higher levels of the co-occurrence of these behaviors (DiCle mente et al., 1991; Rolf, Nanda, Baldwin, Chandra, & Thompson, 1991; Tolou-Shams et al., 2007). For example, using the National Youth Survey data, Elliott and Mo rse (1989) found a strong linear relationship between delinquent activity, drug use, and fre quency of sexual intercourse. The group of adolescents who were labeled “patterned o ffenders” and reported the use of multiple substances were the most likely to report more than six sexual experiences in the past year. More recently, Bryan and Stallings (2002) compared the self-reported substance use and sexual behavior of 200 adolescen t males from the community to 200 male juvenile offenders in substance abuse trea tment. The juvenile offenders displayed significantly higher frequency of intercourse and inconsistent condom use in the past year. For example, the mean number of sexual experiences wit hout a condom was 7.96 for the offender group and 1.6 for the community group. Using substances before or during sexual activity. In addition to studies comparing the sexual practices of delinquent substance users to nondelinquent substance users, a handful of studies suggest that juve nile offenders are also more likely to use substances prior to or during sexual intercou rse (Morris et al., 1998; Robertson et al., 2005; Tolou-Shams et al., 2007). For instance, Crosby et al. (2003) found that adjudicated delinquents were 2.6 times more likely to report using drugs or alcohol during their last sexual expe rience. Specifically, 34% of adjudicated delinquents,

PAGE 64

54 compared to 14% of the nondelinquents, reported using drugs or alcohol during their last sexual experience. Overall, these studies highlight the te ndency for 1) substance using juvenile offenders to engage in a range of risky se xual practices including more frequent sexual intercourse, noncondom use, and multiple partners at a substantially higher rate than nonoffenders, and 2) adolescent offenders to re port the use of substances prior to or during sexual intercourse at a significantly higher rate than nondelinquent adolescents. Substance Use and Risky Sexual Behaviors among Incarcerated Adolescents The large majority of existing knowledge regarding risky sexual practices, STD prevalence rates, and substance use patte rns among juvenile offenders is based on incarcerated adolescents. On one hand, th e generalizability of these studies is questionable due to the focus on detained adol escents who are typically the most serious juvenile offenders. On the other hand, they are quite valuable because they provide an indication of the severity of these problem behaviors among the most serious offenders. In addition, they also bring attention to the alarmingly hi gh co-occurrence of delinquent behavior, substance use, a nd risky sexual practices. Substance users and risky sexual behaviors. A number of studi es suggest that substance-using incarcerated adolescents are si gnificantly more likely to report a range of risky sexual practices compared to non-s ubstance using detainees (Otto-Salaj, GoreFelton, McGarvey, & Canterbury, 2002). For example, in a study of 210 incarcerated adolescents, Castrucci and Ma rtin (2002) found that those who reported regular use (once or more per week during the past thirty days ) of two or more substances (i.e., alcohol, marijuana, cocaine, inhalants, and other street drugs) were 11 times more likely to report

PAGE 65

55 having multiple partners (> 2), 5 times more likely to report exchanging sex for money, and 3 times more likely to report inconsiste nt condom use. Morris et al. (1998) found that significantly fewer incar cerated drug users reported us ing condoms every time they engaged in sexual intercourse compared to non-drug using detainees. Additional studies that focus on the rela tionship between marijuana use and risky sexual practices among incarcerated adolesce nts also highlight a strong association (Barthlow et al., 1995; Kingree et al., 2000; Shafer et al ., 1993). Kingree and Phan (2001), for example, found that detainees who reported the use of ma rijuana in the past 30 days were 3.5 times more likely to have ha d unprotected sex in the past 30 days and nearly twice as likely to test STD positive. Malow et al. (2001) examined the sexual practices of 169 juvenile offe nders court-ordered to substance abuse treatment. The youth labeled “abstainers,” reported an averag e frequency of marijuana use in the past three months of 19.23 times; for youth w ho reported monogamous, protected sex, the mean was 20.70, and of the “multiple partner, unprotected sex” group, the mean was 41.29 times. Teplin et al. (2005) compared the se xual practices of adolescent detainees diagnosed with a substance use disorder (n = 314) to detainees without a substance use disorder (n = 330). Detainees with an SUD were significantly more likely to report a wide range of risky sexual behaviors. For example, 77% of SUD youth and 39% of non SUD youth reported more than two partners in the past three mont hs, 41% of SUD youth and 17% of non SUD youth repor ted unprotected oral sex in the past month, and 47% of SUD youth and 22% of non SUD youth reporte d unprotected vaginal sex in the past month.

PAGE 66

56 Additionally, substance use ha s also been shown to be a consistent predictor of a history of STD (Morris et al., 1995), current STD (Morris et al., 1998; Robertson et al., 2005), teenage pregnancy (Morri s et al., 1998), and negative at titudes toward safe sex practices (Chang et al., 2003) among juvenile offenders. Using substances before or during sexual activity. Not surprising, research also indicates that incarcerated o ffenders report high rates of subs tance use prior to or during sexual activity. For instance, Crosby et al. (2007) examined condom use in the prior two months, among a sample of 134 sexually active female detainees. The mean number of times respondents reported having sex while drunk or high was 2.78. Moreover, use of alcohol and/or drugs during sex was a significant predictor of condom use error (i.e., used a damaged condom, began sex without a c ondom, removed condom before sex was over, re-used a condom, had a condom break). Am ong a sample of 800 juvenile detainees, Teplin et al. (2003) found that over 33% adm itted to engaging in unprotected sex while drunk or high. Similarly, Ki ngree et al. (2000) reported that 40% of African American male detainees reported using marijuana dur ing last sexual experi ence and 16% reported using alcohol. Kingree and Be tz (2003) found that youth wh o reported use of marijuana prior to last sexual intercour se were three times less likely to use a condom during this experience. Interestingly, studies comparing the e ffects of alcohol and marijuana use among incarcerated adolescents suggest that marijuana use is a strong er predictor of risky sexual behavior (Castrucci & Marti n, 2002; Kingree et al., 2000). Based on a sample of 167 incarcerated adolescents, 63.2% reported alwa ys using marijuana in conjunction with a partner who is “not well know n” versus 40.2% who reported always using alcohol. In

PAGE 67

57 regard to condom use, 46.1% reported always using marijuana and only 14.6% reported always using alcohol during unpr otected sex (Rosengard et al., 2006). Similarly, Kingree and Betz found that 45% of Af rican American detainees reported the use of marijuana during last sexual experience and 11% reported th e use of alcohol. Deviuex et al. (2002) found that the average number of unprotected sexual experiences when high on alcohol (in the past three months) was 13, wher eas, the mean was 19 unprotected sexual experiences while high on marijuana. Relatively little research has been conducted on the association between more serious drugs, such as cocaine or heroi n, and risky sexual behaviors among juvenile offenders. Typically, research either focuses exclusively on al cohol and/or marijuana, or combines the use of several different subs tances such as cocaine, hallucinogens, or inhalants into an overall “other substances ” category. The common justification for combining these drugs is the lower rates of the use of these substances among juvenile offenders (Castrucci & Marti n, 2002; Harwell et al., 1999; Robertson et al., 2005). For instance, among juvenile offenders, cocaine pr evalence rates typically range from a low of 2% to a high of 15% for incarcerated sa mples (Castrucci & Mart in, 2002; Harwell et al., 1999; Morris et al., 1998; Oh et al., 1998; Teplin et al., 20 03). Unfortunately, this is a limitation to the current body of research beca use it limits the ability to determine which particular substances are the mo st strongly related to engaging in risky sexual practices. The research reviewed above provides st rong evidence that juvenile offenders are simultaneously engaging in substance use and ri sky sexual behavior at alarming rates. In general, studies indicate th at juvenile offenders who re port the use of at least one substance (mostly marijuana) are three times more likely to engage in risky sexual

PAGE 68

58 practices, most notably noncondom use and havi ng multiple partners. At the same time, juvenile offenders who report the use of one or more substances prior to or during sex are also about three times more like ly to engage in these risk be haviors. Due to the severe health consequences of engaging in such risky sexual practices, these statistics underscore the need for prevention and interv ention services that are able to target delinquency, substance use, and sexual risk -taking in a single, integrated framework. Variations in the Risky Sexual Behavior-Substance Use Link It is also important to consider the vari ations in the strength of the risky sexual behavior-substance use link among juvenile offe nders. Similar to the research reviewed in previous chapters (see Chapters 2 and 3) variations across gender, race, and age have been documented in prior studies and highlight the need to consider these characteristics when examining the link between risky sexual practices, substance us e, and delinquency. Although this body of research is not extensive, and thus to some extent inconclusive, a brief discussion of the varia tion in the risky sexual beha vior-substance use link across socio-demographic categor ies is provided. Gender. On average prior research on the genera l adolescent population suggests that the association between substance use and risky sexual behavior is stronger for females compared to males (Graber, Br ooks-Gunn, & Galen, 1998; Mo tt & Haurin, 1988; Tapert et al., 2001). However, gender differences in the co -occurrence of risky sexual behavior and substance use among juvenile o ffenders is rather mi xed. Several studies have failed to find significant gender differences in regard to the sexual practices of substance using delinquents. For example, Te plin et al. (2003) di d not find significant gender differences regarding a situational a ssociation. Among the 800 detainees included

PAGE 69

59 in their study, 32.8% of males and 33.6% of females reported having unprotected sex while drunk or high. Likewise, Kingree and Phan (2001) found that the association between marijuana use and risky sexual beha vior was not moderated by gender. This finding persisted across both self-reporte d and biological drug test data and for noncondom use and STD test result. Yet, Oh et al. (1998) found that, for females, a positive marijuana test significantly predicted current STD status a nd, for males, a positive cocaine test was a significant predictor of STD status. Robert son et al. (2005) found that having sex while using alcohol was related to STD status for female detainees, but not male detainees. Kingree and Betz (2003) found that marijuana use was significantly linked to a lack of prior discussion regarding sexual risks for fe males but not males. On the other hand, marijuana use was associated with noncondom use in males, but not females. As can be seen, additional research rega rding gender variation in the association of risky sexual practices and substance use is needed. Race. Very few studies have examined raci al differences in the association between risky sexual practices and substan ce use among juvenile offenders. One study that did focus on racial differences found th at a larger number of white adolescent detainees reported having unprotected sex while drunk or high. Among the 800 youth in the sample, 36% of white male and 46% of white female detainees reported ever having unprotected sex while drunk or high compared to 31% of African American male and 28% of African American female detainees (Tep lin et al., 2003). The results of this study are similar to the findings of studies examini ng racial differences in the co-occurrence of risky sexual behavior and substance use among the general adoles cent population (CDC,

PAGE 70

60 2006; Ramisetty-Mikler et al ., 2004; Richter et al., 1993). Among community samples of adolescents, the link between sexual risk-tak ing and substance use is stronger for white adolescents. As stated in previous chapters, resear ch tends to suggest that white juvenile offenders report higher levels of substance us e, but lower levels of risky sexual behaviors (see Chapters 2 and 3). Therefore, based on th e scarce research directly examining racial differences in the co-occurrence of ris ky sexual behavior and substance use among juvenile offenders, coupled with the contradi ctory findings regarding variation in each of the behaviors separately, further research re garding this association is warranted. Age. The few studies that have examined the moderating effect of age on the association between risky sexual behavior and substance use among juvenile offenders provide inconsistent results. Kingree and Phan (2001) found th at juvenile detainees aged 13-15 who reported using marijuana were 8 times more likely to report unprotected intercourse and 5 times more likely to test STD positive; however, marijuana use was not a predictor of unprotected intercourse or STD status for juvenile de tainees aged 16-17. On the other hand, Kingree and Betz (2002) found contradictory results among a sample of incarcerated African American male adolesce nts. This study found that the association between marijuana use and noncondom use wa s stronger for older adolescents. Similarly, Teplin et al. (2003) found that a significantly larger proportion of juvenile detainees aged 16 to 18 reported ever having unprotected sex while drunk or high (females: 46%; males 37%) compared to th e younger detainees aged 14 to 15 (females: 28%; males: 31%) and aged 10 to 13 (females: 10%; males: 12%). These latter findings are consistent with general adolescent studies that reveal a stronger risky sexual behavior-

PAGE 71

61 substance use association for older adol escents (CDC, 2006; Elliott & Morse, 1989; Ramisetty-Mikler et al., 2004), and with the bu lk of studies on juve nile offenders that examine age variation in substance use a nd risky sexual behavior separately (see Chapters 2 and 3). As can be seen, knowledge regarding the variation in the risky sexual behaviorsubstance use link across demographic categories is rather sparse. However, the research that has been conducted is fairly similar to the general direction of the variation in each of the behaviors found in previous studies (see Ch apters 2 and 3). Taken together, female juvenile offenders, youth who report serious criminal invol vement, and older juvenile detainees tend to be more likely to 1) engage in serious substance use, 2) test positive for STDs, and 3) exhibit the co-o ccurrence of risky sexual behavior and substance use. Racial variations in these be haviors are more complex. On average, African American juvenile offenders tend to display elevated levels of risky sexual practices and higher prevalence rates of STDs, but lower levels of substance use and a weaker association between risky sexual behavior and substance use. The lack of definitive results regarding variations in the co-occurrence of risky sexual practices, delinquent behavior, and s ubstance use highlight s the importance of accounting for these demographic characterist ics when examining the nature and/or strength of these relationships. In addition to the research th at demonstrates differences in these specific behaviors, there is also a w ealth of research that highlights racial and gender differences in the cultural expectati ons of behavior, soci alization processes, cognitive development, and mental health pr oblems that have the pot ential to influence behavior (Bennett et al., 2005; Kotchick et al., 2001; Morash & Chesney-Lind, 1991).

PAGE 72

62 Thus, failing to account for demographic differences can lead to an inaccurate or under informed understanding of th e interrelationships among thes e behaviors across specific subgroups. This lack of information will lead to the assumption that the associations among these behaviors, and in turn, the prev ention and/treatment needs of adolescent offenders, are similar across demographic subgr oups. In reality, the needs of adolescent offenders may be quite different across demographic categories. An additional area of research that has not been extensively explored is the causal linkages that account for the tend ency to engage in risky sexu al practices, substance use, and delinquent behavior. Although the public health literature provides extensive information on the strength and direction of the relationship between these behaviors, it gives sparse attention to the causal mechanisms that account for this association. As a result, a definitive conclusion regarding the nature of these associations has not been reached (Rashad & Kaestner, 2004) Therefore, the next chapter reviews the conceptual explanations commonly discussed in regard to adolescent risk-taking, in general. These conceptual explanations are then supported with evidence that specifically addresses risky sexual behavior, substa nce use, and juvenile delinque ncy. Taken as a whole, the commonality in risk factors across these three behaviors highlights th e need to consider these behaviors simultaneously, as part of a general syndrome of deviance.

PAGE 73

63 Chapter 5 Explanations of Risk-Taking Behaviors in Adolescence The period of adolescence is a time of multiple transitions including puberty, change in parent-child interactions, p eer influence, and cognitive and emotional development. At the same time, this time period is also marked by increased autonomy, experimentation, and exploration in a range of behaviors (Bau mrind, 1991; Kuther & Higgins-D’Alessandro, 2000). As a result, th ese factors are posited to lead to an increased tendency to engage in different types of risk-taking behavior. In general, risktaking refers to “p articipation in behavior whic h involves potential negative consequences (or loss) balanced in some way by perceived positive consequences (or gain)” (Gullone & Moore, 2000:393). Thus, th e behaviors that are the focus of this study, risky sexual practices, delinquent behavior and substance use, are all considered risk-taking behaviors. This chapter review s the most commonly discussed explanations for engaging in risk-taking beha viors. The goals of this disc ussion are to 1) facilitate a better understanding of the nature of se xual risk-taking and substance use among adolescent offenders, and 2) highlight the commonalities among the risk factors for risky sexual behavior, substance use, and delinquency. It is important to note that research that has been conducted on the risky sexual practices and substance use patt erns of juvenile offenders ra rely focuses on the cause of this association. Instead, th e bulk of research has only b een descriptive. Thus, the

PAGE 74

64 majority of studies reviewed above were not conducted with the intention of identifying the causal nature of the associations, but rather, were interested in determining the strength of the associations. Despite the lack of research empirically testing the causal nature of the association between risky se xual practices, substance use, and delinquency, a number of explanations fo r this association are rou tinely mentioned throughout the literature. As mentioned earlier, these fact ors are commonly discussed as explanations for the tendency of adolescents to engage in ri sk-taking behavior, in general, rather than focusing on any one specific form of devi ant behavior (e.g., risky sexual practices, substance use, and/or delinquent behavior). These factors can be grouped into five major categories: 1) social/environmental factors, 2) cognitive developmental factors, 3) personality factors, 4) situational fact ors, and 5) problem behavior syndrome. Social/Environmental Factors Social/environmental explanations of risk-taking behavior emphasize the influence of parents and peers. Both peer support and parental involvement are consistent predictors of delinquency, substan ce use, and sexual risk-t aking. As Michael and Ben-Zur (2007) point out, peer groups can function as a source of social support and/or as a source of tempta tion and endangerment. Socia lization into the peer group may encourage involvement in risk-taking behavi ors. In turn, the adolescent will become involved in these risky behaviors because they appear relevant to th e group identity and will aid in the attainment of status with in the group (Diblasio, 1986; Lightfoot, 1992). Research consistently supports these claims. Youth that repor ted having peers that engage in risky sexual behavior are more likely to engage in risky sexual behavior themselves (Metzler, Noell, Biglan, Ary, & Smolkowski, 1994; Nader, Wexler, Patterson, McKusick,

PAGE 75

65 & Coates, 1989; Robertson & Levin, 1999; Sp italnick et al., 2007). Similarly, youth who report having substance-using peers report higher levels and more serious forms of substance use (Fergusson, Swain-Cam pbell, & Horwood, 2002; Jang, 2002; Johnson, Marcos, & Bahr, 1987; Krohn et al., 1996). At the same time, having delinquent peer associations is one of the most robust pred ictors of engaging in delinquent behavior (Warr, 2002). On the other hand, poor family relations, including low attachment, involvement, and monitoring, are also associated with higher levels of risky sexual behavior, substance use, and delinquent involvement (Che n & Thompson, 2007; Crosby, Leichliter, & Brackbill, 2000; DiClemente et al ., 2001; Huebner & Howell, 2003; Kapungu, Holmbeck, & Paikoff, 2006; Mosack, Go re-Felton, Chartier, & McGarvey, 2007; Robertson et al., 2005). Adolescents who do not have a healthy emotional bond to a parent will be less concerned with the conse quences of disobeying rules and letting the parent down. At the same time, low leve ls of monitoring and involvement provide greater opportunities to engage in risk-taking behavior. He nce, positive relationships with parents tend to lead to less deviant be havior, while, involvement with risk-taking peers tends to lead to higher levels of deviant behavior. The combination of these findings highli ghts the importance of social factors as adolescents move through adolescence. During this developmental time period, individuals undergo changes in roles and status that redefi ne their relationships with significant others (Colema n, 1992; Holmbeck, Paokoff, & Brooks-Gunn, 1995; Paikoff & Brooks-Gun, 1991). In particular, this time period is characterized by growing autonomy and emerging individuation from the family, and in parallel, increased interest and

PAGE 76

66 reliance on the peer group (Catalano & Ha wkins, 1996; Igra & Irwin, 1996; Laible, Carlo, Raffaelli, 2000; Thornberry, 1987). Sp ecifically, during adolescence, children tend to break away from their parents, and in turn, the peer group becomes the context in which behavior is determined (Cooper & Ayes-Lopez, 1985). For example, Johnson (1979) states that during th e mid-adolescent period, a youth’ s experiences outside of the home have a greater impact on their confor mity or deviance than do their experiences within the home. A number of studies have found that, during adoles cence, the influence of peers on behavior increases while the in fluence of parents on behaviors decreases (Jang, 1999; Johnson, 1979; Simons et al., 1991; Thornberry, 1996). Cognitive Developmental Factors From a developmental approach, risk-taking behavior is seen as the result of two main areas of cognition. One such factor is impairment in the ability to assess the true extent of risk in a given situation which, in turn, influences the decision-making process. Impairment in the ability to assess the extent of risk may be related to several factors including a lack of knowledge or experience and/or an inaccura te perception of risk. The greater the knowledge or unde rstanding regarding the nega tive consequences and the chances of experiencing such consequences, the less likely an individual is to engage in risk taking behaviors (Chang, Bendel, Koopman, McGarvey, & Canterbury, 2003; DiClemente et al., 1991; Harwell et al., 1999; Kingree & Betz, 2003). Self-reported perception of risk is consistently found to be related to sexual behavior and substance use among juvenile offenders (Belgrave, Randolph, Carter, Braithwaite, & Arrington, 1993; Kingree et al., 2000; Kingree & Betz, 2003; Nader et al., 1989; Robertson et al., 2005; Robertson & Le vin, 1999; Tolou-Shams et al., 2007). The

PAGE 77

67 more a youth believes in the negative conse quences of risk-taking behavior, the less likely they are to engage in such behaviors. These findings have been interpreted using a cost-benefit perspective of decision-making. That is, the higher the perceived costs of engaging in deviant behavior, th e less likely an adolescent is to engage in that behavior (Moore & Gullone, 1996; Moore, Gullone, & Kost anski, 1997; Parsons, Halkitis, Bimbi, & Borkowski, 2000; Pinkerton & Abramson, 1992). The second area is related to egocentrism. Egocentrism em phasizes a specific type of error in judgment that re sults from one’s heightened se nse of self or perception of specialness and uniqueness (Greene et al., 1996). This sense of superiority flows from a cognitive overdifferentiation of self from others, coupled with an underdifferentiation in objective thought. Therefore, “t his source of risk-t aking is not a prob lem of error in judgment; rather, it may be a lack of recogni tion that judgment is needed because the adolescent is ‘blinded’ by feelings of i nvulnerability that ac company feelings of uniqueness…” (Green et al., 2000:441). This invulnerability is found to be negatively associated with perceived susceptibility, inte ntion to avoid risky be havior, and subjective norm (Greene, Rubin, & Hale, 1995; Greene et al., 1996). It is well established that egocentric tendencies are highest during the adolescent years which is also the time period when ri sk-taking behaviors are elevated (Greene, Krcmar, Walters, Rubin, & Ha le, 2000; Greene, Krcmar, Ru bin, Walters, & Hale, 2002; Lapsley, 1993). Therefore, a connection betw een egocentrism and risky-taking behavior has been widely suggested.4 In particular, several st udies have supported the link 4 The literature discussing the link between egocentrism and risk-taking behavior tends to consider this relationship a developmental issue due to the heightened levels of egocentrism found in adolescence (see Green et al. 2000).

PAGE 78

68 between egocentrism and risky sexual behavior in adolescence (Green et al., 1995; Green et al., 1996; Goldsmith, Gabrielson, Gabriels on, Matthews, & Potts, 1972), egocentrism and delinquency (see Greene et al., 2 000), and egocentrism and substance use (Frankenberger, 2004; Green et al., 2000). This evidence suggests that adolescents who take behavioral risks believe themselves to be immune from consequences that might result from risk-taking. Personality Characteristics An additional explanation for the height ened levels of risk-taking found in adolescents is related to the personality characteristics of the adolescents engaging in such behavior. In particular, several resear chers argue that the tendency to engage in risky behavior is the result of a general pr edisposition toward deviance (Gottfredson & Hirschi, 1990; Kingree & Betz, 2003; Zu ckerman, 1979). That is, it is commonly assumed that an underlying personality trait, su ch as impulsivity or low self-control, is responsible for involvement in a ra nge of risk-taking behaviors. The personality trait that has received the most attention in regard to predicting risk-taking behavior is sensation-seek ing (Bonino et al., 2003). As defined by Zuckerman (1979:11), sensation seeking repr esents a “need for varied, novel, and complex sensations and experiences and the wi llingness to take physical and social risks for the sake of such experiences.” Thus, se nsation-seekers tend to ignore or diminish the consequences of risk-taking and focus on th e stimulation obtained from the behavior (Pinkerton & Abramson, 1995). Research indi cates that sensationseeking is a strong predictor of sexual risk-taking, substance us e, and delinquent beha vior in adolescence (Bryan & Stallings, 2002; Devieux et al., 2002; Gillis et al., 1992; Robins, 2004; Rolison

PAGE 79

69 & Scherman, 2002; Spitalnick et al., 2007; Wa gner, 2001; White et al., 1985; Zuckerman, 2008). Taken as a whole, these findings suggest that sensation-seekers seek out a range of stimulating behaviors including delinquenc y, substance use, and sexual risk-taking. Interestingly, Zuckerman (1994) found th at sensation-seeking peaks during the adolescent years. Situational Factors Another popular explanation for the st rong association among substance use and other risk-taking behaviors suggests that s ubstance use directly causes additional riskytaking behavior, particularly risky sexual prac tices. Several mechanisms for this causal process have been mentioned in the litera ture. These include lowering inhibitions, increasing aggression, or diminishing the abil ity to assess risk (Kingree & Betz, 2003; Rees, Argys, & Averett, 2001; Rotheram-Bor us, O’Keefe, Kracker, & Foo, 2000). For example, researchers have posited that, when high on alcohol or dr ugs, an individual’s inhibitions are lowered and th eir capacity to consider costs and rewards is diminished, which in turn, leads to a gr eater likelihood of making poor choices. Substance use has been shown to impair communication and psychomotor skills (Block, Braverman, Farinpour, 1998; Haney, Ward, Comer, Foltin, & Fischma n, 1999), which as a result, could lead to the failure to discuss risks or to use condoms, respectively (Kingree & Betz, 2003). Others have suggested that substance use can serve as an excuse to engage in behavior that would otherwise be considered socially unacceptable (Rees et al., 2001). Risky sexual practices and delinquent behavior are two behaviors ar e that considered socially unacceptable. This argument suggest s that engaging in socially unaccepted

PAGE 80

70 behaviors produces social strain. Substa nce use serves as an excuse or coping mechanism for dealing with this strain (Castrucci & Martin, 2002). Problem Behavior Syndrome A final explanation for risk-taking in adol escence is related to Jessor and Jessor’s problem behavior syndrome. Problem beha vior syndrome explains the association among risky sexual practices, substance use, and delinquency as the manifestation a general syndrome towards deviance (Jessor & Jessor, 1977). For example, it is argued that, due to the strong covariation among th ese behaviors, each behavior should be considered a “symptom” of a larger “syndr ome” of problem behavior rather than considering each behavior separately. Previ ous research examining “problem behavior syndrome” provides preliminary support for the notion that engaging in several different problem behaviors, most notably, delinque ncy, drug use, and risky sexual behavior, constitutes a unidimensional construct (D embo et al., 1992; Donavan & Jessor, 1985; Donovan et al., 1988; Farrington, 1998; Jessor et al., 1998; LeBl anc & Bouthillier, 2003; LeBlanc & Girad, 1997; Osgood, Johnston, O’Ma lley & Bachman, 1988; Stallings et al., 1997; Welte et al. 2004; Young et al., 2000). In further support of problem behavior s yndrome, it is also im portant to note that the large majority of risk f actors for adolescent risk-taking (discussed above) are also considered key risk factors for each of the three specific problem behaviors that are the focus of this study. These shared risk factors include peer influence, family characteristics, personality traits, and cogniti ve development. Due to the commonality in risk factors for all three of these problem be haviors, in addition to the strong covariation

PAGE 81

71 among these behaviors, it has been asserted that engaging in all three behaviors is an “indicator of a unitary construct of unconve ntionality” (Bryan & Stallings, 2002:388). Determining the cause of the co-occu rrence of risky sexual practices, delinquency, and substance use is beyond the sc ope of this study. However, given the strong and consistent relationships found among risky sexual practi ces, substance use, and delinquency reviewed in the preceding chap ters, in addition to the previous empirical support for the existence of problem behavior syndrome, this study is based on the notion that the co-occurrence of these behaviors repres ents a general tendency towards deviance. Based on this premise, the current stu dy focuses on the covariation among these behaviors and seeks to determine if 1) e ngaging in these behaviors forms a latent construct reflective of problem behavior syndrome, 2) if the structure of the latent factor is consistent across demographic subgroups. Accordingly, the next chapter provides an in-depth overview of “problem behavior syndr ome,” the research that does and does not support this concept, and the limitations of the current body of research.

PAGE 82

72 Chapter 6 Problem Behavior Syndrome in Adolescence It is well established that adolescents ar e at a heightened risk for engaging in a range of deviant behaviors, including ri sky sexual practices, substance use, and delinquent behavior. In addi tion, the co-occurrence of variou s forms of deviant behavior throughout adolescence has also been consis tently documented in the literature. Specifically, youth who engage in any one form of deviant behavior (e.g., drug use) are substantially more likely to engage in additional forms of problem behavior (e.g., delinquency). Based on these obs ervations, for years, research ers have suggested that the tendency to simultaneously engage in more than one form of deviant behavior constitutes a general proneness to deviance often referred to as problem behavior syndrome (Jessor & Jessor, 1977). Drawing from the work of Jessor and Jessor (1977:33), problem behavior is defined as “behavior that is socially defi ned as a problem, a source of concern, or as undesirable by the norms of conventional societ y and the institutions of adult authority, and its occurrence usually elicits some kind of social control response.” According to this definition, problem behaviors have the poten tial to be age-graded. A behavior that may be considered deviant in adolescence, for example sexual activity, may not be considered deviant for an adult. Therefore, the focus of this chapter, and the current

PAGE 83

73 study, is on behaviors that are c onsidered problem behaviors in adolescence, particularly risky sexual behavior, substan ce use, and delinquent activity. Drawing from the extensive body of re search discussed th roughout previous chapters, a strong associati on between risky sexual prac tices, substance use, and delinquency has been established. Indeed, a dolescents who engage in any one form of these problem behaviors are substantially more likely to engage in the other two. As such, a detailed review of the existing know ledge regarding problem behavior syndrome, including the definition, empirical evidence, and limitations of prior research, is provided below. Problem Behavior Syndrome A “syndrome” refers to the association of a number of detectable characteristics that often occur together, and when they do, leads to the identifica tion of a particular abnormality (e.g., problem, trait, disease). Based on this definition, problem behavior syndrome can be defined as the tendency to si multaneously engage in a constellation of problem behaviors which form a uni fied disposition towards deviance.5 Osgood et al. (1988:82) refer to this concept as, “a variety of deviant beha viors that form a ‘syndrome,’ which is directly caused by a general late nt variable of unconventionality.” Thus, engaging any one form of deviant behavior, for example substance use, is actually a “symptom” of the larger “syndrome” of pr oblem behavior. Stemming from these definitions, several researchers have argued that explaining this general disposition towards deviance, or in other words, problem behavior syndrome, is sufficient to account 5 Some researchers refer to problem behavior syndrome as general deviance syndrome, generality of deviance, or unconventionality. However, throughout th is document, this concept will be referred to as problem behavior syndrome.

PAGE 84

74 for any one form of deviant behavior. They ar gue that causes specific to any one form of deviant behavior are relatively insignificant because a cause of any one behavior should be a cause of any other form of deviant be havior (Gottfredson & Hi rschi, 1990; Jessor & Jessor, 1977). Several possible explanations for problem behavior syndrome have been developed. The three most widespread explan ations for problem behavior syndrome are reviewed below. Problem Behavior Theory. One of the most commonly discussed theories of problem behavior is Jessor and Jessor’s Pr oblem Behavior Theory (PBT). The PBT framework takes a social-psychological appro ach to the study of a dolescent behavior by focusing on three systems of psychosocial influence: the personality system, the perceived environment system, and the behavior system. The personality system is comprised of motivation, personal beliefs, and personal controls. The motivational component focuses on the value placed on, and the expectation for, the goals of achievement and independence. The personal belief structure focuses on aspects of the self th at may render an adolescent susceptible to problem behaviors, such as internal-external locus of control and self-esteem. Last, the personal control component includes attitudes toward deviance, relig ious beliefs, and the perception of consequences of deviant behavior. The perceived environment system is made up of two components: the distal component and the proximal component. Jessor and Jessor (1977) note that it is important to examine the perceived environmen t, rather than the objective environment, because the perceived environment is the one that the adolescent will most likely react to. The distal environment measures the adoles cent’s social context, which accounts for the

PAGE 85

75 adolescent’s family and peer environment. This includes peer and family support, influence, and expectations. The proximal environment measures the extent to which behaviors are modeled and reinforced by di sapproval and/or appr oval of behavior by parents and peers. The behavior system is made up of tw o types of behaviors: problem behaviors and conventional behaviors. The effects of the personality and perceived environment on the behavior system are the focus of the theo ry. It is suggested that engagement in problem behavior depends on the balance between the personal and environmental influences present in an adolescent’s life. It is hypothesized that 1) multiple forms of deviant behavior will be hi ghly correlated and form a singl e problem behavior index, 2) these deviant behaviors will be negatively corr elated with conventional behaviors, and 3) the personality system and perceived environment system will predict each of the deviant behaviors in similar ways and account for a large portion of the variance in the problem behavior index. Jessor and colleagues have provided consid erable support for PBT. In one of the most comprehensive tests of PBT, Jessor and Jessor (1977) anal yzed 400 high school students and 200 college students over a four-y ear period. Relying on measures of selfreported alcohol use, marijuana use, gene ral deviant behavior, sexual activity, and activism, these authors found support for all th ree of the above cited hypotheses. With the exception of activism, si gnificant positive covariati on among the various forms of deviant behaviors was revealed, while at the same time, a significant negative covariation among the deviant behaviors and conventional behaviors (i.e., involvement in church related activities and grade point average) was also f ound. Furthermore, a problem

PAGE 86

76 behavior index was created which revealed ad equate psychometric properties (see Jessor and Jessor, 1977: Table 3). Both the pers onality system and perceived environment system revealed a similar association to each of the specific deviant behaviors, as well as the problem behavior index. As suc h, Jessor and Jessor concluded that: with respect to problem behavior theo ry, its usefulness has been significantly reinforced. The magnitude of the account it provided for variation in problem behavior was in many cases substantial-about 50% of the variance in the multiple behavior index…the findi ngs provide strong support fo r the concept of problem behavior by revealing an interrelatedness-a syndro me character (pg. 235-236). Further studies conducted by Jessor, Donovan, and colleagues conducted over the past three decades have continued to provide s upport for the tenets of PBT (Costa, Jessor, Donovan, & Fortenberry, 1995; Donovan, Jesso r, & Costa, 1988; Donovan, Jessor, & Costa, 1991; Donovan & Jessor, 1985; Jessor, Donovan, & Costa, 1990). This study represents a partial test of problem behavior theory by examining the first hypothesis, which predicts strong cova riation among several problem behaviors and the formation of a single latent construct of deviance. In addition, similar to the behaviors examined in this study, these aut hors include measures of sexual behavior, substance, and delinquent behavior in thei r studies of problem behavior syndrome. Disposition towards deviance. Several researchers have also suggested that the tendency to engage in multiple forms of devian t behavior is the result of a disposition towards deviant behavior (Bonino, Cattelino, & Ciairano, 2003; Farrington, 1992; Hirschi & Gottfredson, 1993; Kotchick et al., 2001; Newcomb & McGee, 1991). This explanation is based on the no tion that certain in dividual characteristics are able to

PAGE 87

77 predict a variety of deviant behaviors. Some personality characteristics posited to influence the problem behavior syndrome are sensation-seeking, risk-seeking, and impulsivity (Green et al., 2000). These traits are associated with the need to run physical and social risks with the goal of provoking strong sensations, which individuals who possess these traits experience as extremel y exciting and pleasurable. For example, Bonino et al. (2003: 101) point out that, “the wi despread nature of this behavior [deviant behavior] has led to a belief th at there are certain personality types that have a constant need for intense, unusual, new sensations – personalities that require high levels of stimulation…” Thus, sensation-seeking, risk-s eeking, and/or impulsive individuals focus on the immediate rush obtained by engaging in a particular act and fail to consider the long-term consequences. The large majority of deviant behaviors, in cluding risky sexual practices, substance use, and delinquent behavior, provoke the immediate gratification of excitement and sensation. Thus, individuals who possess these traits do not specialize in any one act, but rather, are inclined to engage in any form of devian t behavior that will satisfy these desires. Gottfredson and Hirschi (1990) have suggest ed that the disposition toward deviant behavior is the result of low self-control. That is, that all devi ant behaviors are the product of this latent trait. These authors define low self -control with six elements: immediate gratification of de sires, risk-seeking, lack of empathy, preference for physical activities, laziness, and a lack of future orientation. Accord ing to Gottfredson and Hirschi (1990:91), “our image therefore implies that no specific act, type of crime, or form of deviance is uniquely required by the absence of low self-control.” They maintain that all deviant behaviors provide immediate gratification without concern for long-term

PAGE 88

78 consequences. Thus, these authors argue that individuals low in self-control will engage in multiple forms of deviant behavior includi ng criminal and noncriminal acts (referred to as analogous behaviors). Based on this ar gument, the relationship between drug use, risky sexual practices, and delinquency is not a causal question because they are all manifestations of an underlying tendency to pursue short-term, im mediate pleasure. Gottfredson and Hirschi point to the well established fact th at individuals who engage in one form of deviant behavior, for example delinquency, are much more likely to engage in additional forms of deviant beha vior, such as substance use. In addition, these authors also highlight th e wealth of studies that suppo rt offending versatility (for a review see Gottfredson and Hirs chi, 1990:91-94). To add to this, a large body of research has recently been conducted on the relations hip between individual characteristics, mainly low self-control, and the tendency to engage in various forms of problem behavior. The bulk of empirical evidence suppo rts the tenets of th e theory (see Pratt & Cullen, 2000 for a review). Although some evidence of speciali zation does exist (Farrington, Snyder, & Finnegan, 1988; Os good & Schreck, 2007; Sullivan, McGloin, Pratt, & Piquero, 2006), research tends to suggest that individuals low in self-control report engaging in multiple forms of deviant behavior. For example, based on a metaanalysis of 21 empirical articles examining th e effect of self-contro l on a range of deviant behaviors, Pratt and Cullen (2000:952) f ound that, “consistent with Gottfredson and Hirschi’s contentions, the effect s of self-control appear to be general. Thus, low selfcontrol had a similar effect size for crim e and analogous behaviors…” Based on these findings, they concluded that “low self-c ontrol must be consid ered an important predictor” of deviant behavior (pg. 953).

