Reducing gang involvement through employment: a pilot intervention

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content

REDUCING GANG INVOLVEMENT THROUGH EMPLOYMENT: A PILOT
INTERVENTION

by

Dawn Delfin McDaniel

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)

August 2010

Copyright 2010 Dawn Delfin McDaniel

ii

Dedication
To my parents Lilian Delfin and Thomas McDaniel, and my best friend William Arnold.

iii

Acknowledgments
A number of people have been instrumental in improving the quality of this
dissertation and in bringing it to completion. I extend my particular thanks to my
advisor, Dr. Stanley Huey for his extensive, thoughtful feedback throughout this process.
Also, I would like to thank Ishwar Bridgelal, Amy Shayne, and Sami Klebanoff for years
of dedication as research assistants, Ross Loomis for his statistical advice, and Caitlin
Smith for keeping the BEP torch burning. I would also like to acknowledge Dr. Janet
Schneiderman for her work as a committee member, and also her mentorship. Finally, I
would like to thank my labmates, and Dave Pan in particular for his constant
encouragement.

iv

Table of Contents
Dedication ii
Acknowledgments iii
List of Tables v
List of Figures vi
Abstract vii
Chapter 1: Introduction 1
Chapter 2: Method 14
Chapter 3: Results 30
Chapter 4: Discussion 45
Bibliography 57
Appendices
Appendix A: Session Data for BEP Youth
Appendix B: Example of Full Model

66
79

v

List of Tables
Table 1: Demographic Characteristics at Pretreatment 31
Table 2: Means, Standard Deviations, and Ranges for Within- and Between-
Subjects Analyses

33
Table 3: Correlation Matrix for Between-Subjects Analyses 36
Table 4: Summary of Regression Equations for Mediation Model (n = 25) 41
Table B1: Full Model for Gang Activity
w
79

vi

List of Figures
Figure 1: Flow Chart of Sample through Recruitment, Allocation, and Analyses 16
Figure 2: Esbensen et al.'s (2001) Gang Activity Measure 21
Figure 3: Mediation Model for Between-Subjects Analyses 27
Figure 4: Means for Within-Subjects Hours Employed and Gang Measures
Across Active Phase of BEP

35
Figure 5: Within-Subjects Hours Employed Predicting Changes in the Probability
of Gang Activity

38
Figure A1: BEP Session Data for Youth 1

66
Figure A2: BEP Session Data for Youth 2

67
Figure A3: BEP Session Data for Youth 3 68
Figure A4: BEP Session Data for Youth 4 69
Figure A5: BEP Session Data for Youth 5 70
Figure A6: BEP Session Data for Youth 6 71
Figure A7: BEP Session Data for Youth 7 72
Figure A8: BEP Session Data for Youth 8 73
Figure A9: BEP Session Data for Youth 9 74
Figure A10: BEP Session Data for Youth 10 75
Figure A11: BEP Session Data for Youth 11 76
Figure A12: BEP Session Data for Youth 12 77
Figure A13: BEP Session Data for Youth 13 78

vii

Abstract
Gang-related activity exacts a heavy toll in terms of morbidity and mortality, law
enforcement and corrections resources, and quality of life. Previous intervention efforts
to reduce gang involvement have had negligible effects, and current knowledge about the
processes that redirect youth away from gangs is limited. This dissertation attempts to
contribute to our understanding of gang-related treatment process by examining the
relationship between employment and gang involvement within the context of a pilot
employment-focused intervention for gang-involved youth. A sample of 27 gang-
involved, juvenile offenders were recruited from the Los Angeles Department of
Probation and randomly assigned to either the Behavioral Employment Program (BEP) or
usual probation services (US). Both within-subjects analyses (i.e., pooled time series)
and between-subjects analyses (i.e., multiple regression) examined the impact of BEP-
initiated employment on gang involvement. Increased employment was significantly
related to reductions in gang involvement for within- and between-subjects analyses.
Although no association between treatment condition and increased employment was
found, BEP led to marginally significant reductions in gang involvement at 6-months
post-entry. Employment programs for gang-involved youth are common in community
settings, although empirical support for such programs is rare. Results from this study
offer initial support for BEP and the continued use of employment as a strategy to reduce
gang involvement. Confirmation of these promising findings will require future research
with larger samples.

1
Chapter 1: Introduction
In samples of high-risk youth in large U.S. cities, gang involvement ranges from
14% to 31% (Thornberry, 1998; Thornberry, Huizinga, & Loeber, 2004). Longitudinal
research demonstrates that gang members are more likely than high-risk, non-gang
members to engage in criminal and violent acts (Gatti, Tremblay, Vitaro, & McDuff,
2005; Thornberry, Krohn, Lizotte, & Smith, 2003), exposing these youth to increased risk
of physical violence and death (Peterson, Taylor, & Esbenson, 2004; Spano, Freilich, &
Bolland, 2008). Gang-related violence is particularly devastating in Los Angeles (LA)
County. Historically, gangs account for about 50% of homicides in the county (Tita &
Abrahamse, 2004). Describing this problem as an epidemic, Hutson, Anglin, Kyriacou,
Hart, and Spears (1995) conclude that gang-related homicide and violence represent a
significant public health concern.
In LA, 94% of gang members are either Latino (57%) or African American (37%;
Winton, 2005). The rates are similar across the U.S. with Latinos and African Americans
accounting for 47% and 31% of gang-involved youth, respectively (Egley, 2002; Klein &
Maxson, 2006). Not surprisingly, Latino and African American youth are most often the
victims (50% and 39%, respectively) and perpetrators (48% and 36%, respectively) of
gang-related homicides in LA (Aryan, Jandial, Bennett, Masri, Lavine & Levy, 2005;
Pizarro & McGloin, 2006; Tita & Abrahamse, 2004). Nationwide, homicide is the
leading cause of death for African American males ages 15-24, and it is the second
leading cause of death for Latino males of the same age (Centers for Disease Control,
2007).

2

To address these problems, interventionists have developed programs to reduce
gang involvement and its aftermath (Howell, 2000; Spergel, 1995). Literature reviews of
these interventions find mostly null or indeterminate results (Howell, 2000; Stinchcomb,
2002), which Howell (2000) attributes to a lack of understanding of the processes that
divert youth from gangs. Nevertheless, several interventions appear promising,
especially those that focus on job placement and training (Howell, 2000). These
programs are consistent with theoretical frameworks arguing that increases in social
opportunities, such as employment, should lead to reductions in gang involvement.
However, little is known about the effectiveness of these types of programs since
rigorous evaluation is rare (Howell, 2000).
This dissertation examines the process of gang desistence within the context of an
ongoing employment program for gang-involved youth, the Behavioral Employment
Program (BEP). By examining processes underlying BEP, this dissertation provides a
clearer picture of the relationship between BEP-initiated employment and gang
involvement. Such research is critical for intervention efforts as it increases our
understanding of the treatment mechanisms accelerating gang desistence. To provide a
rationale for this work, the next section delineates the current state of our knowledge on
gang desistance. The study will then review the relationship between employment and
gang involvement. The final section will discuss the rationale for using employment-
based treatment as a method for reducing gang involvement.

3

Gang Desistance
While the literature is replete with investigations on risk factors for joining a
gang, very little is known about the process of leaving a gang. Contrary to the stereotype
of gang membership as a lifelong commitment, research demonstrates that gang
membership is often transitory (Decker & Lauritsen, 2002; Esbensen & Huizinga, 1993;
Thornberry et al., 2003; Vigil, 1988). Indeed, a nationwide study of 1,000 gang members
found that 44% of youth surveyed have attempted to leave the gang (Knox et al., 1995).
Longitudinal studies demonstrate that many youth move in and out of gangs and that the
length of gang membership averages two years or less (Esbensen & Huizinga, 1993;
Klein & Maxson, 2006; Thornberry et al., 2003; Thornberry, 1997). In addition, gang
participation rapidly increases in early adolescence and gradually declines in early
adulthood (Seals, 2009), similar to patterns of delinquency across the lifespan (Hirschi &
Gottfredson, 1983).
In-depth interviews and observations of ex-gang members provide the best
illustration of the process of gang desistance (Decker & Lauritsen, 2002; Moloney,
Mackenzie, Hunt, & Joe-Laidler, 2009; Sanchez-Jankowski, 1991; Vigil, 1988). For
instance, Vigil (1988) describes gang exodus as a process of maturing-out of a gang
through a series of steps related to new commitments and responsibilities. Offering
further support for this perspective, Moloney and colleagues (2009) illustrate how gang-
involved youth reorganize their time, re-orient their setting away from the streets, and re-
evaluate their priorities upon taking on new responsibilities. These authors note a
twofold process to leaving a gang. First, the youth's attitudes change. Then, there are

4

gradual behavioral changes (Moloney et al., 2009). Overall, this ethnographic work
suggests that naturalistic changes in gang involvement are likely to come slowly and
follow new life responsibilities.
There is an additional body of work that examines reasons why youth leave the
gang, providing more detail about the types of new life responsibilities that encourage
gang desistence. These studies commonly describe taking on new adult roles, such as
obtaining employment, getting married, and having a girlfriend or children, as reasons
youth cite for leaving (Spergel, 1990; Vigil, 1988). In addition, youth report violence,
gang-related victimization, family concerns, and incarceration as reasons why they leave
the gang (Decker & Lauritsen, 2002; Sanchez-Jankowski, 1991; Spergel, 1990). These
studies highlight that gang desistance often occurs as the result of normative maturational
processes, such as jobs, marriage, and family. Spergel (1990) provides another
explanation. He writes that "battle fatigue" occurs for youth and their families from the
tolls of violence, victimization, incarceration, and arrests (Spergel, 1990).
Gang desistence research provides valuable information for developing gang
interventions. While this research has been limited, it is a first step in clarifying the
process of gang exodus. These studies demonstrate that gang involvement fluctuates
over time and that reductions in gang involvement result from naturalistic developmental
processes. While many of these factors are difficult to change in an intervention context,
employment appears to be the most promising focus of intervention. Other researchers
agree (Caldwell & Altschuler, 2001; Spergel, 1990; Vigil, 1988). Informed by gang
desistance research, Caldwell and Altschuler (2001) write "career and employment