PAGE 89

79 Learning theories. Other researchers have focused on the influence that an adolescent’s social environment has on the development of problem behavior syndrome. Specifically, this body of research is concerne d with the ways in which adolescents learn deviant behavior (Akers, 1998; El liott et al., 1985). Akers’ so cial learning theory is one example that focuses on learning mechanisms According to social learning theory, adolescents model and imitate the behaviors of significant others in their immediate environment. Referred to as differential a ssociations, these indi viduals not only model behavior for the adolescent, but they also provide reinfo rcement for the adolescent’s behavior. Within these differential associa tions, adolescents learn definitions favorable or unfavorable to a specific behavior ba sed on the modeling and reinforcement that occurs. Therefore, an adolescent’s differen tial associations, which are comprised mainly of family and peers, have the ability to in fluence the development of problem behavior syndrome through a system of imitation, definitions, and reinforcement. For some time, Akers has argued that so cial learning theory has the ability to explain a wide range of de viant behaviors (Akers, 1998). His research has provided support for the tendency to engage in seve ral different forms of deviant behaviors including delinquency, adolescen t smoking, sexual deviance, alcohol use, marijuana use, and academic cheating (Akers, 1985; Akers, 1998; Akers & Cochran, 1985; Akers & Jensen, 2003; Akers & Lee, 1996). However, a major limitation of social learning theory as an explanation for problem behavior syndrome is Akers’ proposition that the definitions and techniques acquired through the learning process tend to be behavior specific. For example, principles four and fi ve of the theory state that “the learning of deviant behavior, including speci fic techniques, attitudes, an d avoidance procedures, is a

PAGE 90

80 function of the effective and av ailable reinforcers…” and “the specific class of behavior learned and its frequency of occurrence ar e a function of the effective available reinforcers, and the direction of the norms, ru les and definitions which in the past have accompanied the behavior” (Akers, 1985:41). Th is means that social learning is the source of each specific form of deviance, and therefore, social learning would account for problem behavior syndrome only to the extent that one’s differential associations model different behaviors. Thus, in order for so cial learning theory to account for problem behavior syndrome, an adolescent would need to be exposed to a variety of different deviant behaviors (Welte, Barnes, & Hoffma n, 2004). Unfortunately, problem behavior syndrome has not been evaluated in this context. There are several additional criminological theories that have not been used to directly explain problem behavi or syndrome, but make strong claims about their ability to explain a variety of deviant behaviors. Hence, they are referred to as “general” theories. Two such theories are Hirschi’s social bond theory and Agnew’s ge neral strain theory. Social bond theory is also concerned with an individual’s social environment. According to this theory, an individual’s bond consists of four elements: attachment to others, commitment to conventional goals, involvement in conventional activities, and belief in conventional values and norms. An indivi dual will engage in nonconforming behaviors when their bonds to conventional society ar e weak or broken (Hirschi, 1969). Thus, social bond theory does not necessarily make claims regarding law violating criminal behavior, but rather focuses on conform ity versus nonconformity to conventional standards.

PAGE 91

81 General strain theory (GST), on the other hand, takes a social-psychological approach to explaining devi ant behavior (Agnew, 1992). According to GST, deviant behavior is an adaptation to strain. Strain comes from three main types of stressors: the removal of positively valued stimuli, the pres ence of negative stimuli, and the failure to achieve goals. These three types of stress produce negative affect, and in turn, deviant behavior is a coping mechanism used to deal with these negative emotions. Delinquent behavior, substance use, and/or risky sexual behavior are all possible behaviors that could be used to cope with strain. Overall, both of these theories have been able to explain a range of deviant behaviors in cluding delinquency and substan ce use (see Akers & Sellers, 2004 for a review), however, they have not been used to directly test problem behavior syndrome. Although identifying the causal mechanisms that lead to the development of problem behavior syndrome is crucial, it is beyond the scope of the current study. Before we can truly identify the causes of problem behavior syndrome, the question of whether or not the concept of general deviance is an empirical reality needs to be definitively answered. On one hand, a large body of evidence exists in support of a general tendency towards deviance. On the other hand, there is also evidence to suggest that specialization in certain deviant behaviors does occur and that multiple forms of deviant behavior do not always form a unified construct. Prior to reviewing previous studies on problem behavior syndrome, two important characteristics of this body of research should be mentioned. First, support either for or against problem behavior syndrome is derived from models that incl ude a wide range of behavioral indicators. Al most all studies on problem behavior syndrome include a

PAGE 92

82 delinquency index, typically re ferred to as general delinquency, which includes a number of delinquent behaviors. In addition, most studies include so me measure of substance use and sexual behavior. However, these variab les differ in their measurement. On one hand, some studies include a general measur e of illicit substance use (e.g., a general index of use of several different forms of substances) (Ary et al., 1999; LeBlanc & Girard, 1997; Newcomb & Bentler, 1991; Os good et al., 1988; Welte et al., 2004; White, 1992), while other studies disaggregate illicit substance use into specific forms of use (e.g., marijuana use only) (Dembo et al., 1992; DeCourville, 1995; Donovan & Jessor, 1985; Donovan et al., 1988; White et al., 1994 ). A common way to measure sexual behavior is to include a measure of virg inity (Donovan et al., 1988; Costa et al., 2005; Jessor & Jessor, 1977) or frequency of sexual intercourse (Farrell et al., 1992; White, 1992). But, these measurement choices are so mewhat questionable. Some researchers would argue that being sexually active is not necessarily a problem behavior. Instead, it is the risks taken during sexual intercourse th at lead to problems. Additional problem behaviors included in studies are gambling (W elte et al., 2004), reckless driving (Osgood et al., 1988), aggression (Cheong & Raudenbush, 2000), suicidal thoughts (White et al., 1992), family rebellion (LeBlanc & Girard, 1997 ) and difficulty with school (Ary et al., 1999; Donovan, 1996; Gillmore et al., 1992; Le Blanc & Girard, 1997; White, 1992). To add to the wide variation in the beha vioral indicators used to measure problem behavior syndrome, there is also a great deal of diversity in the way the behaviors are measured and the number of behaviors included in the statistical models. For instance, studies have measured severity in behaviors using past thirty days (Farrell et al., 1992; Hays et al., 1997), past six months (C heong & Raudenbush, 2000; Donovan et al., 1988;

PAGE 93

83 Newcomb & Bentler, 1991), pa st year (DeCourville, 1995; Welte et al., 2004; White et al., 1994), or a combination of different time frames (e.g. past 30 days substance use and past year delinquency) (Bas ken-Engquist et al., 1996; D onovan & Jessor, 1985; Gillmore et al., 1992; Hemphill et al., 2007; Osgood et al., 1988). Furthermore, the number of behaviors included in the tested models range from four (Donovan & Jessor, 1985; Donovan et al., 1988) to fo rty-five (LeBlanc & Bouthillier, 2003). Secondly, these studies have relied on a variety of analytic techniques, most notably, exploratory factor analysis (E FA)(Donovan & Jessor, 1985; LeBlanc & Girard, 1997), confirmatory factor analysis (CFA ) (DeCourville, 1995; Farrell et al., 1992; Gillmore et al., 1992; LeBlanc & Girard, 1997) and structural equation modeling (SEM) (Ary et al., 1999; Dembo et al., 1992; Newc omb & Bentler, 1991; Osgood et al., 1988; Welte et al., 2004). Although similar in their underlying assumptions regarding manifest and latent constructs, there are also important differences in these approaches. For instance, EFA models are not predetermined, th at is, the links betw een the variables and the latent factors are unknown. Therefore, ther e is more freedom in the number of factors that can be identified. CFA models, on the other hand, are hypothesis-driven and, therefore, the links among the variables ar e specified a priori (Byrne, 2001). Factor analytic models are solely interested in obs erving how the observed variables are linked to their latent factors, whereas SEM models are interested in the causal direction of the variables. Put another way, SEM models assu me that the latent variables are causally related, whereas factor analytic models only ex pect the latent variab les to be correlated. Thus, the type of question that is being investigated is sli ghtly different across the three methods.

PAGE 94

84 As a result of the inconsistencies in both the behavioral indicators and statistical techniques used to examine problem behavi or syndrome, it is difficult to make solid conclusions regarding the evidence in support (or against) problem behavior syndrome. However, regardless of the methodological differences noted above, a great deal of research has provided evidence in favor of a one-factor solution representative of problem behavior syndrome. This study extends this research by examining the covariation in risky sexual practices, deli nquent behavior, and substance use across a variety of CFA and SEM techniques. Evidence in Support of Problem Behavior Syndrome A large amount of evidence in support of adolescent proble m behavior syndrome has been generated. This support comes fr om three types of studies: 1) bivariate correlational analyses, 2) commonality of risk factors, and 3) general factor analytic techniques. Furthermore, the concept of problem behavior syndrome has been supported with samples of children (Cappaldi & Patterson, 1989; Cheong & Raudenbush, 2000; Farrell, Kung, White, & Valois, 2000), adoles cents (Ary et al., 1999; Benda & Corwyn, 2000; Donavan & Jessor, 1988; Jessor & Je ssor, 1977; McGee & Newcomb, 1992) and young adults (Osgood et al., 1988, Ullman & Newcomb, 1999; Welte et al., 2004). This section reviews the empirical ev idence based on these three areas. First, the strong and consistent co rrelation found among a number of deviant behaviors, most notably deli nquency and drug use, suggests th at engaging in a particular form of deviant behavior is a manifestation of a larger syndrome of problem behavior (Barnes & Welte, 1986; Barnes, Welte, Hoffma n, & Dintcheff, 1999; Elliot et al., 1985; Farrell, Danish, & Howard, 1992; Newcomb & McGee, 1991; Proimos, DuRant, Pierce,

PAGE 95

85 & Goodman, 1998; Winters, Stinchfield, Botz et, & Anderson, 2002; also see Donovan & Jessor, 1985 for a review). Specifically, a wealth of studies has consistently revealed that engaging in any one form of deviant behavior is strongly related to engaging in additional forms of deviant behavior. Furthermore, stud ies also indicate that engaging in various forms of deviant behavior (e.g., drugs, delinquency, gambling, sexual activity) is negatively correlated with engaging in conve ntional behaviors (e.g., school achievement, religious activity) (Costa et al., 1995; Farre ll et al., 1992; Hays, Stacy, & Dimatteo, 1987; Newcomb & McGee, 1991; also see Donovan & Jessor, 1985 for a review ). That is, adolescents who engage in conventional activit ies are significantly less likely to engage in any one form of deviant behavior. For example, based on a sample of 1,588 students in grades 7-12, Donovan and Jessor (1988) found that the numbe r of times drunk in the past six months, frequency of marijuana use in the past six months, virginit y, and general deviant behavior (i.e., a scale of ten items including lying, stealing, shoplifting, fighting, prop erty destruction) were all significantly and positively co rrelated to one another, and were significantly and negatively correlated with school performan ce and church attendance. Donovan and Jessor (1985) found similar results across both adolescents and college students. More recently, using six different datasets that include high school students in grades 10 through 12, Donovan (1996) found that marijuan a use was positively and significantly correlated to general deviance, lower school grades, greater intake of alcohol, and a greater frequency of getting drunk. At the same time, a significan t inverse correlation between marijuana use and re ligious activity was found. Th ese significant correlations were stable across the twenty-year time sp an covered by the six different datasets.

PAGE 96

86 In addition to bivariate correlations, Farrington has examined the odds ratios for the relationship between offending and an arra y of problem behaviors at multiple time points. When examining violent behavior, he found that, at ages 12-14, self-reported violent offenders were 3.5 times more likely to report early sexual intercourse, 2.5 times more likely to skip school, and 1.5 times more likely to be frequent liars than nonoffenders. At age 18, self-reported violent of fenders were 2.2 times more likely to be sexually promiscuous, 2.3 times more likely to be heavier gamblers, 2.8 times more likely to drive drunk, 4.3 times more likely to use drug regularly, and 2.4 times more likely to be a heavy drinker than thei r nonoffending counterpart s (Farrington, 1998). When examining property crime (Farrington, 1992), he found that, at ages 12-14, selfreported burglars 5.6 times more likely to repor t sexual intercourse, 3.5 times more likely to skip school, and 5 times more likely to be frequent liars. At age 18, self-reported burglars were 3.4 times more likely to report risky sexual practices, 2.5 times more likely to be heavy gamblers, 2.7 times more likely to drive drunk, and 3.1 times more likely to be a heavy drinker compared to nonoffende rs. Based on these results, Farrington (1992:266) notes that “all of th ese types of acts could be regarded as different ageappropriate manifestations of an underlying antisocial tendency.” Secondly, variables that are able to pr edict the occurrence of any one of these forms of deviant behaviors, such as risky se xual practices, are genera lly able to predict the occurrence of additional forms of deviant behavior, such as delinquent involvement (Ary et al., 1999; Elliott & Huizinga, 1984; Elliott et al., 1985; Ensminger, 1990; Jessor & Jessor, 1977; Metzler, Biglan, Ary, Noe ll, & Smolkowski, 1993). Therefore, it is argued that various forms of deviant behavior share a numbe r of common risk factors.

PAGE 97

87 The most common risk factors mentioned in the literature are related to family environment and peer relations. Specifically, these factors ha ve been shown to predict a wide range of problem behavi or including substance use an d misuse, virginity versus nonvirginity, high-risk sexual behaviors, and a range of delinquent behaviors including violent and nonviolent offending. Finally, a number of studies on adolescen t problem behaviors have indicated that, when examining the structure of multiple deviant behaviors, they tend to form a unified construct (Ary et al., 1999; DeCourville, 1995; Dembo et al., 1992; Newcomb & McGee, 1991; Ullman & Newcomb, 1999; Welte et al., 2004; see LeBlanc & Bo uthillier, 2003 for a review). These findings provide further evid ence for the argument that the tendency to engage in any one form of deviant behavior in adolescence is actually a manifestation of a unidimensional syndrome of deviance. For example, using the Monitoring the Future (MTF) data, Osgood et al. (1988) found that past year criminal behavior, hea vy alcohol use, marijuana use, use of other illicit drugs, and dangerous driving formed a single latent variable for both crosssectional and longitudinal (three waves of data) relationships. Their results indicated that criminal behavior was the form of problem be havior most closely related to the latent factor, followed by serious illicit drug use and marijuana use. Acro ss the three waves of data, the proportion of explained variance fo r the observed behavior s ranged from 74% for criminal behavior to 27% for dangerous driving. Donovan et al. (1988) also found that the four problem behaviors used in their study (i.e., times drunk in the past six months, frequency of marijuana use in the past six months, virginity, and general deviant behavior) formed a single underlying construct.

PAGE 98

88 Similarly, using confirmatory factor analys es, Farrell et al. (1992) found a one factor solution based on five deviant behaviors (cig arette use alcohol use, marijuana use, delinquency, and sexual intercourse) using a sample of low income 7th and 9th graders. In both of these studies, the f actor variance was less than 50% and the substance use measures revealed the strongest a ssociation to the latent factor. Moreover, LeBlanc and Bouthillier (2003) reviewed the findings from twenty-one published studies using various forms of devian t behavior and a range of factor analytic techniques (e.g., factor analyses structural equation modeling). In regards to the problem behaviors examined in these studies, all 21 studies includ ed some form of delinquent behavior, 18 included some fo rm of substance use, and ei ght included behaviors related to sexual activity. In sum, a latent construct represen tative of problem behavior syndrome was identified in every study. Base d on their review, the authors stated that “this result was independent of informant, th e set of deviant behaviors, the statistical method, the nature of the sample, the historical period, and the site of the study” (pg. 83). Thus, they concluded that the unidimensional c onstruct of deviance is universal. As can be seen, a large number of studies provide s upport for the concept of adolescent problem behavior syndrome. Yet, a number of additi onal studies conducted on adolescents fail to support the concept of problem behavior syndrome. Evidence against Problem Behavior Syndrome The empirical evidence against the c oncept of problem behavior syndrome surrounds the issue of unidimensionality. Specif ically, a number of studies have failed to find a unitary latent factor based on vari ous forms of deviant behaviors (Grube &

PAGE 99

89 Morgan, 1990; Shaw et al., 1992). Taken toge ther, these studies cal l for a degree of skepticism regarding problem behavior syndrome. For instance, based on a sample of 11-12 years olds, Gillmore et al. (1992) examined the covariation among past year school trouble, aggressiveness at school, delinquent behavior (six-item scale incl uding stealing, vandalism, fighting, throwing objects, shoplifting, and burglary), and substanc e use (tobacco, alcoho l, and other illicit drugs). Results suggested that a three-f actor solution differentiating between school problems, substance use, and delinquency f it the data best. In addition, White and Labouvie (1994) and Hemphill et al. (2007) found that delinquency and drug use represent two distinct dimensions of problem behavior in adolescence. Additional researchers have found that a second-order factor model captures the notion of problem behavior syndrome more accu rately than a first order latent model (LeBlanc & Girard, 1997; Resinow, RossGaddy, & Vaughan, 1995; Vingilis & Adlaf, 1990). For example, McGee and Newcomb (1992) analyzed confirmatory factor analysis (CFA) models from four different waves of data spanning early adolescence to adulthood. Each model included measures of drug use, social conformity, academic orientation, sexual involvement, and criminal behavior. Results re jected a first-order construct in all four models and confirmed a higher second-order late nt construct. Based on these results, it has been suggested th at deviance is not a unitary phenomenon, but instead should be organized into types of behavior within which the specific behaviors are more closely related to each other than th ey are to other forms of deviance in other groups. Accordingly, McGee and Newcomb ( 1992:773-774) stated that the finding of a

PAGE 100

90 second-order latent construct is a “more comprehensive, but not contradictory, picture of the syndrome” which captures “both the spec ific and shared aspects of deviance.” Limitations of the Current Body of Research In addition to the empirical evidence against problem behavior syndrome, there are three additional shortcomings of the current body of research that need to be addressed. These limitations are related to the samples used to study problem behavior syndrome and the statistical techniques employe d. Specifically, the th ree issues that are discussed below, and addressed by the current study, are: 1) a lack of studies examining samples of adolescent offenders, 2) an insu fficient understanding of the variation in problem behavior syndrome across socio-demogr aphic categories, and 3) the reliance on general factor analytic methods which reveal relatively low levels of explained variance. This study attempts to address these issues using a sample of newly arrested juvenile offenders and a group-based modeling strategy. Addressing these limitations is essent ial because understand ing the nature of problem behavior syndrome has important im plications for prevention and intervention services that target at-risk adol escents. If the tendency to engage in a variety of problem behaviors constitutes a unitary syndrome and, ther efore, can be explained as a whole, as implied by the researchers discussed above, then it is crucial that prevention and intervention efforts begin to target the syndr ome as a whole, rather than focusing on any one particular behavior. Alte rnatively, if specialization in deviant behaviors is the norm then specific prevention and intervention services should continue to target the needs associated with each particular behavior. Al so, identifying differences in the structure of

PAGE 101

91 problem behavior across demographic categorie s provides valuable information regarding the specific needs of different subpopulations of adolescents. Lack of studies involv ing adolescent offenders. The majority of evidence in support of problem behavior syndrome is based on community samples of adolescents which contain low rates of adolescent offende rs. Indeed, it is well established that adolescent offenders report heightened levels of various forms of problem behaviors, most notably substance use and risky sexual pr actices. Therefore, relying on adolescent offending populations to study problem behavior syndrome will enhance the variability in risky sexual practices, substance use, and de linquent involvement which will enable us to effectively study their interrelationships by providing a more powerful test of the syndrome. If, for example, there is substant ial variation in the pr oblem behaviors and the covariation among them remains strong, great er support for the significance of these relationships will be provided, in turn, suppl ementing the results from general population studies (Dembo et al., 1992). To date, only a small number of studies examining problem behavior syndrome have been conducted on adolescent offenders. Dembo et al. (1992) examined the concept of problem behavior syndrome using a samp le of 201 adolescent detainees. These authors examined five separate structural e quation models based on past year marijuana use, alcohol use, and one of these five delinqu ency variables: genera l theft, index crimes, drug sales, person crimes, and total delinquency. Results indicated that all of the models revealed a one-factor solution. However, the general factor failed to account for all or most of the variance in e ach specific behavior.

PAGE 102

92 LeBlanc and Girard (1997) examined problem behavior syndrome on two separate samples of adjudicated boys residi ng in Canada (first sample, n = 470; second sample, n = 506). Using principal component an alysis, these authors examined the factor structure of six deviant behavi ors: vandalism, family and school rebellion, minor theft, serious theft, aggression, and use of drugs Results confirmed a second-order latent factor fit the data best. More recently, LeBlanc and Bouthillier (2003) examined 45 deviant behaviors categorized in to four types of deviant be haviors: overt (interpersonal violence), covert (property crimes), aut hority conflict (e.g., rebellious, stubborn, defiant at school and home), and reckless (e.g., substa nce use, sexual activit y, reckless driving) on a sample of 656 adjudicated delinquents also from Canada. Using CFA, results provided support for a unidimensional latent co nstruct of deviance, but with fairly low levels of explained variance. Overall, these studies provide prelimin ary evidence of a unidimensional syndrome of problem behavior among juve nile offenders. However, they also provide a level of uncertainty regarding the nature of the syndrome. Dembo et al. and LeBlanc and Bouthillier found a one-factor solution, whereas LeBlanc an d Girard found that a secondorder latent factor was more appropriate. Furthermore, all three studies highlighted relatively low levels of vari ation in the specific deviant behaviors accounted for by the latent construct. This study contributes to the adolescent offender literature by examining the structure of problem behavior syndrome on a sample of newly arrested juvenile offenders. By relying on newly arrested juve nile offenders, this study is an improvement over previous studies relying on both community samples and incarcerated adolescents.

PAGE 103

93 On one hand, community samples only include a small number of adolescent offenders, and an even smaller number of serious juve nile offenders. Thus, the variability in problem behaviors among community samples is much smaller compared to the offender samples, which means that the ability to pow erfully examine the interrelationships among the behaviors is limited. On the other hand, incarcerated samples are based on the most serious juvenile offenders at the back end of the system. But, only 20% of adolescents are sent to detention following arrest (Stahl et al., 2007). And, thes e are the adolescents who report the highest levels of problem be haviors. Thus, the prevalence of problem behaviors in samples of deta ined adolescents will be much higher than general offender samples, revealing a higher occurrence of problem behavior syndrome. Therefore, neither of these samples can be generalized to the entire adolescent offending population. The sample used in this study guards against these issues by includi ng adolescents at the front-end of the juvenile justice system including first-time offenders and youth being sent to secure detention. Inconclusive evidence on variations in problem behavior syndrome. Although the extent of variation in problem behavior syndrome across demographic categories has not received much empirical atte ntion, and therefore is extrem ely inconclusive, variations across gender, race, and age have been documen ted. Specifically, differences in both the strength of the correlation among specific deviant be haviors and in the structure of the latent construct have been observed. Th ese findings warrant consideration when examining problem behavior syndrome because they imply that the structure of problem behavior syndrome may differ across dem ographic subgroups. If substantiated, identifying which subgroups of adolescents are most at-risk for developing problem

PAGE 104

94 behavior syndrome is critical information for preventive programs that seek to target problem behavior syndrome. The failure to account for these varia tions could lead to erroneous conclusions regarding the nature of the syndrome by gene ralizing findings to all adolescents, when they may be truly representing only one pa rticular subgroup of youth. Of the three areas of variation, gender differences have been studied most extensively (Williams, Ayers, Abbott, Hawkin s, & Catalano, 1996). A number of studies that compare problem behavior syndrome across gender groups have failed to find a significant gender gap (Costa et al., 1995; Dembo et al., 199 2; Donovan & Jessor, 1985; Farrell et al., 1992; Gillmore et al., 1991; LeBlanc & Bouthillier, 2003; Newcomb & McGee, 1991). However, other studies have found that the constellation of problem behaviors varies by gender (D eCourville, 1995; Ensminger, 1990). For example, Bartlett, Holditch-Davis, and Belyea (2005) found a si gnificantly higher num ber of males than females display problem behavior syndrom e in adolescence. Moreover, using 11 different forms of problem behavior (i.e., al cohol use, marijuana use, other drug use, cigarette use, violence, general delinquenc y, school delinquency, gr ades, sexual behavior, psychological problems, and suic ide), White (1992) analyzed separate factor analytic models for males and females. Although bot h gender-specific models revealed a one factor solution, the behaviors th at loaded on each factor diffe red. For males, suicide and psychological problems failed to load on the factor; for females, violence was the only variable that did not load on the factor. The examination of racial variation in problem behavior syndrome has not received a great deal of empirical attenti on. And, the very few studies that have

PAGE 105

95 examined racial differences provide mixed re sults. On one hand, two studies have found that the covariation among problem behavi ors is stronger for white adolescents, compared to minority adolescents. Welte et al. (2004) found that the covariation among gambling, drug use, alcohol use, and deli nquency was lower for African American adolescents. Similarly, Co sta et al. (1995) found that early sexual intercourse was associated with less involvement in school activity, more involvement in delinquent behavior, problem drinking, and marijuana us e for White and Hispanic adolescents. However, they failed to find any significant differences in these behaviors across virgins and nonvirgins among the African American re spondents. On the other hand, Dembo et al. (1992) failed to find any si gnificant differences in probl em behavior syndrome across racial categories among a sample of justic e involved youth, but Bartlett et al. (2005) found that Black adolescents were 1.5 times more likely to display problem behavior syndrome. In regards to age differences, a handful of longitudinal studies indicate that the covariation among problem behaviors is str onger in early to mid adolescence (Newcomb & McGee, 1991; White, 1992). These findings suggest that, as yout h progress into young adulthood, specialization in problem behavior s tends to become more common and the covariation among deviant behaviors weakens (Newcomb & Bentler, 1986; Osgood et al., 1988). Taken together, these studies provide ge neral support for the existence of problem behavior syndrome across socio-demographic ca tegories. However, they also suggest that the strength of the a ssociation among various problem behaviors may differ across race, gender, and/or age. Given the varia tion that has been documented across race,

PAGE 106

96 gender, and age in the associ ation between substance use and delinquency (Chapter 2), risky sexual practices and de linquency (Chapter 3), and the co-occurrence of all three problem behaviors (Chapter 4), in additi on to the preliminary evidence regarding variations in problem behavior syndrome di scussed above, additional research regarding demographic variations in the structure of problem behavior syndrome is clearly warranted. Identifying this variation is important for two main reasons. First, identifying the similarities and/ or differences in the co-occurrence of problem behaviors has important implications for prevention and inte rvention services. If in fact, differences in the structure of problem behavior syndrome exist across demographic subgroups then such individua l-level characteristics will aid in the identification of adolescents most at-risk fo r problem behavior syndrome. At the same time, if the interrelationshi ps among behaviors differ acro ss groups, then prevention and treatment programs will need to be tailored to meet the need s of each particular group. Secondly, understanding the ways in wh ich race, gender, and age influence the tendency to engage in various forms of devian t behavior will also provide a more detailed understanding of the syndrome. Typically, st udies examine problem behavior syndrome on an entire sample of adolescents, often school-based samples, and fail to examine subsamples based on race, age, and/or gender. This has the potential to lead to errors in generalizing problem behavior syndrome to specific demographic subgroups. For example, Study A examines the factor structure of five deviant behaviors on an entire sample of high-school students a nd concludes that these behaviors load on a single factor representative of problem beha vior syndrome. Consequently, these results are generalized to high school students. Bu t, Study A failed to examine race and/or

PAGE 107

97 gender-specific models and the sample is co mprised of a large number of white male students. As a result, whether or not th e covariation among the behaviors is similar across demographic subcategories is unknown. Therefore, these results can not be reliably generalized across raci al and/or gender categories. How can we be sure that the covariation and factor struct ure is similar across racial categories for high school students? Hence, by failing to acco unt for differences across demographic characteristics, we are assuming that the results apply equally to all subjects in the sample. But, given the documented variati ons observed across race and/or gender in a variety of specific forms of problem behavi or (i.e., substance use, delinquency, sexual behavior), it is quite possible that the covariation among the fi ve behaviors va ries across these demographic categories. Based on a sample of 5,537 high-sch ool students, Basen-Engquist, Edmundson, and Parcel (1996) examined the structure of 25 health risk beha viors including sexual behavior, substance use, school involvem ent, violence, social environmental characteristics, accidental and intentional inju ry, and dietary behavior. Their preliminary analyses revealed a four dimensional model fit the data best. However, when they examined this model across demographic subgroups representing Black females, white females, Black males, and white males, they found significant differences. The model fit the data well for the white male group, but, a poor fit was obtained for the Black male, Black female, and white female groups. Furthe rmore, white females displayed the highest loadings on Dimension 4 (illegality), whereas Black male and female participants revealed the lowest values on this dimensi on. Dimension 2 (health protective action) was

PAGE 108

98 more relevant for the Black male and fema le groups and both male groups showed higher loadings on Dimension 1 (p roblem behaviors). As can be seen, accounting for variations across race and gender in the study of problem behavior syndrome is crucial to obtaining an accurate understanding of this concept. The current study extends this research by examining how the structure of problem behavior syndrome varies acro ss demographic subgroups among a sample of newly arrested juvenile offenders. Statistical methods used. As summarized above, the larg e majority of studies that rely on factor analytic tec hniques to assess problem beha vior syndrome tend to find a one-factor solution representative of problem be havior syndrome. However, the bulk of studies that do provide suppor t for problem behavior syndrome tend to reveal 1) moderate correlations among the latent cons truct and the specific forms of deviant behaviors and 2) relatively lo w levels of explained variance. In particular, across the prior studies reviewed above, the explained variance of each specific deviant behavior included in the models ranged from the low tw enties to the mid seventies. For example, Osgood et al. (1988) reported explained va riance values ranging from 24% to 74%, LeBlanc and Girard (1997) reported a range of 48% to 54%, Donovan and Jessor (1985) reported a range of 39% to 44%, McGee a nd Newcomb’s (1992) findings ranged from 55% to 57%, and White (1992) reported a range of explained variance of 27% to 32%. As reviewed above, these are studies that are used to provide strong support for the existence of problem behavior syndrome. Howe ver, as can be seen, on average, less than half of the variability in each problem behavior is accounted for by this general construct. As such, a number of the researchers that have found a single unidime nsional construct of

PAGE 109

99 problem behavior syndrome conclude that devi ant behavior is, in pa rt, a manifestation of a general syndrome, and in part, a uni que phenomenon (Dembo et al., 1992; Osgood et al., 1988; White, 1992). One reason for the low levels of expl ained variance may be related to the statistical techniques used to examine this conc ept. Standard factor analytic techniques examine the tendency to engage in various fo rms of problem behavi or across the entire sample. Thus, these types of methods pr ovide us with the weighted average of covariation across the entire data set and lead to the assumption that the average level of covariation applies equally to all individuals (Luke, 2004). As a result, these standard techniques ignore the influence that individual-level factors may have on the tendency to engage in problem behaviors. For example, it may be possible that: 1) a certain group of adolescents tends to display the problem be havior syndrome, 2) additional subgroups of adolescents are engaging in onl y one or two specific forms of deviant behavior (e.g., only delinquency or only substance use), and 3) th e strength of the association among various forms of problem behavior differs across importa nt individual-level factors. If so, relying on a weighted average has the potential to lead to the estimates being carried by a particular group with either very high or very low levels of behaviors, or a large number of cases belonging to a particul ar group(Widaman & Reise, 1997). This likely possibility may account for the relatively low levels of explained variance found in previous st udies that fail to account for such socio-demographic variation. In other words, the low levels of explained variance found in the current body of research examining problem behavior syndrome may be due to the existence of different subgroups of adolescents in regard to their engagement in multiple forms of

PAGE 110

100 deviant behaviors. General factor analytic techniques are unable to account for these different groups. As such, examining differe nces in the interrelationship among a variety of deviant behaviors across race, gender, a nd age has the potential to significantly improve our understanding of problem behavior syndrome, as well as help overcome the issue of low explained variance. Taken together, the heightened levels of co-occurring deviant behaviors among juvenile offenders, the well established vari ation in specific problem behaviors across demographic categories, and the low levels of explained variance that are revealed with standard factor analytic met hods, suggests that examining diffe rences in the structure of problem behavior syndrome across demogra phic subgroups is clearly warranted. In particular, it is possible that: 1) adolescent offenders differ in their tendency to engage in risky sexual practices, delinque ncy, and substance use, and 2) these subgroups differ on important demographic characteristics. By examining these possibilities, the cu rrent study extends previous research regarding variations in pr oblem behavior syndrome acro ss demographic subgroups. At the same time, this study overcomes the curr ent shortcomings of the problem behavior syndrome literature that are cau sed by relying on standard fa ctor analytic techniques by applying a group-based factor an alytic technique to the concept of general deviance. Briefly, with multiple-group SEM, one can compar e the factor structure, or covariation, among a number of observed variables acro ss specified subgroups (Muthn & Muthn, 2007). Thus, this method of analysis will enable the identification of the variation in the strength of the covariation among delinque nt behavior, risky sexual practices, and substance use across demographic subgroups rath er than considering the average level of

PAGE 111

101 strength among the variables. Widaman and Re ise (1997) point out that, when there is evidence to suggest that groups within th e population exist for whom the indicator variables are differentially related to the latent variable, group-based modeling is a powerful tool for accounting for these differences. Current Study The purpose of the current study was to 1) examine whether or not the tendency to engage in risky sexual practices, substance use, and delinquency form a unidimensional construct reflective of problem behavior syndrome among a sample of newly arrested juvenile offenders, and 2) iden tify variations in the structure of problem behavior syndrome across individual-level factors. To accomplish these goals, a sample of newly arrested juvenile offenders pro cessed at a centralized intake facility was analyzed. Exploring these associations exte nds previous research on problem behavior syndrome by building upon the limitations de scribed above. Based on the literature reviewed in the previous chap ters, three main research quest ions guide the current study. 1) Does the tendency to engage in risky se xual practices, substance use, and delinquent behavior form a unified la tent construct representati ve of problem behavior syndrome among a sample of newly arrested juvenile offenders? 2) If so, does age have a direct effect on this latent factor? 3) Does the factor structure of this latent construct, as well as the association between age and the latent factor, differ across demographic s ubgroups of newly arrested juvenile offenders?