5

options, job skill development, finances, extended family responsibilities and
attachments, and inferential thinking" should be the focus of interventions for gang-
involved youth in late adolescence (p. 31). Taken together, this body of work warrants a
closer examination of the association between employment and gang involvement.
Employment and Gang Involvement
Considerable research supports the link between unemployment and gang-related
activity (Curry & Thomas, 1992; Fagan, 1990; Hagedorn, 1988; Huff, 1990; Klein, 1995;
Seals, 2009; Thornberry et al., 2003; Vigil, 1988). In an examination of metropolitan
areas across the U.S., Wells and Weisheit (2001) found that increases in the
unemployment rate were related to the presence of gangs. Similarly, for youth of a legal
age to work, Seals (2009) discovered that a change from 5-10% in the local
unemployment rate corresponded to a 34% increase in the predicted probability of gang
participation. Consistent with this correlational work, Knox et al. (1995) found that 25%
of surveyed gang members report that the reason they joined a gang was "to make
money" (p. 58).
The relationship between employment and gang involvement is complex.
Although the aforementioned research suggests that youth unemployment may lead to
gang involvement, other work suggests that gang involvement contributes to youth
unemployment as well. Indeed, entry into the labor market is difficult for gang-involved
youth for a number of reasons (Klein, 1995; Seals, 2009). First, gang-involved youth
more often engage in antisocial behavior, which often leads to police contact, arrests, and
criminal records (Thornberry et al., 2003). Experimental research shows that job

6

applicants with criminal records receive dramatically fewer job callbacks and interviews
than non-felon applicants with identical credentials (Pager, 2007; Pager, Western, &
Bonikowski, 2008). Second, even when youth disengage from the gang, the youth’s
history may leave a negative imprint on the community, police, rival gangs, and their
own gang's perceptions of this youth (Decker & Lauritsen, 2002), which may in turn limit
the youth’s employment opportunities and ability to maintain employment. For example,
police or rival gang members may stop the youth on his or her way to work, making it
difficult for this youth to be a reliable employee. This description is consistent with a
recent study of gang-involved youth that found that employment is a risk factor for
victimization (Spano, Freilich, & Bolland, 2008). Third, the gang constrains the
behaviors of its members by monopolizing their social affiliations and activities, and
limiting access to prosocial networks that may facilitate access to employment
(Thornberry et al., 2003). Fourth, the individual image these youth develop to fit into the
gang (e.g., tattoos, dress, and demeanor) may hamper their ability to fit into a
conventional work environment. Finally, surveys show that gang-involved youth
experience weak school attachment, educational frustration, low levels of school
participation, less educational attainment, and school failure (Curry & Spergel, 1992;
Knox et al., 1995; Seals, 2009). This lack of academic success may limit the type
employment available to these youth (Spergel, 1990).
In summary, research suggests a reciprocal relationship between employment and
gang involvement (Klein, 1995). While unemployment appears to facilitate gang
involvement, gang-involved youth also face serious barriers to sustained employment.

7

To further clarify the relationship between employment and gang involvement, the next
section provides theoretical explanations for these findings.
Explanatory Theories of Gang Involvement
Starting in the 1950s, many sociological theories were developed to explain the
existence and persistence of gangs (Goldstein, 1991). The two most prominent are social
control theory (Hirschi, 1969) and strain theory (Cloward & Ohlin, 1960). While social
control theory attempts to explain why some youth do not join gangs, strain theory
attempts to explain why some youth are gang-involved (Goldstein, 1991). These theories
have many similarities. First, both theories view gang involvement as a normal response
to diminished social opportunities in these youths' neighborhoods (Cloward & Ohlin,
1960; Hirschi, 1969). Second, these theories argue that altering a youth’s social
conditions, for instance through employment opportunities, will mitigate gang
involvement. Finally, according to both theories, gangs and employment have a similar
socializing function for youth; both gangs and employment provide youth with
opportunities for social support, social control, social status, and fiscal independence
(Spergel, 1995).
Increasingly, researchers are using psychological theories such as social learning
theory, alone or in combination with sociological theory, to explain gang involvement
(Esbensen & Huizinga, 1993; Esbensen, Winfree, He, & Taylor, 2001; Goldstein, 1991;
Winfree, Backstrom, & Mays, 1994). Social learning theory postulates that youth learn
antisocial values vicariously from observing other gang-involved youth (Goldstein,
1991). To reduce gang involvement, social learning theory argues for providing youth

8

with an alternative social context that promotes prosocial learning, such as a workplace
where fellow employees reject antisocial attitudes (Bendixen, Endresen, & Olweus, 2006;
Thornberry et al., 2003; Wright & Cullen, 2004).
Social control, strain, and social learning theories all focus on the youth's social
context, and each suggests that employment can serve as a catalyst for positive change in
antisocial behavior. However, these theories implicate very different mechanisms in their
explanations of how employment reduces gang involvement. Social control theory
emphasizes how employment fulfills the youth's needs and strengthens their bond to
conventional ideas and institutions, making gang involvement unnecessary and
conflictual with their beliefs (Hirschi, 1969). Strain theory emphasizes how employment
provides a legitimate opportunity for the youth to obtain status and income in their
community, which means that the youth no longer needs the gang to fulfill their
economic aspirations (Cloward & Ohlin, 1960). Finally, social learning theory
emphasizes how employment contributes to the youth developing new social supports
who may reinforce prosocial behaviors while reproving antisocial acts that promote gang
involvement (Wright & Cullen, 2004).
According to these theories, employment addresses many of the social concerns
that perpetuate gang involvement. For this reason, interventions often use employment as
a strategy to reduce delinquency and gang involvement. The following section will
briefly review the empirical literature on this topic.

9

Employment Programs for Delinquent and Gang-Involved Youth
A recent meta-analysis of 52 controlled evaluations of employment-related
programs for delinquent youth revealed that the treatment effects on youth recidivism
were small, with a mean odds ratio effect size of 1.31 (McDaniel, Bridgelal, & Huey,
2010). However, moderator analyses showed dramatic variation in outcomes across
studies (McDaniel et al., 2010). Generally, programs that promoted job retention through
continued monitoring of youth after job placement produced the most robust effects
(McDaniel et al., 2010). In addition, program effectiveness varied by risk and
racial/ethnic status, suggesting that employment-related programs may need to make
adaptations in order to meet the needs of the diverse populations they serve (McDaniel et
al., 2010).
Although only four percent of the programs included in this meta-analysis
targeted gang-involved youth (McDaniel et al., 2010), the literature suggests that
employment interventions may be particularly suitable for gang members (Houston,
1996; Huff, 1998; Spergel, 2007). For example, survey data shows that 74% of gang-
involved youth indicate that "work is good and necessary" (Knox et al., 1995, p. 45), and
49% report that "job training and employment is the answer to gang membership"
(Houston, 1996, p. 7). Also, 25% of gang-involved youth report that they would give up
illicit financial opportunities (e.g., selling drugs) for employment that offered minimum
wage (Huff, 1998). Further, if asked what type program is most effective in their
opinion, 39% of gang-involved youth report employment programs (Houston, 1996).

10

Gang programs run by community agencies recognize the importance of
employment, as they often provide job referrals, job placement, and job training for gang-
involved youth (Howell, 2000; Spergel, 2007). However, the effectiveness of these
programs is largely unknown (Howell, 2000; Stinchcomb, 2002). Indeed, when Fisher,
Montgomery, and Gardner (2008) attempted to conduct a meta-analysis on employment
programs for gang-involved youth, these authors found no evaluations with a randomized
design. Nevertheless, several studies offer preliminary support for the use of
employment programs for gang-involved youth (Howell, 2000).
One such study, the Group Guidance Project (Klein, 1971), emphasized
employment as one approach to reducing gang cohesiveness. During periods of
employment, Klein (1971) found that program participants were 30% less likely to "hang
out" on the streets and 9% less likely to be with their gang-involved friends than they
were during periods of unemployment (Klein, 1971). While overall results indicated that
gang-related delinquency increased over the course of the program (from pre-test to post-
test), some positive outcomes were found for youth who were employed (Klein, 1971).
In the 1980s, a police department implemented the Gang Employment Program,
which offered pre-employment services consisting of (a) obtaining job leads, (b)
contacting youth when there is prospective employment, (c) familiarizing youth with job
applications and interviews, and (d) arranging transportation for interviews (Willman &
Snortum, 1982). Compared to a control group, participants in the Gang Employment
Program showed no differences in police detentions at a 7-month follow-up (Willman &

11

Snortum, 1982). However, a 30-month follow-up with program participants found that
53% of youth were still working and only 9% were in jail (Willman & Snortum, 1982).
In the 1990s, the Little Village Gang Project began offering a comprehensive
community-wide approach to intervening with a gang in Chicago (Spergel, 2007). Using
a matched comparison group, Spergel (2007) found that the more time gang-involved
youth engaged in legitimate employment versus illegal forms of obtaining money, the
lower their levels of self-reported total offenses, serious violent offenses, drug use, and
drug selling. In addition, the higher the level of legal income, the lower the level of self-
reported total offenses, serious violent offenses, drug use, and drug selling (Spergel,
2007).
This brief review of employment programs for gang-involved youth demonstrates
that only a handful of evaluations exist (Klein, 1971; Spergel, 2007; Willman & Snortum,
1982). Further, as previous reviews note, rigorous methodology is absent from this
literature (Fisher et al., 2008; Howell, 2000; Stinchcomb, 2002). While Spergel (2007)
found that employment has a small effect on delinquency within the context of a gang
program, the intervention's impact on gang involvement is unknown. Also, aside from
some anecdotal evidence, it is unclear from past evaluations whether these programs
actually increased employment for gang-involved youth. In sum, while there is strong
theoretical support for the use of employment programs, more empirical support is
needed.