PAGE 112

102 Chapter 7 Methods Sample The current study focuses on the relations hip between three different forms of problem behavior among a sample of newly arre sted juvenile offenders. Specifically, this study has three main objectives. The first objec tive is to examine the covariation in risky sexual practices, substance use, and delinquent behavior and to dete rmine whether or not these behaviors form a unitary latent construc t reflective of problem behavior syndrome. The second objective is to determine whether ag e has a direct effect on the latent factor. The third objective involves comparing th e covariation among the observed variables (i.e., latent structure), as well as the effect of age on th e latent factor, across demographic subgroups. Data for this study were collected in a National Institute on Drug Abuse (NIDA) funded research project located in Hills borough County, FL (Bele nko et al., 2008). The project involved a successful collabora tion among the Hillsborough County Juvenile Assessment Center (HJAC), the Florida Department of Health (DOH), Hillsborough County Health Department (HCHD), and the Florida Department of Juvenile Justice (DJJ). Project goals included estimating the prevalence of Chlamydia and gonorrhea among a sample of newly arrested juvenile offenders, examining the relationship between

PAGE 113

103 drug use and risky sexual behaviors (inc luding STD infection), and assessing the feasibility of providing public health services to newly arrested juvenile offenders. All youth under the age of 18 who ar e arrested in Hillsborough County are transported to the HJAC for intake screen ing and assessment procedures. Standard HJAC processing involves trained HJAC st aff completing a Detention Risk Assessment Instrument (DRAI) to determine if an arrested youth will be released to the community, placed on nonsecure home detention, or sent to secure detention. The DRAI takes into consideration the youth’s most serious current offense, other current offenses, pending charges, prior offense history, current legal status, and aggravating or mitigating circumstances (see Dembo et al., 1994). In addition, youth are asked to voluntarily participate in a psychosocial risk assessmen t and to provide a urine specimen for drug testing. Figure 1 provides a schematic diagram of the study protocol. Data collection procedures for the project began in June 2006.6 To be eligible to participate in the research project, youth had to: 1) be twelve years of age or older7 and 2) agree to provide a urine specimen for drug testing. Once an el igible youth was identified, data collection procedures occurred in the following steps. 1. Project-trained HJAC assessors prov ided brief STD pre-counseling to the youth. Counseling covered the prevalence of STDs among adolescents, importance of getting tested for Chlamydia and gonorrhea, 6All study protocols were approved and monitored by the project’s oversight Institutional Review Board. 7 Under Florida law, youth 12 or older are protected from STD test disclosure to parents and do not need parental consent for an STD test.

PAGE 114

104 how these diseases are spread, and pr ocedures for getting treatment if positive. 2. HJAC staff asked youth to voluntarily consent to having their urine specimen split and tested for Chlamydia and gonorrhea. 3. Those consenting to STD testing completed a supplemental risk assessment interview, which consiste d of several questions regarding sexual behavior, previous testing an d treatment experiences, and prior sex education. They were informed that, if STD-positive, a HCHD worker would provide free, confidential treatment. 4. Urine specimens were collected a nd tested at the DOH laboratory. 5. Coordination was established among DOH staff and HCHD Disease Intervention Specialists (DIS) to pr ovide confidential treatment to youth testing positive for Chlamydia or gono rrhea. To assist with contacting STD-positive youth for treatment, HJAC assessors completed a Supplemental Contact Form with socio-demographic and locator information, as well as post-HJAC placement status (release to community, nonsecure home detention, or secure detention). For STDpositive youth, this form was sent by DOH staff via secure fax to the HCHD. DIS would then seek to lo cate and treat positive youth.8 8 Any identifying information was removed from all data that was given to the research staff to secure the youths’ confidentiality.

PAGE 115

105 Figure 1. Data Collection Protocol Among the eligible 759 males and 634 females who were assessed and asked to participate in the study, 83% of each gender consented to provide the initial urine specimen. Of these, 85.3% of males and 87.5% of females also consented to be tested for STDs (70.5%, 72.7%, and 71.5% of assessed male s, females, and youth overall). No significant differences were found in cons ent rates by gender, race, age, HJAC operational shift (7AM-3PM, 3PM-11PM, 11PM-7AM), or post-HJAC placement. In total, 948 youth, 506 males and 442 females, ag reed to participate in the project. The data used for this study were weighted. In general, females account for about 25% of the overall HJAC population. Howeve r, to ensure sufficient power for genderspecific analyses, they were ove r-enrolled in the project to account for approximately half of study participants. Therefor e, weighting was needed to adjust the sample to represent Juvenile Assessment Center (HJAC) Eligibility Criteria Pre-test Counseling & Consent Department Of Health (DOH) Laboratory Split and Test Urine Fax Positive Results to HCHD Hillsborough County Health Department (HCHD) Intervention Specialist Contact STD Positive Youth Provide Free Treatment Risk Questions & Urine Collection

PAGE 116

106 the targeted population (i.e., the juvenile offending populat ion in Hillsborough County). Accordingly, the proportion of potential ma le enrollees per month from June through September 2006 was used to estimate the number of eligible males booked over the entire recruitment period and to calculate a weighti ng factor of 1.901 for eligible males (n=506, weighted n=961). In all analyses, the male cohort was weighted to provide estimates for the full population during the recruitment period. The female cohort, based on all eligible females, was not weighted (n=442). Thus the total weighted sample involves 1,403 newly arrested adolescents. Individual Level Measures UA Drug test results. For the current study, substance use was measured with two separate indicators: marijuana use and cocaine use Drug use data are based on the urine specimen that was voluntarily provided by st udy participants. At th e testing lab, the split urine specimens were tested for drugs using the EMIT procedure. The cutoff level for a positive marijuana test result was 50 ng/ml of urine and the cutoff level for a positive cocaine test result was 300 ng/ml. The survei llance window for marijuana is five days for moderate users, 10 days for heavy users, and 20 days for chronic users. For cocaine, the surveillance window is 96 hours (Dembo et al., 1999). Both variables are coded 0 = negative and 1 = positive. Risky sexual behavior. Data pertaining to youths’ sexual practices was obtained from the risk assessment questionnaire. Hen ce, this information is self-reported data. Three types of risky sexual beha vior are included in this study. Sex without a condom was measured with the question “Have you ever ha d unprotected sexual in tercourse with the opposite sex?” Sex on drugs/alcohol was measured with the question “Have you ever had

PAGE 117

107 sexual intercourse while usi ng alcohol or other drugs?” Multiple partners was an open ended question asking respondents to reveal the nu mber of sex partners in their lifetime. Following the work of previous public health research (Teplin et al., 2003), responses were recoded into a dichotomous variable re presenting “four or more partners in their lifetime.” Each risky sexual behavior item is a dichotomous variable coded 0 = no and 1 = yes. The dichotomous nature of the sexual risk items does not allow for the examination of differing levels of sexual risk -taking. Therefore, the three sexual risk items were summed into one overall risky se xual behavior index ra nging from 0 (no risky sexual behavior) to 3 (all thr ee risky sexual behaviors). The be nefit of using this index is the ability to examine the strength of the ite m-factor relations acro ss differing levels of sexual risk-taking. The mean for the sexual risk index was 0.61 (SD = .80). Criminal history. Criminal involvement wa s measured us ing the youths’ officially recorded criminal history. This information was obtained via the Department of Juvenile Justice’s tracking system (JJIS). Arrest History represents the number of all prior arrests listed for each juvenile included in the study. This includes arrests for felony and misdemeanor charges, as well as vi olent, property, publi c disorder, and/or noncriminal arrests (e.g., violati on of probation). Descriptive st atistics revealed a skewed distribution. Therefore, “prior arrests” was truncated at the 90th percentile, which equaled 7 arrests (10% of the sample had seven or more arrests in their lifetime). Hence, this item is a categorical variab le ranging from 0 to 7. The value 7 represents 7 or more lifetime arrests. Demographic characteristics. Three demographic characteristics were used in the analyses. Gender is a dichotomous variable coded 1 = male and 0 = female. Age is a

PAGE 118

108 continuous variable. The racial/e thnic variation in the data was quite small. Less than 1% of the respondents were identified as a race other than white or African American (e.g., Native American, Asian, other). Therefore, race is dichotomized with 1 = African American and 0 = Non African American. Table 2 provides a description of the individual-level char acteristics of the weighted sample. As can be seen, over 68% of the sample was male. A little over half identified themselves as African America n. The average age of the respondents was 15.5. Over three-fourths were sent home follo wing arrest (i.e., released or placed on non secure home detention). Thirty -eight percent of the sample tested marijuana positive and just over 5% tested cocaine positive. Analyses Steps Analyses for the current study were carried out in several step s. Throughout these steps, a variety of bivariate, confirmatory factor analysis, an d structural equation modeling techniques were performed. First, a series of bivari ate analyses were completed to determine the level of associa tion among the four observed variables. These descriptive analyses laid the groundwork for the overall confirmatory factor analyses. Problem behavior syndrome is de fined as an underlying, unidimensional disposition toward deviance and is identified by the tendency to engage in multiple forms of problem behavior (Jessor & Jessor, 1977). Based on this definition, problem behavior syndrome represents a latent variable. A late nt variable is described as an unobservable or unmeasurable concept that helps explain the association among two or more observed variables (Bollen, 2002). For the purposes of the current study, risky sexual practices, marijuana and cocaine test result, and deli nquent behavior are th e observed indicators

PAGE 119

109 expected to form a latent variable reflectiv e of a general disposition towards deviance. Table 2. Sample Characteristics of th e Weighted Sample (n = 1403) Variable N % of Sample Gender : Male 961 68.5 Female 442 31.5 Race : African American 736 52.4 Non African American (82% Caucasian) 664 47.3 Age : 12 40 2.8 13 132 9.4 14 194 13.8 15 273 19.5 16 345 24.6 17 364 26.0 18 54 3.9 Mean Age = 15.5 (SD = 1.48) Post HJAC Placement : Release 849 60.5 Non-Secure Home Detention 217 15.5 Secure Detention 335 23.9 Prior Arrests: 0 584 41.6 1 234 16.7 2 161 11.5 3 110 7.9 4 77 5.5 5 62 4.4 6 32 2.3 7 or more 141 10.0 Risky Sexual Practices : Sex without a Condom 302 21.5 More than 3 Lifetime Partners 392 28.0 Sex while Using Drugs/Alcohol 110 7.9 Drug Test Positive : Marijuana Positive 529 37.7 Cocaine Positive 75 5.3 Factor analytic methods are the most common techniques used to identify latent factors. The basic assumption of factor anal ytic techniques is that observed variables are

PAGE 120

110 a combination of some underlyi ng latent construct (Byrne, 2001) As such, these types of methods examine how a set of observed variable s are interrelated and form dimensions of one or more underlying constructs. Joreskog (1979:5) asserts that this multivariate approach attempts to “explain” the correla tions among a set of observed items through an analysis that “yields a smaller number of unde rlying factors that cont ain all the essential information about the linear interrelat ionships among the observed variables.” There are three main types of factor analytic methods used to identify latent factors, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and structural equation modeling (SEM). The dist inction between EFA and CFA is related to the specification of the relationships among the observed indicators (Byrne, 2001). In EFA models, the number of factors to be extr acted and the items that are reflective of each factor are not pre-specified. Thus, EFA models are not testing any specific hypotheses. The purpose of CFA models, on the other hand, is hypothesis testing (Kim & Mueller, 1978). In CFA, a model that reflects certain assumptions regarding the interrelatedness of the observed variables is pr e-specified. That is, the number of latent factors and the item-factor relations are specifie d first. The analysis then determines how well the model, reflecting the particular fact or structure, fits the data (Long, 1983). The main difference between CFA and SEM is th at, in SEM, paths specifying directional relationships are included in the model (K line, 2005). Thus, SEM often involves CFA modeling and path analysis. Specifically, SE M involves a standard CFA, which focuses on the factors and their observed indicator s (measurement model), and path analysis, which focuses on the directional relationships between two or more latent factors and/or additional covariates posited to be related to the latent factor (structural model) (Byrne,

PAGE 121

111 2001). CFA and SEM involve a set of regression equations that determine how well each observed variable is explained by the latent f actor. The factor loading for each variable represents the regression coefficient for that particular variable which represents the correlation between the variable and the factor (Kline, 2005). Although somewhat arbitrary, a general rule of thum b is that a factor loading greater than 0.4 indicates that an observed variable adequately “loads” onto th e latent factor (Raubenheimer, 2004). In other words, the higher the factor loading, the better the observed variable is explained by the latent factor. (For SEM modeling, additi onal regression coefficients are used to determine the strength of the association among the covariates included in the model [e.g., age] and the latent factor [path coefficients]). In addition to the factor loadings, seve ral fit indices are used to assess how well the pre-specified model fits the data. The first model fit statistic is the chi-square of model fit ( ). A non-significant p-value indicates that the null hypothesis cannot be rejected because the specified model’s covarian ce structure is not si gnificantly different from the observed covariance matrix (Byrne 2001). The Root Mean Square Error of Approximation (RMSEA) represents a measur e of the goodness-of-fit that could be expected if the model were tested on th e entire population (Stamatis, 2001). RMSEA values at .05 or less indicate a close model fit, and values between .05 and .08 indicate an adequate model fit (Browne & Cudek, 1993; Hu & Bentler, 1999). The Comparative Fit Index (CFI) and the Tucker Lewis Index (TLI) measure the covariation among the observed variables in the data (Bentler, 1990; Tucker & Lewis, 1973). The typical range for both TLI and CFI is between 0 and 1 w ith values greater than .90 indicating an

PAGE 122

112 acceptable fit (Arbuckle & Wothke, 1999; Browne & Cudek, 1993). Finally, with categorical variables, the weighted root m ean square residual (WRMR) represents the weighted difference between the predicted and observed variances and covariances in the model (Muthn & Muthn, 2007). WRMR values of less than .90 indicate a good model fit (Yu & Muthn, 2001). Standard CFA and SEM models rely on general maximum likelihood (ML) estimation, a technique that assumes that the observed variables are continuous and normally distributed (Bollen, 1989). These assumptions are not met when the observed variables are categorical. As a result, significant problems can result when fitting a standard CFA or SEM model with ML estimat ion using categorical data including an inflated chi-square test, underestimated para meters, and biased standard errors (Muthn & Kaplan, 1985; 1992). An alternative to standard CFA and SEM estimation techniques (i.e., ML estimation) is CFA and SEM with weighted least squares regression (WLS) (Flora & Curran, 2004). This type of estimation rela xes the assumptions of ML by accounting for the non-normal, categorical nature of the variables (Kaplan, 1996). Flora and Curran (2004:469) indicate that this procedure "pro vides asymptotically unbiased, consistent, and efficient parameter estimates as well as a correct chi-square test of fit with dichotomous or ordinal observed variables." In addition to relying on WLS, when factor indicators are categorical, the regression coefficients (i.e., factor loadings ) are probit estimates. Probit regression is a log-linear approach to handling categorica l variables that is based on the Poisson distribution. Also, for categori cal models, thresholds are mode led instead of intercepts or

PAGE 123

113 means (which are estimated when the factor in dicators are continuous). A threshold is a z-score reflecting the probability of falling in a given category, based on the latent factor. The number of thresholds for each observed in dicator is the number of categories minus one. The first threshold represents the lowe st value of the indicator (Muthn & Muthn, 2007). All of the models were estimated us ing Mplus version 5.1 (Muthn & Muthn, 2007). Mplus offers a number of differen t WLS estimation procedures. Muthn, Du Toit, & Spisic (1997) recommend using WL SMV, involving weighted least square parameter estimates using a diagonal weight ma trix with standard errors and a meanand variance-adjusted chi-square te st statistic. This type of estimation accounts for nonnormality, as well as sample size (Muthn et al., 1997). Based on the literature reviewed in the prev ious chapters, it is clear that a priori assumptions regarding the factor structur e of problem behavior syndrome guide the current study. Specifically, th e foundation of this study rests on the basic assumption that engaging in risky sexual pract ices, substance use, and de linquent behavior form an underlying factor reflecting problem behavior syndrome. As such, CFA is the appropriate factor analytic technique to address the firs t research question because it allows for the specification of these expected relationships prior to analyzing the model (Joreskog, 1979). Thus, the next step in the analyses examined a CFA model involving the four observed variables using all 1,403 study participants. The resu lts of this model provide support for the first research question, a nd laid the groundwork for all subsequent analyses reported in this study.

PAGE 124

114 A second goal of the current study is to examine how age influences problem behavior syndrome. Therefore, SEM tech niques were used to incorporate a path reflecting the effect of age on the latent f actor. Prior to analyzing this SEM, the relationship between age and the four obser ved variables was analyzed. Due to the continuous nature of the age variable, a t-test was used to examine the variability in age across these behaviors. The results of these analyses supported the in clusion of age in the basic CFA model. Figure 2 represents the general SEM model that serves as the foundation for this study. Figure 2. General Structural Equation Model As can be seen, it is hypothesized th at self-reported risky sexual behavior, officially recorded criminal history, and ma rijuana and cocaine test results load on one underlying factor (F) reflective of problem behavior syndrome. The arrows from the Marijuana Test Result Cocaine Test Result Criminal History Sexual Risk Index e e e e F Age

PAGE 125

115 latent factor to each of the observed indicators re present a direct effect of the latent factor on each specific behavior. The “e” associated with each of the observed variables represents measurement error. The arrow from age to the latent factor represents a direct effect of age on the latent factor. Thus, this arrow represents th e direct influence that age has on the latent factor. The path coefficient is estimated using multiple regression analysis. The value of the path estimate indicates the direction and st rength of the effect of age on the latent factor for a one unit increase in age (Schumaker & Lomax, 2004). The decision to include age as a covari ate was based on several reasons. First, a wealth of research has documented a str ong, positive linear relati onship between age and each of the observed variables. Additional research suggests that as youth progress through adolescence the covariation among these behaviors strengthens and then peaks in mid to late adolescence (New comb & McGee, 1991; White 1992). Including age as a covariate, rather than a grouping variable, enables the preservation of the continuous nature of the variable, and in turn, allows fo r the estimation of this linear relationship. In addition, breaking age into meaningf ul categories would have substantially increased the number of subgroups analyzed in the SEM models. As a result, the complexity of the models would have substa ntially increased. Further, several of the subgroups, particularly thos e involving youth at younger ages would have contained a low number of youth. Results based on a sm all number of youth may be unreliable, leading to skepticism when drawing conclusions about these groups. Accordingly, accounting for the direct effect of age on the latent factor, while also maintaining the continuous nature of the variable, seem s crucial to providing the most accurate

PAGE 126

116 information regarding the effect of age on problem behavior syndrome. Building on the results of the overall CFA model, the direct path from age to the latent factor was introduced into the overall model. This resulted in the overall SEM. Therefore, the next step in the analyses involved examining the overall SEM based on the weighted 1,403 newly arrested youths. Th is model provides support for the second research question. Multiple-group analyses. The third objective of this study is to compare the factor structure of the latent factor discussed above across different subgroups of juvenile offenders based on important demographic characteristics. Group membership is determined by the participant’s race and gender. The subgroups used in these analyses are based on prior research, re viewed in preceding chapters, regarding variations in risky sexual practices, substance use, and de linquent behavior, in addition to the meaningfulness of the groups. When there is reason to believe that groups within the population exist for whom the indicator variable s are differentially related to the latent variable, group-based CFA and/or SEM techni que is the appropriate technique to use because these similarities and differences can be taken into account (Widaman & Reise, 1997). For instance, a wealth of research hi ghlights important diffe rences in problem behavior for males and females. In genera l, adolescent males are overrepresented in all forms of juvenile delinquency. It has been estimated that the ratio of male to female delinquency is nearly 4 to 1 for violent crim es and 2 to 1 for property crimes (Snyder, 2006). Additionally, studies indicate that fe male offenders are more likely to report serious drug use, while male juvenile offenders are more likely to report marijuana use

PAGE 127

117 (Belenko et al., 2004; Neff & Waite, 2007). Gende r differences in risky sexual behavior, as well as the association between risky se xual practices and substance use, reveal inconsistent patterns. However, it is well established that female offenders are more likely to test STD positive and/or report a history of an STD (Joesof et al., 2006). Furthermore, although a number of studies have failed to find significant gender differences in the structure of problem beha vior, a handful of studi es have highlighted variation in the strength of the association among different forms of deviant behavior (Bartlett et al., 2005; Donovan & Jessor, 1985; Gillmore et al., 1991; LeBlanc & Bouthillier, 2003; White, 1992). Taken togeth er, the research reviewed in previous chapters indicates that the strength of the covariation among risky sexual practices, substance use, and criminal involvement may differ for male and female juvenile offenders. As such, the current study disti nguishes between male and female juvenile offenders. Racial differences in problem behavior have also been consistently documented. White juvenile offenders report higher levels and more serious forms of substance use (Belenko et al., 2004; Teplin et al., 2005). African American juvenile offenders, on the other hand, tend to report higher levels of ri sky sexual practices and are more likely to test STD positive (Kahn et al., 2005; Lofy et al., 2006; Risser et al., 2001). Relatively few studies have examined racial difference s in the risky sexual practices-substance use link. However, the studies that have been c onducted indicate that the link between these two behaviors is stronger for white offenders (Chapter 4). Furthermore, the small number of studies that has asse ssed the structure of problem be havior across racial groups suggests that the covariation among different forms of problem behaviors is stronger for

PAGE 128

118 white adolescents (Costa et al ., 1995; Welte et al., 2004). Ho wever, this evidence is far from conclusive. As a whole, the evidence re garding racial differenc es clearly highlights the need to account for racial variatio n when examining problem behavior among adolescent offenders. Thus, this study fu rther disaggregates the gender subgroups into African American and non African American (mostly Caucasian). The intersection of race and gender also has the potential to influence problem behavior syndrome. A handful of studies have documented differences in problem behavior across race and gender. For instan ce, studies tend to suggest that African American male juvenile offenders report hi gher levels of risky sexual practices, STD infection, and delinquent behavior, but lower le vels of substance use, compared to white male juvenile offenders (Canterbury et al., 1995; Kahn et al., 2005; Neff & Waite, 2007; Teplin et al., 2003). More recently, attention has been given to the special needs of African American female adolescents. In particular, this de mographic subgroup has been identified as an important risk group for problem behavior and poor health outcomes (CDC, 2008). Compared to non African American female offe nders, African American female juvenile offenders have been found to be more likely to report serious juve nile delinquency, report higher levels of unprotected sexual intercourse, and are substantially more likely to test STD positive (Brown et al., 1992; Evans et al ., 2004; Kahn et al., 2005). It has been suggested that African American female o ffenders experience a “double disadvantage” due to both their racial minority and ge nder status (MacDonald & Chesney-Lind, 2001). Both race and gender have multiple correlates that may account for these differences, including cultural expectations experience of sexual victim ization, SES, education, and

PAGE 129

119 poor family environment (Kotchick et al., 2001 ). Therefore, the intersection of these characteristics may lead to a greater li kelihood of problem behavior syndrome among African American female offenders. Furthermore, studies that examine beha vior across race and/or gender subgroups fail to take into account the ag e of the respondent. Therefore, this study attempts to fill this gap in the literature by examining how age influences the covariation among risky sexual behavior, criminal history, and mariju ana and cocaine use across race and gender. Specifically, the subgroups exam ined in this study are: 1. African-American males (n = 517) 2. African-American females (n = 219) 3. Non African American males (n = 445) 4. Non African American females (n = 223) Due to the noted differences in each of the problem behaviors across race and gender, as well as the research (reviewe d above) regarding the interc eption of race and gender, it was decided to rely solely on the race-ge nder subgroups. Given th e evidence, examining subgroups based on only race or only gender w ould have masked important differences due to the failure to consider the other demographic characteristic. The four subgroups were specified a priori as part of the initia l modeling process. Then, the fit of the model to the data, as well as the relationship of the observed variables to the latent factor, were examined across the specified groups. In order to compare subgroups of indi viduals on a latent trait (e.g., problem behavior syndrome), it must first be dete rmined that the observed variables under consideration are measuring the same tra it across the different groups (Drasgow, 1987).

PAGE 130

120 Thus, the goals of group-based factor analytic techniques ar e to determine 1) if the observed indicators are measuring the same late nt construct, and 2) whether the groups differ in interpretable ways on the latent factor. These questions are answered by testing the invariance of th e relations among the observed variables across the groups (W idaman & Reise, 1997). Specifically, measurement invariance “involves the study of similarities and differences in the covariation patterns of item -factor relations” (Windle, Iwawaki, & Lerner, 1988:551). Model invariance reflects the assumption that the relationship between the latent factor and the observed variables is equal across the groups (Widaman & Reise, 1997). Thus, if the best fitting model is found to possess meas urement invariance, then the structure of the latent factor is th e same across the groups. Identifying the appropriate level of m easurement invariance, involves comparing varying levels of invariance. Several steps are required to carry out this process. The first step involves identifying a baseline model that adequately fits the data. This model specifies minimal constraints to identify th e model in each group (Widaman & Reise, 1997). The baseline CFA, therefore, is an unconstrained (i.e., free) CFA model allowing the model parameters to be freely estimated for each group. In subsequent analyses, this baseline model serves as a benchmark to co mpare the more restricted (i.e., invariant) models. The next step involves testing measur ement invariance. This step entails examining a constrained CFA, which consists of fixing parameter estimates, such as factor loadings, factor variances, factor means or in tercepts, to be equal across the groups. The constrained model tests the assu mption that relations between the observed

PAGE 131

121 variables and the latent fact or are similar across the groups The modification indices based on the results of the constrained mode l provide suggestions for ways to improve the model fit. In particular, these sugges tions indicate which parameter estimates should be allowed to vary across the groups (i.e ., reducing the restrictions of the model parameters), in other words, identifying disparity in the obs erved variable-latent factor relationships. The value of th e modification indices represents the expected drop in chisquare if the parameter in questi on is freed (Muthn & Muthn, 2007). Model fit indices are used to determine which level of invariance fits the data best. In a group-based model, measures of model fit are inva riant across groups. Satisfactory values on the m odel fit indices discussed above (i.e., nonsignificant chisquare, CFI and TLI greater than 0.90, low RMSEA and WRMR) signify that the specified model for each group fits the data well If the model fit in dices indicate a good fit for two or more of the models (e.g., free model and constrained model), a chi-square test of model difference is used to determine which model is the best fitting model. This statistic tests the baseline model (i.e., unconstr ained model) against the more restricted (i.e., invariant) model. A nonsignificant chi-square value indicates that constraining the model parameters does not worsen the fit of the model to the data (Muthn and Muthn, 2007), thus, suggesting that a hi gher degree of measurement i nvariance is appropriate. As can be seen, the major advantage of group-based factor analytic techniques is the ability to account for heterogeneity in th e observed variable-laten t factor relationship that exists within the sample. Accordi ng to Widaman & Reise (1997:316), the primary benefit of group-based technique s is that “these procedures provide simple and direct ways of testing crucial hypothese s related to factor invariance.” That is, such techniques

PAGE 132

122 provide a useful way to investigate similarities and differences in the factor structure across groups. Accordingly, when empirical, as well as substantive, evidence highlights the potential for heterogeneity in the structure of a latent factor across groups within a population, these techniques are the appropria te analytic tools for examining these variations. In the current study, the multi-group analyses proceed in several steps. First, a number of significance tests comparing the f our problem behavior measures across race and gender were conducted to determine if, in f act, there is variability in these behaviors across the demographic characteristics. Th e sample was then divided into the four demographic subgroups. Chi-square tests of significance were conducted for these comparisons because they are categorical vari ables (Agresti & Finlay, 1997). Significant chi-square values suggested that these subgr oups differ on each of the problem behaviors, thus, providing support for the group-based SEM analyses. For exploratory purposes, Muthn and Muthn (2007) recommend perfor ming a separate CFA for each subgroup to assess whether the CFA fits the data for each group separately. Therefore, four separate CFA models were examined, one for each demographic subgroup. Then, three group-based CFA models were examined. First, the unconstrained group-based CFA was performed. In this m odel: 1) the factor loading for the first observed variables (sexual risk index) was set to one and all other fa ctor loadings were freely estimated across the groups, 2) all thre sholds were freely estimated across the groups, 3) factor variance was freely estimated across the groups, and 4) the factor mean was set at zero for all groups. The sec ond CFA model tested measurement invariance by constraining the factor loadings and the thresholds for each of the observed variables. The

PAGE 133

123 fit of this model determined whether or not there is important vari ation in the factor structure across African American females, African American males, non African American females, and non African American males. Based on the modification indices of the constrained CFA, an additional CFA was performed freeing the arrest history factor loading for African American males and the thresholds of the sexual risk index for Non Af rican American females. Specifically, this model involves: 1) constraining the factor loadings for sexual risk, marijuana test result, and cocaine test result to be equal across the four groups, 2) c onstraining the factor loading for arrest history to be equal fo r African American females, Non African American females, and Non African American males, 3) allowing the factor loading for arrest history to be freely estimated for Afri can American males, 4) holding the arrest history, marijuana test result, and cocaine te st result thresholds equal across the four groups, 5) constraining the sexual risk thresh olds to be equal for African American females, African American males, and Non African American males, 6) allowing the sexual risk thresholds to be freely estimated for Non African American females, and 7) allowing the factor variances to be freely estimated acro ss the four groups. Then, a chi-square difference test was us ed to identify which of these models fit the data best. The best fitting model served as the measurement model of the final groupbased SEM that is the focus of this study. Pr ior to analyzing the final, group-based SEM, the relationship between age a nd each of the observed variables was examined across the demographic subgroups. These results provided further support for the inclusion of the direct effect of age on the latent factor in the group-based analys es. Finally, the groupbased SEM was performed. The results of th is model provide moderate support for the

PAGE 134

124 third research question. The next chapter pres ents a step-by-step discussion of the results of each phase of the analytic pro cess outlined in this chapter.

PAGE 135

125 Chapter 8 Results The objectives of this stu dy are to: 1) determine whet her risky sexual practices, substance use, and criminal involvement fo rm a unidimensional latent factor using a sample of newly arrested juvenile offenders, 2) assess the effect that age has on the latent factor, and 3) compare the structure of the latent factor, as well as the effect of age on the latent factor, across four demographic subgroups. To accomplish these goals, a number of bivariate relationships, followed by factor analytic and structural equation models, were analyzed. This chapter presents the findings of the current study, which are based on a sample of 948 (weighted: n = 1,403) newly arrested juvenile offenders. Preliminary Analyses Prior to examining the basic SEM that serves as the founda tion of this study (displayed in Figure 2), a series of bivari ate analyses were performed to examine the association among the four observed variables. The results of these analyses laid the groundwork for the CFA model, which form s the measurement model of the SEM. Bivariate Analyses. First, the bivariate correl ations among the four observed variables were analyzed to determine wh ether the four obser ved variables were associated. All correlations are significant a nd positive; indicating that participants who had higher values on any one of the four obser ved variables were more likely to have a higher value on the other three behavioral i ndicators. Table 3 presents the polychoric

PAGE 136

126 correlation matrix.9 As can be seen, the relationship among the substance use variables is the strongest association. Table 3 Polychoric Correlations between th e Observed Indicators (n = 1403) 1 2 3 4 1. Marijuana -2. Cocaine 0.559 -3. Sexual Risk Index 0.356 0.322 -4. Arrest History 0.277 0.273 0.203 -Tables 4-7 report the bivari ate relationships among the four problem behaviors. As can be seen in Table 4, youth who tested marijuana positive were significantly ( p < .001) more likely to test positive for cocaine, ha ve a higher number of prior arrests, and report a higher number of ris ky sexual practices. In part icular, 83% of the cocaine positive youth were marijuana positive, compared to 35% of the cocaine negative youth. Only 29% of the youth with no prior criminal history tested positive for marijuana, but 54% of the youths with four arrests and 59% of the youths with 7 or more arrests tested marijuana positive. Similarly, 59% of the youths who reported all three of the risky sexual behavior items tested marijuana pos itive, whereas only 27% of the youth who reported no risky sexual practices tested positive. Based on these results, it is clear that marijuana use is associated with additional problem behaviors among the youths included in this study. 9 Polychoric correlations are used wh en the variables are categorical. Po lychoric correlation extrapolates what the categorical variables' distributions would be if continuous, adding tails to the distribution. As such it is an estimate based on the assumption of an underlying continuous bivariate normal distribution (Flora & Curran, 2004).

PAGE 137

127 Table 4. Bivariate Relationships between Ma rijuana Test Result and Risky Sexual Behavior, Arrest History, and Cocaine Test Result (n = 1,403) Variables Marijuana Positive (n = 529) Cocaine : Negative 35.2% Positive 82.9% (4) = 1474.64, p < .001 Arrest History: 0 29.1% 1 30.8% 2 45.3% 3 43.2% 4 53.8% 5 42.9% 6 46.9% 7 or more 58.9% (16) = 84.76, p < .001 Sex Risk Index : 0 26.8% 1 50.9% 2 54.5% 3 58.5% (8) = 95.07, p < .001 Table 5 indicates that, similar to mariju ana use, youth who tested cocaine positive were significantly more likely ( p < .001) to engage in each of the problem behaviors. Of the marijuana positive youths, 12% tested cocaine positive; of the marijuana negative youth, 1.5% tested cocaine positive. An in teresting curvilinear relationship between cocaine use and arrest history was revealed. For example, yout h with six arrests revealed the higher proportion of cocaine positive youth. However, four teen percent of the youths with seven or more arrests tested cocaine positive, whereas 3% of the youths with no prior arrests and 2% of the youths with one arre st tested cocaine posit ive. In regard to risky sexual practices, 12% of the youth who reported all th ree behaviors, compared to nearly 3% of the youths reporting no be haviors, tested cocaine positive.