12

Behavioral Employment Program
The ongoing BEP evaluation attempts to fill this gap in the literature. This
evaluation partially replicates a previously successful program, the Behavioral
Employment Intervention Program (BEIP), developed by Walter and Mills (1980). The
original BEIP evaluation (Mills & Walter, 1977; Walter & Mills, 1980) was well-
specified, which facilitated replication in the present pilot study. In addition, BEIP
outcomes were examined in a controlled evaluation, including an ethnically diverse
sample of delinquent youth (Mills & Walter, 1977; Walter & Mills, 1980). At the one-
year follow-up, results showed that 90% of youth in BEIP had no further arrests
(compared to 30% in the control condition; Walter & Mills, 1980). In addition, a recent
meta-analysis found that BEIP was one of the most effective employment programs for
reducing re-arrest among delinquent youth (McDaniel et al., 2010). One of the factors
that distinguished the original BEIP from many other employment programs was its
success in placing youth in jobs (McDaniel et al., 2010). Walter and Mills (1980)
showed that 100% of BEIP youth were employed during treatment, compared to 39% of
the control youth. At the one-year follow-up, 34% of BEIP youth still had a job,
compared with 0% of control youth (Mills & Walter, 1977).
Hypotheses
This dissertation was part of an ongoing pilot evaluation of BEP as modified for
juvenile gang offenders. The evaluation provided a unique opportunity to collect two
types of data for within- and between-subjects analyses. The within-subjects data were
derived from in-session measures, with BEP participants reporting regularly on events

13

over the past week. In contrast, the between-subjects measures were completed during
the formal, quarterly assessments of BEP and usual service (US) participants. The main
objective of this study was to investigate the impact of BEP-initiated employment on
gang involvement. This study explored distinct hypotheses for each analytic approach.
The sole within-subjects hypothesis was that increases in employment will be
related to reductions in gang involvement over time. Information gathered from these
within-subjects analyses contribute to the understanding of processes occurring within
BEP. A shortcoming of this approach is that treatment effects cannot be examined. In
order to address this limitation, the between-subjects method compared outcomes across
both treatment conditions: BEP and usual services (US). This approach provides further
clarification of the role of employment in BEP by testing for mediation. A series of four
hypotheses were explored to examine the presumed treatment mechanism:
(1) Compared to US, BEP will lead to greater reductions in gang involvement,
(2) Compared to US, BEP will lead to greater increases in employment,
(3) Increases in employment will be associated with reductions in gang
involvement, and
(4) The association between treatment condition and gang involvement will be
mediated by increases in employment.
While Walter and Mills (1980) described employment as an essential element of the
original BEIP, there is no empirical evidence, as yet, to support this assumption.
Through two alternative analytic approaches, this study aims to provide further
clarification on this potential treatment mechanism.

14

Chapter 2: Method
Participants
Participants met the following inclusion criteria: (a) 16 to 20 years of age, (b) at
least one arrest within the past year, (c) affiliation with a street gang in Los Angeles, (d)
residence in LA County, (e) English fluent, and (f) interested in acquiring employment.
Sixteen years was the minimum age of inclusion due to employment restrictions that exist
for younger adolescents in the state of California. Based on previous work, gang-
involvement was determined by an affirmative response to the screening question, “Are
you a member of a street gang or tagger crew?” (Maxson, Whitlock, & Klein, 1998). No
youth was excluded based on preexisting mental or physical health conditions.
Procedures
A probation officer at the LA Department of Probation referred participants.
Each month the probation officer provided BEP staff with a list of youth who were
identified as gang members by law enforcement and who were scheduled for imminent
release from juvenile camps. A BEP staff member screened these youth over the phone
to assess whether they met the inclusion criteria, and eligible youth were informed of the
study. If youth (and caregivers in the case of minors) agreed to participate, informed
consent was obtained, and the youth completed the pretreatment (T1) assessment within
72 hours of consent. BEP staff administered the assessment in the family’s home at a
time convenient for them.
Between March 2007 and March 2010, a total of 27 youth consented to the study
and participated in the T1 assessment. Upon completion of the T1 assessment, youth

15

were randomly assigned to either BEP or US. Fifteen participants were assigned to BEP
and 12 to US. Figure 1 shows the flow of participants through randomization, allocation,
and analyses (adapted from Moher, Schulz, & Altman, 2001). As depicted in this flow
chart, two sub-samples were used for the analyses. The within-subjects approach used
the sub-sample of treated BEP participants (n = 13) and the between-subjects approach
used data from the treated BEP and US participants (n = 25). The data collection
procedures differed by approach.
The within-subjects method focused only on in-session data from BEP youth.
During each session, BEP participants were asked questions related to employment and
gang involvement. The counselor administered these questions at the beginning of the
session, which took approximately five minutes to complete. Of the 15 participants
randomized to BEP, only the 13 treated participants were included in the within-subjects
analyses. The excluded participants did not attend any BEP sessions; one participant was
re-incarcerated the day of his T1 assessment, and another participant withdrew from the
study before counseling sessions began. Two other participants dropped out, or
terminated treatment between 8 and 14 weeks. Treatment completers and drop-outs were
included in the within-subjects analysis to provide a conservative test of BEP effects.
Therefore, a total of 13 BEP youth were examined.
The between-subjects analyses were based on a 2 (treatment type: BEP vs. US) x
3 (time: pretreatment [T1], 3-months post-entry [T2], 6-months post-entry [T3]) factorial
design. A structured interview was administered to study participants at T1 and T3. This
assessment battery was two hours in duration and was administered at the participant’s

16
Figure 1

Flow Chart of Sample through Recruitment, Allocation, and Analyses
Note. This flow chart was adapted from Moher, Schulz, and Altman (2001). BEP = Behavioral Employment Program.
US = Usual Services.

17
home. At T2, an abbreviated assessment was conducted over the phone. To compensate
for their time, youth were paid $25 for each completed in-home assessment and $10 for
each abbreviated phone assessment. The untreated BEP youth (n = 2) were excluded
from the between-subjects analyses. Thus, these analyses included the 13 treated BEP
youth and the 12 US youth.
Treatment Conditions
Behavioral Employment Program. Youth assigned to BEP received counseling
services provided by masters-level graduate students in clinical psychology or marriage
and family counseling at the University of Southern California. A total of six counselors
(4 female and 2 male) treated BEP youth. All counselors were trained by and received
weekly supervision from Dr. Stanley Huey (the Principal Investigator). During
supervision meetings, the therapists and supervisor reviewed the goals and progress for
each youth. Counselors had low caseloads (1 to 5 youth per clinician) to allow for
necessary intensity of treatment services and the provision of comprehensive services
delivered in community settings, such as the home, school, neighborhood, and social
service agencies.
BEP counseling was designed to be flexible to adjust to the youth's strengths and
weaknesses. The main goal of counseling was to assist the youth in obtaining and
maintaining employment. In order to accomplish this goal, counselors used a number of
behavioral strategies (e.g., contingency contracting, behavioral enactment, positive
reinforcement/praise, prompts, and planned ignoring) over the course of treatment. For
example, all youth completed mock job interviews, which required youth to dress and
18

prepare as though they were actually interviewing. These interviews were video-
recorded and then reviewed in-session to allow for youth self-evaluation, as well as
counselor feedback and praise.
Following a procedure similar to Walter and Mills (1980), BEP staff contacted
small- to medium-sized business owners via phone or door-to-door solicitation to find
employers interested in mentoring high-risk youth. Employers were offered incentives
(e.g., reimbursement of ½ the youth's salary for their first three months of employment)
to hire BEP youth. These incentives were to offset any additional expenses required for
training these youth and to compensate employers for their time working with BEP.
Unfortunately, because this study occurred during the worst economic recession since the
Great Depression, this approach was unsuccessful in the current replication as no
program-sanctioned employers hired BEP youth. Given the failure to duplicate Walter
and Mills' (1980) job placement strategy, the program adapted BEP to incorporate Azrin
and Besalel's (1980) job-seeking approach. Azrin and Besalel's (1980) approach focused
on developing the youth's pre-employment skills by requiring the youth to fill out
applications, return phone calls, prepare for job interviews, and find leads. At 3-months
post-entry, only 54% of BEP youth secured employment, which was approximately half
the rate found in Walter and Mills' (1980) original study. Thus, BEP was less successful
than expected in assisting youth secure employment.
BEP counseling consisted of four possible phases. In the first phase, pre-
employment skills training was the primary focus of treatment. This phase involved
twice-weekly meetings with the BEP counselor. Youth remained in this phase until
19