PAGE 138

128 Table 5. Bivariate Relationships between Cocaine Test Result and Risky Sexual Behavior, Arrest History, and Marijuana Test Result (n = 1,403) Variables Cocaine Positive (n = 75) Marijuana : Negative 1.5% Positive 11.9% (4) = 1474.64, p < .001 Arrest History: 0 3.4% 1 2.1% 2 6.2% 3 8.1% 4 3.8% 5 6.3% 6 18.8% 7 or more 13.5% (16) = 57.84, p < .001 Sex Risk Index : 0 2.7% 1 7.7% 2 12.6% 3 12.2% (8) = 39.40, p < .001 Table 6 reports the bivariat e association among each of the problem behaviors and risky sexual practices. Sixt y percent of the cocaine ne gative youth reported no risky sexual practices, whereas only 29% of the cocaine positive youth reported no risky sexual practices. Nearly 7% of the cocaine positive youth reported all three behaviors, compared to only 2.7% of the cocaine negativ e youth. Marijuana use revealed a similar pattern: 68% of the negativ e youth and 43% of the pos itive youth reported no risky sexual practices; however, nearly 2% of th e negative and 5% of the positive youth reported all three behaviors. Prior arrests reve aled an inconsistent pattern of association. The percentage of youth reporting all three behaviors increased with number of prior arrests. But as can be seen, this linear association stopped at six arrests and then

PAGE 139

129 substantially declined at 7 or more. This finding is some what contradictory to a wealth of research that suggests that the most serious adolescent offenders tend to report the highest levels of risky sexual practices (Tolou-Simmons et al., 2007). Regardless, these findings highlight a significant ( p < .001) association suggesting that youths involved in this study that reported a higher number of risky sexual pr actices were more likely to test marijuana and/or cocaine positive and have a higher number of arrests (up to six arrests). Table 6. Bivariate Relationships between Risky Sexual Behavior and Arrest History, Marijuana Test Result, Cocaine Test Result Sexual Risk Index Variables 0 1 2 3 Cocaine : Negative 60.4% 27.4% 9.5% 2.7% Positive 29.3% 40.0% 24.0% 6.7% (6) = 34.50, p < .001 Marijuana : Negative 68.3% 22.2% 7.6% 2.0% Positive 43.1% 37.6% 14.7% 4.5% (6) = 88.12, p < .001 Arrest History : 0 64.7% 23.5% 9.9% 1.9% 1 59.4% 32.5% 7.3% 0.9% 2 60.5% 24.1% 11.7% 3.7% 3 61.3% 30.6% 4.5% 3.6% 4 51.3% 38.5% 3.8% 6.4% 5 39.7% 28.6% 20.6% 11.1% 6 42.4% 21.2% 27.3% 9.1% 7 or more 43.3% 38.3% 15.6% 2.8% (24) = 89.34, p < .001 Table 7 reports similar results regarding the associations between arrest history and the other three problem behaviors. All associations are significant at the p < .001 level. Interestingly, a larger percentage of marijuana posi tive (13%) and cocaine positive (14%) positive had two prior arrests, compar ed to youths with three to six arrests.

PAGE 140

130 However, the largest percentage of marijuan a positive (15.7%) had seven or more arrests. Similarly, a larger percentage of youths with two arrests reported a ll three risky sexual practices (15.1%), compared to youths with th ree or four arrests. But, the largest proportion of youths reporting all three sexual behaviors ha d five arrests (16.7%). Table 7. Bivariate Relationships between Arrest History and Risky Sexual Behavior, Marijuana Test Result, and Cocaine Test Result Arrest History Variables 0 1 2 3 4 5 6 7 or more Cocaine : Negative 42.5% 17.2% 11.4% 7.5% 5.6% 4.4% 2.0% 9.2% Positive 26.3% 6.6% 13.2% 11.8% 3.9% 5.3% 7.9% 25.0% (16) = 57.84, p < .001 Marijuana : Negative 47.4% 18.6% 10.0% 7.0% 4.1% 4.1% 1.9% 6.6% Positive 32.1% 13.6% 13.8% 9.1% 7.9% 5.1% 2.8% 15.7% (16) = 84.76, p < .001 Sex Risk : 0 45.8% 16.8% 11.9% 8.2% 4.8% 3.0% 1.7% 7.4% 1 34.7% 19.2% 9.9% 8.6% 7.6% 4.6% 1.8% 13.7% 2 39.7% 11.6% 13.0% 3.4% 2.1% 8.9% 6.2% 15.1% 3 26.2% 4.8% 14.3% 9.5% 11.9% 16.7% 7.1% 9.5% (24) = 88.34, p < .001 Indeed, these preliminary results sugges t that participation in any one of the behaviors included in this study is significantly related to par ticipation in th e other three problem behaviors. Therefore, these findi ngs support the assump tion that these four observed behaviors form a latent construct reflective of problem behavior syndrome. This a priori assumption guides the general SEM model that serves as the foundation of this study. Confirmatory Factor Analysis. The next step in the SEM process was to examine a CFA model including the four observed variab les based on the sample as a whole. The

PAGE 141

131 results of this model are presented in Table 8.10 The model fit indices highlight a good fit of the model to the data. The chi-square test of model fit is nonsignificant ( = 0.54, p = 0.76) indicating that the null hypot hesis (i.e., the model fits th e data) cannot be rejected. Both CFI and TLI are close to 1 (CFI = 1.00; TLI = 1.021), and the WRMR is low (WRMR = 0.15). RMSEA equals zero, which me ans that the model fit the data so well the RMSEA could not be estimated. Th e Mplus program did not suggest any modifications to the CFA model. Based on these results, it can be concluded that the hypothesized model fits the data well. The regression estimates for each of the observed variables are significant indicating that each behavioral indicator loads onto the latent factor. However, notable differences in the effect size of the factor loadings were revealed. The standardized loadings for both measures of substan ces use were much higher (marijuana: bstdYX = 0.746; cocaine: bstdYX = 0.725, compared to the sexual index ( bstdYX = 0.480) and arrest history ( bstdYX = 0.385).11 These values indicate that, for the newly arrested offenders included in this study, the latent factor wa s better able to pr edict substance use. It is also important to note that the re sidual variance of the latent factor in the CFA model is signifi cant at the .001 level. Accordingly, there is still a significant amount of variation in the latent factor that is not accounted for by the observed variables. This finding suggest s that additional factors may be useful in understanding problem behavior syndrome among newl y arrested juvenile offenders. 10 The Regression estimates and standard errors re ported in the tables throughout this study are understandardized estimates. 11 The standardized parameters reported in this chap ter were estimated using the stdYX option in Mplus. This standardization process relies on the variances of the latent variable for standardization (Muthn & Muthn, 2007). Tables including these results are available upon request.

PAGE 142

132 Table 8. Confirmatory Factor Analysis (n = 1403) Variable Estimate S.E. Critical Ratio F1 by : Sex Risk 1.000 --Marijuana 1.554* 0.286 5.427 Cocaine 1.511* 0.243 6.222 Arrest History 0.801* 0.149 5.370 Residual Variance : F1 0.231* 0.054 3.780 Model Fit Statistics: = 0.539, p = 0.76; CFI = 1.000; TLI = 1.021; RMSEA = 0.00; WRMR = 0.151. p < .001 These results provide support for the fi rst research question. Among the newly arrested youths included in this study, the risky sexual behavior index, marijuana test result, cocaine test result, and arrest hist ory form a unidimensional factor reflecting problem behavior syndrome. Accordingly, th is CFA model serves as the measurement model for all subsequent SEM an alyses reported in this study. Basic Structural Equation Model Due to the large body of evidence that s uggests that age influences the tendency to engage in each of the observed behaviors, as well as the structure of problem behavior syndrome, including age in the model reported in Table 8 is necessary to obtaining an accurate understanding of the latent factor (W hite, 1992; Teplin et al., 2003; Loeber et al., 1999; Newcomb & McGee, 1991). As displayed in Figure 2, this model hypothesizes that the risky sexual behavior index, marijuana test result, cocaine test result, and arrest history form a unidimensional la tent factor and that age has a direct effect on the latent factor. Prior to analyzing this model, the bivariate relationship be tween age and each of the four observed vari ables was assessed.

PAGE 143

133 Bivariate analyses with age. Table 9 displays the bivari ate relationships between age and the risky sexual behavi or index, arrest history, mariju ana test result, and cocaine test result. Table 9. Sexual Behavior, Arrest History, Ma rijuana and Cocaine Test Result by Age Variables Mean Age Marijuana Positive : Negative 15.23 Positive 15.87 F(1,402) = 32.58, p < .001 Cocaine Positive : Negative 15.43 Positive 16.19 F(1,402) = 9.53, p < .001 Sexual Risk Index : 0 15.05 1 15.99 2 16.06 3 15.88 F (1,402) = 36.78, p < .001 Arrest History : 0 15.37 1 15.40 2 15.42 3 15.51 4 15.47 5 15.47 6 15.81 7 or more 15.84 F (1,402) = 1.72, p = .05 As can be seen, older adoles cents were significantly more likely to test marijuana and cocaine positive ( p < .001). The mean age for youths who reported two risky sexual behaviors was the highest ( p < .001). In regard to arrest history, older ad olescents were also somewhat more likely to have a higher number of arrests ( p = .05). Because the

PAGE 144

134 latent factor is comprised of these four obs erved variables, these findings provide support for the hypothesis that age will have a direct effect on the latent factor. Structural equation model. The basic structural equation model that forms the foundation of this study is presented in Figure 2 and Table 10. Introduc ing age as a direct effect on the latent factor reduced the quality of the fit of the model to the data (compared to the CFA results in Table 8). The fit of the overall model was marginally acceptable, although not desirable. Both CFI and TLI were somewhat low, although they were greater than the cut-off level of 0.90 (C FI =0.948, TLI = 0.906). The RMSEA (.052) and WRMR (0.841) were somewhat high. Th e chi-square test of model fit ( = 17.84) was significant at the .05 level which indicates that the null hypothesis (i.e ., the model fits the data) should be rejected. Table10. Structural Equation Model (n = 1403) Variable Estimate S.E. Critical Ratio F1 by : Sex Risk 1.000 --Marijuana 1.168* 0.166 7.028 Cocaine 1.252* 0.194 6.436 Arrest History 0.599* 0.103 5.805 F1 on : Age 0.182* 0.027 6.812 Residual Variance : F1 0.302* 0.062 4.851 R-Square : Sex Risk 0.349 Marijuana 0.464 Cocaine 0.526 Arrest History 0.131 Model Fit Statistics: = 17.84, p = 0.003; CFI = 0.948; TLI = 0.906; RMSEA = 0.052; WRMR = 0.841. p < .001

PAGE 145

135 However, the direct effect of age on the la tent factor revealed a significant effect. This effect suggests that age has a positive eff ect on the latent factor which indicates that the covariation among the four observed indica tors is stronger for older adolescents. Thus, the second research question is also s upported. Age does have a direct effect on the latent factor. The standardized factor loadings, as well as the r-square values, highlighted important variation in the rela tionship between each of the f our behavioral indicators and the latent factor. Although significant the effect size for arrest history ( bstdYX = 0.362) was substantially lower than the other three indicators (sexual risk index: bstdYX = 0.591; marijuana test result: bstdYX = 0.682; cocaine test result: bstdYX = 0.725). Furthermore, the r-square values in Table 10 indicate that 46 % of variation in marijuana use, 53% of variation in cocaine use, 35% of variation in the risky sexual behavior index, but only 13% of the variation in arrest history was accounted for by the latent factor. Thus, the ability of the latent factor to explain th e observed indicators va ried among the newly arrested juvenile offenders. Similar to th e results reported in Ta ble 9, the relationship between the latent factor and substance us e was the strongest for the newly arrested juvenile offenders involved in this study. These findings coinci de with prior research that highlights marked differences in the observed be haviors-latent factor relations (Osgood et al., 1988; Welte et al., 2004). It is important to note that the residual variance of the latent factor in the CFA model is significant at the .001 level. Accordingly, ther e is a significant amount of variation in the latent factor that is not accounte d for by the observed variables. This

PAGE 146

136 finding suggests that additional factors may be useful in understanding problem behavior syndrome among newly arrested juvenile offenders. As argued throughout this study, one potential reason for the undesirable fit of the overall SEM model to the data, as well as the large amount of unexplained variance, may be related to differences in the relatio nships examined in the SEM model across subgroups nested in the overall sample. That is, it is possible that the association among risky sexual behavior, cocaine a nd marijuana test result, and a rrest history, as well as the effect of age on this association, may differ across demographic subgroups. Failing to account for this variation has the potential to influence the results of the SEM analyses conducted on the sample as a whole. Theref ore, the rest of this chapter focuses on variations in the observed vari ables, the latent factor, a nd the effect of age on these variables, across four demographic subgroups : African American females, Non African American females, African American male s, and Non African American males. Group-Based Confirmatory Factor Analysis Building on the findings above, the next phase of the analyses involved examining whether the factor structure found in Table 8 is consistent across the four demographic subgroups. The group-based analyses proceeded in several steps. First, a series of bivariate analyses were performed to determine whether each of the four observed indicators varied acr oss race and gender. Then, the sample was broken down into four demographic subgroups: African American females (n = 219), Non African American females (n = 223), African Amer ican males (n = 517), and Non African American males (n = 445). Differences in th e risky sexual behavior index, arrest history, marijuana test result, and cocaine test resu lt were examined across these four groups.

PAGE 147

137 Next, a separate CFA was performed for each of the four demographic subgroups. These results provided preliminary suppor t for the group-based analyses. Last, four group-based models were pe rformed. First, an unconstrained CFA model was tested across the four groups. In the unconstrained model, the factor loadings and thresholds were free to vary across the four subgroups. Next, a constrained groupbased CFA was performed to determine if the st ructure of the latent factor was the same across the four groups. The goal of the constrained model was to test for CFA measurement invariance and equality in the structural parameters (e.g., factor loadings, thresholds) to determine whether the model fit the data similarly across the four subgroups. This CFA model involved holding th e factor loadings and thresholds equal across groups. Then, based on suggestions in the modification indi ces, an additional group-based CFA was performed freeing 1) the factor load ing for arrest history for African American males and 2) freeing the th resholds for the sexual risk index for Non African American females. A chi-square test of model difference was used to determine which model fit the data best. The best fitt ing CFA served as the measurement model in the final group-based SEM analyses. Bivariate analyses. Tables 11-13 present the bivari ate relationships between the four problem behaviors incl uded in this study and gende r, race, and race-gender subgroup, separately. As Table 11 indicat es, males included in this study were significantly more likely to test marijuana positive (27% of females, 43% of males, p < .001). However, no significant gender differen ces were found for cocaine test result (4% of females, 6.1% of males, p = .36). Males were also more likely to have a higher number of arrests (7 or more arrest s = 6% of females, 12% of males, p < .001).

PAGE 148

138 Significant differences in risky sexual practices were also re vealed. As can be seen, a larger percentage of females reported no ris ky sexual practices and all three behaviors; however, a larger percent of males repor ted one or two risky sexual practices. Table 11. Observed Indicators by Gender Variables Female (n = 442) Males (n = 961) Marijuana Positive : Negative 73.3% 56.9% Positive 26.5% 42.9% (2) = 34.69, p < .001 Cocaine Positive : Negative 95.7% 93.9% Positive 4.1% 6.1% (2) = 2.06, p = .36 Sexual Risk Index : 0 65.4% 55.7 % 1 22.9% 30.5% 2 8.6% 11.0% 3 3.2% 2.8% (3) = 12.92, p < .01 Arrest History : 0 53.4% 36.2% 1 18.1% 16.0% 2 11.5% 11.4% 3 4.8% 9.3% 4 2.0% 7.1% 5 2.5% 5.3% 6 2.0% 2.4% 7 or more 5.7% 12.1% (8) = 63.05, p < .001 Table 12 reports the bivariate relationsh ips between each of the four behavior indicators and race. A significantly higher num ber of Non African American participants tested cocaine positive (3.5% of African American partic ipants, 7.5% of Non African American participants, p < .001), and reported a higher number of risky sexual practices (all three risky sexual behavi ors = 0.8% of African Americ an participants, 5.3% of Non

PAGE 149

139 African American participants p < .001) than African Amer ican youths. There were no significant differences in marijuana test resu lt (36% of African American participants, 40.0% of Non African American participants, p = .25). Similar to previous research, African American youths revealed more serious arrest histories (7 or more = 12.2% of African American particip ants, 7.7% of Non African American participants, p < .001) compared to Non African American participants. Table 12. Observed Variables by Race Variables African American (n = 7 36) Non African American (n = 664) Marijuana Positive : Negative 63.9% 59.8% Positive 36.0% 39.9% (2) = 2.78, p = .249 Cocaine Positive : Negative 96.5% 92.2% Positive 3.5% 7.5% (2) = 12.26, p < .01 Sexual Risk Index : 0 60.6% 57.1% 1 31.5% 24.2% 2 7.1% 13.4% 3 0.8% 5.3% (3) = 44.91, p < .001 Arrest History : 0 31.8% 52.0% 1 18.1% 15.2% 2 13.6% 9.3% 3 10.2% 5.4% 4 6.0% 5.0% 5 5.3% 3.5% 6 2.9% 1.7% 7 or more 12.2% 7.7% (8) = 66.81, p < .001

PAGE 150

140 Next, the demographic characteristics of the youth were broken down into the four subgroups. Differences in the risky sexua l behavior index, arre st history, marijuana test result, and cocaine test result across these four groups are presented in Table 13. Table 13. Observed Indicators by Demographic Subgroup Variables Non AA FemaleAA Female Non AA Male AA Male Marijuana : Negative 68.2% 78.5% 57.6% 56.0% Positive 31.8% 21.4% 42.4% 43.8% (6) = 43.84, p < .001 Cocaine : Negative 93.7% 97.7% 91.5% 95.9% Positive 6.3% 2.3% 8.6% 4.1% (6) = 17.46, p < .01 Sexual Risk Index : 0 58.3% 72.6% 56.0% 55.5% 1 24.7% 21.0% 23.8% 36.0% 2 11.7% 5.5% 15.1% 7.7% 3 5.4% 0.9% 5.2% 0.8% (9) = 69.79, p < .001 Arrest History : 0 61.4% 45.2% 47.8% 26.1% 1 17.0% 19.2% 14.1% 17.6% 2 7.2% 16.0% 10.3% 12.6% 3 4.0% 5.5% 6.1% 12.2% 4 0.4% 3.7% 7.2% 7.0% 5 3.6% 1.4% 3.4% 7.0% 6 1.3% 2.7% 1.8% 2.9% 7 or more 4.9% 6.4% 9.0% 14.7% (24) = 143.64, p < .001 Important differences across the four de mographic subgroups were revealed for each of the four variables. African Amer ican males, followed closely by Non African American males, were more likely to test marijuana positive. Non African American males, followed by Non African American females were more likely to test cocaine positive. In regard to risky sexual practi ces, Non African American females and Non

PAGE 151

141 African American males were substantially more likely to report all three risk-taking behaviors. A substantially la rger portion of African American males had seven or more arrests, compared to the other three groups. African American females were the least likely test drug positive. Non African American females revealed the lowest proportion of youth with seven or more arrests (4.9%). These preliminary findings coincide with prior research that highlights marked differences in problem behaviors across ge nder (Belenko et al., 2004; Neff et al., 2007), race (Belenko et al., 2004; Dembo et al., 2007a), and gender-race subgroups (Teplin et al., 2005) among adolescent offenders. Furtherm ore, by comparing the results in Tables 10-12, it is clear that examini ng behaviors across only one de mographic factor does not provide the most accurate information. For example, the findings in Table 11 suggest that females in this study were more likely to report all three risky sexual practices. But, as can be seen in Table 13, this finding is being carried by the higher number of Non African American females repor ting all three behaviors. Th e African American females in this study reported the relatively levels of a ll three sexual risk beha viors. These results are somewhat contradictory to previous re search that highlights African American females as a high-risk group for problem be haviors (CDC, 2008). One possible reason for these findings could be the types of m easures used in this study (e.g., risky sexual practices, officially recorded delinquency, dr ug test results) compared to the measures used in other studies (e.g., STD infection, se lf-reported delinquency a nd substance use). Also, the results in Table 12 suggest that Non African American participants were more likely to test marijuana positive. Yet, as displayed in Table 13, this finding is based on the large number of Non African American ma les that tested positive. A substantially

PAGE 152

142 larger percentage of African American males tested marijuana positive, compared to Non African American females. Throughout this st udy, it has been suggested that both race and gender need to be taken into account when examining the structure of problem behavior syndrome among juve nile offenders. These findings underscore the importance of considering demographic characteristics a nd provide a level of support for the groupbased SEM. Separate CFA models. Next, four separate CFA m odels were analyzed to determine whether or not the overall model fit the data for each of the groups separately. The results for each of the four models are pr esented in Table 14. The data fit the model in three of the four groups. The residual variance for three groups was nonsignificant indicating that the obse rved indicators accounted for a si gnificant amount of the variation in the latent factor (unexplained va riance: AAF = 26%, NAAM = 24%, AAM = 13%). However, the results of the separate CFA fo r Non African American females revealed a questionable fit of the model to the data. In addition, the residual variance for Non African American females was significant; 37% of the variation in the latent factor was not explained by the observed variables in the CFA model for this subgroup. In all four groups, the factor loadings fo r each of the four observed variables were positive and significant. But, two discrepancies in the strength of the relationships should be mentioned. Although significant, the stre ngth of the relationshi p between marijuana test result and the latent fact or for the African American ma les is not as strong, compared to the other three groups. Similarly, for th e African American fe males, the significance of the relationship between th e latent factor and arrest history is somewhat weaker, compared to the other three groups.

PAGE 153

143 Table 14. Separate CFA Models for Each of the Four Demographic Subgroups Estimate S.E. Critical Ratio Non African American Females (n = 223) F1 by : Sex Risk 1.000 0.000 -Marijuana 0.891** 0.266 3.343 Cocaine 1.428** 0.359 3.978 Arrest History 0.833** 0.253 3.296 Residual Variance : F1 0.371** 0.133 2.798 = 6.959, p = 0.00; CFI = 0.924; TLI = 0.810; RMSEA = 0.105; WRMR = 0.475 African American Females (n = 219) F1 by : Sex Risk 1.000 0.000 -Marijuana 1.593** 0.739 2.154 Cocaine 1.644** 0.400 4.100 Arrest History 0.569* 0.263 2.163 Residual Variance : F1 0.264 0.146 1.803 = 0.708, p = 0.70; CFI = 1.000; TLI = 1.036; RMSEA = 0.000; WRMR = 0.173 Non African American Males (n = 445) F1 by : Sex Risk 1.000 0.000 -Marijuana 1.258** 0.412 3.054 Cocaine 1.454** 0.442 3.290 Arrest History 0.982** 0.343 2.868 Residual Variance : F1 0.240 0.126 1.904 = 0.513, p = 0.77; CFI = 1.000; TLI = 1.079; RMSEA = 0.000; WRMR = 0.142 African American Males (n = 517) Variable F1 by : Sex Risk 1.000 0.000 -Marijuana 2.694* 1.220 2.209 Cocaine 1.638** 0.546 3.001 Arrest History 0.655* 0.286 2.289 Residual Variance : F1 0.131 0.076 1.716 = 0.608, p = 0.74; CFI = 1.000; TLI = 1.081; RMSEA = 0.000; WRMR = 0.171 Overall, these preliminary CFA models s uggest that the same latent factor is being measured in each of the four groups ; however the factor structure for the Non

PAGE 154

144 African American females may differ from the factor structure of the other three groups. These findings support further investigation of the comparison of the factor structure across these groups. Unconstrained group-based CFA. The first model that was analyzed involves an unconstrained model in which the factor loadin gs and thresholds are free to vary across the groups, while the intercepts are held at zero. Results of this model indicated a good fit of the model to the data. The chi-s quare test of model fit was nonsignificant ( = 7.52, p = 0.48), CFI and TLI were greater than 0.90 (CFI = 1.00, TLI = 1.01), RMSEA was low (0.000), and the WRMR was less than 0.90 (0.55). All of the factor loadings were significant (p < .05) and in the same direction across the four groups. Therefore, the next step in the analytic pha se was to determine if allowing these parameters to vary across the groups is meaningful to the overall fit of the model to the data. Constrained group-based CFA. The next step in the group-based analyses involved examining CFA measurem ent invariance. In the c onstrained CFA, the factor loadings and thresholds were held equal ac ross the groups. The results of this model reveal a poor fit of the model to the data. The chi-square test of model fit was significant ( = 58.94, p = .0002), CFI and TLI were less th an 0.90 (CFI = 0.87, TLI = 0.89), and RMSEA and WRMR were high (RMSEA = 0.07, WRMR = 1.74). These results indicate that important difference(s) in the factor structure of the late nt factor exist across the four groups. The next step in the group-based modeli ng process was to review the modification indices to determine if allowi ng any of the model parameters to vary across the subgroups will improve the fit of the model. The modification indices provided two important

PAGE 155

145 suggestions. The first suggestion involved fr eeing the factor loading for the African American males (MI = 15.043). Drawing from Table 13, it is clear that this group displays the largest proporti on of youths with a high numbe r of prior arrests (three or more arrests). The second suggestion involve d freeing the threshold of the sexual risk index for Non African American females (MI = 15.943). Referring back to Table 13, it is also clear that this group represents a high-risk group fo r risky sexual behavior. Thus, these associations may be influencing the re lationships between arrest history and the latent factor for African American males and risky sexual behavior a nd the latent factor for Non African American females. Final group-based CFA. Based on the modification indi ces, the next, and final, CFA that was performed involved freeing the th resholds of the sexual risk index for the Non African American females, as well as freei ng the factor loading of arrest history for the African American males. Results for this model are presented in Table 15. Overall, the model fit the data well ( = 26.95, p = 0.26; CFI = 0.985; TLI = 0.986; RMSEA = 0.027; WRMR = 1.150). Significant differences in the factor mean across the four groups were revealed. As can be seen, the Non African American males served as the reference group. Thus, the African American ma les revealed a significantly higher factor mean, whereas the Non African American fema les revealed a significantly lower factor mean, compared to the Non African American males. However, ther e were no significant differences found in the factor mean for the Non African American males and the African American females. Similar to the other three groups, the fact or loading of arrest history for the African American males was significant. Howe ver, freeing the thresholds of the sexual

PAGE 156

146 risk index for the Non African American fe males produced a nonsignificant association between the first and second thresholds and the latent factor. This finding indicates that important differences in the association between the extent of risky sexual practices (i.e., number of behaviors reported) and the latent factor exist fo r this subgroup. In contrast, all three sexual risk thresholds remain ed significant for the other three groups. The last step in determining the best fitting CFA model was the chi-square difference test. This statistic compares the l east restrictive model to a more restrictive model (Muthn & Muthn, 2007). Because the fu lly constrained model revealed a poor fit, it was not necessary to include this model in the chi-square difference test. Thus, this step involved comparing the unconstrained m odel to the model presented in Table 15. A significant p-value means the restriction worsen s model fit. Results of the chi-square difference test indicated that the restricted model does not worsen the fit of the model ( = 20.27, p = 0.21). Therefore, the final CFA model used in the SEM analyses is the model reported in Table 15. Three implications can be drawn from thes e results. First, the differences in the values of the substance use factor loadi ngs found in the unconstrained model do not reflect any meaningful differen ces in the structure of the la tent factor. Co nstraining the factor loadings to be equa l across the groups produced a better model fit. Secondly, although significant, freeing the factor loadi ng for the African American males resulted in an improved model fit. Finally, allowing the thresholds for the sexual risk index to vary for the Non African American female s also improved the fit of the model.

PAGE 157

147Table 15. Final Group-based CFA Non AA Females (n = 223) AA Females (n = 219) Non AA Males (n = 445) AA Males (n = 517) Estimate S.E. CR Estimate S.E. CR Estimate S.E. CR Estimate S.E. CR F1 by : Sex Risk 1.000 --1.000 --1.000 --1.000 --Marijuana 1.044** 0.251 4.159 1.044** 0.251 4.159 1.044** 0.251 4.159 1.044** 0.251 4.159 Cocaine 1.752** 0.329 5.332 1.752** 0.329 5.332 1.752** 0.329 5.332 1.752** 0.329 5.332 Arrest History 0.819** 0.206 3.974 0.819** 0.206 3.974 0.819** 0.206 3.974 2.066** 0.487 4.244 Mean F1 -0.623** 0.231 -2.698 -0.124 0.094 -1.317 0.000 --0.202** 0.056 3.587 Variance: F1 0.602 0.322 1.868 0.125* 0.049 2.555 0.257** 0.085 3.974 0.025 0.014 1.817 Thresholds: Marijuana1 0.240** 0.064 3.748 0.240** 0.064 3.748 0.240** 0.064 3.748 0.240** 0.064 3.748 Cocaine1 1.347** 0.115 11.701 1.347** 0.115 11.701 1.347** 0.115 11.701 1.347** 0.115 11.701 Sex Risk1 -0.333 0.194 -1.718 0.217** 0.058 3.760 0.217** 0.058 3.760 0.217** 0.058 3.760 Sex Risk2 0.550 0.328 1.676 0.849** 0.083 10.283 0.849** 0.083 10.283 0.849** 0.083 10.283 Sex Risk3 1.368* 0.597 2.291 1.477** 0.130 11.397 1.477** 0.130 11.397 1.477** 0.130 11.397 Arrest History1 -0.150* 0.073 -2.072 -0.150* 0.073 -2.072 -0.150* 0.073 -2.072 -0.150* 0.073 -2.072 Arrest History2 0.290** 0.069 4.190 0.290** 0.069 4.190 0.290** 0.069 4.190 0.290** 0.069 4.190 Arrest History3 0.615** 0.075 8.236 0.615** 0.075 8.236 0.615** 0.075 8.236 0.615** 0.075 8.236 Arrest History4 0.851** 0.082 10.320 0.851** 0.082 10.320 0.851** 0.082 10.320 0.851** 0.082 10.320 Arrest History5 1.034** 0.092 11.260 1.034** 0.092 11.260 1.034** 0.092 11.260 1.034** 0.092 11.260 Arrest History6 1.217** 0.102 11.954 1.217** 0.102 11.954 1.217** 0.102 11.954 1.217** 0.102 11.954 Arrest History7 1.342** 0.109 12.274 1.342** 0.109 12.274 1.342** 0.109 12.274 1.342** 0.109 12.274 Model Fit Statistics: = 26.95, p = 0.26; CFI = 0.985; TLI = 0.986; RMSEA = 0.027; WRMR = 1.150. Significance Levels: p < .05; ** p < .01

PAGE 158

148 Group-Based Structural Equation Model The last set of bivariate analyses involved assessing the relationship between age and the observed variables acro ss the four demographic subgroups. These results are presented in Table 16. Table 16. Bivariate Relationship between the obse rved indicators and age across the four demographic subgroups Mean Age Non AA Females (n = 223 ) AA Females (n = 219 ) Non AA Males (n = 445) AA Males (n = 517) Marijuana: Negative 15.36 14.95 15.54 15.04 Positive 15.80 15.85 15.97 15.81 F(222) = 4.73* F(218) = 5.87** F (443) = 5.36** F(515) = 35.80*** Cocaine: Negative 15.46 15.14 15.69 15.33 Positive 16.29 15.25 16.21 16.27 F(222) = 4.88* F(218) = 0.15 F(443) = 2.54 F(515) = 8.24** Sex Risk Index: 0 15.21 14.74 15.35 14.89 1 15.98 16.27 16.16 15.82 2 16.16 15.58 16.20 15.90 3 15.58 16.00 16.17 15.00 F(222) = 5.39*** F(218) = 9.41*** F(443) = 9.69*** F(515) = 13.69*** Arrest History: 0 15.31 15.11 15.63 15.18 1 15.68 15.05 15.73 15.44 2 15.69 15.14 16.00 15.09 3 15.78 15.67 15.93 15.27 4 16.00 15.38 15.24 15.68 5 16.38 14.67 15.88 15.16 6 15.00 14.83 16.50 16.00 7 or more 16.545 15.36 16.00 15.75 F(222) = 1.94 F(218) = 0.32 F (443) = 1.45 F(515) = 2.18* Significance Levels: p < .05; ** p < .01; *** p < .001. Some interesting associations emerge d. For all four groups, a significant relationship between age and marijuana use, as well as age and risky sexual behavior, was revealed. However, a significant associ ation between age and cocaine use was found

PAGE 159

149 for African American males and Non African American females. On average, cocaine positive youths in these two groups were signifi cantly older than cocaine negative youths. Last, the only group to reveal a significant a ssociation between age a nd arrest history was African American males. African American males with a higher number of prior arrests tended to be older than Afri can American males with a lo w number of prior arrests. These bivariate analyses underscore the importa nce of considering the effect that age has on problem behavior, as well as highlight marked variation in the effect that age has on these behaviors across the demographic s ubgroups. Thus, these findings support the inclusion of the direct effect of age on the final CFA model reported in Table 15. The results of the final group-based SEM are presented in Table 17. As can be seen, the model fit the data. Both CFI and TLI were greater than the cut-off level of 0.90 (CFI =0.986, TLI = 0.970). The RMSEA (.034) was acceptable; WRMR (1.134) was somewhat high.12 The chi-square test of model fit ( = 21.68, p = 0.20) was nonsignificant at the .05 level. The factor loadings for each of the obs erved variables were positive and highly significant, except for the free factor loadi ng of arrest history for African American males, which remained significant, although not as strong ( bstdYX = 0.329). This finding suggests that, compared to the other three groups, the relationship between number of prior arrests and the latent f actor is not as strong for newl y arrested African American males. Importantly, all three sexual risk fr eed thresholds for the Non African American females revealed nonsignificant effects, whereas all three sexual risk thresholds for the 12 Muthn & Muthn (2002) indicate that WRMR is highly sensitive to sample size and can produce unreliable estimates. They suggest that, if all other fit indices are satisfactory based on the recommendation by Hu & Bentler (1999), then the fit of the model is appropriate.