employment was obtained. Consequently, many youth in the program did not move
beyond this first phase. Once a youth secured employment, the focus of treatment shifted
to assessment and maintenance of appropriate job-related behaviors (e.g., keeping
workspace clean, wearing uniform, answering phone appropriately, etc.) and the second
phase began. Phase II included weekly sessions with the BEP counselor. Phase III
focused on the generalization of skills learned in the prior phases. In this phase, BEP
counseling sessions were reduced to bi-weekly meetings. Phase III ended approximately
six months after treatment was initiated, marking the conclusion of the more intensive or
active phases of BEP. In the final BEP phase, Phase IV, contact between the BEP
counselor and youth faded to monthly phone contact, which was accompanied by in-
person sessions when needed.
Usual Services. Youth assigned to the US condition received a range of services
typically available to youth on probation. Youth were initially assigned to a juvenile
probation officer who supervised the youth in order to facilitate compliance with court
orders. Probation officers often monitored the youth by tracking progress in school and
other relevant community-based services. In addition, probation officers provided case
management services to the youth. Often probation officers provided youth referrals for
mental health services, vocational services, recreational services, or other services.
Follow-up surveys were used to assess the services utilized by US participants.
Mental health services were utilized by 25% of the youth. These services were provided
through Medi-Cal funded community-based agencies. One youth participated in
Homeboy Industries, a community-based job placement, training, and education program
20

developed specifically to meet the needs of gang-involved youth. Additionally, 17% of
the youth reported participating in youth programs comprised of afterschool educational
services that aimed to ease the transition back into the community and school.
Measures
Despite using the same gang involvement and employment measures for the
within- and between-subjects analyses, the time frames varied by approach. The within-
subjects outcomes were derived from in-session data, with participants reporting on
events over the previous week. In contrast, the between-subjects measures were
completed during the formal, quarterly assessments, which required participants to report
on events spanning the last three months. To help differentiate these variables, a
subscript (
w
) is used to identify within-subjects measures.
Employment. To examine changes in employment, hours employed per week
was assessed through youth self-report (Staff & Uggen, 2003). Hours employed refers to
the average number of hours worked per week (Staff & Uggen, 2003), and is frequently
used as a measure of employment (Paternoster, Bushway, Brame, & Apel, 2003; Wright
& Cullen, 2004; Wright, Cullen, & Williams, 2002). For the within-subjects data, the
question "How many hours did you work this week?" was used to assess hours
employed
w
; for the between-subjects data, the question "In the past 3 months, how many
hours did you work every week, on average?" was used.
Gang involvement. To determine level of gang involvement, several measures
were used. First, participants answered a question based on Esbensen et al.'s (2001)
concentric circle diagram of gang activity (see Figure 2). If administered in-person,
21
participants were shown Figure 2 and asked, "Thinking about your group of friends in the
gang, in terms of your involvement in group activities, how close or far from the center of
this group are you now?" Possible responses were as follows: "never in this group," "no
longer in a gang," "not active in a gang," "rarely active in a gang," "active in a gang,"
"very active in a gang," and "gang leader" (Esbensen et al., 2001). In the within-subjects
analyses, categorical variables violate the assumption of homogeneous variance.
Figure 2

Esbensen et al.'s (2001) Gang Activity Measure

1 No Longer in Group
2 Not Active
3 Rarely Active
4 Active
5 Very Active
6 Leader
0 Never in this
Group

Therefore, to address this issue, the gang activity
w
responses were converted into a
dummy coded variable (coded 0 = never in this group, no longer in a gang or not active
in a gang; 1 = rarely active, very active, or gang leader). Because categorical variables
are not problematic for the statistical models used in the between-subjects analyses, the
original response categories were maintained for this approach.

22

Second, participants were asked to report the number of days spent with gang
peers. To assess the within-subjects days with gang
w
, the question “How many days a
week did you hang out with friends in a gang?” was asked during each session. For the
between-subjects days with gang, the question "In general, how many days a week do
you hang out with friends in the gang?" was asked at the post-entry time periods.
Third, participants were administered the Gang Membership Inventory (GMI:
Pillen & Hoewing-Roberson, 1992), the only measure of gang membership available with
established psychometric properties (Herrmann, McWhirter, & Sipsas-Herrmann, 1997).
The GMI consists of 15 items (coded 1 = true, 0 = false) that assess gang member
activities that range from peripheral (e.g., “Friends joined a gang”) to core (e.g., “Fought
representing a gang”). Each "true" response is summed to yield a total score. The GMI
has good internal reliability ( α = .83) and correlates well with high-risk indicators such as
substance abuse and school problem behavior (Pillen & Hoewing-Roberson, 1992). The
original GMI index of "in the last year" was changed to "in the last week" for the within-
subjects items and "in the last three months" for the between-subjects measures.
Statistical Methods
Separate analytic techniques were used to examine the hypotheses. The within-
subjects data were analyzed using Pooled Times Series Analyses (PTSA). The between-
subjects data were examined through a series of regression analyses that test for treatment
effects and treatment mediation.
Within-Subjects Analyses. PTSA is a regression-based approach that pools data
from several participants into one time-series (Hoeppner et al., 2008), combining the
23

strengths of ordinary time-series with multiple regression (Moore et al., 1994). While
many psychologists endorse the scientific merit of PTSA, Moore et al. (1994) note that it
is rarely used in the psychology literature. The underutilization of this method may be
attributable to its association with case studies or ordinary time-series analyses, which
have no basis for generalization (Moore et al., 1994). Also, time-series analysis requires
more observation points than available in most psychological studies. In addition,
statistical software commonly used in psychology do not allow for a straightforward
application of PTSA (Johnson, 1995). Nevertheless, in other fields (e.g., econometrics
and engineering), PTSA has become increasingly popular (Hoeppner et al., 2008) and
statistical packages used in these fields, such as EVIEWS 7.0 (Quantitative Micro
Software, 2009) and STATA (StataCorp, 2009), have the capability to handle PTSA
models.
Due to potential violations of the standard linear model, PTSA involves
specifying and testing a series of tentative statistical models that provide alternative
assumptions for identifying and correcting sources of error (West & Hepworth, 1991;
Sayrs, 1989). For example, in the present study, a base model was used, which depends
on an ordinary least squares (OLS) model for continuous data, and a linear probability
model (LPM) for binary variables (West & Hepworth, 1991). This base model was used
because it provides unbiased parameter estimates (i.e., β statistics). However, because of
the likely violations of assumptions of this model due to heteroskedacity and serial
dependence, the significance tests (i.e., t and F statistics) can be artificially inflated (West
& Hepworth, 1991). As the following sections describe, a series of steps were taken to
24

specify and test a final model that meets the underlying assumptions of PTSA (West &
Hepworth, 1991).
A concern that arises when using time-sequenced data is that the error terms are
not independent or, in other words, the serial dependence of residuals exists across time.
The Durbin-Watson statistic is used to compute the serial dependence of the base model
(West & Hepworth, 1991). If serial dependence is detected, then it means that the
statistical parameters of the base model are incorrect. In this case, a feasible generalized
least squares (FGLS) model is used in addition to the base model. In using a FGLS
model, a correction is applied to estimate the degrees of freedom called the Prais-Winsten
estimation procedure. After the Prais-Winsten procedure is applied, another Durbin-
Watson statistic is calculated to test for serial dependence. Correcting for serial
dependence using FGLS and the Prais-Winsten estimation procedure is expected to
provide adequate control for serial dependence and also allow for reliable hypothesis
testing. The parameter estimates of both the base model (i.e., OLS) and final model (i.e.,
FGLS) are examined for a more accurate check of robustness; however, the unbiased
regression coefficients of the base model are reported in the results.
To further address issues related to potential sources of bias, the regression model
controlled for time-related trends (i.e., seasonality) because previous work has shown a
connection between summer months and gang activity (Hutson, Anglin, & Pratts, 1994;
West & Hepworth, 1991). Thus, dummy codes for the months of the year were created to
control for seasonal effects of gang involvement in the time-series. In addition, PTSA
models can control for other important variables if they are entered in the regression
25

model. For example, since a previous study (Seals, 2009) found that a youth's age is
related to changes in gang involvement, this variable was controlled for by using dummy
codes for different age groups. Similarly, to identify sources of contamination expected
from pooling data from multiple participants, a least squares with dummy variables
(LSDV) approach was used. Specifically, to capture the similarity expected across all
time points of one participant, each participant was identified with a dummy code when
they were entered into the regression equations. The total number of dummy codes
accounting for individual differences was (n - 1) or 12 for the present study.
In sum, the following model was used for the within-subjects analyses:
Υ
it
= β
0
+ β
1
Χ
it
+ αΑ
i
+ γD
it
+ ε
it

In this model, Υ
it
represents the dependent variable with i = individual participants and t
= time. β
0
is the model intercept and β
1
is the coefficient that describes the size of the
effect of the independent variable (represented by Χ
it
) on the dependent variable. α is a
matrix of coefficients for each individual control dummy (Α
i
), γ is a matrix of
coefficients for seasonal and age controls (D
it
), and ε
it
is the residual value in the
regression.
Between-subjects analyses. In a Monte Carlo experiment using simple
mediation models, Hoyle and Kenny (1999) found that even a very large coefficient
could be non-significant with small samples. In order to address this issue, these authors
proposed that Baron and Kenny's (1986) original steps for mediation be examined in
terms of zero and nonzero coefficients to provide the conceptual framework for
26

mediation. Once there is conceptual evidence for mediation, then a formal statistical test
(e.g., Sobel [1982] test) is conducted that utilizes non-parametric bootstrapping
procedures, which are robust even with small samples (Shrout & Bolger, 2002).
Thus, following the approach recommended by Hoyle and Kenny (1999), four
conditions must be met to demonstrate mediation. First, the independent variable must
be associated with the dependent variable (see path c in Figure 3A). Second, the
independent variable must be associated with the hypothesized mediator (see path a in
Figure 3B). Third, the mediator must be associated with the dependent variable (see path
b in Figure 3B). Finally, the impact of the independent variable on the dependent
variable must be zero after controlling for the mediator (path c' in Figure 3B). The final
step of this procedure determines whether full or partial mediation has occurred (Hoyle &
Kenny, 1999). Full mediation is indicated when c' = 0. If the first three steps in the
mediation analyses are satisfied and the effect of the independent variable on the
dependent variable is reduced, but not to zero, then partial mediation is indicated (Hoyle
& Kenny, 1999).
As just outlined, mediation requires that the effect of the independent variable on
the dependent variable be a result of a mediating variable. This relationship is called the
indirect effect, which contrasts with the direct effect (Figure 3A) of the independent
variable on the dependent variable. Statistically, the indirect effect is assessed by
multiplying the effect of the independent variable on the mediator (also known as a) and
the mediator on the dependent variable (also known as b); therefore, the indirect effect is

27
Figure 3

Mediation Model for Between-Subjects Analyses

A

A) Direct effect

B

B) Indirect (or Mediated) effect
Treatment
Condition
(BEP v. US)
T3 Gang
Involvement
c
a b
c'
Treatment
Condition
(BEP v. US)
T3 Gang
Involvement
T2 Hours Employed
28

equivalent to ab (see Figure 3B). To determine the significance of ab, the Sobel (1982)
test is conducted.
While the Sobel (1982) test is a very common approach to testing the significance
of the indirect effect, it may not be appropriate for use with small samples. Indeed,
simulation studies show that the confidence intervals of the Sobel test based on small
samples do not represent true confidence intervals (MacKinnon, Lockwood, & Williams,
2004). Further, Preacher and Hayes (2004) note that as sample size decreases, the Sobel
test becomes less conservative. Thus, to test the significance of ab, a bias-corrected
bootstrap analysis was used. Shrout and Bolger (2002) recommend this approach when
sample sizes are small to obtain better power when estimating the standard error. This
method is a parametric approach that makes no assumptions about the shape of the
distribution and corrects for skewness of distributions (Preacher & Hayes, 2004). A
bootstrap involves re-sampling from the original dataset (thousands of times) and
estimating the indirect effect of each re-sampled dataset (Shrout & Bolger, 2002). The
estimate of ab used in this approach is simply the mean indirect effect computed over the
re-sampled data. Preacher and Hayes (2008) have developed SPSS macros for
bootstrapping that were used in the present study. These macros provide an estimate of
the indirect effect, an estimated standard error, and the 95% confidence interval (Preacher
& Hayes, 2008).
An assumption of mediation is that the outcome does not cause the mediator
(Johansson & Per Hoglen, 2007). Because research suggests that employment and gang
involvement have a reciprocal relationship (Klein, 1995), the mediator and outcome were
29

assessed at sequential time points to rule out the possibility that the outcome changed
before the mediator (Johansson & Per Hoglen, 2007). More specifically, T2 hours
employed (denoted as: hours employed
T2
) and T3 gang measures were used.
Also, to control for individual differences between participants and for
measurement error that occurs with the use of repeated measures, residual gain scores
were computed (Steketee & Chambless, 1992). These scores were created for the gang
measures because hours employed was constant at T1. The residual gain scores were
computed by standardizing the variables and determining the correlation of T1 and T3
scores (Steketee & Chambless, 1992). This correlation was multiplied by the T1 score,
then subtracted from the T3 score (Steketee & Chambless, 1992). A subscript (
T3|T1
) is
used to identify these change scores. This procedure is equivalent to using a multiple
regression approach, where the post-entry score is the dependent variables and
pretreatment score, on the same measure, is entered in an equation before the predictor
(Sketetee & Chambless, 1992). Johansson and Per Hoglend's (2007) review of mediators
of treatment outcome highlights the use of residual gain scores to examine change in
outcome variables.
30

Chapter 3: Results
Demographic Characteristics
Because subjects were randomly assigned to conditions, demographic differences
between the groups were not expected. However, given the small sample size, it is
possible that randomization was not completely effective. Therefore, ANOVAs and chi-
square tests were used to test for differences in demographic variables across conditions
at T1 (Table 1). No statistically significant differences were found between the two
conditions.
Across both conditions, the youth were predominately male (96.0%), with an
average age of 17.4 years (SD = 0.8). Over three-quarters (76.0%) were Latino and
24.0% were African American. Thirteen percent of the youth had children of their own.
Thirty-six percent of the caregivers were married, 28.0% were single, and 36.0% were
divorced or separated. Most (60.0%) caregivers had less than a high school education,
and the majority (58.3%) had either part-time or full-time jobs. Of those caregivers
without jobs, 20.8% were homemakers and 20.8% were unemployed. The self-reported
arrest histories of the referred youth demonstrate a high level of criminal involvement.
The youth averaged 4.6 (SD = 3.7) previous juvenile arrests, ranging from 1 to 15 arrests
per participant. The mean age of the youths at the time of first arrests was 14.9 (SD =
1.4). Using DSM-IV criteria, youth had 1.9 (SD = 1.9) mental health diagnoses on
average, which most often included a drug use disorder (70.0%).
31

Table 1

Demographic Characteristics at Pretreatment
Total
(n = 25)
BEP
(n = 13)
US
(n = 12)

p
Age (M, SD) 17.4 (0.8) 17.6 (0.9) 17.2 (0.7) .18
Gender (n, %) .33
Male 24 (96.0) 12 (92.3) 12 (100.0)
Female 1 (4.0) 1(7.7) 0 (0.0)
Race (n, %) .91
African American 6 (24.0) 3 (23.1) 3 (25.0)
Latino 19 (76.0) 10 (76.9) 9 (75.0)
Has children (n, %) .39
Yes 3 (13.0) 1 (7.7) 2 (20.0)
No 22 (87.0) 12 (92.3) 8 (80.0)
Caregiver marital status (n, %) .85
Married 9 (36.0) 4 (30.8) 5 (41.7)
Single 7 (28.0) 4 (30.8) 3 (25.0)
Divorced or separated 9 (36.0) 5 (38.5) 4 (33.3)
Caregiver education (n, %) .69
< high school 12 (60.0) 7 (53.8) 5 (71.4)
High school 5 (25.0) 4 (30.8) 1 (14.3)
> high school 3 (15.0) 2 (15.4) 1 (14.3)
Caregiver work status (n, %) .71
Full-time 11 (45.8) 7 (53.8) 4 (36.4)
Part-time 3 (12.5) 2 (15.4) 1 (9.1)
Unemployed 5 (20.8) 2 (15.4) 3 (27.3)
Homemaker 5 (20.8) 2 (15.4) 3 (27.3)
Number of siblings (M, SD) 2.4 (1.7) 2.6 (1.9) 2.1 (1.6) .45
Age at first arrest (M, SD) 14.9 (1.4) 14.5 (1.5) 15.3 (1.2) .19
Number of arrests (M, SD) 4.6 (3.7) 5.5 (4.3) 3.2 (2.0) .17
Number of mental health
diagnoses (M, SD)
1.9 (1.9) 1.8 (1.6) 1.9 (2.3) .93
Note. P-values were computed using either one-way ANOVAs or Pearson Chi-Square.
M = mean. SD = standard deviation. n = sample size. The two variables, caregiver
education and work status, have sample sizes less than 25 because some youth did not
report this information.

32

Initial Analyses
Table 2 displays the means, standard deviations, and ranges for key outcome
variables at the within-subjects and between-subjects level. While Table 2 presents
between-subjects data from all time periods, it is important to note that T2 employment
and T3 gang involvement data were the focus of the analyses. The within- and between-
subjects approaches require distinctive initial analyses, which are reported in separate
sections.
Within-subjects analyses. An advantage of PTSA is its ability to handle
variations in data due to attrition and missing data (Johnson, 1995; Elgar et al., 2004).
Participants who dropped out were included up to the point when they ended treatment.
The mean number of sessions per youth was 11.96 with a range from 8 to 45 sessions.
BEP sessions occurred relative to treatment phase, on a twice weekly, weekly, or bi-
weekly basis. The average time between sessions was 8.88 days (SD = 16.41). Nine
percent of the data on hours employed
W
was missing, which was most often due to youth
skipping this item when unemployed. Thus, missing data for employment was estimated
using the following decision rule. First, if the missing data point fell prior to
employment, or during a period of known unemployment, a zero was added. Second, for
missing data that came at a transition between employment and unemployment, it was
assumed the youth was unemployed and a zero was added to the data. Between 2-3% of
the data were missing for the gang involvement measures. These missing data were
dropped from the analyses because unlike the employment data they were missing at
random.

33
Table 2

Means, Standard Deviations, and Ranges for Within- and Between-Subjects Analyses
Within-Subjects Between-Subjects
BEP US
Variable M SD Range n M SD Range n M SD Range n F
Employment
Hours employed

7.68 12.88 0 - 52 337
T1 0 0 0 13 0 0 0 12 0.00
T2 12.16 18.14 0 - 56 12 12.50 17.56 0 - 48 8 0.00
T3 7.27 15.08 0 - 50 11 6.25 12.75 0 - 35 8 0.02
Gang Involvement
Gang activity 2.78 1.27 1 - 6 330
T1 3.38 1.33 1 - 5 13 3.64 1.36 1 - 6 11 0.21
T2 2.92 1.17 1 - 5 12 2.86 1.35 1 - 4 7 0.01
T3 2.82 1.60 0 - 5 11 3.17 1.47 1 - 5 6 0.19
Days with gang 1.45 1.91 0 - 7 330
T1 3.42 2.89 0 - 7 12 3.00 2.41 0 - 7 12 0.14
T2 2.60 2.37 0 - 7 10 1.67 2.73 0 - 7 6 0.52
T3 2.22 2.33 0 - 7 9 3.16 2.64 0 - 7 6 0.53
Gang membership 1.51 3.62 0 - 10 210
T1 4.85 3.07 0 - 9 13 5.45 2.34 2 - 9 11 0.60
T2 3.91 3.42 0 - 11 11 2.50 2.35 1 - 7 6 0.80
T3 3.20 2.39 1 - 8 10 4.86 3.53 0 - 9 7 1.34
Note. M= mean. SD = standard deviation. n = sample size. For the within-subjects approach, the means represent data collapsed across BEP
sessions and participants. ANOVAs were conducted to examine differences in the between-subjects data. T1 = pretreatment, T2 = 3-months post-
entry and T3 = 6-months post-entry.