PAGE 160

150 other three groups were highly significant ( p < .001). This finding indicates that the relationship between the response categories for the sexual risk index for Non African American females is different than the ot her three groups. The nu mber of risky sexual behaviors reported did not influence the strengt h of the relationship between risky sexual behavior and the latent factor for newly arrested Non African American females. The effect of age on the latent factor wa s significant in all four groups. As age increased, so did the covariation among the risky sexual behavior index, marijuana and cocaine test result, and arrest history for all four of the demographic subgroups. However, as can be seen in Table 16, the effect of ag e on the latent factor was somewhat stronger for the Non African American males, compared to the other groups. These findings are important because it suggests that early inte rvention efforts targeted at problem behavior syndrome may be more appropriate for certain demographic subgroups and either less or ineffective for other demographic subgroups. In terestingly, the signifi cant differences in the factor means found in Table 15 disappeared in the group-based SEM. Thus, the introduction of age into the model accounted fo r differences in the average level of the latent factor across the four groups. This finding provides additional support for the importance of accounting for age when examining the latent factor. The residual variances repor ted in Table 17 also reveal some important findings. For Non African American males, there is a si gnificant amount of va riance in the latent factor that is not accounted for by the risky sexual behavior index, marijuana and cocaine test result, and arrest history. This suggest s that additional behaviors are important to fully understanding problem behavior s yndrome for this subgroup. Conversely, the

PAGE 161

151 residual variance is nonsignifican t for the other three groups, which means that, for these three groups, a large amount of the variance in the latent factor is explained by the observed variables. Across the four groups, the amount of variation in each of the four observed variables that is explained by the latent fact or ranged from a low 9% (arrest history for African American females to a high 75% (coc aine test result for Non African American females). These values are similar to the re sults of prior studies involving juvenile offenders that reveal a wide range of expl ained variance across the observed variables (Dembo et al., 1992; LeBlanc & Girard, 1997). One important finding, however, is the difference in the amount of explained varia tion across the four groups. For example, for both female groups, a substantia l portion of the va riation in cocaine use was explained by the latent factor (70-75%). However, a so mewhat smaller portion of cocaine use was explained for the male groups (50-51% fo r both male groups). For both African American groups, the latent fact or explained over 55% of the variation in marijuana use, however, only 24% and 33% of the variation in marijuana use was explained by the latent factor for Non African American females and males, respectively. Freeing the sexual risk index thresholds for the Non African American females and the arrest history factor loading for Af rican American males also provides support for differences in the structure of the late nt factor across the demographic subgroups. Although arrest history remained significant fo r all four of the gr oups, the size of the coefficient for the African American males was somewhat smaller, indicating that the association among the latent fa ctor and arrest hi story was not as strong for African

PAGE 162

152 American males, compared to the other th ree groups. Due to th e substantially higher proportion of African American males with a larg e number of prior arrest s, this finding is not surprising. Once the sexual risk threshol ds for the Non African American females were freed, they all became nonsignificant. However, all three of the sexual risk thresholds for the other three groups were significant ( p < .01). Accordingly, the number of risky sexual practices that a Non African American female engages in is not related to the latent factor. Taken as a whole, these findings indicat e that, although the overall fit of the model is similar across groups, the associa tion between the latent factor and each observed variable is somewhat different ac ross the groups. The next, and final, chapter discusses the implications of these results, as well as the limitations of the current study and directions for future research.

PAGE 163

153 Table 17. Group-based Structural Equation Model Non AA Females AA Females Non AA Males AA Males Estimate S.E. CR Estimate S.E. CR Estimate S.E. CR Estimate S.E. CR F1 by : Sex Risk 1.000 --1.000 --1.000 --1.000 --Marijuana 0.942** 0.239 3.941 0.942** 0.239 3.941 0.942** 0.239 3.9410.942** 0.239 3.941 Cocaine 1.166** 0.308 3.781 1.166** 0.308 3.781 1.166** 0.308 3.7811.166** 0.308 3.781 Arrest History 0.685** 0.180 3.796 0.685** 0.180 3.796 0.685** 0.180 3.796 0.698* 0.226 2.622 Mean F1 -3.602 2.532 -1.423 0.484 1.446 0.334 0.000 --0.578 1.191 0.485 Residual Variance F1 0.608 0.429 1.417 0.182 0.144 1.261 0.345** 0.118 2.921 0.119 0.085 1.404 Thresholds: Marijuana1 2.405** 0.801 3.033 2.405** 0.801 3.033 2.405** 0.801 3.0332.405** 0.801 3.033 Cocaine1 3.081** 1.159 2.658 3.081** 1.159 2.658 3.081** 1.159 2.6583.081** 1.159 2.658 Sex Risk1 1.246 1.747 0.713 3.868** 0.848 4.564 3.868** 0.848 4.5643.868** 0.848 4.564 Sex Risk2 2.305 1.791 1.287 4.715** 0.922 5.115 4.715** 0.922 5.1154.715** 0.922 5.115 Sex Risk3 3.163 1.896 1.668 5.466** 1.025 5.334 5.466** 1.025 5.3345.466** 1.025 5.334 Arrest History1 0.833 0.751 1.109 0.833 0.751 1.109 0.833 0.751 1.109 0.833 0.751 1.109 Arrest History2 1.268 0.745 1.703 1.268 0.745 1.703 1.268 0.745 1.703 1.268 0.745 1.703 Arrest History3 1.605* 0.748 2.146 1.605* 0.748 2.146 1.605* 0.748 2.146 1.605* 0.748 2.146 Arrest History4 1.842* 0.754 2.443 1.842* 0.754 2.443 1.842* 0.754 2.443 1.842* 0.754 2.443 Arrest History5 2.028* 0.764 2.655 2.028* 0.764 2.655 2.028* 0.764 2.655 2.028* 0.764 2.655 Arrest History6 2.202** 0.774 2.847 2.202** 0.774 2.847 2.202** 0.774 2.8472.202** 0.774 2.847 Arrest History7 2.328** 0.782 2.978 2.328** 0.782 2.978 2.328** 0.782 2.9782.328** 0.782 2.978 F1 on: Age 0.284* 0.117 2.426 0.122* 0.055 2.209 0.160** 0.048 3.375 0.133* 0.053 2.523 R-Square : Sex Risk 0.404 0.446 0.377 0.239 MJ 0.238 0.581 0.336 0.549 Coc 0.702 0.747 0.504 0.511 Arrest History 0.329 0.087 0.181 0.108 Model Fit Statistics: = 21.68, p = 0.20; CFI = 0.986; TLI = 0.970; RMSEA = 0.034; WRMR = 1.134.

PAGE 164

154 Chapter 9 Discussion and Conclusion A wealth of previous resear ch has indicated that engagi ng in a number of different problem behaviors forms a unidimensional cons truct commonly referred to as problem behavior syndrome (Dembo et al., 1992; Jesso r & Jessor, 1979; LeBlanc & Bouthillier, 2003; Welte et al., 2004). However, three limitations of this body of research have inhibited our understandi ng of the nature of this concep t across different populations. These limitations include the reliance on commun ity or incarcerated samples, a lack of information regarding variation in the st rength of the association among problem behaviors across important indi vidual-level factors, and the use of standard factor analytic techniques. This study sought to overcome these limita tions by assessing the covariation among risky sexual practices, mariju ana and cocaine use, and criminal history among newly arrested juvenile offenders, as well as examining differences in the covariation among these behaviors across de mographic subgroups. Specifically, three main objectives guided this study: 1) to determine whether risky sexual behavior, marijuana and cocaine test result, and crim inal history form a unidimensional latent construct reflective of problem behavior syndrome among a sample of newly arrested juvenile offenders, 2) to examine the direct e ffect of age on the latent factor, and 3) to identify differences in the factor structure, as well as the effect of age on the latent factor,

PAGE 165

155 across four demographic subgr oups (African American fema les, Non African American females, African American males, and Non African American males). The findings of this study provide suppor t for all three object ives. Confirmatory factor analysis, using the full sample of ne wly arrested adolescents, provided moderate support for the first research question. The model fit indices indicated that risky sexual behavior, marijuana and cocaine use, and cr iminal history do form a unidimensional latent factor with significant item-factor relations. However, referring to the standardized loadings, as well as the r-square values, it is clear that the latent factor was a better predictor of substance use, compared to ris ky sexual practices and arrest history. This finding suggests that, although the latent factor represents a unidimensional construct, the interrelationships involved in the di mensionality are somewhat different. The structural equation model introducing the direct effect of age onto the latent factor revealed a positive and significant eff ect of age. This finding suggests that, across the entire sample of newly arrested adol escents, the relationship between age and the latent factor was str onger for older adolescents involved in the study. Hence, the second research objective was also supported. Then, a number of group-based modeling techniques were performed to assess the variation in the structure of the latent factor across the four demographic subgroups. A fully constrained model did not fit the data well This means that the latent factor was not completely invariant across the four groups Rather, the best fitting model involved partial invariance, indicating that although the factor stru cture was somewhat similar across the four groups, there were also im portant differences across the groups.

PAGE 166

156 One important difference found in this st udy was in regard to the relationship between arrest history and the latent factor for African Am erican males. Freeing the arrest history factor loading for this s ubgroup improved the fit of the model, which signifies differences in the relationship between arrest history and the la tent factor for this group, compared to the other three groups. A lthough the arrest hist ory factor loading remained significant, the value of the critical ratio, as well as the standardized estimate, suggested that the association between arrest history and th e latent factor was not as strong for African American males. Refe rring back to Table 13, it is clear that a disproportionate number of Af rican American males had a high number of arrests. Because getting arrested is somewhat more common among this group, it is possible that this behavior is not as strong of an indicator of additional behavioral problems, compared to the other three groups. In addition, as can be seen in Table 15, th e latent factor only accounted for 11% of the variation in arrest history for this age group. Thus, for African American males, number of prior arrests ma y not be a critical component of problem behavior syndrome as it is for the other three groups. Another important difference found in this study was related to the relationship between risky sexual practices and the latent factor. Resu lts indicated that freeing all three sexual risk thresholds for the Non Af rican American females further improved the fit of the model to the data. Once the thresholds were freed for this group, they became nonsignificant, yet the thresholds for the other three groups re mained significant ( p < .001). It is important to note that, although the thresholds were nonsignificant, the sexual risk factor loading remained significant. This means that, the overall strength of the association between risky sexua l practices and the latent factor was similar across the

PAGE 167

157 four groups; however, for Non African American females, the specific number of risky sexual practices reported did not influence th is association. Thus, for the Non African American females included in this sample distinguishing between those who reported one versus three risky sexual practices did not influence the strength of the relationship between the sexual behavior index and the la tent factor. Altern atively, the number of risky sexual practices reported by participants in the other three groups did influence this relationship. This study sought to explore the issue of relatively low le vels of explained variance reported in previous st udies. In particular, it was argued that the low levels of explained variation could be a result of the failure to account for group differences in the associations among the observed variables and the latent factor. The findings of this study provide a measure of support for this ar gument. Similar to a number of previous studies examining problem behavior syndr ome (see LeBlanc & Bouthillier, 2003 for a review), the overall SEM model using the entire sample (Table 10) re vealed a significant factor variance. But, the re sults of the group-based SEM hi ghlighted important variation in the ability of the observed variables to explain the late nt factor which supports the need to examine the factor across subgroups ne sted within a sample Specifically, non African American males were the only group for which there was a significant amount of variation that was not accounted for by the four observed variab les. That is, risky sexual practices, marijuana and cocaine test result, and arrest history e xplained a substantial proportion of the variation in the latent factor for African American females, African American males, and Non African American fe males. Yet, it seems that for Non African American males, additional problem behavior s are important in expl aining variation in

PAGE 168

158 problem behavior syndrome. Thus, in or der to fully understand problem behavior syndrome among this demographic subgroup, future research is needed to identify which additional problem behaviors are impor tant components of the syndrome. Subgroup differences in the amount of vari ation in each of the problem behaviors accounted for by the latent factor were also re vealed. For instance, the r-square values found in the current study ranged from 0.24 to 0.45 for risky sexual practices, 0.24 to 0.58 for marijuana test result, 0.50 to 0.75 for cocaine test result, and 0.09 to 0.33 for arrest history. Across the groups, the latent factor explained a greater proportion of variation in cocaine use and risky sexual practices for females, compared to males. However, compared to the Non African Amer ican subgroups, the latent factor accounted for a greater proportion of vari ation in marijuana use for bot h African American groups. Furthermore, the latent factor explained the largest amount of variation in both substance use measures for the African American female s; however this group displayed the lowest levels of explained variation in arrest history. These diffe rences suggest that examining levels of explained variance based on entire samples has the poten tial to significantly inhibit a true representation of the latent factor-observed va riable relationship that is specific to each particular subgroup. Overall, the differe nces revealed in this study provide moderate support for the third re search question. As a result, several implications can be drawn from these results. Implications of the Results Indeed, the findings of this study coincide with the large body of research that suggests that engaging in multiple problem behaviors forms a unidimensional construct reflective of problem behavior syndrome (for a review, see LeBlanc & Bouthillier, 2003).

PAGE 169

159 Taken together, this body of research warra nts the attention of current prevention and intervention strategies that continue to target a single problem behavior. Presently, intervention strategies that focus on reducing a single proble m behavior are the norm in most juvenile justice systems across the country (OJJDP, 2008). It is crucial that these programs involve a comprehensive strategy th at targets all three behaviors in an integrated fashion. Not only should these prog rams begin to target multiple behaviors, but more importantly, they should focus on th e ways in which one problem behavior is related to another. Furthermore, the strong the relationshi p between the latent factor and the substance use measures found in the overall model suggests that intervention programs that target adolescent substan ce use may be particularly e ffective in reducing additional problem behaviors. For example, one co mponent of many substance abuse prevention programs is to educate youth on the negative co nsequences of using substances. It is widely acknowledged that adolescent offenders are more likely to engage in risky sexual practices while using substances and s ubstance-using adolescent offenders are disproportionately more likely to test STD positive (DiClemente et al., 1991; Teplin et al., 2005; Tolou-Shams et al., 2007). Ther efore, substance use prevention programs represent an effective avenue to provid e at-risk youth with lessons on sexual responsibility, as well as to target the association between risky sexual behavior and substance use. On one hand, the invariance of the mode l parameters for marijuana and cocaine test results highlight important similarities in the treatment needs of newly arrested juvenile offenders. For all four demographi c subgroups, the strength of the association

PAGE 170

160 between substance use and the late nt factor was very similar. Thus, these results indicate that identifying substance-using offenders who most likely manife st problem behavior syndrome based on racial and gender characteris tics may not prove to be effective. On the other hand, the weaker (but still significant) relationship between arrest history and the latent factor for the Af rican American males suggests that delinquent behavior may not be as much of an indicator of problem behavior syndrome for this group compared to the other three groups. Two pl ausible explanations for this finding exist. First, the disproportionate number of study participants who had a high number of arrests suggests that juvenile delinquency is more severe among this subgroup. Therefore, serious criminal involvement may not represent the same risk for engaging in additional problem behaviors. That is, if engaging in delinquent behavior is consid ered “less deviant” behavior for this group of adolescents, then it may be that the association between juvenile delinquency and additi onal behaviors that are consid ered “problemed” is not as strong, in turn, reducing the a ssociation between arrest hi story and problem behavior syndrome. The second possible explanation is related to the reliance on officially recorded juvenile delinquency. It is widely ar gued that racial bias exists in the arrest procedures of juvenile offenders (Lieber, 2003; OJJDP, 1999). Thus because of the use of officially record delinquency, delinquency measured in this study may not necessarily be “more normative” behavior for this subgroup, but rather an indicator of bias in arrest procedures. Unfortunately, this study was unable to examine self-reported juvenile delinquency. Future research is needed to tease out the differences in the association between delinquent behavior and proble m behavior syndrome based on different

PAGE 171

161 measurement strategies, as well as to determ ine if the findings of this study can be replicated using self -reported delinquency. The nonsignificant sexual risk thresholds for the Non African American females also provide suggestions for the identifica tion of treatment needs. Whether a Non African American female reported one or three sexual risk-taking behaviors did not influence the relationship between the risky sexu al behavior index and the latent factor. For this subgroup, what is important is whether or not she engaged in th e behavior at all. Risk of problem behavior syndrome was simila r across the extent to which a Non African American female was involved in risky sexual practices. Based on th ese results, level of risky sexual behavior may not be an effec tive tool for identifying Non African American females most likely to display problem beha vior syndrome. However, the extent of engagement in risky sexual practices seems to be an appropriate tool for identifying youth who manifest problem behavior syndrom e for the other three groups. For Non African American females, initiation of ri sk-taking sexual behavior may be a more appropriate indicator of the syndrome. The concept of relative deviance may at least partially explain the subgroup differences found in this study (Dembo & Shern, 1982; Kaufman, 1978). According to this view, persons who are more deviant fr om the norms of their social and cultural setting tend to exhibit more serious behavior problems. As menti oned earlier, because being arrested was more common among the African American males in this study, it is likely that this behavior is considered more normative to the cultural and social surrounding of these adolescents, in turn, reduc ing the strength of its relationship with more serious problem behaviors. At the sa me time, relative deviance could also account

PAGE 172

162 for the differences in the association between risky sexual practices a nd the latent factor. There are important gender differences in the cultural expectations of sexual behavior during adolescence. It is substantially mo re common for young men to talk or boast about their sexual encounters, and less likely that they will be judged, ridiculed, or scolded for engaging in these behaviors, co mpared to their female counterparts. Furthermore, it is well established that Afri can American female adolescents are more likely to experience teenage pregnancy, as well as test STD positive (CDC, 2008), compared to the other three groups. Thus, engaging in risky sexual practices for Non African American females may be somewhat “m ore deviant” compared to the other three groups. This would explain the finding that whether or not a Non African American female engaged in risky sexual practices, not the level of engagement, was related to problem behavior syndrome. For this subgroup of study pa rticipants, even minimal involvement in risky sexual practices is a sign of more serious problem behaviors. Finally, the significant effect of age on the latent factor across the four groups, as well as the disappearance of the significant factor mean differences (found in Table 15) once age was included in the model, undersco res the importance of early intervention for at-risk adolescents. The strongest effect of age on the latent factor was found among the African American males. For example, the sta ndardized estimate for the effect of age on the latent factor ranged from 0.36 to 0.44 for the other three groups ( p < .05), however the standardized estimate for th e effect of age on the latent factor was nearly 0.60 for the African American males ( p < .01). Thus, it appears that early prevention strategies may be particularly effective for this subgroup of adolescents. Ba sed on a review of effective intervention strategies, Farri ngton and Welsh (2006) recomme nd that early intervention

PAGE 173

163 should 1) begin in the early years of a child 's life, by early adol escence, 2) intervene before the onset of misbehavior, and 3) target children who are identified as being at-risk for behavioral problems. The findings of this study support these recommendations, as well as represent a preliminary step in dete rmining which subgroups of adolescents (i.e., African American males) would benefit the most from early intervention efforts. In summary, important subgroup differenc es in the intervention needs of adolescent offenders were revealed in th is study. A wealth of empirical evidence suggests that intervention programs that are specifically tailored and delivered to a particular subgroup of adolescents are the most successful in decreasing high-risk behaviors (e.g., Non African American female s) (DiClemente et al., 2004; Jemmott, Jemmott, & Fong, 1998; Orr, Langefeld, Katz, & Caine, 1996; St. Lawrence, Brasfield, Jefferson, Alleyne, & O’Bannon, 1995). The reas on for the effectiveness of tailored intervention programs stems from the acknow ledgement that “adolescents are not a homogeneous population; rather, adolescents are a he terogeneous mosaic of subgroups of different ethnicities/cultures, behavioral risk characteristics developmental levels, sexual preferences, and gender differences” (DiC lemente et al., 2008:600). Thus, adapting prevention and intervention strate gies to meet the developmenta l and social needs of each particular subgroup of adolescen ts at risk for problem behavior syndrome is the most effective intervention strategy. Accordi ngly, if the subgroup differences in the intervention needs of newly arrested adolesce nt offenders implied by the findings in this study are replicated in other jurisdictions these differences cannot be ignored. In addition, the findings of this study also carry important theoretical implications. A number of well-known crimi nological theories claim to be “general”

PAGE 174

164 theories because they are able to explain a wide range of deviant behaviors including risky sexual practices, substa nce use, and delinquency. For example, Gottfredson and Hirschi (1990) argue that low self-control is the cause of de linquent, as well as analogous, behaviors. They maintain that diffe rences in these behaviors across individual characteristics (e.g., race or gender) are the product of differences in low self-control (Nakhaie, Silverman, & LaGrange, 1999). A nother example is social learning theory. Akers (1985) argues that an adol escent learns to engage in all forms of deviant behavior in the course of interactions with sign ificant others, through a process of imitation, reinforcement, and exposure to definitions that support deviant beha vior. According to this perspective, any differences in devian t behavior (e.g., racial or gender differences) are the result of differences in the direction of the le arning process. The strong association found among these behaviors supports the notion that a common set of risk factors may be causing al l three forms of probl em behavior; hence, providing preliminary support for a “general” cau se of several forms of deviant behavior. Unfortunately, data limitations prohibited the theoretical examination of the cause(s) of problem behavior syndrome in the current st udy. Future research is needed to examine the ability of these general theories to predict the latent factor, as well as determine if the relationship between the theoretical constructs and the latent factor is consistent across the demographic subgroups. If, in fact, one or more of these general criminological theories is able to predict problem beha vior syndrome across the four demographic subgroups then the claim of being a “general” theory will be upheld.

PAGE 175

165 Limitations The use of already collected data limited the measurement of key variables used in the current study. For example, asking partic ipants about previous delinquent behavior was beyond the goals of the original rese arch project. Alternatively, this study operationalized delinquent behavior with officially recorded arrests. As a result, the measure of delinquent behavior used in this study only captures behaviors for which the participant was actua lly caught and apprehended. It is well established that self-reported delinquency is much higher than officially re corded delinquency (E lliott & Ageton, 1980; Farrington, 1985; Farrington, Jo llife, & Hawkins, 2003; Hindelang, Hirschi, & Weis, 1981). In addition, it is commonly argued that officially recorded delinquency is influenced by the bias that exists in the j uvenile justice system (Lieber, 2003). Based on these shortcomings, the validity of relying on officially record ed arrests, as a measure of juvenile delinquency, is questionable. On th e other hand, relying on officially recorded delinquency guards against problems associated with self-report measures including differences in recall, willingness to re port, and/or participation levels. In addition, although using bi ological data measures of substance use overcomes issues related to inaccurate self-reported information (not ed above), it also has its shortcomings. For example, the short time pe riod for which drug use is detectable in urine is one important limitation to relying on drug test results as a measure of substance use. For heavy users, marijuana only stays in a youth’s system for approximately 20 days and cocaine remains in the system for le ss than four days (Dembo et al., 1999). Therefore, the drug tests were only able to capture recent or current drug use, which limited the number of drug users in the sample Relying on self-reported data would have

PAGE 176

166 allowed for an extension in the observation time frame for drug use (e.g., past year use), which in turn, could have increased the num ber of drug users included in the sample. On the other hand, the risky sexual behavi or measures used in this study were based on self-reports. Obtaining valid informa tion regarding these items is particularly daunting because the empirica l investigation of such a personal and often nonpublic, nondisclosed behavior is logis tically complicated (DiCleme nte et al., 2008). Because engaging in sexual behavior, pa rticularly unsafe sexual behavi or, is considered a private matter, the participants involved in this study may not have felt comfortable fully disclosing their sexual history to intake screeners whom they have never met before. Therefore, underreportin g of this behavior at the asse ssment center is quite likely. Another issue related to relying on self -reported sexual practices is the potential for subgroups differences in the willingness to report their behavior. Research suggests that African Americans and females are less likely to be honest about their behavior (Rosay, Najaka, & Herz, 2007). Furthermore, as stated above, female adolescents are subjected to a greater leve l of ridicule and judgment for engaging in risky sexual practices, which may have led the females in this study to underre port their sexual history. Such differential tendency in the wil lingness to report behavior may also have influenced the findings. Unfortunately, asking ad olescents about their sexual practices is the only plausible means of obtaining this in formation. Given that the items used to measure risky sexual practices used in this st udy are similar to measures used in previous studies on adolescent offenders (e.g., Teplin et al., 2003), it is likely that these problems are consistent across studies examining ris ky sexual behavior. Also, the dichotomous nature of the sexual risk items limited our abi lity to examine the severity of involvement

PAGE 177

167 in sexual-risk taking. Future research s hould attempt to examin e problem behavior syndrome by including improved measures for this behavior. The failure to include ethnicity as an additional demographic factor was also a limitation to the current study. As a result of the data collection procedures, as well as HIPPA safeguards prohibiting the research team to see any confidential information, the participant’s ethnicity was una ble to be collected in a va lid manner. For example, descriptive analysis indicated that less than 10% of the final sample was Hispanic. Given that over 26% of the population between the ages of 10 and 17 in Hillsborough County is Hispanic, the proportion of study participants identified as Hispanic did not seem representative (US Census Bureau, 2008). Research does indicate that Hispanic a dolescents represent a high-risk group for risky sexual practices, subs tance use, and delinquent be havior (Crepaz et al., 2007; Herbst et al., 2007; Loue, 2006; Ramisetty-Mikler et al., 2004). Therefore, in addition to examining racial and gender differences in problem behavior syndrome, it is also important to identify ethnic differences. It has been suggested that Hispanic adolescents experience a different socialization process, due to ethnic differences in cultural values and expectations, family rela tions, and socio-economic fact ors, compared to white and African American adolescents (Marin, 1989; Pan & Farrell, 2006; McLoyd, Cauce, Takeuchi, & Wilson, 2004; Saint-Jean & Cranda ll, 2004). These differences are likely to lead to differences in the development and structure of problem behavior syndrome. Thus, future research would benefit from further disaggregating the demographic subgroups examined in this study by race, gender, and ethnicity.

PAGE 178

168 Another important limitation to the current study is the incons istency in the time frame included in the measurement of the observed behaviors. For example, officially recorded delinquency encompasses delinquent behavior across the youths’ lifespan. Similarly, the risky sexual practice items aske d respondents if they had “ever” engaged in the behavior, again, referring to lifetime beha vior. Yet, relying on drug test results to measure substance use only includes recent or current use. Therefore, it is uncertain whether or not the youths i nvolved in this study were simultaneously engaging in these behaviors. Furthermore, because the arre st history and risky sexual practice items measure lifetime participation, th e meaningfulness of the sign ificant effect of age on the latent factor is also questi onable. It was determined that older study participants were engaging in the four problem behaviors at a substantially higher rate than younger study participants. On one hand, this signifies an im portant direct effect of age. On the other hand, however, it is also possible that the ol der adolescents had more time to engage in such behaviors. This possibility is impor tant to understanding th e relationship between problem behavior syndrome and age. Future research is needed to tease out these possibilities and to understand the relative effect of age on the latent factor across different developmental time periods. Moreover, the data were collected at one site, which limits the generalizability of these results. There is a need to determin e if the findings obtained in this study can be replicated in centrali zed intake centers in other locatio ns, serving different populations of juvenile arrestees. Last, the data were cr oss-sectional which prohibited the examination of the temporal sequencing of involvement in each of the problem behaviors. Thus, the analyses in this study focused only on the st rength and direction of the association among

PAGE 179

169 the problem behaviors; no causal statements about any of the relationships can be made (Cook & Campbell, 1979). Despite these lim itations, this study contributes to the existing body of literature on adolescent problem behaviors in two important ways. Contributions to the Literature This study adds to the existing resear ch on problem behavior syndrome because the findings are based on newly arrested juvenile offenders. Prior research has focused on either general adolescent samp les or incarcerated samples, which represents two ends of the “juvenile offender continuum.” Thus, the generalizability of the findings of prior studies to the entire juvenile offending population is not possibl e. By supporting the first research objective (i.e., unidimensional latent factor), using a sample of newly arrested offenders that involved first time offenders to more serious, chronic offenders, this study strengthens the evidence in support of pr oblem behavior syndro me among adolescent offenders as a whole. More importantly, this study also pr ovides preliminary information on the similarities and differences in the interv ention needs of newly arrested adolescent offenders who are at the front-end of the juven ile justice system. B ecause intake centers represent the front-end of the juvenile justice system, all or most juvenile offenders experience the intake phase of the juvenile justice system process. Intervention efforts targeted at problem behavior syndrome at this phase of the juvenile justice process may prove to be more effective than services at the back-end of the sy stem where only a small portion of adolescents, who are typically the most serious offenders, end up. A wealth of research has indicated that seri ous juvenile offenders are the most resistant to intervention strategies (Welsh, 2005). Therefore, intake screening centers provide a great avenue for

PAGE 180

170 the screening and assessment of a large, di verse number of youths who may be more responsive to intervention strategies targ eted at problem behavior syndrome. Also, a major limitation of previous probl em behavior research concerned the use of standard factor analytic techniques that examine the study sample as a whole. Relying on such general analytic strate gies inhibited our understandin g of the structure of problem behavior syndrome across important individual level characteristics. Such group-specific information is critical to obtaining a comple te understanding of the syndrome. This study overcame this limitation by using group-based fa ctor analytic techni ques that identified similarities and differences in the structure of problem behavior syndrome across four demographic subgroups. Given the consistent individual-level differences in problem behaviors revealed in previous studies, in addition to the differences found in this study, group-based modeling seems to be a more appropriate technique for obtaining an accurate understanding of problem behavior syndrome. Widaman and Reise (1997:316) point out that, when there is evidence to s uggest that groups within the population exist for whom the indicator variables are differen tially related to the latent variable, groupbased CFA and/or SEM techniqu es provide “ an adaptable an d powerful set of tools for investigating similarities and differences across groups in measurement structures.” Accordingly, it is my hope that future rese archers will begin to utilize this group-based approach when studying problem behavior synd rome. In addition to the continued use of group-based modeling, this study also lays the groundwork for a number of additional areas for future research on group differe nces in problem behavior syndrome.

PAGE 181

171 Directions for Future Research First, it is important to include additiona l problem behaviors in an assessment of problem behavior syndrome. Although an accep table (i.e., nonsignificant) amount of the variation in the latent factor was explained by risky sexual practices, substance use, and arrest history in three of the four groups still 20% or more was unexplained (depending on the group). Thus, future research should strive to identify wh ich additional problem behaviors are related to these three risk-t aking behaviors. Alcohol use, cigarette smoking, skipping school, gambling, reckless driv ing, bullying, and/or suicidal behavior are some examples of behaviors that have been found to be components of problem behavior syndrome (Gottfredson, 2001; Junger et al., 1995; LeBlanc & Bouthillier, 2003; Welte et al., 2004; White, 1992). Furthermore, it is quite possible th at different problem behaviors may be important to different subgr oups of adolescent offenders. Therefore, the inclusion of these behaviors may improve th e ability to explain th e latent factor, as well as provide further insight into group diffe rences of problem behavior syndrome. In addition to considering the relative importance of including additional problem behaviors, future research should also strive to validate the unidimensionality of problem behavior syndrome by comparing the f it of the model th at hypothesizes unidimensionality to alternative models th at hypothesize two or more dimensions. Previous studies, although few in number, have found a two-fact or (Hemphill et al., 2007; White & Labouvie, 1994) or three-factor solution (Gillmore et al., 1992) fits the data better than a one fact or solution when examining problem behaviors among the general adolescent population. Given the small number of problem behaviors included in this study, the examination of a multiple-fact or model was not prac tical. The findings of

PAGE 182

172 this study do support a unidimensional constr uct. However, the stronger relationship found between the latent factor and both s ubstance use measures (i.e., standardized loadings and r-square values), compared to th e relationship between th e latent factor and risky sexual practices and the latent factor and criminal i nvolvement, yield a measure of skepticism regarding the validity of unidimens ionality. Thus, not onl y is it important to identify additional problem behaviors that cont ribute to problem behavior syndrome, it is also important to validate the unidimensionality of the construct using a greater number of problem behaviors. Compared to pr evious studies on community samples of adolescents, the number of behaviors used in this study was relatively low. On average, about five to eight problem behaviors have been included in the latent factor model (LeBlanc & Bouthillier, 2003). Regardless, the findings of this study supplement the findings of previous studies reveal ing a unidimensional construct. Also, contemporary criminologi cal research has provided evidence of a variety of offender types. This body of research invol ves the longitudinal examination of offending levels (Fergusson, Horwood, Nagin, 2000; Mo ffitt, 1993; Nagin & Land, 1993; Patterson & Yoerger, 1993; Piquero, Brezina, & Turner 2005; Sampson & Laub, 2003) as well as the identification of subgroups based on t ype of offending (Brame, Mulvey, & Piquero, 2001; Farrington et al., 1988; Francis, Soothi ll, & Fligelstone, 2004; Mazerolle, Brame, Paternoster, Piquero, & Dean, 2000; Piquero, Paternoster, Mazerolle, Brame, & Dean, 1999). More recently, additional studies have identified different subgroups of individuals based on a variety of deviant be haviors including crimin al behavior and drug use (Dembo & Schmeidler, 2002; Massoglia, 20 06). The bulk of these studies suggest that the offending population is characteri zed by distinctive s ubgroups based on the

PAGE 183

173 frequency and/or continuity of engaging in these behaviors. Studies that have supported the notion of distinct subtypes of offenders indicate that anywhere from two (Moffit, 1993; Roeder, Lynch, & Nagin, 1999) to se ven (Bushway, Thornberry, & Krohn, 2003) subgroups characterize the criminal population. Drawing on this work, it seems plausibl e that different subgroups of juvenile offenders exist based on their involvement in a variety of problem behaviors. On one hand, the co-occurrence of problem behaviors is extremely prevalent among juvenile offenders, but on the other hand, it is also clear that not all juvenile offenders engage in risky sexual behavior and/or substance use. Accordingly, th e possibility of the existence of different subgroups based on their involve ment in these behaviors seems warranted. For example, it may be that a portion of juve nile offenders is frequently engaging in delinquent behavior, risky sexua l behavior, and serious drug us e. However, it is also possible that some juvenile offenders are engaging in risky se xual behavior, but not substance use, and vice versa. Thus, it is pos sible that different t ypes and/or levels of problem behavior syndrome exist among adolescent offenders. Latent class analysis provides the opport unity to examine whether these variations in problem behavior syndrome exist. Briefl y, latent class analysis distinguishes between different patterns of covariation found in a single dataset and identifies subgroups based on these different patterns (McCutcheon, 1987). Thus, this method of analysis accounts for the variation in the strength of the asso ciations found among the different forms of deviant behaviors and identifies different “classes” based on this variation. The important distinction between fact or analytic techniques and late nt classes analyses lies in the scaling of the latent variable (Muthn, 2006). Factor analysis relies on continuous

PAGE 184

174 latent variables and is used to examine unde rlying dimensions by explaining patterns of association among observed indicators. Late nt class analysis, on the other hand, uses categorical latent variables to identify homogenous groups of individuals. The identification of subgroups of offenders base d on different pattern s of association among the problem behaviors also has important intervention implications. By identifying “classes” based on different patt erns of covaration, we can then seek to identify important risk factors for membership into each group a nd create services de signed to target the varying interrelationships among the problem behaviors specific to each subgroup. The use of longitudinal data will also enhance our knowledge of problem behavior syndrome by allowing the examinati on of temporal ordering. Studies that include a diverse sample of elementary-age d children and follow them through early adulthood will be able to examine both the development and continuity of the syndrome. Identifying which problem behaviors tend to precede other problem behaviors will facilitate the identification of youths most at risk for the development of the syndrome. That is, at-risk youths in need of holistic intervention targeted at problem behavior syndrome could begin to be identified by engagement in these "problem-starting" behaviors. Furthermore, alt hough the findings of this study s uggest that the structure of the latent factor is relativel y similar across demographic su bgroups, this does not suggest that the developmental process of the latent construct is similar across groups. Prior research has revealed marked differences in the onset of risky sexual practices, substance use, and delinquent behavior across race and ge nder (Belenko et al., 2004; Kelley et al., 1997; Rosenthal et al., 1999; Santelli et al., 2000; Stevens et al., 2004). Thus, identifying differences in the temporal ordering of the development of problem behavior syndrome is