34
Preliminary examination of the data was conducted to examine potential
deviations from the assumptions of the PTSA model (e.g., BEP phase-related trends,
curvilinear trends, or time-related trends). First, the BEP session data were plotted for
each youth. Appendix A provides these figures, which depict changes in hours
employed
w
, gang activity
w ,
days with gang
w,
and gang membership
w
across sessions.
While plotting data in this way provided important descriptive information on individual
differences in the data, overall trends were difficult to capture. Therefore, the average
scores across youth were also inspected. Figure 4 shows the average scores for hours
employed
w
and the gang measures at each session. Upon visual inspection, a positive
linear effect for hours employed
w
and two variables, gang activity
w
and days with gang
w
,
was apparent. No clear pattern of effects for hours employed
w
and gang membership
w

was identified.
Between-subjects analyses. Prior to statistical analyses, all data were examined
for outliers and missing values. No outliers were found; however, missing data was an
area of concern. Often data was missing because a participant was incarcerated. Rather
than considering all these data points missing, the participants were given a value of 0 for
hours employed
T2
when official records confirmed their incarceration. After
incarceration was considered, the missing data rate for hours employed
T2
was 20%.
Missing data for the T3 gang measures ranged from 32% to 40%, with the highest
missing data rates for days with gang. Missing data varied because of disrupted data
collection (e.g., youth did not complete full assessment) or assessor error (e.g., failure to
administer every measure).
35

Figure 4

Means for Within-Subjects Hours Employed and Gang Measures Across Active Phase of BEP

Hours employed
Gang measure

Note. Only data from the
active phase of BEP were
plotted (approximately 27
sessions). Beyond this
treatment phase, an expected
decline in participation
occurred. At 8-months post-
entry, only 5 participants
remained in treatment, and at
10-months 1 participant
remained in treatment.

36
Missing data were imputed to preserve a sample size of 25. Multiple imputation
was performed using NORM software (Schafer, 1997). NORM uses an estimation
procedure that takes other data collected from the same sample into account. This
method is preferable to alternative approaches, such as Last Value Observed Carried
Forward, because of its ability to handle small sample sizes and large amounts of missing
data (Barnes, Lindborg, & Seaman, 2006; Graham & Shafer, 1999).
Correlations among hours employed
T2
and the gang measures are presented in
Table 3. No correlations reached the generally accepted 0.90 cut-off point for
multicollinearity (Tabachnick & Fidell, 2007). A number of significant correlations were
found in the positive direction for the gang measures. Two marginally significant
negative relationships were found for hours employed
T2
and the two gang variables: gang
activity
T3|T1
and gang membership
T3|T1
.

Table 3

Correlation Matrix for Between-Subject Analyses
Variable 1 2 3 4
1. Hours employed
T2
1.00
2. Gang activity
T3|T1
-0.35
+
1.00
3. Days with gang
T3|T1
-0.23 0.74
***
1.00
4. Gang membership
T3|T1
-0.40
+
0.53
**
0.40
+
1.00
Note. Pearson's correlation coefficients were used to test significance.
+
p < .10. *p < .05. **p < .01. ***p < .001. T2 = 3-months post-entry.
T3|T1 = 6-months post-entry (T3) residual gain score controlling for
pretreatment (T1) scores.

37

Within-Subjects Analyses: Pooled Time Series
PTSA was used to test the hypothesis that increases in employment will be related
to decreases in gang involvement over the course of BEP. Prior to performing the main
analyses, a number of alternative models were run to ensure correct model specification.
Overall, these findings indicated that (a) time between sessions, (b) treatment drop out,
and (c) hours employed squared (to test for nonlinear relationships) provided no
additional explanatory power; therefore, these variables were not included in the main
analyses. Given the pilot nature of the overarching study, effect sizes (Cohen's ƒ
2
) were
also computed to provide information on the magnitude of effects (Cohen, 1992). As a
“rule of thumb,” 0.02, 0.15, and 0.35 indicate a small, medium, and large effect,
respectively (Cohen, 1992).
Gang activity
w
. The first dependent variable examined was the binary gang
activity
w
measure. The base model (i.e., LPM) was run with the following predictor
variables: (1) hours employed
w
and (2) dummy codes for months, youth's age, and
individual differences. The Durbin-Watson statistic was used to compute the serial
dependence statistic, then the Prais-Winsten correction was applied. In Appendix A, a
table is presented that summarizes the parameter estimates from both the base and final
models.
The final test of this model was statistically significant, F(28, 302) = 12381.45, p
< .001. The regression coefficient for hours employed
w
was -.0059, t(281) = -4.81, p <
.001. As displayed in Figure 5, these results indicate that a 1-hour increase in
38
employment decreases the probability of being active in a gang by 0.59%. The R
2
change
value of 0.04 is equivalent to a small effect (ƒ
2
= 0.09).
An ANOVA test was used to compare partial sum of squares of the predictors in
the regression model. Month and youth's age significantly predicted gang activity
w
,
F
change
(11, 281) = 9.29, p < .001, and F
change
(3, 281) = 6.43, p < .001, respectively,
accounting for 11% of the variance in the model. Broadly, these results indicate that
older youth have a lower probability of gang activity and that summer months increase
the probability that youth are active in a gang. In addition, individual differences
significantly predicted gang activity
w
, F
change
(12, 317) = 81.37, p < .001, and accounted
for 75% of the variance in the model.
Figure 5
Within-Subjects Hours Employed Predicting Changes in the Probability of
Gang Activity
w

0.40
0.45
0.50
0.55
0.60
0.65
0 1020 304
Hours Employed
0

Note. Control variables were held constant.

39

Days with gang
w
. The second variable examined was days with gang
w
. Serial
dependence was addressed using methods described earlier. The final test of this model
was significant, F(28, 302) = 34.24, p < .001. The base model (i.e., OLS) indicated that
hours employed
w
was a statistically significant predictor of days with gang
w
, -.023, t
(302) = -2.84, p < .01. These results show that a 1-hour increase in employment is
associated with a 0.02 decrease in days spent with gang friends (i.e., a reduction of 3.36
hours/week). The R
2
change value of 0.02 is equivalent to a small effect (ƒ
2
= 0.03). In
the final model (i.e., FGLS), this regression equation was only marginally significant (p =
.07). As described in the methods section, while the tests of significance of the final
model are more reliable, the parameter estimates of the base model are unbiased;
therefore, both models were examined for a more accurate check of robustness. Given
the preliminary nature of the present study, these marginal findings are reported, but
should be interpreted with caution.
Based on an ANOVA test used to compare the predictors, month, F
change
(11, 330)
= 2.69, p < .01, and youth's age, F
change
(3, 330) = 6.91, p < .001, significantly predicted
days with gang
w
. The variance accounted for by both variables was 10%. The dummy
codes for individual differences were also a significant predictor, F
change
(12, 330) = 18.15,
p < .001, accounting for 38% of the variance in the model.
Gang membership
w
. The last variable examined was gang membership
w
. After
applying the correction for serial dependence, the results from the final model were
significant, F(24, 183) = 29.33, p < .001. Despite this finding, the regression coefficient
for hours employed was nonsignificant, -0.02, t (183) = -0.97, p = 0.33. The R
2
change
40

value of 0.03 is equivalent to a small effect (ƒ
2
= 0.01). Although not statistically
significant, the regression coefficient indicates that a 1-hour increase in employment
corresponds to a 0.20 decrease in units of gang membership, and that the association
between these two variables is mainly related to individual differences. These results
likely reflect limited variation in the gang membership measure (M = 1.51, SD = 2.02,
Range = 0 - 10). The ANOVA test of the model revealed that individual differences
accounted for 25% of the variance in the model. Also, this test confirmed that the
dummy codes for individual differences were a statistically significant predictor,
F
change
(9, 183) = 4.56, p < .001. The variables for month and youths age did not
contribute significantly to the model.
Between-Subjects Analyses: Meditation
A series of regression equations was used to test for mediation. In these
regressions, treatment condition served as the independent variable, hours employed
T2
,
served as the potential mediating variable, and the T3 gang measures served as the
dependent variables. Results from these analyses are summarized in Table 4.
Hypothesis 1: BEP will lead to greater reductions in gang involvement. The
first step in mediation is to establish that the independent variable is related to the
outcome. Thus, the direct effect of treatment condition on the three gang measures was
explored (see path c in Figure 3A). A marginally significant treatment effect was found
for gang membership
T3|T1
. Specifically, the regression coefficient was -0.67, t (24) = -
1.89, p < 0.10, indicating that BEP decreases gang membership by 0.67 standardized
units.

41
Table 4

Summary of Regression Equations for Mediation Model (n = 25)

Paths in mediation model
β SE β
t Value Effect
Size (ƒ
2
)
Effect of treatment condition on gang measures (Path c)
Treatment condition Æ Gang activity
T3|T1
-0.14 0.36 -0.39 0.01
Treatment condition Æ Days with gang
T3|T1
-0.16 0.40 -0.39 0.01
Treatment condition Æ Gang membership
T3|T1
-0.67 0.35 -1.89
+
0.15
Effect of treatment condition on employment (Path a)
Treatment condition Æ Hours employed
T2
-4.57 6.44 -0.71 0.02
Effect of employment on gang involvement with control for treatment condition (Path b)
Hours employed
T2
Æ Gang activity
T3|T1
-0.02 0.01 -1.86
+
0.02
Hours employed
T2
Æ Days with gang
T3|T1
-0.02 0.01 -1.19 0.01
Hours employed
T2
Æ Gang membership
T3|T1
-0.03 0.01 -2.63* 0.28
Effect of treatment condition on gang measures with control for employment (Path c')
Treatment condition Æ Gang activity
T3|T1
-0.24 0.35 -0.68 0.16
Treatment condition Æ Days with gang
T3|T1
-0.22 0.40 -0.56 0.07
Treatment condition Æ Gang membership
T3|T1
-0.79 0.32 -2.48* 0.31
Note. Treatment condition = coded (0 = US and 1 = BEP).
+
p < .10. *p < .05. The effect sizes of 0.02, 0.15, and 0.35, according to
Cohen's (1992) "rules of thumb", indicate a small, medium, and large effect, respectively

42
While some researchers (Frazier, Tix, & Barron, 2004) argue that a significant effect for
path c is an essential component to mediation, Hoyle and Kenny (1999) note that only
evidence of a causal influence reflected in a nonzero value for path c is required for small
samples. Thus, the hypothesized mediation model was still tested.
The effect sizes of this relationship were also explored. The R
2
value for gang
membership
T3|T1
was 0.13, representing a medium effect (ƒ
2
= 0.15). However, a small
effect was found for the other gang measures, gang activity
T3|T1
(ƒ
2
= 0.01) and days with
gang
T3|T1
(ƒ
2
= 0.01).
Hypothesis 2: BEP will lead to greater increases in employment. The second
step of mediation is to establish that the independent variable is related to the mediator
(see path a in Figure 3B). This relationship was not statistically significant, and only a
small effect (ƒ
2
= 0.02) was found, with a regression coefficient ( β = -4.57) opposite of
what was expected. Since employment was hypothesized as the mechanism of change in
BEP, we actually expected positive effects of large magnitude.
Hypothesis 3: Increased employment will be associated with reductions in
gang involvement. This between-subjects hypothesis mirrors the already described
within-subjects analyses. Hours employed
T2
was regressed on the gang measures,
controlling for treatment condition (see path b, Figure 3B). First, a significant
relationship was found, with increases in hours employed
T2
predicting decreases in gang
membership
T3|T1
. The regression coefficient was -0.03, t (24) = -2.63, p < 0.05, which is
a medium-to-large effect (ƒ
2
= 0.28). This finding indicates that when treatment
condition is controlled, gang membership is reduced by 0.03 standardized units for each
43

additional hour employed. Second, a marginally significant effect for gang activity
T3|T1

was found. The regression coefficient was -0.02, t (24) = -1.86, p < 0.10, representing a
small effect (ƒ
2
= 0.02). This finding indicates that after treatment condition is
controlled, gang activity is reduced by 0.02 standardized units for each additional hour
employed. A small effect (ƒ
2
= 0.07) in the negative direction was also observed for days
with gang
T3|T1
.
Hypothesis 4: The association between treatment condition and gang
involvement will be mediated by increases in employment. The final step in testing
for mediation requires examining the association of the independent variable on the
dependent variable, when controlling for the mediator (see path c' Figure 2B). A
significant relationship was found for treatment condition on gang membership
T3|T1
after
controlling for hours employed
T2
. The regression coefficient was -0.79, t (24) = -2.48, p
< 0.05. This finding indicates that when the variable hours employed is controlled, BEP
decreases gang membership by 0.79 standardized units. Effect sizes for the gang
measures were as follows: a medium-to-large effect (ƒ
2
= 0.31) for gang membership
T3|T1
,
a medium effect for gang activity
T3|T1
(ƒ
2
= 0.16), and a small effect (ƒ
2
= 0.07) for days
with gang
T3|T1
. Conceptually, since path c' is not zero or |c'| < |c|, there is not sufficient
evidence for full or partial mediation (Hoyle & Kenny, 1999). However, since the direct
effect (path c) and indirect effect (path a multiplied by path b) had opposite signs, there is
a possibility of inconsistent mediation or suppression effects.
Suppression effects were tested for gang membership
T3|T1
using a bias-corrected
bootstrapping procedure (Preacher & Hayes, 2004). Results revealed that the indirect
44

effect was 0.12 (SE = 0.21). The 95% confidence interval, -0.17 to 0.72, contained zero.
Therefore, the indirect effect was not significant, which indicates that suppression effects
are not present.
45

Chapter 4: Discussion
Although early sociological theory suggests a link between employment and gang
involvement (Hirschi, 1969; Cloward & Ohlin, 1960), empirical work on this topic has
been sparse. Also, insufficient attention has been given to the role of employment in the
context of gang intervention. The present study used data from a pilot randomized trial
of BEP, an employment-based intervention for gang-involved youth. The main objective
of the present study was to evaluate the extent to which BEP-initiated employment
influenced gang involvement. Within- and between-subjects analyses were used to test
several questions related to this goal.
The within-subjects analyses examined the hypothesis that increases in
employment are related to decreases in gang involvement over the course of treatment.
The results support this hypothesis, with analyses indicating that employment was a
significant or marginally significant, negative predictor of the variables gang activity and
days with gang. However, the magnitude of this relationship was consistently small.
The between-subjects analyses supplemented these within-subjects results. A
series of regression analyses were used to test the hypothesis that employment mediates
BEP treatment effects. Contrary to expectations, results revealed no evidence for
mediation. Nevertheless, a marginally significant treatment effect was found, showing
that BEP led to reductions in gang membership. This relationship remained after
controlling for hours employed. Effect sizes of the three gang measures indicate that
BEP had small-to-medium effects on gang involvement. Additionally, the between-
subjects results showed that increases in employment led to significant decreases in gang
46

membership and marginally significant decreases in gang activity. The magnitude of
these effects varied from a small effect for gang activity to medium-to-large for gang
membership.
Using two different analytic techniques to measure similar relationships merits
discussion. The primary goal of the present study was to examine the impact of increases
in BEP-initiated employment on gang involvement; the results from these two
approaches converge to clarify this treatment process. While both approaches generally
revealed a similar pattern of results, several noteworthy differences should be mentioned.
Specifically, the within-subjects results showed that hours employed predicted changes in
the variables gang activity and days with gang, but not gang membership. In contrast, 3
of the 4 marginally or statistically significant between-subjects findings were for the gang
membership measure. This discrepancy is likely reflective of issues related to the
sensitivity of the gang measures.
Another possible explanation for the differences in the within- and between-
subjects results is related to power. Mediator analyses are commonly used in the
literature, but this approach likely does not have the power to detect important effects in
samples of this size (for further discussion of these issues see Hoyle & Kenny, 1999).
Indeed, results from post hoc power analyses revealed power ranging from 0.11 to 0.54
for detecting small-to-medium effects (Erdfelder, Faul, & Buchner, 1996). In contrast,
the within-subjects analyses had power of 0.89 to detect significant effects. That stated,
the between-subjects method is considered the "gold standard" in evaluation research
because it includes a randomized comparison group allowing for the examination of
47

treatment effects. Thus, while within- and between-subjects approaches both have
strengths and limitations, taken together these analytic strategies complement each other,
allowing for a broader understanding of the relationships between these variables than
would be possible using one approach alone.
Employment and Gang Involvement in BEP
The present study provides evidence that increases in employment lead to
reductions in gang involvement among BEP participants. While dominant theory places
unemployment at the root of gang formation (Hirschi, 1969; Cloward & Ohlin, 1960), no
studies have rigorously examined the impact of obtaining employment on self-identified
gang involvement. Ethnographic studies examining the natural process of gang
desistence often implicate employment as a strategy to extricate youth from gangs
(Spergel, 1990; Vigil, 1988). Informed by this work, many gang mitigation efforts
include employment as a key intervention component (Howell, 2000). The present
results support the continued use of employment in gang intervention efforts, although
with the caveat that employment generally had only a small effect on gang involvement.
Encouragingly, the present study also demonstrated that BEP had a marginally
significant, medium effect on gang membership. Reviews of gang interventions find very
few programs that demonstrate promise (Howell, 1998; Howell, 2000; Spergel et al.,
1994; Stinchcomb, 2002). Additionally, a meta-analysis by Fisher and colleagues (2008)
found no rigorous evaluations of employment-focused gang interventions. Therefore, the
present results are an important contribution to the evidence base. They challenge the
notion that "nothing works" for gang involvement (Howell, 2007), while concurrently
48

attempting to identify how BEP works. A potential explanation of these promising
results is that BEP uses theory-driven intervention strategies. Furthermore, BEP
approaches gang involvement as a chronic problem, requiring a long-term intervention.
Indeed, relevant reviews and meta-analyses underscore that treatment
effectiveness rests on sound theory and the selection of appropriate treatment targets,
modalities, and duration (Andrews et al., 1990; McDaniel, Bridgelal, & Huey, 2010;
Spergel et al., 1994; Stinchcomb, 2002; Uggen & Staff, 2003). As an example, Uggen
and Staff's (2003) review highlights that duration (i.e., long-term) and intensity (i.e.,
high) are critical components of effective employment interventions for delinquent youth.
In the same vein, the literature on gang desistance suggests that changes in gang
involvement occur gradually (Decker & Lauritsen, 2002; Moloney et al., 2009; Sanchez-
Jankowski, 1991; Vigil, 1988), and intervention efforts may need to make a long-term
commitment to youth in order to target the different aspects of gang involvement
(Stinchcomb, 2002). Certainly, multiple socioeconomic stressors, mental health
problems, and family-based strains and conflicts complicate the lives of gang-involved
youth (Thornberry et al., 2003). Such ecological challenges may hamper intervention
efforts, limiting the efficacy of interventions with gang youth. Given these challenges, it
may be that gradual remediation of gang involvement is a more realistic goal for gang-
focused interventions. Indeed, the present results showed that in general there were
small, incremental reductions in gang involvement over the course of 6 months. Further,
visual inspection of the within-subjects data illustrated that acquiring employment
generally took several months to achieve. McDaniel, Smith, and Huey (2010) plan to
49