PAGE 185

175 also critical to developing the most effectiv e strategies for identif ying and treating at-risk adolescents. With the use of longitudinal data, the ne xt step in improving our understanding of group differences in problem behavior syndrome is to apply a socioecological approach to this concept. A socio-ecological approach involves examining individuals’ behavior within the context of their social and physical environment, inclusive of familial, peer, and community influences (DiClemente et al., 2008). The integration of these psychosocial influences reciprocally shape one another and collectivel y affect the balance of risk for problem behavior syndrome. Parents and peers are considered two of the most significant factors in determining the direction of a youth’s socializ ation process. Indeed, different kinds of family environment and/or peer groups pr ovide differing standards for behaviors (Wilson & Donnermeyer, 2006). Poor family relati ons, including low attachment, involvement, monitoring, and inconsistent or harsh discipline are consistent predictors of a range of problem behaviors, including sexual behavior, substance use, and criminal involvement (Chen & Thompson, 2007; Crosby et al., 2001; DiClemente et al., 2001; Gorman-Smith et al., 2000; Huebner & Howell, 2003; Ka pungu et al., 2006; Mosack et al., 2007; Robertson et al., 2005). Developm ental research also reveals important racial and gender differences in parenting pract ices (Bulcroft, Carmody, & Bulcroft, 1996; Park & Bauer, 2002; Pinderhughes et al., 2000). This body of evidence suggests that African-American parents are less likely to supervise their child ren’s behavior and more likely to provide harsh, physical, and inconsistent discipline. In regard to ge nder differences, females tend

PAGE 186

176 to be supervised more closely, receive less physical discipline, and report stronger attachment to parents. At the same time, it is well established that peers have a strong influence on the development and continuity of problem beha vior. Peer groups provi de the context for learning problem behaviors through modeling be havior, the transmission of definitions and neutralizations for engaging in such behaviors, and positive and/or negative reinforcement for behavior. Thus, the ratio of positive peer factors to negative peer factors may influence the lik elihood of an adolescent mani festing problem behavior syndrome. A number of studies have found th is to be true for risky sexual practices (Robertson & Levin, 1999; Spitaln ick et al., 2007), substance use (Fergusson et al., 2002; Jang, 2002), and delinquent behavior (Warr, 2002) Studies also suggest that males are more likely to have peers that engage in problem behavior (Mears, Ploeger, & Warr, 1998; Simons, Miller, & Ageton, 1980) and the e ffect of deviant peers on behavior is substantially stronger for males compared to females (Johnson, 1979; Smith & Paternoster, 1987). Compared to white adol escents, African Ameri can adolescents report having a greater number of friends who engage in problem behaviors, however, the effect of having deviant peers on problem behavior is stronger for white adolescents (Brannock, Schandler, & Oncley, 1990; Matsueda & Heimer, 1987; Williams et al., 1999). Accordingly, identifying the types of parent ing and peer groups th at lead to problem behavior syndrome, as well as the similariti es and differences in these relationships across demographic subgroups, will substantially improve our understanding of the social influences related to problem behavior syndrom e. This information will aid in the quest

PAGE 187

177 for the most comprehensive prevention and intervention strategies tailored to the specific needs of different subg roups of offenders. The neighborhood in which an adolescent resides also has the potential to influence problem behavior syndrome. It is well established that ad olescents who reside in disadvantaged areas, characterized by poverty, residential in stability, population heterogeneity, and family disruption, are at a heightened risk for engaging in problem behavior, including delinquency, substance use, and risky sexual practices (Brooks-Gunn et al., 1993; Leventhal & Brooks -Gunn, 2000; Sampson et al., 2002). Specifically, it is suggested that these structur al factors influence the 1) costs and rewards assigned to engaging in deviant behavior, 2) presence of negative role models in the neighborhood, and 3) amount of unsupervised socializing in the neighborhood. The interaction of these three factors influences the direction of the socializati on process, and in turn, the tendency for an adolescent to engage in multiple problem behaviors. Furthermore, a number of previous studi es have found that the intersection of neighborhood characteristics a nd individual-level demographi c factors influences the tendency to engage in risky sexual practi ces, substance use, and deviant behavior (examined separately). Such findings reflec t important interactions between structural factors and demographic charac teristics (Brewster, 1994; Br owning et al., 2004; Cubin et al., 2005; Choi, Harachi, and Catalano, 2006; Frank et al., 2007; Kulis et al., 2007; Peeples & Loeber, 1994; Reardon, Brennan, and Buka, 2002; Turner et al., 2007; Wallace & Murroff, 2002). These studies, coupled with the prel iminary findings of this study, lay the groundwork for future research to focus on the variations in the association among

PAGE 188

178 neighborhood characteristics and proble m behavior syndrome across specific demographic categories. To date, studies th at examine the influence of the social and economic characteristics of the neighborhood on the tendency to simultaneously engage in multiple forms of deviant behavior, as well as the interaction between community characteristics and demographic factors are re latively nonexistent. The lack of research regarding these associations is unfortunate be cause this information could help local and state governments identify which communities are in need of services for at-risk youth, in addition to which types of services are most needed. Research also suggests that community conditions indirectly influence problem behavior through their impact on a youth’s immediate social environment (i.e., family and peers) (Chuang, Ennett, Bauman, & Foshee, 2005; Jang & Johnson, 2001; Li, Feigelman, & Stanton, 1999; Tarter, Vanyul ov, & Kirisci, 2006; Vazsonyi, TrejosCastillo, & Young, 2007). These studies suggest that the interaction of communityand individual-level factors de termines an adolescent’s developmental pathway to problem behavior (Ingoldsby & Shaw, 2002). Communities influence the ability of the family to provide opportunities to adolesce nts to be exposed to, and l earn about, deviant behavior. At the same time, community characteristics also influence the opportu nities to associate with different types of peer groups (W ilson & Donnermeyer, 2006). Therefore, the interaction between community, family, and peer factors may not only determine if a child will engage in one or more forms of problem behavior, but may also influence which behaviors the child will engage in depending on exposure, community perception, and reinforcement in a given neighborhood.

PAGE 189

179 In additional to social environmental f actors, psychological factors may also be important in explaining problem behavior syndrome. Indee d, a wealth of studies has revealed a consistent relationship between me ntal health and problem behaviors. For example, conduct disorder, depression a nd suicidal ideation, attention deficit hyperactivity disorder (ADHD), and posttraumatic stress disorder (PTSD) are consistently found to negatively influence the decision to engage in risky sexual behavior, substance use, and delinquent beha vior (Bardone et al., 1996; 1998; Brooks et al., 2002; Halfors et al., 2004; Dembo et al., 2007; Kaltiala-Heino, Losunen, Rimpela, 2003; Teplin et al., 2005; McClelland et al., 20 04; Waller et al., 2006). In addition, racial and gender differences in the prevalence of mental health problems among juvenile offenders are routinely found in the literatu re (McClelland et al., 2004; Teplin et al., 2003). These variations may also prove impor tant to understanding variations in the structure of problem behavior synd rome across demographic subgroups. Unfortunately, such social and psychologi cal factors were unable to be included in the analyses, which is a major limitation to the current study. In order to fully understand the developmental process of pr oblem behavior syndrome, and in turn, develop effective prevention and intervention strategies, this is a critical task for future research. Identifying the mechanisms that lead to problem behavior syndrome will aid in the quest for the most comprehensive interv ention strategies for problem behavior syndrome. Integrating informati on regarding the effect of the interaction of 1) individuallevel psychological factors, 2) parenting and p eer influences, 3) struct ural characteristics, and 4) demographic characteristics on problem behavior syndrome will lead to the development of effective, socio-ecologi cal prevention and intervention strategies.

PAGE 190

180 Another avenue for future research is th e theoretical examination of the causal processes that lead to the de velopment of problem behavior syndrome. In addition to enabling the examination of the temporal ordering of different problem behaviors, longitudinal data will also allow researchers to examine the "generality" of criminological theories. This line of research should entail examining the ability of a number of individual-level theoretical c onstructs to predic t the development of problem behavior syndrome, as well as determine if this ability is consistent across demographic subgroups. For years, the capability of such general theo ries to explain behavi or across demographic characteristics has been a popular deba te among criminologists (Akers, 1998; Gottfredson & Hirchi, 1990; Hay, 2001). Howe ver, this body of research tends to examine the ability of different "general theories to predic t problem behaviors separately. When a number of problem beha viors are studied in a single study, they are typically studied separately, each behavior serving as a dependent variable and then reporting the results of each of the different models. Examining the ability of these general theories to predict the latent constr uct, as well as identif ying any differences in this ability across individual ch aracteristics, will enhance our understanding of the causal processes that lead to problem behavior syndrome, while at the same time, provide additional information toward a re solution to this long-standing debate. If future research reveals that the so-called “general” theoretica l concepts are able to predict the latent construct of deviance across the different s ubgroups, additional support for the generality of these theories will be provided. Finally, it is important to note that the notion of problem behavior syndrome as a "real" or concrete syndrome has yet to be es tablished. The definition of a latent factor

PAGE 191

181 implies that it is a hypothetical construct that accounts fo r associations among observed indicators (Bollen, 2002). Thus, problem behavior syndrome remains an empirical abstraction. However, the tendency to simulta neously engage in a number of risk-taking behaviors is more than just an empirical concept; it is a seri ous social problem. According to Brosboom et al. (2003:207), what is needed is the adherence to realism. This perspective holds that the latent variable signifies a real entity. In other words, the latent factor is assumed to exist independent of measurement. “The assumption that the model is true must be taken literally, more literally, than many latent variable theorists would be comfortable with” (Brosboom et al., 2003:216). Thus, in addition to the continued empirical examination of problem be havior syndrome, it is also important for juvenile justice professionals, teachers, a nd parents, to recognize and accept problem behavior syndrome as a concrete reality. W ithout such recognition, the critical need for treatment strategies targeting multiple problem behaviors in an integrated framework will continue to go unrecognized by the juvenile justice system. Conclusion The findings of this study underscore the importance of integr ating criminological and public health efforts to reduce problem behavior among adolescent offenders. Risky sexual practices, substance use, and criminal involvement have been long considered public health, as well as criminological, issues However, cross-disc iplinary efforts to combat these behaviors are relatively rare. This is unfortunate because collaboration among public health and juvenile justice system agencies would be quite beneficial to improving the lives of juvenile offenders.

PAGE 192

182 For example, such collaboration would s ubstantially widen the resource bank of information available to both disciplines, which as result, could result in the introduction of fresh perspectives on techniques for reducing problem behavior syndrome among juvenile offenders. For example, the juvenile justice system has recently begun to emphasize law and order. Dispositions are increasingly reactive and punitive in nature, focusing on punishment rather than interven tion (Feld, 1998; OJJDP, 2001). However, there is little empirical evidence that punitiv e sanctions are effective in reducing problem behaviors among adolescent offenders. On the contrary, studies suggest that incarceration can often produce ad verse effects such as higher recidivism rates, more serious substance use, and emotional tr auma (MacKenzie, 2002; Spohn & Holleran, 2002). Thus, as Welsh (2005:24) points out, this trend toward becoming adversarial "represents an unsustainable approach to th e prevention" of misbehavior among juvenile offenders. The public health approach to dealin g with adolescent problem behavior represents a healthy alternative to preven ting and reducing problem behavior syndrome. Public health strategies focus on primary pr evention prevention in the first instance, prior to developing serious problem behavior (Welsh, 2005). This involves identifying and targeting the root causes of the devel opment of problem behavior syndrome, rather than reacting once it has been developed. The significant effect of age found in this study supports the importance of taking such an approach by relying on early intervention strategies. There are a number of such public health, early intervention strategies that target the risk factors for pr oblem behaviors that have show n considerable promise (see DiClemente et al., 2008 and Farrington & Welsh, 2007 for a review) for reducing

PAGE 193

183 problem behavior syndrome among adolescent offenders. However, the integration and implementation of these intervention strategies in the juvenile justice system is rarely accomplished. This is not to suggest that public health st rategies should be seen as a challenge to the standard juvenile justice focus on law and or der, but rather as a complement to it, part of an effort to create a more balanced and comprehensive strategy in reducing problem behavior syndrome among j uvenile offenders (Prothrow-Stith, 1992; Welsh, 2005). Reducing the development of problem behavior syndrome "…will require an interdisciplinary approach. Professionals from sociology, criminology, economics, public policy, psychology, anthropology, and public health must work together to understand the causes and develop the solutions" (Rose nberg & Mercy, 1991:11). Taking a socioecological, interdisciplinary approach to understanding, and more importantly preventing, the development of problem behavior syndrome will not only improve the lives of the adolescents involved, but will also improve th e health and well-being of the community as a whole.

PAGE 194

184 References Abram, K.M., Teplin, L.A., McClelland, G.M., Dulcan, M.K. (2003). Comorbid psychiatric disorders in youth in juvenile detention. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 419-426. Abrantes, A.M., Hoffman, N.G., & Anton, R. (2005). Prevalence of co-occurring disorders among juveniles committed to detention centers. International Journal of Offender Therapy and Comparative Criminology, 49 (2), 179-193. Agnew, R. (1991). A longitudinal test of social theory and delinquency. Journal of Research in Crime and Delinquency, 28, 126-156. Agnew, R. (1992). Foundations for a general strain theory of crime and delinquency. Criminology, 28, 126-156. Agresti, A. & Finlay, B. (1997). Statistical methods for the social sciences. Upper Saddle River, NJ: Prentice Hall. Akers, R. (1985). Deviant behavior. Belmont, CA. Wadsworth Publishing. Akers, R. (1998). Social learning and social structure: A general theory of crime and deviance. Boston, MA. Northeastern University Press. Akers, R.L. & Cochran, J.K. (1985). Adolescent ma rijuana use: A test of three theories of deviant behavior. Deviant Behavior, 6, 323-346. Akers, R., & Lee, G. (1996). A longitudinal test of social learning theory: adolescent smoking. Journal of Drug Issues, 26 (2), 317-343. Akers, R., & Jenson, G. (2003). Social learning theory and the explanation of crime: A guide for the new century. New Brunswick, NJ: Transaction. Akers, R.L. & Sellers, C.S. (2004). Criminological theories : Introduction, evaluation, and application. Roxbury: Los Angeles. Anaya, H.D., Swendeman, D., & Rotheram-Borus, M.J. (2005). Differences among sexually abused and nonabused youth living with HIV. Journal of Interpersonal Violence, 20 (12), 1547-1559.

PAGE 195

185 Arbuckle, J.L. & Wothke, W. (1999). Amos 4.0 User’s Guide. Chicago, ILL: SPSS. Armour, S. & Haynie, D.L. (2007). Adol escent sexual debut and later delinquency. Journal of Youth and Adolescence, 36, 141-152. Ary, D.V., Duncan, T.E., Biglan, A.., Metzle r, C.W., Noell, J.W., & Smolkowski, K. (1999). Development of adol escent problem behavior. Journal of Abnormal Child Psychology, 27 (2), 141-150. Bachanas, P.J., Morris, M.K., Lewis-Gess, J.K ., Sarett-Cuassay, E.J., Sirl, K., Ries, J.K., & Sawyer, M.K. (2002). Predictors of ris ky sexual behavior in African American adolescent girls: Implicati ons for prevention interventions. Journal of Pediatric Psychology, 27 (6), 519-530. Bailey, S.L., Camlin, C.A., & Ennett, S.T. (1998). Substance use and risky sexual behavior among homele ss and runaway youth. Journal of Adolescent Health, 23 (6), 378-388. Bardone, A.M., Moffitt, T.E., Caspi, A., Dickson, N., & Silva, P.A. (1996). Adult mental health and social outcomes of adol escent girls with depression and conduct disorder. Developmental Psychology, 8, 811-829. Bardone, A.M., Moffitt, T.E., Caspi, A., Dick son, N., Stanton, W., & Silva, P.A. (1998). Adult physical health outcomes of ad olescent girls with conduct disorder. Journal of the American Academy of Child and Adolescent Psychiatry, 37, 594-601. Barnes, G.M. & Welte, J.W. (1986). Adolescen t alcohol abuse: Subgr oup differences and relationships to other problem behaviors. Journal of Adolescent Research, 1, 7994. Barnes, G.M., Welte, J.W., Hoffman, J.A. & Dintcheff, B.A. (1999). Gambling and alcohol use among youth: Influences of demographic, socialization, and individual factors. Addictive Behaviors, 24 (6), 749-767. Barnes, G.B., Welte, J.W., & Hoffman, J.H. (2002). Relationship of alcohol use to delinquency and illicit drug use in adoles cents: Gender, age, and racial/ethnic differences. Journal of Drug Issues, 153-178. Barthlow, D.J., Horan, P.F., DiClemente, R. J., & Lanier, M.M. (1995). Correlates of condom use among incarcerated adol escents in a rural state. Journal of Criminal Justice Behavior, 22, 295-306. Bartlett, R., Holditch-Davis, D., & Belyea, M. (2005). Clusters of problem behaviors. Research in Nursing & Health, 28, 230-239.

PAGE 196

186 Basen-Engquist, K., Edmundson, E.W., & Parcel, G.S. (1996). Structure of health risk behavior among high school students. Journal of Consulting & Clinical Psychology, 74 (4), 764-775. Baskin-Sommers, A. & Sommers, I. (2006) The co-occurrence of substance use and high-risk behaviors. Journal of Adolescent Health, 38, 609-611. Baumrind, D. (1991). The influence of pare nting style on adolescent competence and substance use. Journal of Early Adolescence, 11, 56-95. Beachy, G.M., Peterson, D.M., Pearson, F.S.(1979). Adolescent drug use and delinquency: A research note. Journal of Psychedelic Drugs, 11 (4), 313-316. Belenko, S., Dembo, R., Weiland, D., Rollie, M., Sa lvatore, C., Hanlon, A., & Childs, K. (2008). Recently arrested adolescents ar e at high-risk for sexually transmitted diseases. Sexually Transmitted Diseases, 35 (8), 758-763. Belenko, S. & Logan, T.K. (2003). Deliveri ng effective treatment to adolescents: Improving the juvenile drug court model. Journal of Substance Abuse Treatment, 25, 189-221. Belenko, S., Sprott, J.B., & Peterson, C. (2004). Drug and alcohol involvement among minority and female juvenile offenders: Treatment and policy issues. Criminal Justice Policy Review, 15 (1), 3-36. Belgrave, F. Z., Randolph, S. M., Carter, C., Braithwaite, N., & Arrington, T. (1993). The impact of knowledge, norms, and self-effi cacy on intentions to engage in AIDSpreventive behaviors among young incar cerated African-American males. Journal of Black Psychology, 19, 155-168. Bell, T.A., Farrow, J.A., Stamm, W.E., Critchlow, C.W., & Holmes, K.K. (1985). Sexually transmitted diseases in females in a juvenile detention center. Sexually Transmitted Diseases, 12, 140-144. Benda, B.B. & Corwyn, R.F. (2000). A test of the validity of delinquency syndrome construct in a homogenous sample. Journal of Adolescence, 23, 497-511. Bennett, S., Farrington, D.P., & Huesmann, L.R. (2005). Explaining gend er differences in crime and violence: The importance of social cognitive skills. Aggression and Violent Behavior: A Review Journal, 10 (3), 263-288. Bentler, P.M. (1990). Comparative f it indices in structural models. Psychological Bulletin 107 238-246.

PAGE 197

187 Bentler, P.M. & Newcomb, M.D. (1986). Pe rsonality, sexual beha viors, and drug use revealed through latent variable methods. Clinical Psychology Review, 6, 363385. Biglan, A., Metzler, C.W., Wi rt, R., Ary, D., Noell, J., Ochs, L., French, C. & Hood, D. (1990). Social and behavior al factors associated with high-risk sexual behavior among adolescents. Journal of Behavioral Medicine, 13 (3), 245-261. Block, R., Braverman, K., & Farinpour, R. (1998). Acute effects of marijuana on cognition: Relationships to chro nic effects and smoking techniques. Pharmacology, Biochemistry, and Behavior, 43, 907-917. Bollen, K. (1989). Structural equations w ith latent variables. New York: Wiley. Bollen, K.B. (2002). Latent variables and in psychology and the social sciences. Annual Review of Psychology, 53, 605-634. Bonino, S., Cattelino, E., & Ciariano, S. (2003). Adolescents and risk: Behaviors, functions, and protective factors. New York: Springer. Bornovalova, MA., Gwadz, M.A., Kahler, C., Aklin, W.M.& Lejuez, C.W. (2008). Sensation-seeking and risk-taking propens ity as mediators in the relationship between childhood abuse and HIV-related risk behavior. Child Abuse & Neglect, 32 (1), 99-109. Borsboom, D., Mellenbergh, G.J. & Van Heerden, J.V. (2003). The theoretical status of latent variables. Psychological Review, 110 (2), 203-219. Boyer, C.B., Shafer, M., Wibbelsman, C.J., S eeberg, D., Teitle, E., & Lovell, N. (2000). Associations of sociodemographic psychos ocial and behavior factors with sexual risk and sexually transmitted diseases in teen clinic patients. Journal of Adolescent Health, 27 (2), 102-111. Boyer, C.B., Tschann, J.M., & Shafer, M.A. (1999). Predictors of risk for sexually transmitted diseases in ninth grade urban high school students. Journal of Adolescent Research, 14 (4), 448-465. Boyle, M.H., Offord, D.R., Racine, Y.A., Szatmari, P., Fleming, J.E., & Links, P.S. (1992). Predicting substance use in late adolescence: Results from the Ontario Child Health Study follow-up. American Journal of Psychiatry, 149, 761-767. Brooks-Gunn, J., Duncan, G. J., Klebanov, P. K., & Sealand, N. (1993). Do neighborhoods influence child and adolescent development? American Journal of Sociology, 99 353-395.

PAGE 198

188 Brown, L.K., DiClemente, R.J., & Park, T. ( 1992). Predictors of c ondom use in sexually active adolescents. Journal of Adolescent Health 13 651-657. Brown, L.K., Lourie, K.J., Zlot nick, C., & Cohn, J. (2000). Imp act of sexual abuse on the HIV-risk-related behavior of adolesce nts in intensive psychiatric treatment. American Journal of Psychiatry, 157 (9), 1413-1415. Browne, M.W. & Cudeck, R. (1993). Alternativ e ways of assessing model fit. In K. Bollen and S. Long (Eds.), Structural Equation Models (pp. 136-162). Newbury Park, CA: Sage. Browning, C. R., Leventhal, T., & Brooks-G unn, J. (2004). Neighborhood context and racial differences in early adolescent sexual activity. Demography, 41 (4), 697720. Bryan, A. & Stallings, M.C. (2002). A case control study of adolescent risky sexual behavior and its relationship to person ality dimensions, conduct disorder, and substance use. Journal of Youth and Adolescence, 31 (5), 387-393. Bui, K.V.T., Ellickson, P.L., & Bell, R.M. (2000). Cross-lagged relationships between adolescent problem drug use, delinquen t behavior, and emotional distress. Journal of Drug Issues, 30, 283-303. Bulcraft, R.A., Carmody, D.C., & Bulcroft K.A. (1996). Patterns of parental independence giving to adolescents: Varia tions by race, age, and gender of child. Journal of Marriage & the Family, 58 (4), 866-883. Bushway, S.D., Thornberry, T.P., & Krohn, M.D. (2003). Desistance as a developmental process: A comparison of st atic and dynamic approaches. Journal of Quantitative Criminology, 19 (2), 129-153. Byrne, B.M. (2001). Structural equation modeling with AMOS. New Jersey: Lawrence Elbaum Associates. Canterbury, R.J., McGarvey, E.L., SheldonKeller, A.E., Waite, D., Reams, P. & Koopman, C. (1995). Prevalence of HIV related risk behaviors and STDs among incarcerated adolescents. Journal of Adolescent Health, 17 (3), 173-177. Capaldi, D.M. & Patterson, G.R. (1989). Psychometric properties of fourteen latent constructs from the Oregon Youth Study. New York: Springer. Castrucci, B.C. & Martin, S.L. (2002). The association between substance use and risky sexual behaviors among incarcerated adolescents. Maternal and Child Health Journal, 6 (1), 43-47.

PAGE 199

189 Catalano, R.F. & D.J. Hawkins. (1996). The social development model: A theory of antisocial behavior ” In D.J. Hawkins (Ed.), Delinquency and Crime: Current Theories (pp. 149-197). Cambridge, UK: Ca mbridge University Press. Centers for Disease Control and Prevention (CDC). (1996). Sexually transmitted disease surveillance, 1995. Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. Centers for Disease Control and Prevention (CDC). (1998). HIV prevention through early detection and treatm ent of other sexually transm itted diseases: United States recommendations of the Advisory Committee for HIV and STD Prevention Atlanta: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. Centers for Disease Control and Prevention (CDC), Division of HIV/AIDS Prevention. (2000). Surveillance Report, 11 (2) Atlanta: US Department of Health and Human Services, Centers for Disease Control and Prevention. Center for Disease Control. (2006). Youth Risk Behavior Surveillance – United States, 2005. Surveillance Summaries, MMWR, 55, #SS-5. Centers for Disease Control and Prevention (CDC). (2006b). Sexually transmitted disease surveillance 2006 supplement, Chlamydia prevalence monitori ng project annual report 2005. Atlanta: US Department of Health and Human Services, Centers for Disease Control and Prevention. Centers for Disease Control and Prevention (CDC). (2007). Sexually transmitted disease surveillance 2006 supplement, Chlamydia prevalence monitori ng project annual report 2006. Atlanta: US Department of Health and Human Services, Centers for Disease Control and Prevention. Chacko, M., Barnes, C., Wiemann, C., & DiCl emente, R. (2004). Implementation of urine testing for chlamydia (CT) and gonorrhea (NGC) in a community clinic. Journal of Adolescent Health 34, 1460-153. Chaiken, J.M. & Chaiken, M.R. (1990). Drugs a nd predatory crime. In M. Tonry & J.Q. Wilson (Eds.), Drugs and crime (pp. 203-239). Chicago: University of Chicago Press. Chang, V.Y., Bendel, T.L., Koopman, C., Mc Garvey, E.L. & Canterbury, R.J. (2003). Delinquents’ safe sex attitudes: Rela tionships with demographics, resilience factors, and substance use. Criminal Justice & Behaviors, 30 (2), 210-229.

PAGE 200

190 Chapple, C., Johnson, K.D., & Whitbeck, L.B. (2004). Gender and arrest among homeless and runaway youth: An analysis of background, family, and situational factors. Youth Violence and Juvenile Justice, 2 (2), 129-147. Chen, A.C.C. & Thompson, E.A. (2007). Preven ting adolescent risky sexual behavior: Parents Matter! Journal for Specialists in Pediatric Nursing, 12 (2), 119–122. Cheong, Y.F. & Raudenbush, S.W. (2000). Meas urement and structural models for children’s problem behaviors. Psychological Methods, 5 (4), 477-495. Choi, Y., Harachi, T.W., & Catalano, R.F. (2006). Neighborhoods, family, and substance use: Comparisons of the relations across racial and ethnic groups. Social Service Review, 80 (4), 675-704. Costa, F.M., Jessor, R., Donovan, J.E., & Fo rtenberry, J.D. (1995). Early initiation of sexual intercourse: The influence of psychosocial unconventionality. Journal of Research in Crime and Delinquency, 5 (1), 93-121. Chuang, Y.C., Ennett, S.T., Bauman, K.E., & Foshee, V.A. (2005). Neighborhood influences on adolescent cigarette an d alcohol use: Mediating effects through parent and peer behaviors. Journal of Health & Social Behavior, 46 (2), 187-204. Coleman, J.C. (1992). The nature of adoles cence. In J.C. Coleman and C. WarrenAdamson (Eds.), Youth policy in the 1990s: The way forward (pp. 8-27). Florence, KY: Taylor & Francis. Cook, T.D. & Campbell, D.T. (1979). Quasi-experimentation: Desi gn and analysis issues for field settings. Boston: Houghton Mifflin Company. Cooper, M.L., Peirce, R.S., & Huselid, R.F. (1994). Substance use and sexual risk-taking among Black and White adolescents. Health Psychology, 13, 251-262. Crepaz, N., Horn, A.K., Rama, S.M., Griffin, T., Deluca, J.B., Mullins, M.M., & Aral, S.O. (2007). The efficacy of behavioral interventions tin reducing HIV risk sex behaviors and incident sexually transm itted disease in Black and Hispanic sexually transmitted disease clinic patients in the United States: A meta-analytic review. Sexually Transmitted Diseases, 34, 319-332. Crosby, R.A., DiClemente, R.J., Wingood, G.M., Rose, E., & Levine, D. (2003). Adjudication history and African Amer ican adolescents’ risk for acquiring sexually transmitted diseases: An exploratory analysis. Sexually Transmitted Diseases, 30 (8), 634-638.

PAGE 201

191 Crosby, R.A., Leichlitter, J.S. & Brackbill, R. (2000). Longitudinal prediction of sexually transmitted diseases among adolescents: Results from a national study. American Journal of Preventative Medicine, 18, 312-317. Crosby, R., Salazar, L.F., DiClemente, R. J., Yarber, W.L., Caliendo, A.M., StaplesHorne, M. (2007). Condom misuse among ad judicated girls: A ssociations with laboratory confirmed Chlamydia and Gonorrhea. Journal of Pediatric Adolescent Gynecology, 20, 339-343. Cubin, C., Santelli, J., Brindis, C.D., & Br aveman, P. (2005). Neighborhood context and sexual behaviors among adolescents: Fi ndings from the National Longitudinal Study of Adolescent Health. Perspectives on Sexual and Reproductive Health, 37 (3), 125-134. Cocozza, J.J. & Skwoyra, C. (2000). Youth with mental disorders in the juvenile justice system: Trends, issues, and emerging responses. Juvenile Justice, 7, 3-13. De Genna, N.M., Cornelius, M.D., & Cook, R.L. (2007). Marijuana use and sexually transmitted infections in young wo men who were teenage mothers. Womens Health Issues, 17 (5), 300-309. Deas-Nesmith, D., Brady, K.T., White, R., & Campbell, S. (1999). HIV-risk behaviors in adolescent substance abusers. Journal of Substance Abuse Treatment, 16 (2), 169172. DeCourville, N.H. (1995). Testing the appli cability of problem behavior theory to substance use in a longitudinal study. Psychology of Addictive Behaviors, 9 (1), 53-66. Derzon, J.H. & Lipsey, M.W. (1999). A synthe sis of the relationship of marijuana use with delinquent and problem behaviors. School Psychology International, 20, 5768. Dembo, R., Jainchill, N., Turner, C., Fong, C., Farkas, S., & Childs, K. (2007). Levels of psychopathy and its correlates: A study of incarcerated youths in three states. Behavioral Sciences and the Law, 25, 1-22. Dembo, R., Pacheco, K., Schmeidler, J., Fish er, L., & Cooper, S. (1997). Drug use and delinquent behavior among high risk youths. Journal of Child and Adolescent Substance Abuse, 6 1-25. Dembo, R. & Shern, D. (1982). Relative deviance and the process of drug involvement among inner-city youths. International Journal of Addictions, 17, 1373-1399.

PAGE 202

192 Dembo, R., & Schmeidler, J. (2002). Family empowerment inte rvention: An innovative service for high-risk yo uths and their families New York: The Haworth Press. Dembo, R., Turner, G., Sue, C.C., Schmeidler J., Borden, P., & Manning, D. (1994). An assessment of the Florida Department of Health and Rehabilitative Services Detention Risk Assessment Instrument on Youths Screened and Processed at the Hillsborough County Juvenile Assessment Center. Journal of Child & Adolescent Abuse, 4 (1), 45-77. Dembo, R., Wareham, J., Poythress, N ., Cook, B., & Schmeidler, J. (2006). The correlates and consequences of dr ug involvement among youths entering a juvenile justice diversion program. Journal of Child and Adolescent Substance Use, 15 (4), 1-36. Dembo, R., Wareham, J., & Schmeidler J. (2007a). Drug use and delinquent involvement: A growth model of para llel processes among high risk youths. Criminal Justice and Behavior, 34 (5), 680-696. Dembo, R., Wareham, J., & Schmeidler, J. ( 2007b). A longitudinal st udy of cocaine use among juvenile arrestees. Journal of Child and Adolescent Substance Use, 17 (1), 83-109. Dembo, R., Washburn, M., Wish, E.D., Schmei dler, J., Getrau, A., Berry, E., Williams, L., & Blount, W.R. (1987). Further examin ation of the association between heavy marijuana use and crime among youths entering a juvenile detention center. Journal of Psychoactive Drugs, 19 (4), 361-373. Dembo, R., Williams, L., Fagan, J., & Schmei dler, J. (1993). The relationships of substance involvement and other delinquenc y over time in a sample of juvenile detainees. Criminal Behavior and Mental Health, 3 158-179. Dembo, R., Williams, L., Schmeidler, J., Ge treu, A., Berry, E., Genung, L., Wish, E., & LaVoie, L. (1991). A longitudinal study of the relationships among marijuana/hashish use, cocaine use and delinquency in a cohort of high risk youths. Journal of Drug Issues, 21 271-312. Dembo, R., Williams, L., Schmeidl er, J. & Howitt, D. (1991). Tough cases: School outreach for at-risk youth Washington, D.C.: U.S. Department of Education, Office of the Assistant Secretary for E ducational Research and Improvement. Dembo, R., Williams, L., Wothke, W., Schmeidl er, J., Getreu, A., Berry, E., & Wish, E. (1992). The generality of deviance: Replic ation of a structural model among high risk youths. Journal of Research in Crime and Delinquency, 29 200-216.

PAGE 203

193 Dembo, R., Williams, L., Wothke, W., & Schmei dler, J. (1994). The relationships among family problems, friends' troubled behavi or and high risk youths' alcohol/other drug use and delinquent beha vior: A longitudinal study. International Journal of the Addictions, 29 1419-1442. Devieux, J., Malow, R., Stein, J.A., Jenni ngs, T.E., Lucenko, B.A., Averhart, C., & Kalichman, S. (2002). Impulsivity and HIV risk among adjudicated alcoholand other drug-abusing adolescent offenders. AIDS Education and Prevention, 14, 2435. Devine, D., Long, P., & Forehand, R. (1993). A prospective study of adolescent sexual activity: Description, corr elates, and predictors. Advances in Behavior Research and Therapy, 15, 185-209. Diblaso, F. (1986). Drinking alcohol on the roads. Journal of Youth and Adolescence, 15, 173-188. DiClemente, R.J. (1991). Predictors of HIVpreventive sexual behavior in a high-risk adolescent population: The influe nce of perceived norms and sexual communication on incarcerated adoles cents’ consistent use of condoms. Journal of Adolescent Health, 12, 385-390. DiClemente, R.J., Crittenden, C.P., Rose, E ., Sales, J.M., Wingood, G.M., Crosby, R.A., & Salazar, L.F. (2008). Psychosocial pr edictors of HIV-associated sexual behaviors and the efficacy of prevention interventions in adolescents at-risk for HIV infections: What works and what doesn’t work? Psychosomatic Medicine, 70, 598-605. DiClemente, R.J., Durbin, M., Siegel., D ., Krasnovsky, F., Lazarus, N., & Comacho, T. (1992). Determinants of condom use among junior high school students in a minority, inner-city school district. Pediatrics, 89, 197-202. DiClemente, R.J., Lanier, M.M., Horan, P.F. & Lodico. M. (1991). Comparison of AIDS knowledge, attitudes, and behaviors am ong incarcerated adolescents and a public school sample in San Francisco. American Journal of Public Health, 81, 628-629. DiClemente, R.J., Lodico, M., Grinstead, O. A., Harper, G., Rickman, R. Evans, P. & Coates, T.J. (1996). African-American a dolescents residing in high-risk urban environments do use condoms: Correlates and predictors of condom use among adolescents in public housing developments. Pediatrics, 98, 269-277. DiClemente, R.J., Wingood, G.M., Crosby, R., Sionean, C., Cobb, B.K., Harrington, K., Davies, S., Hook, E.W., & Kim, M. (2001). Parental mon itoring: Association with adolescents’ risk behavior. Pediatrics, 107, 1363-1368.

PAGE 204

194 DiClemente, R.J., Wingood, G.M., Harringt on, K.F., Lang, D.L., Davies, S.L., Hook, E.W., Oh, M.K., Crosby, R.A., Hertzber g, V.S., Goardon, A.B., Hardin, J.W., Parkers, S., & Robillard, A. (2004). E fficacy of an HIV prevention intervention for African American adolescent girls: A randomized clinical trial. Journal of the American Medical Association, 292, 171-179. DiClemente, R.J., Wingood, G.M., Sionean, C ., Crosby, R., Harrington, K., Davies, S., & Hook, E.W. (2002). Association of adol escents’ history of sexually transmitted disease (STD) and their current high-ri sk behavior and STD status: A case for intensifying clinic-based prevention efforts. Sexually Transmitted Diseases, 29 (9), 503-509. Dishion, T.J. (2000). Cross-setting consiste ncy in early adolescent psychopathology: Deviant friendships and pr oblem behavior sequalae. Journal of Personality, 68, 1109-1126. Donovan, J.E. (1996). Problem-behavior theo ry and the explana tion of adolescent marijuana use. Journal of Drug Issues, 26, 379-404. Donovan, J.E. & Jessor, R. (1985). Structure of problem behavior in adolescence and young adulthood. Journal of Consulting and Clinical Psychology, 53(6), 890-904. Donovan, J.E., Jessor, R., & Costa, F.M. (1988). Brief report: Syndrome of problem behavior in adolescence: A replication. Journal of Consulting & Clinical Psychology, 56 (5), 762-765. Donovan, J.E., Jessor, R., & Costa, F.M. (1990) Personality, perceived life chances, and adolescent health behavior: An applica tion of problem behavior theory. In K. Hurrelmann and F. Losel (Eds.), Health hazards in adolescence (pp. 25-41). New York: Aldine. Donovan, J.E., Jessor, R., & Costa, F.M. (1991). Adolescent health behavior and conventionality-unconventionality: An extension of problem-behavior. Health Psychology, 10 (1), 52-61. Donenberg, G.R., Emerson, E., Bryant, F. B., Wilson, H., & Weber-Shifrin, E. (2001). Understanding AIDS-risk behavior among a dolescents in psychiatric care: Links to psychopathology and peer relationships. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 642-653. Drasgow, F. (1987). Study of the measurement bias of two standardized psychological tests. Journal of Applied Psychology, 72, 19-29.

PAGE 205

195 Duncan, S.C., Strycker, L.A., & Duncan, T.A. (1999). Exploring associations in developmental trends of adolescent subs tance use and risky sexual behavior in a high-risk population. Journal of Behavi oral Medicine, 22, 21-34. Ellickson, P.L., Hays, R.D., & Bell, R. M. (1992). Stepping through the drug use sequence: Longitudinal scalogram analys is of initiation and regular use. Journal of Abnormal Psychology, 102, 441-451. Ellickson, P.L. & McGuigan, K.A. (2000). Earl y predictors of adolescent violence. American Journal of Public Health, 90, 566–572. Elliott, D.E. & Ageton, S. (1980). Reconciling ra ce and class differences in self-reported and official estimates of delinquency. American Sociological Review, 45 (1), 95110. Elliott, D.S. & Huizinga, D. (1984). The relationship between delinquent behavior and ADM problems. Boulder, CO: Behavioral Research Institute. Elliott, D.S., Huizinga, D., & Menard, S. (1989). Multiple problem youth. New York: Springer. Elliott, D.S. & Morse, B. (1989). Delinquency and drug use as risk factors in teenage sexual activity. Youth & Society, 21 (1), 32-60. Elze, D.E., Auslander, W., McMille n, C., Edmond, T., & Thomspon, R. (2001). Untangling the impact of sexual abus e on HIV risk behaviors among youth in foster care. AIDS Education and Prevention, 13 (4), 377-389. Ensminger, M.E. (1990). Sexual activity and problem behavior among black, urban adolescents. Child Development, 61, 2032-2046. Evans, A.E., Sanderson, M., Griffin, S.F., Rein inger, B., Vincent, M.L., Parra-Medina, D., Valois, R.F., & Taylor, D. (2004). An exploration of the relationship between youth assets and engagement in risky sexual behaviors. Journal of Adolescent Health, 35 (5), 424-430. Fagan, J., Weis, J.G., & Chang, Y. (1990). Delinquency and substance use among innercity students. Journal of Drug Issues, 20, 351-402. Farrell, A.D., Danish, S.J., & Howard, C.W. (1992). Relationship between drug use and other problem behavior sin urban adolescents. Journal of Consulting and Clinical Psychology, 60 (5), 705-712.

PAGE 206

196 Farrell, A.D., Kung, E.M., White, K.S., & Valo is, R.F. (2000). The structure of selfreported aggression, drug use, and deli nquent behavior during early adolescence. Journal of Clinical Child Psychology, 29, 282-293. Farrington, D.P. (1985). Predic ting self-reported and offici al delinquency. In D.P. Farrington and R. Tarling (Eds.), Prediction in Criminology (pp. 150-173). Albany, NY: State University of New York Press. Farrington, D.P. (1993). Motivations for conduct disorder and delinquency. Development and Psychopathology, 5, 225-241. Farrington, D.P. (1998). Predictors, causes, a nd correlates of male youth violence. In M. Tonry and M.H. Moore (Eds.), Youth Violence (Crime and Justice: A Review of the Research Vol. 24: pp. 421-475). Chicago: Un iversity of Chicago Press. Farrington, D.P., Jolliffe, D., & Hawkins, J.D. (2003). Comparing delinquency careers in court records and self-reports. Criminology, 41 (3), 933-958. Farrington, D.P., Snyder, H., Finnegan, T. (1988). Specialization in juvenile court careers. Criminology, 26, 461-487. Farrington, D.P. & Welsh, B.C. (2006). Saving Children from a Life of Crime: Early Risk Factors and Effective Interventions. New York: Oxford University Press. Feld, B.C. (1998). Juvenile and criminal justic e systems’ response to youth violence. In M. Tonry & M.H. Moore (Eds.), Youth Violence: A Review of Research (Vol. 24, pp. 189-261). Chicago: Universi ty of Chicago Press. Fergusson, D.M. & Horwood, L.J. (1997). Ea rly onset cannabis use and psychosocial adjustment in young adults. Addiction, 92, 279-292. Fergusson, D.M. & Horwood, L.J. (2000). Does cannabis use encourage other forms of illicit drug use? Addiction, 95 (4), 505-520. Fergusson, D.M. & Lynskey, M.T. (1996). Alcohol misuse and adolescent sexual behaviors and risk taking. Pediatrics, 98, 91-96. Fergusson, D.M., Lynskey, M.T., & Horwood, L. J. (1996). Alcohol misuse and juvenile offending in adolescence. Addiction, 91, 483-494. Fergusson, D.M., Swain-Campbell, N.R. & Horwood, L.J. (2002). Deviant peers affiliations, crime, and substance use: A fixed effects regression analysis. Journal of Abnormal Child Psychology, 30 (4), 419-430.

PAGE 207

197 Flora, D.B. & Curran, P.J. (2004). An empiri cal evaluation of alte rnative methods of estimation for confirmatory factor analysis. Psychological Methods, 9 (4), 466491. Francis, B., Soothill, K., & Fligelstone, R. (2004). Identifying pattern s and pathways of offending behavior. European Journal of Criminology, 1 (1), 47-87. Frank, R., Cerda, M. & Rendon, M. (2007). Barr ios and burbs: Residential context and health-risk behaviors among Angeleno adolescents. Journal of Health and Social Behavior, 48 (3), 283-300. Frankenberger, K.D. (2004). Adolescent egocen trism, risk perceptions and sensation seeking among smoking and nonsmoking youth. Journal of Adolescent Research, 19 (5), 576-590. Fullilove, M.T., Golden, E, Fullilove, R.E., Lennon, R., Porterfield, D., Schwarcz, S. & Bolan, G. (1993). Crack cocaine use a nd high-risk behaviors among sexually active black adolescents. Journal of Adolescent Health, 14, 295-300. Gfoerer, J.C., Wu, L.T., & Penne, M.A. (2002). Initiation of marijuana use: Trends, patterns, and implications. Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Gillmore, M.R., Butler, S.S., Lohr, M.J. & G ilchrist, L. (1992). Substance use and other factors associated with risky sexua l behavior among pre gnant adolescents. Family Planning Perspectives, 24, 255-268. Gillmore, MR., Hawkins, J.D., Catalano, R.F., Day, L.E., Moore, M., & Abbott, R. (1991). Structure of problem behaviors in preadolescence. Journal of Consulting and Clinical Psychology, 59 (4), 599-506. Goldsmith, S., Gabrielson, I., Matthews, V., & Potts, L. (1972). Teenagers, sex, and contraception. Family Planning Perspectives, 4, 32-38. Gorman-Smith, D., Tolan, P.H. & Henry, D. B. (2000). A developmental-ecological model of the relation of family f unctioning to patterns of delinquency. Journal of Quantitative Criminology, 16 (2), 169-198. Gottfredson, D.C. (2001). Schools and delinquency. Cambridge: Cambridge University Press. Gottfreson, M. & Hirschi, T. A general theory of crime Stanford, CA: Stanford University Press.

PAGE 208

198 Graber, J., Brooks-Gunn, J., & Galen, B. ( 1998). Betwixt and between : Sexuality in the context of adolescent transitions. In R. Jessor (Ed.), New perspectives on adolescent risk behavior (pp. 270-318). New York: Cambridge. Greene, K., Krcmar, M., Walters, L.H., Rubin, D.L., Hale, J. (2000). Targeting adolescent risk-taking beha viors: The contributions of egocentrism and sensationseeking. Journal of Adolescence, 23, 439-461. Greene, K., Krcmar, M., Rubin, D.L., Walters, L.H., & Hale, J.L. (2002). Elaboration in processing adolescent health messages: Th e impact of egocentrism and sensation seeking. Journal of Communication, 52 (4), 812-831. Greene, K., Rubin, D.L., & Hale, J.L. (1995) Egocentrism, message explicitness, and AIDS messages directed toward adolescen ts: An application of the theory of reasoned action. Journal of Social Behavior and Personality, 10, 547-570. Greene, K., Rubin, D.L., Walters, L.H., & Hale J.L. (1996). The utility of understanding adolescent egocentrism in desi gning health promotion messages. Health Communications, 8, 131-152. Greenwood, P.W. (1992). Substance abuse pr oblems among high-risk youth and potential intervention. Crime & Delinquency, 38, 444.458. Grube, J.W. & Morgan, M. (1990). The stru cture of problem behaviors among Irish adolescents. British Journal of Addiction, 85, 667-675. Gullone, E. & Moore, S. (2000). Adolescent ri sk-taking and the five -factor model of personality. Journal of Adolescence, 23, 393-407. Haasapalo, J. & Tremblay, R.E. (1994). Physic ally aggressive boys from ages 6 to 12: Family background, parenting behavi or, and prediction of delinquency. Journal of Consulting and Clinical Psychology, 62 (5), 1044-1052. Hallfors, D., Cho, H., & Brodish, P.H. (2006) Identifying high school students “at-risk” for substance use and other behavioral problems: Implications for prevention. Substance Use & Misuse, 41 (1), 1-15. Hallfors, D.D., Waller, M.W., Ford, C.A., Halpern, C.T., Brodish, P.H., & Iritani, B. (2004). Adolescent depression and su icide risk: Associ ation with sex. American Journal of Preventative Medicine, 27 (3), 224-231. Halpern, C.T., Hallfors, D., Bauer, D.J., Iritani, B., Waller, M. W., & Cho, H. (2004). Implications of racial and gender diff erences in patterns of adolescent risk behavior for HIV and other sexually transmitted diseases. Perspective on Sexual Reproductive Health, 36 (6), 239-247.

PAGE 209

199 Haney, M., Ward, A., Comer, S., Foltin, R., & Fischman, M. (1999). Abstinence symptoms following smoked marijuana in humans. Psychopharmacology, 141, 395-404. Harwell, T.S., Trino, R., Bret, R., Yorkma n, S. & Gollub, E.L. (1999). Sexual activity, substance use, and HIV/STD knowledge among detained male adolescents with multiple versus first admissions. Sexually Transmitted Diseases, 26 (5), 265-271. Hawkins, J., Catalano, R., & Miller, J. (1992). Risk and protective fa ctors for alcohol and other drug problems in adolescence a nd early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112 (1), 64-105. Hawkins, D.F., Laub, J.H., Lauritsen, J.K., & Cothern, L. (2000). Race, ethnicity, and serious and violent juvenile offending. Juvenile Justice Bulletin, June 2000. Washington, DC: Office of Juvenile Justice Delinquency Prevention. Hay, Carter. (2001). Parenting, self-control, and delinquency: A test of self-control theory. Criminology, 39 (3), 707-736. Hays, R.D., Stacy, A.W., & DiMatteo, M.R. (1987). Brie f report: Problem behavior theory and adolescent alcohol use. Addictive Behavior, 12 189-193. Helstrom, A., Bryan, A., Hutchison, K.E ., Riggs, P.D., & Blechman, E.A. (2004). Tobacco and alcohol use as an explanation for the association between externalizing behavior and illicit drug use among delinquent adolescents. Prevention Science, 5 (4), 267-277. Hemphill, S.A., Herrenkohl, T.I., LaFazia, A.N., McMOrris, B.J., Toumbourou, J.W., Arthur, M.W., Catalano, R.F., Hawkins, J.D., & Bond, L. (2007). Comparison of the structure of adolescent problem beha vior in the United States and Australia. Crime & Delinquency, 53 (2), 303-321. Herbst, J.H., Kay, L.S., Passin, W.F., Lyles, C.M., Crepaz, N., & Marn, B.V. (2007). A systematic review and meta-analysis of behavioral interventions to reduce HIV risk behaviors of Hispanics in th e United States and Puerto Rico. AIDS & Behavior, 11, 23-47. Hindelang, M., Hirschi, T., & Weis, J.G. ( 1981). Measuring delinque ncy. Beverly Hills, CA: Sage. Hingson, R.W., Strunin, L., Berlin, B.M. & H eeren, T. (1990). Beliefs about AIDS, use of alcohol and drugs, and unprotected sex among Massachusetts adolescents. American Journal of Public Health, 80, 295-299.

PAGE 210

200 Hirschi, T. (1969). Causes of delinquency. Transaction Publishers: New Brunswick, NJ. Hirschi, T. & Gottfredson, M. (1993). “Co mmentary: Testing the general theory of crime.” Journal of Research in Crime and Delinquency, 30 (1), p.47-54. Hoff, T., Greene, L. & Davis, J. (2003). National survey of adolescents and young adults: Sexual health knowledge, attitudes, and experiences. Washington, DC: Kaiser Family Foundation. Horney, J., Osgood, D., & Marshall, I.H. (1995). Criminal careers in the short term: Intraindividual variability in crime and its relation to local life circumstances. American Sociological Review, 60, 665-673. Huebner, A.J. & Howell, L.W. (2003). Exam ining the relationship between adolescent sexual risk-taking and perceptions of monitoring, communication, and parenting styles. Journal of Adolescent Health, 33 (2), 71-78. Holmbeck, G.N., Paikoff, R.L., & Brooks-Gunn, J. (1995). Parenting adolescents. In M.H. Bornstein (Ed.), Handbook of parenting: Children and parenting (pp. 3551). Hillsdale, NJ: Lawrence Erlbaum Associates. Hu, L. & Bentler, P.M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criter ia versus new alternatives. Structural Equation Modeling 6 (1), 1-55. Huizinga, D. & Jakob-Chien, C. (1998). The contemporaneous co-occurrence of serious and violent juvenile offender and other pr oblem behaviors. In R. Loeber & D.P. Farrington (Eds.), Serious and violent juvenile o ffenders: Risk factors and successful interventions (pp. 47-67). Thousand Oaks, CA:Sage. Huizinga, D., Loeber, R., & Thornberry, T.P. (1993). Longitudinal study of delinquency, drug use, sexual activity, and pregan ancy among children and youth in three cities. Public Health Reports: Journal of the US Public Health Service, 108, 9096. Huizinga, D., Loeber, R., & Thornberry, T.P. (1995). Recent findings from the program of research on the causes and correlates of delinquency (US Department of Justice, Office of Justice Programs, Office of Juvenile Justice Delinquency Prevention, NCJ 159042). Washington, DC : US Government Printing Office. Huizinga, D., Loeber, R., Thornberry, T.P. & Cothern, L. (2000). Co-occurrence of delinquency and other problem behaviors. Juvenile Justice Bulletin November 2000. US Washington, DC: Office of Juveni le Justice Delinqu ency Prevention.

PAGE 211

201 Huizinga, D., Loeber, R., & Thornberry, T.P. (1994). Urban delinquency and substance abuse: Initial findings. Wa shington, DC: Office of J uvenile Justice Delinquency Prevention. Igra, V. & Irwin, C.E. (1996). Theories of adolescent risk-taking behavior. In R.J. DiClemente, W.B. Hansen, & L.E. Ponton (Eds.), Handbook of adolescent health risk behavior (pp. 35-51). New York: Plenum. Inciardi, J.A, Pottieger, A.E., Forney, M. A., Chitwood, D.D., & McBride, D.C. (1991). Prostitution, IV drug use, and se x-for-crack exchanges among serious delinquents: Risks for HIV infection. Criminology, 29 (2), 221-236. Ingoldsby, E.M. & Shaw, D.S. (2002). Nei ghborhood contextual factors and earlystarting antisocial pathways. Clinical Child & Family Psychology Review, 5 (1), 21-55. Jang, S.J. (1999). Age-varying effects on fa mily, school, and peers on delinquency: A multilevel modeling test of interactional theory. Criminology, 37 (3), 643-685. Jang, S.J. (2002). The effects of family, school peers, and attitudes on adolescents’ drug use: Do they vary by age? Justice Quarterly, 19 (1), 97-26. Jang, S.J. & Johnson, B.R. (2001). Neighborhood disorder, individual religiosity, and adolescent use of illicit drugs: A test of multilevel hypothesis. Criminology, 39 (1), 109-143. Jemmott, J.B., & Jemmott, L.S. (1993). Al cohol and drug use during sexual activity: Predicting the HIV-related risk behavior s of inner-city black male adolescents. Journal of Adolescent Research, 8, 41-57. Jemmott, J.B., Jemmott, L.S., & Fong, G.T. (1998). Abstinence and safer sex HIV riskreduction interventions for African Ameri can adolescents: A randomized clinical trial. Journal of the American Medical Association, 292, 1529-1536. Jessor, R., Donovan, J. E., & Costa, F. (1991). Beyond adolescence: Problem behavior and young adult development. New York: Cambridge University Press. Jessor, R., & Jessor, S. L. (1977). Problem behavior and psyc hosocial development: A longitudinal study of youth New York: Academic Press. Joesof, M.R., Kahn, R., & Weinstock, H.S. (2006). Sexually transmitted diseases in incarcerated adolescents. Current Opinion in Infectious Diseases, 19 (1), 44-48. Johnson, R.E., Marcos, A., & Bahr, S.J. (1987). Th e role of peers in the complex etiology of adolescent drug use. Criminology, 25 (2), 323-340.

PAGE 212

202 Johnson, B.D., Wish, E.D., Schmeidler, J., & Huizinga, D. (1991). Concentration of delinquent offending: Serious drug in volvement and high delinquency rates. Journal of Drug Issues, 21 (2), 205-229. Johnson, R. 1979. Juvenile Delinquency and Its Origins Cambridge: Cambridge University Press. Johnston, L. D., O'Malley, P. M., Bachma n, J. G., & Schulenberg, J. E. (2007). Monitoring the Future national results on adolescent drug use: Overview of key findings, 2006 (NIH Publication No. 07-6202). Be thesda, MD: National Institute on Drug Abuse. Joreskog, K.G. (1979). Basic ideas and com ponent analysis. In K.G. Joreskog & D. Sorbom (Eds.), Advances in Factor Analysis and Structural Equation Modeling (pp. 5-20). Cambridge, MA: Abt. Kahn, R.H., Mosure, D.J., Blank, S., Kent, C.K., Chow, J.M., Boudov, M.R., Brock, J., & Tullough, S. (2005). Chlamydia tr achomatis and Neisseria gonorrhea prevalence and coinfection in adolescents entering selected US juvenile detention centers, 1997-2002. Sexually Transmitted Diseases, 32 (4), 255-259. Kandal, D.B. & Anderws, K. (1987). Processes of adolescent socializ ation by parents and peers. Substance Use & Misuse, 22 (4), 319-342. Kaltiala-Heino, R., Kosunen, E., & Rimpela, M. (2003). Pubertal timing, sexual behavior, and self-reported depres sion in middle adolescence. Journal of Adolescence, 26 (5), 531-545. Kandel, D.B., & Logan, J.A. (1984). Pattern s of drug use from adolescence to young adulthood: Periods of risk for initia tion, continued use, and discontinuation. American Journal of Public Health, 74, 660-666. Kandel, D., Yamaguchi, K., & Chen, K. ( 1992). Stages of drug involvement from adolescence to adulthood: Furthe r evidence for the gateway theory. Journal of Studies on Alcohol, 53, 447-457. Kapungu, C.T., Holmbeck, G.N., & Paikoff, R.L. (2006). Longitudinal association between parenting practices and early sexual risk behavior among urban African American adolescents: The moderating role of gender. Journal of Youth and Adolescence, 35 (5), 787-798. Kaufman, R. (1978). The relationship of social cl ass and ethnicity to drug abuse. In D.E. Smith, S.M. Anderson, M. Burton, N. Gottlieb, W. Harney, and T. Chung (Eds.), A Multicultural View of Drug Abuse, pp. 158-164. Cambridge, MA: Schenkman.

PAGE 213

203 Kellam, S.G., Ling, X., Merisca, R., Brown, C. H. & Ialongo, N. (1998). The effect of the level of aggression in the first grade cl assroom on the course and malleability of aggressive behavior into middle school. Development & Psychopathology, 10, 165-185. Kelley, B.T., Huizinga, D., Thornberry, T.P ., & Loeber, R. (1997). Epidemiology of serious violent offenders. Juvenile Justice Bulletin June 1997. Washington, DC: US Department of Justice: Office of Justice Programs. Kim, J.Y.S. & Fendrich, M. ( 2002). Gender differences in juve nile arrestees’ drug use, self-reported dependence, and pe rceived need for treatment. Psychiatric Services, 53, 70-75. Kim, J. & Mueller, C.W. (1978). Factor analysis: Statistica l methods and practical issues. Beverly Hills: Sage. Kingree, J.B. & Betz, H. (2003). Risky sexua l behavior in relation to marijuana and alcohol use among African American, male adolescent detainees and their female partners. Drug and Alcohol Dependence, 72, 197-203. Kingree, J.B., Braithwaite, R., & Woodring, T. (2000). Unprotected sex as a function of alcohol and marijuana use among adolescent detainees. Journal of Adolescent Health, 27 (3), 179-185. Kingree, J.B. & Phan, D.L. (2001). Mariju ana use and HIV risk among adolescent offenders: The moderating effect of age. Journal of Substance Abuse, 13, 59-71. Kline, R.B. (2005). Principles and practice of st ructural equation modeling New York: Guilford Press. Koniak-Griffin, D. & Brecht, M. (1995). Li nkages between sexual risk-taking, substance use, and AIDS knowledge among pregnant adolescents and young mothers. Nursing Research, 44, 340-346. Kornhauser, R. (1978). Social sources of delinquency. Chicago: University of Chicago Press. Kosterman, R., Hawkins, J.D., Guo, J., Ca talano, R.F., & Abbott, R.D. (2000). The dynamics of alcohol and marijuana initiatio n: Patterns and predictors of first use in adolescence. American Journal of Public Health, 90 (3), 360-366. Kosunen, E., Kaltiala-Heino, R., Rimpela, M., & Laippala, P. (2003). Risk-taking sexual behavior and self-reported depr ession in middle adolescence, Child Care, Health, and Development, 29 (5), 337-344.

PAGE 214

204 Kotchick, B.A., Shaffer, A., Forehand, R., & Mi ller, K.S. (2001). Adolescent sexual risk behavior: A multi-system perspective. Clinical Psychology Review, 21 (4), 493519. Krohn, M., Lizotte, A., Thornberry, T.P., Smith, C., & McDowall, D. (1996). Reciprocal causal relationships among drug use, peers, and beliefs: A five wave panel model. Journal of Drug Issues, 26, 405-428. Kulis, S., Marsiglia, F.F., Sicotte, D., & Nieri, T. (2007). Neighborhood effects on youth substance use in a southwestern city. Pacific Sociological Review, 50 (2), 273301. Kuther, T.L. & Higgins-D’Alessandro, A. (2000). Bridging the gap between moral reasoning and adolescent enga gement in risky behavior. Journal of Adolescence, 23, 409-422. Laible D.J., Carlo, G., & Raffaelli, M. (2000) The differential relations of parent and peer attachment to adolescent adjustment. Journal of Youth and Adolescence, 29 (1), 45-59. Lapsley, D.K. (1993). Toward an integrated theory of adolescent ego development: The “new look” at adol escent egocentrism. American Journal of Orthopsychiatry, 63 (4), 562-571. Lebeau-Craven, R., Stein, L., Barnett, N ., Colby, S.M., Smith, J.L., & Canto, A.L. (2003). Prevalence of alcohol and drug use in an adolescent training facility. Substance Use and Misuse, 38 (7), 825-834. LeBlanc, M.L. & Bouthillier, C. (2003). A de velopmental test of the general deviance syndrome with adjudicated girls and boys using hierarchical confirmatory factor analysis. Criminal Behavior and Mental Health, 13, 81-105. LeBlanc, M. & Girard, S. (1997). The gene rality of deviance: Replication over two decades with a Canadian sample of adjudicated boys. Canadian Journal of Criminology, 171-183. LeBlanc, M. & Loeber, R. (1998). Developm ental criminology revisited. In M. Tonry (Ed.), Crime and Justice (Vol. 23, pp. 115-197). Chicago: University of Chicago Press. Lessem, J.M., Hopfer, C.J., Haberstick, B.C ., Timberlake, D., Ehringer, M.A., Smolen, A., & Hewitt, J.K. (2006). Relationship between adolescent marijuana use and young adult illicit drug use. Behavior Genetics, 36 (4), 498-506.

PAGE 215

205 Leventhal, T. & Brooks-Gunn, J. (2000). The ne ighborhoods they live in: The effects of neighborhood residents on child and adolescent outcomes. Psychological Bulletin, 126 (2), 309-337. Li, X., Feigelman, S., & Stanton, B. (1999). Pe rceived parental monitoring and health risk behaviors among urban low inco me African American children and adolescents. Journal of Adolescent Health, 27, 43-48. Lieber, M.J. (2003). The contexts of juvenile justice d ecision making: When race matters. Albany, NY: State University of New York Press. Lightfoot, C. (1992). Constucting self and peer culture: A narrative perspective on adolescent risk-taking. In L.T. Wagner and J. Valsiner (Eds.), Children’s development within social c ontext: Research and methodology (pp. 229-245). Hillsdale, NJ: Lawrence Erlbaum Associates. Lo, C.C. (2000). Timing of drinking initi ation: A trend study predicting among high school seniors. Journal of Drug Issues, 30 (3), 525-554. Loeber, R. (1988). Natural histories of c onduct problems, delinquency, and associated substance use: Evidence for developmen tal progressions. In B.B. Lahey & A.E. Kazdin (Eds.), Advances in clinical child psychology (Vol. 11, pp. 73-124). New York: Plenum. Loeber, R. & Farrington, D.P. (Eds.). (1998). Serious & violent offending: Risk factors and successful interventions. Thousand Oaks, CA: Sage. Loeber, R., Green, S.M., Lahey, B.B., Fric k, P.J., McBurnett, K. (2000). Findings on disruptive disorders from the first deca de of the Develpmental Trends Study. Clinical Child and Family Review, 3, 37-60. Loeber, R., Stouthamer-Loeber, & Raskin-Wh ite, H. (1999). Developmental aspects of delinquency and internalizing problems a nd their association with persistent juvenile substance use between ages 7 and 18. Journal of Clinical Child Psychology, 28 (3), 322-332. Lofy, K.H., Hofmann, J., Mosure D.J., Fine, D.N., & Marrazzo, J.M. (2006). Chlamydial infections among female adolescents scr eened in juvenile detention centers in Washington State, 1998-2002. Sexually Transmitted Diseases, 33 (2), 63-67. Long, J.S. (1983). Confirmatory factor analysis Beverly Hills: Sage publications. Loue, S. (2006). Preventing HIV, eliminating disparities among Hispanics in the United States. Journal of Immigration and Health, 8, 313-318.

PAGE 216

206 Lowry, R., Holtzman, D., Truman, B.I., Kann, L ., Collins, J.L., & Kolbe, L.J. (1994). Substance use and HIV-related sexual behaviors among U.S. high school students: Are they related? American Journal of Public Health, 84, 1116-1120. Luster, T. & Small, S.A. (1994). Factors a ssociated with sexual risk-taking behaviors among adolescents. Journal of Marriage and the Family, 56, 622-632. MacDonald, J. & Chesney-Lind, M. (2001). Gender bias and juvenile ju stice revisited: A multi-year analysis. Crime & Delinquency, 47 (2), 173-195. MacKenzie, D.L. (2002). Reducing the crim inal activities of known offenders and delinquents: Crime prevention in the cour se and corrections. In L.W. Sherman, D.P. Farrington, B.C. Welsh, & D.L. MacKenzie (Eds.), Evidence Based Crime Prevention (pp. 330 – 404). New York: Routledge. Magura, S., Shapiro, J.L. & Kang, S.Y. (1994). Condom use among criminally-involved adolescents. AIDS Care, 6 (5), 595-603. Malow, R.M., Devieux, J.G., Jennings, T., Lucenko, B.A. & Kalichman, S.C. (2001). Substance-abusing adolescents at vary ing levels of HIV risk: Psychosocial characteristics, drug us e, and sexual behavior. Journal of Substance Abuse, 13, 103-117. Malow, R.M., Devieux, J.G., & Lucenko, B. A. (2006). History of childhood sexual abuse as a risk factor for HIV risk behavior. Journal of Psychological Trauma, 5 (3), 1332. Mann, A. (2003). Relationships matter: Impact of parental, peer factors on teen, young adult substance abuse. NIDA Notes, 18 (2). Retrieved January 17, 2007, from www.drugabuse.gov/NIDA_notes.html Marin, G. (1998). AIDS prevention among Hispanic s: Needs, risk behaviors, and cultural values. Public Health Reports, 104 (5), 411-415. Matsueda, R.L. & Heimer, K. (1987). Race, fam ily structure, and delinquency: A test of differential association and social control theories. American Sociological Review, 52 (6), 826-840. Mazzerolle, P., Brame, R., Paternoster, R ., Piquero, A., & Dean, C. (2000). Onset age, persistence, and offending versat ility: Comparisons across gender. Criminology, 38 (4), 1143-1172. McClelland, G.M., Elkington, K.S., Teplin, L.A., & Abram, K.M. (2004).Multiple substance use disorders in juvenile detainees. The Journal of the American Academy of Child and Adolescent Psychiatry, 43 (10), 1215-1224.

PAGE 217

207 McClelland, G.M., Teplin, L.A., & Abram, K.M. (2004). Detection and prevalence of substance use among juvenile detainees. Juvenile justice bulletin. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. McCutcheon, A.L. (1987). Latent class analysis. Newbury Park: CA: Sage. McGee, L. & Newcomb, M.D. (1992). General deviance syndrome: Expanded hierarchical evaluations at four ages from ear ly adolescent to adulthood. Journal of Consulting and Clinical Psychology, 60 (5), 766-776. McLoyd, V.C., Cauce, A.M., Takeuchi, D. & Wilson, L. (2004). Mar ital processes and parental socialization in families of color: A decade review of research. Journal of Marriage and Family, 62 (4), 1070-1093. Mears, D.P., Ploeger, M., & Warr, M. (1998) Explaining the gender gap in delinquency: peer influence and moral evaluations of behavior. Journal of Research in Crime and Delinquency, 35(3), 251-266. Menard, S., Mihalic, S., & Huizinga, D. (2001). Drugs and crime revisited. Justice Quarterly, 18 (2), 269-299. Mertz, K.J., Voigt, R.A., Hutchins, K ., & Levine, W.C. ( 2002). Findings from STD screening of adolescents and adul ts entering corrections facilities. Sexually Transmitted Diseases, 29 (12), 834-839. Metzler, C.W., Biglan, A., Ar y, D., Noell, J., & Smolkowski, K. (1993). The relationship among high-risk sexual behavior and other adolescent problem behaviors. Journal of Behavioral Medicine, 17, 419-438. Metzler, C.W., Noell, J., Biglan, A., Ar y, D., & Smolkowski, K. (1994). The social context for risky sexual behaviors among adolescents. Journal of Behavioral Medicine, 17, 419-438. Mezzich, A.C., Tarter, R.E., Giancoloa, P.R ., Lu, S., Kirisci, L. & Parks, S. (1997). Substance use and risky sexual be havior in female adolescents. Drug and Alcohol Dependence, 44, 157-166. Michael, K. & Ben-Zur, H. (2007). Risk-t aking among adolescents: Associations with social and affective factors. Journal of Adolescence, 30, 17-31. Millstein, S.G. & Moscicki, A.B. (1995). Sexually-transmitted disease in female adolescents: Effects of psychosoc ial factors and high risk behavior. Journal of Adolescent Health, 17 (2) 83-90.

PAGE 218

208 Moffit, T.E. (1993). Adolescencelimited and life-course persiste nt antisocial behavior: A developmental taxonomy. Psychological Review, 100 (4), 674-701. Moore, S. & Gullone, E. (1996). Predicting a dolescent risk behavior using a personalized cost-benefit analysis. Journal of Youth and Adolescence, 25, 343-359. Moore, S., Gullone, E., & Kostanski, M. (1997). An examination of adolescent risktaking using a story completion task. Journal of Adolescence, 20, 369-379. Morash, M. & Chesney-Lind, M. (1991). A refo rmulation and partial test of the power control theory of delinquency. Justice Quarterly, 8 (3), 347-377. Morris, R.E., Harrison, E.A., Knox, G.W., Tr omanhauser, E., Marquis, D.K. & Watts, L.L. (1995). Health risk behavioral surv ey from 39 juvenile correctional facilities in the United States. Journal of Adolescent Health, 17, 334-344. Morris, R.E., Baker, C.J., Valentine, M.V., & Pennisi, A.J. (1998). Variations in HIV risk behaviors of incarcerated juvenile s during a four-year period: 1989-1992. Journal of Adolescent Health, 23 (1), 39-48. Morrison, D.M., Gillmore, M.R., Hoppe, M.J., Gaylord, J., Leigh, B.C. & Rainey, D. (2003). Perspectives on Sexual and Reproductive Health, 35 (4), 162-168. Mosack, K.E., Gore-Felton, C., Chartier, M ., & McGarvey, E. (2007) Individual, peer, and family variables a ssociated with risky sexua l behavior among male and female incarcerated adolescents. Journal of HIV/AIDS Prevention & Education for Adolescents & Children, 8 (1), 115-133. Mosher, W.D. & McNally, J.W. (1991). Contrace ptive use at first premarital intercourse: United States, 1965-1988. Family Planning Perspectives 23 (3), 108-116. Mott, F. & Haurin, R. (1988). Linkages betwee n sexual activity and alcohol and drug use among American adolescents. Family Planning Perspectives, 20, 128-136. Muthn, B. (1991). Multilevel factor anal ysis of class and st udent achievement components. Journal of Educational Measurement, 28, 338-354. Muthn, B. (2006). Should substance use diso rders be considered as categorical or dimensional? Addiction, 101 (1), 6-16. Muthn, B., DuToit, S.H.C., & Spisic, D. ( 1997). Robust inference us ing weighted least squares and quadratic estimating equati ons in latent variable modeling with categorical and continuous outcomes. Psychometrika

PAGE 219

209 Muthn, B.& Kaplan, D. (1985). A comparison of some methodologies for the factor analysis of non-normal Likert variables. British Journal of Mathematical and Statistical Psychology 38, 171-189. Muthn, L. & Muthn, B. (2002). How to use a Monte Carlo Study to decide on sample size and determine power. Structural Equation Modeling, 9 (4), 599-620. Muthn, L.K. & Muthn, B.O. (2007). Mplus user’s guide (5th edition). Los Angeles, CA: Muthen & Muthen. Nader, P.R., Wexler, D.B., Patterson, T. L., McKusick, L., & Coates, T. (1989). Comparison about beliefs about AIDS among urban, suburban, incarcerated, and gay adolescents. Journal of Adolescent Health Care, 10, 413-418. Nagin, D.S. & Farrington, D.P. (1992). The onset and persistence of offending. Criminology, 30 (4), 501-524. Nagin, D.S. & Land, K.S. (1993). Age, criminal careers, and populat ion heterogeneity: Specification and estimation of a non parametric, mixed Poisson model. Criminology, 31, 327-362. Nakhaie, M., Silverman, R. & LaGrange, T. (1999). Self-control and social control: An examination of gender, ethnicity, class, and delinquency. Canadian Journal of Sociology, 25 (1), 35-59. National Campaign to Prevent Teen Pregnancy (2000). The next best thing: Encouraging contraceptive use among sexual active teens. Retrieved on March 3, 2008 from < www.teenpregnancy.org >. National Center of Addiction and Substance Abuse. (2002). Trends in substance use and treatment needs among inmates (Final Report to the National Institute of Justice). New York. National Center of Addiction and Substan ce Abuse (2004). Criminal neglect: Substance use, juvenile justice, and th e children left behind. New York. National Institute of Justice (2000). 1999 Annual report on drug use among adult and juvenile arrestees. Washington, DC: Office of Justice Programs. Neff, J.L. & Waite, D.E. (2007). Male vers us female substance abuse patterns among incarcerated juvenile offenders: Compari ng strain and social learning variables. Justice Quarterly, 24 (1), 106-132.

PAGE 220

210 Newcomb, M.D. & Bentler, P.M. (1986). Fre quency and sequence of serious drug use: A longitudinal study from early adolescence to young adulthood. Journal of Drug Education, 16, 101-120. Newcomb, M.D. & McGee, L. (1989). Adoles cent alcohol use a nd other delinquent behaviors: A one-year long itudinal analysis control ling for sensation seeking. Criminal Justice and Behavior, 16 (3), 345-369. Newcomb, M.D. & McGee, L. (1991). Infl uence of sensation seeking on general deviance and specific problem behavi ors from adolescence to young adulthood. Journal of Personality and Social Psychology, 61 (4), 614-628. Noell, J., Rohde, P., Seeley, J., & Ochs, L. (2001). Childhood sexual abuse, adolescent sexual coercion, and sexually transmitted infection acquisition among homeless female adolescents. Child Abuse & Neglect, 25 (1), 137-148. O’Donnell, B.L., O’Donnell, C.R., & Stue ve, A. (2001). Early sexual initiation and subsequent sex-related risks among urban minority youth: The Reach for Health study. Family Planning Perspective, 33 (6), 268.275. Office of Applied Studies. (2003). Results from the 2002 National Survey on Drug Use and Health: National findings. Rockville, MD: Substance Abuse and Mental Health Services Administration. Office of Applied Studies. (2004). Results from the 2003 National Survey on Drug Use and Health: National findings. Rockville, MD: Substance Abuse and Mental Health Services Administration. Office of Juvenile Justice Delinquency Preven tion (OJJDP). (1999). Mi norities in the juvenile justice system. Juvenile Justice Bulletin, 1999. Office of Juvenile Justice Deli nquency Prevention (OJJDP). (2008). Model Programs Guide Retrieved October 21, 2008 at http://www.dsgonline. com/mpg2.5/mpg_index.htm Office of National Drug Control Policy. (2004). Predicting heavy drug use. Washington, DC: Executive Office of the President (Publi cation No. NCJ 208382). Oh, K.M., Smith, K.R., O’Cain, M., Kilm er, D., Johnson, J., & Hook, E.W. (1998). Urine-based screening of adolescents in detention to guide treatment for Gonococcal and Chlamydial infections: Translating research into intervention. Archives of Pediatrics & Adolescent Medicine, 152 (1), 52-56.

PAGE 221

211 Orr, D.P., Langefeld, C.D., Katz, B.P., & Caine, V.A. (1996). Behavioral intervention to increase condom use among high-risk female adolescents. Journal of Pediatrics, 128, 288-295. Osgood, D.W. & Chambers, J.M. (2000). Social disorganization outside the metropolis: An analysis of rural youth violence. Criminology, 38 (1), 81-115. Osgood, D.W., Johnston, L.D, O’Malley, P.M. & Bachman, J.G. (1988). The generality of deviance in late adol escence and early adulthood. American Sociological Review, 53, 81-93. Osgood, D.W. & Schreck, C.J. (2007). A new method for stu dying the extent, stability, and predictors of individual specialization in violence. Criminology, 45(2), 273311. Otto, R.K., Greenstein, J.J., Johnson, M.K ., & Friedman, R.M. (1992). Prevalence of mental disease among youth in the juveni le justice system. In J.J. Cocozza (ed.), Responding to the mental health needs of youth in the juvenile justice system, (pp. 7-48). National Coalition fo r the Mentally Ill in the Criminal Justice System: Seattle. Otto-Salaj, L.L., Gore-Felton, C., McGarvey, E., & Canterbury, R.J. (2002). Psychiatric functioning and substance use: Fact ors associated with HIV risk among incarcerated adolescents. Child Psychiatry and Human Devleopment, 33 (2), 91106. Pack, R.P., DiClemente, R.J., Hook, E.W. & Oh, K.M. (2000). High prevalence of asymptomatic STDs in incarcerated mi nority male youths: A case for screening. Sexually Transmitted Diseases, 27 (3), 175-177. Paikoff, R.L. & Brooks-Gunn, J. (1991). Do pa rent-child relationships change during pregnancy? Psychological Bulletin, 110 (1), 47-66. Pan, E. & Parrell, M.P. (2006). Ethnic differenc es in the effects of intergener ational relations on adolescent problem behavi or in U.S. single-mother families. Journal of Family Issues, 27 (8), 1137-1158. Park, H.S. & Bauer, S. (2002). Parenting pract ices, ethnicity, socioeconomic status, and academic achievement in adolescents. School Psychology International, 23 (4), 386-395. Parsons, J.T., Halkitis, P.N., Bimbi, D. & Borkowski, T. (2000). Perceptions of the benefits and costs associated with condom use and unprotected sex among late adolescent college students. Journal of Adolescence, 23, 377-391.

PAGE 222

212 Parsons, J.T., Norman, L., Kocik, S., Butler, R., Nuss, R. & the CDC HBIEP Adolescent Study Group. (1998). The role of the fam ily system in HIV risk reduction: Adolescents with hemophilia and HIV-infection and their parents. Journal of Pediatric Psychology, 23, 57-65. Patterson, G.R. & Yoerger, K. (1993). Devel opmental models for delinquent behavior. In S. Hodgins (Ed.), Mental disorder and crime (pp. 140-172). Newbury Park, CA: Sage. Peeples, F. & Loeber, R. (1994). Do individu al factors and neighbor hood context explain ethnic differences in juvenile delinquency? Journal of Quantitative Criminology, 10, 141-157. Pinderhughes, E.E., Bates, J.E., Dodge, K.A., Petit, G.S., Zelli, A. (2000). Discipline responses: Influences of parents’ socio economic status, ethnicity, beliefs about parenting, stress, and cognitive-emotional processes. Journal of Family Psychology, 14 (3), 380-400. Pinkerton, S.D. & Abramson, P.R. (1995). Deci sion-making and personality factors in sexual risk-taking for HIV/AIDS: A theoretical integration. Personality and Individual Differences, 19 (5), 713-723. Piquero, A.R., Brezina, T., & Turner, M. G. (2005). Testing Moffitt’s account of delinquency abstention. Journal of Research in Crime and Delinquency, 42 (1), 27-54. Piquero, A.R., Paternoster, R., Mazerolle, P ., Brame, R., & Dean, C.W. (1999). Onset age and offense specialization. Journal of Research in Crime and Delinquency, 36 (3), 275-299. Piquero, A.R. & Tibbetts, S.G. (eds.). (2002). Rational choice and cr iminal behavior: Recent research and future challenges. London: Routledge. Potter, C.C & Jenson, J.M. (2003). Cluster prof iles of multiple problem youth: Mental health problem symptoms, substance use, and disorderly conduct. Criminal Justice and Behavior, 30 (2), 230-250. Pratt, T., & Cullen, F. (2000). The empirical status of Gottf redson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38 (3), 931-64. Priomos, J., Durant, R.H., Pierce, J.D., & Goodman, F. (1998). Gamb ling and other risk behaviors among 8th-12th grade students. Pediatrics, 102, 1-6. Prinz, R.J. & Kerns, S.E. (2003). Earl y substance use by juvenile offenders. Child Psychiatry and Human Development, 33 (4), 263-377.

PAGE 223

213 Prothrow-Smith, D. (1992). Can physicia ns help curb adolescent violence? Hospital Practice, 27, 193-207. Puzzanchera, C., Adams, B., Snyder, H., and Kang, W. (2007). Easy Access to FBI Arrest Statistics 1994-2005 Accessed February 10, 2008 at http://ojjdp.ncjrs.gov/ojstatbb/ezaucr/ Puzzanchera, C. & Kang, W. (2007). Juve nile Court Statistics Databook. Accessed February 10, 2008 at: http://ojjdp.ncjrs.gov/ojstatbb/jcsdb/ Ramisetty-Mikler, S., Caetano, R., Goebert, D., & Nishimura, S. (2004). Ethnic variation in drinking, drug use, and sexual be havior among adolescents in Hawaii. Journal of School Health, 74 (1), 16-22. Ramrakha, S., Bell, M.L., Paul, C., Dickson, N., Moffitt, T.E. & Caspi, A. (2007). Childhood behavior problems linked to sexual risk taking in young adulthood: A birth cohort study. Journal of the American Academy for Child and Adolescent Psychiatry, 46 (10), 1272-1279. Rashad, I. & Kaestner, R. (2004). Teenage sex, drugs, and alcohol use: Problems identifying the cause of risky behaviors. Journal of Health Economics, 23, 493503. Reardon, S.F., Brennan, R.T. & Buka, S.L. ( 2002). Estimating multi-level discrete time hazard models using cross-sectional da ta: Neighborhood effects on the onset of adolescent cigarette use. Multivariate Behavioral Research, 37 (3), 297-330. Rees, D.I., Argys, L.M., & Averett, S.L. (2001). New evidence on the relationship between substance use and adolescent sexual behavior. Journal of Health Economics, 20, 835-845. Reitman, D., St. Lawrence, J., & Jefferson, K. (1996). Predictors of African American adolescents’ condom use and HIV risk behavior. AIDS Education and Prevention, 8 (6), 499-515. Richter, D.L., Valois, R.F., McKeown, R.E ., & Vincent, M.L. (1993). Correlates of condom use and number of sexual part ners among high school adolescents. Journal of School Health, 63 (2), 91-96. Risser, J., Risser, W.L., Gefter, L.R., Br andstetter, D.M., & Cromwell, P.F. (2001). Implementation of a screening program fo r Chlamydial infecti on in incarcerated adolescents. Sexually Transmitted Diseases, 28 (1), 43-46.

PAGE 224

214 Robertson, A.A., Dill, P.L., Husain, J. & U ndesser, C. (2004). Prevalence of mental illness and substance abuse disorders among incarcerated juvenile offenders in Mississippi. Child Psychiatry and Human Development, 35 (1), 55-74. Robertson, A. & Levin, M.L. (1999). AIDS kno wledge, condom attitudes, and risk-taking sexual behavior of substance-abusing j uvenile offenders on probation or parole. AIDS Education and Prevention, 11 (5), 450-461. Robertson, J. A. & Plant, M.A. (1988). Alcohol, sex and risks of HIV infection. Drug and Alcohol Dependence, 22 (1-2), 75-78. Robertson, A.A., Thomas, C.B., St. Lawrence, J., & Pack, R. (2005). Predictors of infection with Chlamydia or Gonorrhea in incarcerated adolescents. Sexually Transmitted Diseases, 32 (3), 115-122. Rodgers, J.L. & Rowe, D.C. (1990). Adolescent sexual activity and mildly deviant behavior. Journal of Fam ily Issues, 11 (3), 274-293. Roeder, K. Lynch, K.G., & Nagin, D.S. (1999) Modeling uncertainty in latent class membership: a case study in criminology. Journal of the Am erican Statistical Association, 94 (447), 766-776. Rolf, J., Nanda, J., Baldwin, J., Chandra, A., & Thompson, L. (1991). Substance misuse and HIV/AIDS risks among deli nquents: A prevention challenge. The International Journal of Addictions, 25, 253-259. Rosay, A.B., Najaka, S.S. & Herz, D.C. ( 2007). Differences in the validity of selfreported drug use across five factors: Gender, race, age, type of drug, and offense seriousness. Journal of Quantitative Criminology, 23 (1), 41-58. Rosenbaum, E. & Kandel, D.B. (1990). Early onset of adolescent se xual behavior and drug involvement. Journal of Marriage and the Family, 52, 783-798. Rosenberg, M. & Mercy, J.A. (1991). Introductio n. In M.L. Rosenberg and M.A. Fanley (Eds.), Violence in America: A Public Health Approach (pp. 1-12). New York: Oxford University Press. Rosengard, C., Stein, L.A.R., Barnett, N.P ., Monti, P.M., Golembeske, C., & LebeauCraven, R. (2006). Co-Occurring sexual risk and substance use behaviors among incarcerated adolescents. Journal of Correctional Health Care, 12 (4), 279-287. Rosenthal, D.A., Smith, A., & De Visser, R. (1999). Personal and social factors influencing age at first sexual intercourse. Archives of Sexual Behavior, 28 (4), 319-333.

PAGE 225

215 Rotheram-Borus, M.J., O’Keefe, Z., Kracker R., & Foo, H.H. (2000). Prevention of HIV among adolescents. Prevention Science, 1 (1), 15-30. Rowe, D.C., Rodgers, J.L., Meseck-Busbey, S. & St. John, C. (1989). Sexual behaviors and nonsexual deviance: A sibli ng study of their relationship. Developmental Psychology, 25 (1), 61-69. Saint-Jean, G. & Crandall, L.A. (2004). Ethni c differences in the salience of risk and protective factors for alcohol and marijuan a: Findings from a statewide survey. Journal of Ethnicity in Substance Abuse, 3 (1), 11-27. Sampson, R. & Laub, J. (2003). Shared beginnings, divergent lives: Delinquent boys to age 70. Cambridge, MA: Harvard University Press. Santelli, J.S., Lindberg, L.D., Abma, J., Mc Neely, & Resnick, M. (2000). Adolescent sexual behavior: Estimates and trends from four nationally representative survey. Family Planning Perspective, 32 (4), 156-165. Scaramella, L.V., Conger, R.D., Simons, R. L., & Whitbeck, L.B. (1998). Predicting risk for pregnancy by late adolescence: A social contextual perspective. Developmental Psychology, 34, 1233-1245. Sealock, M.D., Gottfredson, D., & Gallagher, C.A. (1997). Drug treatment for juvenile offenders: Some good and bad news. Journal of Research in Crime and Delinquency, 34, 210-236. Seidman, S. N., & Rieder, R. O. (1994). A revi ew of sexual behavior in the United States. American Journal of Psychiatry, 151 330-341. Shafer, M.A. & Boyer, C.B. (1991). Psychosocia l and behavioral factors associated with risk of sexually transmitted diseases, including human immunodeficiency virus infection, among urban high school students. Journal of Pediatrics, 119 (5), 826833. Shafer, M. A., Hilton, J. F., Ekstrand, M., Ke ogh, J., Gee, L., DiGiorgio, L., Shalwitz, J., & Schachter, J. (1993). Relationship be tween drug use and sexual behaviors and the occurrence of sexually transmitted diseases among high-risk male youth. Sexually Transmitted Diseases, 20, 307-313. Shahmanesh, M., Gayed, S., Ashcroft, M., Smith, R., Roopnarainsingh, R., Dunn, J., & Ross, J. (2000). Sexually Transmitted Infections, 76, 268-272. Shaw, D.S., Wagner, E.F., Arnett, J., & Aber W.S. (1992). The factor structure of the reckless behavior questionnaire. Journal of Youth & Adolescence, 21 (3), 305-323.

PAGE 226

216 Sheer, V.C. & Cline, R.J. (1994). The devel opment and validation of a model explaining sexual behavior among college students : Implications for AIDS communication campaigns. Human Communication Research, 21, 280-304. Shrier, L.A., Emans, S.J., Woods, E.R., & DuRant, R.H. (1997). The association of sexual risk behaviors and problem dr ug behaviors in high school students. Journal of Adolescent Health, 20 (5), 377-383. Simons, R.L., Miller, M.G., & Aigner, S.M. ( 1980). Contemporary th eories of deviance and female delinquency: an empirical test. Journal of Research in Crime and Delinquency, 17 (1), 42-57. Simons, R, Whitbeck, L.B., Conger, R.D. & K. J. Conger. (1991). Parenting factors, social skills, and value commitments as precursors to school failure, involvement with deviant peers, a nd delinquency behavior.” Journal of Youth and Adolescence 20(6), 645-663. Smith, C. (1997). Factors associated with ear ly sexual activity among urban adolescents. Social Work, 42 (4), 334-346. Smith, D. & Paternoster, R. (1987). The gender gap in theories of de viance: issues and evidence. Journal of Research in Crime and Delinquency, 24(2), 140-172. Snyder, H.N. (2006). Juvenile Arrests 2004. Washington, DC: Office of Juvenile Justice Delinquency and Prevention. Spitalnick, J.S., DiClemente, R.J., Wingood, G.M., Crosby, R.A., Milhausen, J.M., McCarty, F., Rose, E. & Younge, S.N. (2007). Brief report: Sexual sensation seeking and its relationship to risky sexual behavior among African-American adolescent females. Journal of Adolescence, 30, 165-173. Spohn, C. & Holleran, D. (2002). The effect of imprisonment on recidivism rates of felony offenders: A focus on drug offenders. Criminology, 40, 329-358. St. Lawrence, J.S. (1993). African-Ameri can adolescents’ know ledge, health-related attitudes, sexual behavior, and contr aceptive decisions: Implications for the prevention of adolescent HIV infection. Journal of Consulting and Clinical Psychology, 61, 104-112. St. Lawrence, J.S., Brasfield, T.L., Jeffe rson, K.W., Alleyene, E., O’Bannon, R.E., & Shirley, A. (1995). Cognitive-behavioral in tervention to reduce African American adolescents’ risk for HIV infection. Journal of Consulting and Clinical Psychology, 63, 221-237.

PAGE 227

217 St. Lawrence, J.S., Crosby, R.A., O’Bannon, R.O. (1999). Adolescent risk for HIV infection: Comparison of four high-risk samples. Journal of HIV/AIDS Prevention for Adolescents and Children, 3, 63-85. Stamatis, D.H. (2001). Six sigma and beyond: Statistics and probability. Boca Raton, FL: CRC Press. Stenmark, D.E., Wackwitz, J.H., Peffrey, M.C., & Dougherty, F. (1974). Substance use among juvenile offenders: Relationships to parental substance use and demographic characteristics. Addictive Diseases: An International Journal, 1 (1), 43-54. Stevens, S.J., Estrada, B., Murphy, B.S., McKnight, K.M.,& Tims F. (2004). Gender differences in substance use, mental h ealth, and criminal justice involvement of adolescents at treatment entry and a three, six, twel ve, and thirty-month follow up. Journal of Psychoactive Drugs, 36 (1), 13-25. Stevens, S.J., Hasler, J., Murphy, B.S., Tayl or, R., Senior, M., Ba rron, M., et al. (2003). La Canada adolescent treatment program: Addressing issues of drug use, gender, and trauma. In S.J. Stevens, & A.R. Morral (Eds.), Adolescent substance abuse treatment in the United States: Exempl ary models from a national evaluation study (pp. 183-209). New York: Haworth Press Stevens, S.J., Murphy, B.S. & McKnight, K. (2003). Traumatic stress and gender differences in relationship to substance abuse, mental health, physical health, and HIV risk behavior in a sample of ad olescents enrolled in drug treatment. Child Maltreatment, 8 (1), 46-57. Stice, E., Myers, M.G., & Brown, S.A. (1998) Relations of delinquency to adolescent substance use and problem use: A prospective study. Psychology of Addictive Behaviors, 12 (2), 136-146. Stouthamer-Loeber, M. & Wei, E.H. (1998) The precursors of young fatherhood and its effects on delinquency of teenage males. Journal of Adolescent Health, 22, 56-65. Strunin, L. & Hingson, R. ( 1992). Alcohol, drugs, and a dolescent sexual behavior. International Journal of the Addictions, 27, 129-146. Sullivan, C.J., McGloin, J.M., & Pratt, T.C. (2006). Rethinking the “norm” of offender generality: Investigating specialization in the short-term. Criminology, 44 (1), 199233. Tapert, S.F., Aarons, G.A., Sedlar, G.R., & Brown, S.A. (2001). Adolescent substance use and sexual risk-taking behavior. Journal of Adolescent Health, 28 (3), 181189.

PAGE 228

218 Tarter, R.E., Vanyukov, M. & Kirisci, L. ( 2006). Predictors of marijuana use in adolescents before and after licit dr ug use: Examination of the gateway hypothesis. American Journal of Psychiatry, 163 (12), 2134-2140. Teplin, L., Abram, K., McClel land, G., & Dulcan, M. (2002) Psychiatric disorders in youth in juvenile detention. Archives of General Psychiatry, 59 1133-1143. Teplin, L.A., Elkington, K.S., McClelland, G.M., Abram, K.M., Mericle, A.A. & Washburn, J.J. (2005). Major mental di sorders, substan ce use disorders, comorbidity, and HIV-AIDS risk be haviors in juvenile detainees. Psychiatric Services, 56 (7), 823-828. Teplin, L.A., Mericle, A.A., McClelland, G. M., & Abram, K.M. (2003). HIV and AIDS risk behaviors in juvenile detainees: Implications for pub lic health policy. American Journal of Public Health, 93 (6), 906-912. Thornberry, T.P. (1987.) Toward an in teractional theory of delinquency. Criminology, 25 (4), 863-891. Thornberry, T.P. (1996). Empirical support for interactional theory: A review of the literature. In J. David Hawkins (Ed.), Delinquency and Crime: Current Theories (pp. 198-235). Cambridge, UK: Cambridge University. Thornberry, T.P., Huizinga, D., & 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, & J. Wilson (Eds.), Sourcebook on serious, violent and chronic juvenile offenders (pp. 213-237). Thousand Oaks, CA: Sage. Timmermans, M., Van Lier, P. A. & Koot, H.M. (2007). Whic h forms of child/adolescent externalizing behaviors account for late adolescent risky se xual behavior and substance use? Journal of Child Psychology & Psychiatry. Timmons-Mitchell, J., Brown, C., Schulz, C., Webster, S.E., Underwood, L.A. & Semple, W.E. (1997). Comparing the need s of female and male incarcerated juvenile delinquents. Behavioral Sciences and the Law, 15, 195-202. Tittle, C.R., Ward, D.A., & Grasmick, H.G. ( 2003). Gender, age, and crime: A challenge to selfcontrol theory. Journal of Research in Crime and Delinquency, 40 (4), 426-453. Tolou-Shams, M., Brown, L.K., Gordon, G., & Fe rnandez, I. (2007). A rrest history as an indicator of adolescent/young a dult substance use and HIV risk. Drug and Alcohol Dependence, 88, 87-90.

PAGE 229

219 Tracy, P.E. & Kempf, K. (1996). Continuity and discontinuity in criminal careers. New York: Plenum Press. Tracy, P.E., Wolfgang, M.E. & Figlio, R.M. (1990). Delinquency careers in two birth cohorts. New York: Plenum. Tubman, J.G., Gil, A.G., Wagner, E.E. & Ar tigues, H. (2003). Patterns of sexual risk behaviors and psychiatric disorders in a community sample of young adults. Journal of Behavioral Medicine, 26, 473-500. Tubman, J.G., Gil, A.G., & Wagner, E.F. (2004). Co-occurring substance use and delinquent behavior du ring early adolescence. Criminal Justice and Behavior, 31, 463-488. Tubman, J.G., Windle, M., & Windle, R. C. (1996). Cumulative sexual intercourse patterns among middle adolescents: Problem behavior precursors and concurrent health risk behavior. Journal of Adolescent Health, 18, 182-191. Tucker, L. R., & Lewis, C. (1973). A relia bility coefficient for maximum likelihood factor analysis. Psychometrika, 38 1-10. Turner, M.G., Hartman, J.L., & Bishop, D.M. (2007). The effects of prenatal problems, family functioning, and neighborhood disa dvantage in predicting life-course offending. Criminal Justice and Behavior, 34 (10), 1241-1261. Tyler, K.A., Whitbeck, L.B., Hoyt, D.R. & Yoder, K.A. (2000). Predictors of selfreported sexually transmitted diseases among homeless and runaway adolescents. Journal of Sex Research, 37 (4), 369-377. Ullman, J.B. & Newcomb, M.D. (1999). The tran sition from adolescent to adult: A time of change in general and specific deviance. Criminal Behavior and Mental Health, 9, 74-90. U.S. Department of Commerce, Bureau of the Census. (2000). Census 2000, Summary file 3-sample data ( Hillsborough County ). http://www.census.gov Van Kammen, W.B. & Loeber, R. (1994). Ar e fluctuations in delinquent behaviors related to the onset and offset of ju venile illegal drug us e and drug dealing? Journal of Drug Issues, 24 9-24. Van Kammen, W.B., Loeber, R., & Stouthamer-L oeber, M. (1991). Substance use and its relationship conduct problems and delinquency in young boys. Journal of Youth and Adolescence, 20 (4), 399-413.

PAGE 230

220 Vaughn, M.G., Freedenthal, S., Jenson, J., & Howard, M.O. (2007). Psychiatric symptoms and substance use among juvenile offenders. Criminal Justice and Behavior, 34 (10) 1296-1312. Vaughn, M.G., Wallace, J.M., Davis, L.E., Fernandes, G.T., & Howard, M.O. (2007). Variations in mental health proble ms, substance use and delinquency between African American and Caucasian juvenile offenders. International Journal of Offenders Therapy and Comparative Criminology. Vazsonyi, A.T., Trejos-Castillo, E., & Young, M.A. (2007). Rural and non-rural African American youth: Does context matter in the etiology of problem behaviors? Journal of Youth & Adolescence, 36 (8). Vingilis, E. & Adlaf, E. (1990). The structur e of problem behavior among Ontario high school students: A confirmatory factor analysis. Health Education Research, 5 (2), 151-160. Voisin, D.R. (2005). The relationship between violence exposure and HIV sexual risk behaviors: Does gender matter? American Journal of Orthopsychiatry, 75 (4), 497-506. Wagner, M.K. (2001). Behavioral characterist ics related to substa nce abuse and risktaking, sensation-seeking, anxiety sensitivity, and self-reinforcement. Addictive Behaviors, 26, 115-120. Wagner, E.F. & Kassel, J.D. (1995). Substance use and abuse. In R.T. Ammerman & M. Hersen (Eds.), Handbook of child behavior therapy in the psychiatric setting (pp. 367-388). New York: John Wiley. Wagner, E.F. (1996). Substance use and violent behavior in adolescence. Aggression and Violent Behavior: A Review Journal, 1, 375-387. Wallace, J.M. & Muroff, J.R. (2002). Prev enting substance abuse among African American children and youth: Race differe nces in risk factor exposure and vulnerability. The Journal of Primary Prevention 22 235-261. Waller, M.W., Hallfors, D.D., Halpern, C.T., Iritani, B.J., Ford, C.A., & Guo, G. (2006). Gender differences in associations betw een depressive symptoms and patterns of substance use and risky sexual behavior among a nationally representative sample of U.S. adolescents. Archives of Women’s Mental Health, 9, 139-150. Warr, M. (2002). Companions in crime: The social aspects of criminal conduct. Cambridge: Cambridge University Press.

PAGE 231

221 Warren, C.W., Kann, L., Leavy Small, M., Santelli, J.S., Collins, J.L., & Kolbe, L.J. (1997). Age of initiating selected he alth-risk behaviors among high school students in the United States. Journal of Adolescent Health, 21 (4), 225-231. Wei, Z., Makkai, T., & McGregor, K. (2003) Drug use among a sample of juvenile detainees. Trends and Issues in Crime and Criminal Justice, June 2003. Australian Institute of Criminology. Welsh, B.C. (2005). Public health and the pr evention of juvenile criminal violence. Youth Violence and Juvenile Justice, 3, 23-40. Welte, J.W., Barnes, G.M., Hoffman, J.H., & Dintcheff, B.A. (1999). Trends in adolescent alcohol and other substance use: Relationships to trends in peer, parent, and school influences. Substance Use and Misuse, 34, 1427-1449. Welte, J.W., Barnes, G.M., & Hoffman, J.H. (2004). Gambling, substance use, and other problem behaviors among youth: A test of general deviance models. Journal of Criminal Justice, 32, 297-306. Welte, J.W., Barnes, G.M., Hoffman, J.H., Wieczorek, W.F., & Zhang, L. (2005). Substance involvement and the trajector y of criminal offending in young males. The American Journal of Drug and Alcohol Abuse, 31, 267-284. White, H.R. (1992). Early problem be haviors and later drug problems. Journal of Research in Crime & Delinquency, 29 (4), 412-429. White, H.R. (1997). Alcohol, illicit drugs, and vi olence. In D. Stoff, J. Brieling, & J.D. Masser (Eds.), Handbook of antisocial behavior (pp. 511-523). New York:Wiley. White, H.R., & Labouvie, E.W. (1994). Genera lity vs. specificity of problem behavior: Psychological and functional differences. Journal of Drug Issues, 24, 55-74. White, H.R., Loeber, R., Stouthamer-Loe ber, M., & Farrington, D.P. (1999). Developmental associations between substance use and violence. Development and Psychopathology, 11, 785-803. Widaman, K.F. & Reise, S.P. (1997). Expl oring the measuremen t invariance of psychological instruments: Applications in the substance use domain. In K.J. Bryant, M. Windle, and S.B. West (Eds.), The Science of Prevention. Washington, DC: American Psychological Association. Williams, J.H., Ayers, C., Abbott, R.D., Hawkins, J.D., & Catalano, R.F. (1996). Structural equivalence of involvement in problem beha vior by adolescent across racial groups using multiple group confirmatory factor analysis. Social Work Research, 20 (3), 179-188.

PAGE 232

222 Wilson, J.M. & Donnermeyer, J.F. (2006). Urba nity, rurality, and a dolescent substance use. Criminal Justice Review, 31 (4), 337-356. Windle, M., Iwawaki, S., & Lerner, R.M. (1988). Cross-cultural comparability of temperament among Japanese and American preschool children. International Journal of Psychology, 23, 547-567. Winters, K.C., Stinchfield, R.D., Botzet, A. & Anderson, N. (2002). A prospective study of youth gambling behaviors. Psychology of Addictive Behaviors, 16, 3-9. Winters, K.C., Weller, C.L., & Meland, J.A. (1993). Extent of drug abuse among juvenile offenders. Journal of Drug Issues, 23 (3), 515-525. Wood, V.D. & Shoroye, A. ( 1993). Sexually transmitted disease among adolescents in the juvenile justice system of the District of Columbia. Journal of National Medical Association, 85 (6), 435-439. Wulfert, E. & Wan, C.K. (1993). C ondom use: A self-efficacy model. Health Psychology, 12, 346-353. Yamaguchi, K., & Kandel, D.B. (1987). Drug us e and other determinants of premarital pregnancy and its outcome. Journal of Marriage and the Family, 49, 257-290. Yan, A.F., Yu-Wen, C., Stoesen, C.A., & Wa ng, M.Q. (2007). STD/HIV-related sexual risk behaviors and substance use among US rural adolescents. Journal of National Medical Association, 99 (12), 1386-1394. Yu, C.Y. & Muthen, B. (2001). Evaluation of model fit indice s for latent variable models with categorical and continuous outcomes (Technical Report). Los Angeles: University of California, Los Angele s, Graduate School of Education and Information Studies. Zhang, Z. (2004). Drug and alcohol use and related matters among arrestees, 2003. Washington, DC: U.S. Department of Ju stice, National Institute of Justice. Zhang, L., Wieczorek, W.F., & Welte, J.W. ( 1997). The impact of age of onset of substance use on delinquency. Journal of Research in Crime and Delinquency, 34, 253-269. Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal. Hillsdale, NJ: Erlbaum.

PAGE 233

About the Author Kristina K. Childs received her Bachelor s of Science in Psychology and Criminal Justice from Michigan State University in 2002 and her Mast ers of Arts in Criminology from the University of South Florida in 2005. She has worked on a number of research projects involving delinquent and nondelinque nt adolescents. Her research interests include risky sexual behavior and sexually tr ansmitted diseases among juvenile offenders and juvenile justice system reform. Sh e is currently conducting a post doctoral fellowship with the University of New Orl eans, Department of Psychology. This project involves the Louisiana Models for Change ini tiative, which is res ponsible for both state and parish-level juvenile justice system reform.