examine data from 9- and 12-months post-entry to provide further information about the
stability of treatment effects.
Unexpectedly, BEP did not lead to increases in hours employed. At 3-months
post-entry, BEP youth were employed an average of 12.16 hours per week, compared to
12.50 hours for US youth. BEP is a partial replication of BEIP (Walter & Mills, 1980), a
program which achieved 100% employment for its treated youth. The current replication
was almost half as successful with 54% of BEP youth securing employment. A number
of programmatic challenges led to this low employment rate. The intervention was
conducted between 2007 and 2010, a time when the California unemployment rate was at
its highest level since 1940 (Steinhauer, 2009). Potentially related to this economic
downturn, Walter and Mills' (1980) approach of matching program-sanctioned employers
to youth proved unsuccessful. Therefore, as an alternative strategy, BEP counselors used
Azrin and Besalel's (1980) job-seeking approach. Yet, this approach also had limited
success in part due to the youths' low motivation for formal employment, lack of family
support, and drug use. Further, the impression of counselors was that BEP was more
effective for "fringe" gang members with more transient gang involvement than with
"core" members who reported daily engagement in gang activities (Spergel, 1990). Azrin
and Besalel's (1980) strategies were laborious, requiring more motivation and effort than
seemed acceptable for "core" members. Indeed, Moffitt (1993) identifies a similar
taxonomy for antisocial behavior, distinguishing between "adolescent-limited" and "life-
course persistent" delinquents, and argues that intervention efforts for the latter group are
generally ineffective. It is likely that the youth's core versus fringe status moderates BEP
50

treatment effects. Therefore, in future refinements of BEP, a more targeted focus on
fringe gang members might not only be a more appropriate match for Azrin and Besalel's
(1980) job-seeking strategies, but should likely result in larger treatment effects as well.
Employment was the presumed mediator of BEP treatment effects. While the
present results did not find evidence for this hypothesis, the findings do suggest that there
are likely other employment-oriented processes that mediate treatment outcomes. Unlike
gang interventions that target gang involvement explicitly (i.e., suppression efforts that
focus on dismantling gangs through policing and incarceration; Howell, 2000), BEP
targets gang involvement indirectly through two treatment foci: (a) helping youth obtain
and maintain employment and (b) reducing their risk for re-arrest. Since BEP did not
directly attempt to discourage youth from affiliating with gang peers, reductions in gang
involvement may have occurred through employment facilitating prosocial relationships
(e.g., positive peer friendships and role models) and prosocial attitudes (e.g., work
ambitions and values). This explanation is in line with the social learning perspective
(Esbensen & Huizinga, 1993; Esbensen, Winfree, He, & Taylor, 2001; Goldstein, 1991;
Winfree, Backstrom, & Mays, 1994). Another possible explanation from a life-course or
developmental perspective is that changes in gang involvement may have occurred
through BEP's focus on maturational processes related to employment (e.g., increased
responsibility or desire for independence; Decker & Lauritsen, 2002; Moloney,
Mackenzie, Hunt, & Joe-Laidler, 2009; Sanchez-Jankowski, 1991; Vigil, 1988). Finally,
social control theory implicates increases in supervision or involvement in new social
institutions as potential employment-oriented mediators (Hirschi, 1969). Such treatment
51

processes underlying this evaluation are important to study as they help refine program
theory and improve program efficacy (Fairchild & Mackinnon, 2009). These processes
are important to test in future studies with larger sample sizes.
The Impact of Age and Seasonality on Gang Involvement
While the primary goal of this study was to examine the impact of employment on
gang involvement, the results indicate that both the youth's age and the month of the year
are also important predictors. First, the within-subjects results showed that youth age
was significantly related to changes in gang activity and days with gang. These findings
are not surprising considering prior research (Seals, 2009; Thornberry et al., 2003). Gang
researchers are increasingly using developmental and life-course perspectives as a
framework for explaining gang-related behavior (Howell & Egley, 2005; Thornberry et
al., 2003). These perspectives highlight that patterns of gang involvement are related to
physical, cognitive, emotional and social changes occurring across development
(Caldwell & Atschuler, 2001; Howell & Egley, 2005; Thornberry et al., 2003). To
account for age-related changes, Caldwell and Atschuler (2001) recommend matching
risk and protective factors, which are informed by a youth's developmental status, to
intervention strategies. For example, these authors propose that interventions targeting
mid-adolescence gang-involved youth focus on "coping with peer pressure, social skill
development, conflict resolution, self-concept, adventure, avocational pursuits and
vocational aptitude" (Caldwell & Altschuler, 2001, p. 31). In contrast, these authors
recommend that interventions with late adolescents focus on processes related to the
"youth's strong sense of independence" and maturing relationships with peers, family,
52

and adults (Caldwell & Altschuler, 2001, p. 31). A developmental or life-course
perspective has also been used to explain differential age-related outcomes in
employment programs (Uggen & Staff, 2003). Uggen and Staff (2003) hypothesize that
employment may have different meaning for younger youth than older youth (Uggen &
Staff, 2003). Indeed, these authors found that employment with high job autonomy,
wage, and social status increased adolescent delinquency even though these job
characteristics were expected to decrease arrest, recidivism, and substance abuse in older
youth (Uggen & Staff, 2003).
Second, the within-subjects results showed that the months of the year were
significantly related to changes in the variables gang activity and days with gang.
Specifically, these results indicate that increases in gang involvement occur during
summer months. These findings are consistent with prior work, which shows that
summer months have the highest number of gang-related drive-by shootings and injuries
(Hutson, Anglin, & Pratts, 1994). Summer temperatures and the stifling conditions of
low-income housing likely promote an increase in the youth's street orientation during
these months. Beyond the obvious connection with weather, summer months concur with
fewer academic activities and, subsequently, more competition for youth employment.
Gang intervention efforts, especially those that are employment-focused, may be critical
during this time.
Implications for Research and Policy
The modest sample size of pilot studies often precludes the use of advanced
statistical modeling approaches (i.e., latent growth curve modeling, hierarchical linear
53

modeling, and latent transition analyses). PTSA is a methodologically-sound strategy for
analyzing data when the sample size is small (Soliday, Moore, & Lande, 2002). Yet, few
psychological studies (Moore, Osgood, Larzelere, & Chamberlain, 1994) have used the
PTSA approach. As shown in the present study, this approach has great utility for pilot
evaluations. For example, PTSA yielded clinically relevant effect sizes (Moore et al.,
1994). For pilot interventions, these effect sizes are critical for determining the sample
size required for adequate power in further evaluation (Cohen, 1992). Based on the
modest effects documented, a sample size of 481 would be required for a large-scale trial
evaluating the relationship of employment and gang involvement using standard multiple
regression techniques (Cohen, 1992).
Policy should be geared towards funding approaches to reduce gang involvement
that are backed by sound theory and strong methodology (Andrews et al., 1990). BEP is
one of only a handful of gang interventions that have been evaluated. At best, prior
evaluations of gang interventions show small effects on gang involvement; however, at
worst, these evaluations show iatrogenic effects (Howell, 1998; Howell, 2000; Spergel et
al., 1994; Stinchcomb, 2002). The present study found small-to-medium effects for BEP.
This preliminary finding is not trivial and even a small decrease in gang involvement
could have a major impact in reducing gang-related crime. Indeed, previous research
suggests that youth delinquency is cut in half when a youth leaves a gang (Thornberry et
al., 2004). Also, these positive changes could create enormous savings in incarceration
and probation costs.
54

Historically, evaluation research has not been viewed as an integral component of
gang-related policy initiatives (Stinchcomb, 2002). Yet, every major review of gang
interventions highlights the need for rigorous evaluation (Howell, 1998; Howell, 2000;
Spergel et al., 1994; Stinchcomb, 2002). In order for this important work to be done,
funding should be set aside for appropriate assessment. While the budget for evaluation
is often small compared to the entire budget for gang programs, rigorous program
evaluation is essential in enabling policymakers to make informed decisions (Klein,
1995).
Limitations
Several methodological limitations were apparent in this study. The first involved
the very modest sample size. To account for this limitation, the present study used
PTSA, which provides a methodologically-sound approach for analyzing data with small
samples (Soliday, Moore, & Lande, 2002). A consequence of the increased power of
PTSA is increased degrees of freedom and heterogeneity. The OLS model addresses the
problems of heterogeneity, although not without additional constraints. Measures that are
time-invariant, such as gender and race, must be excluded from the model because these
measures will have no explanatory power (Johnson, 1995). Even though researchers can
control for age-related effects in PTSA, its influence is overestimated in the final model,
which makes significance tests misleading (Soliday, Moore, & Lande, 2002). Also,
although the internal validity of the design is strong, the small sample size limits the
generalizability and stability of the findings. Thus, replication is very important.
55

The second major limitation stems from the gang measures used in the present
study. There is increasing recognition in the gang literature that existing assessment tools
and techniques for capturing gang processes are inadequate (Bjerregaard, 2002). Despite
using one of the only gang measures with reported psychometric properties, the GMI
(Pillen & Hoewing-Roberson, 1992), there was difficulty encountered due to this
measure's lack of sensitivity to weekly changes in gang behavior. Walker-Barnes and
Mason (2004) developed a modified version of this measure that uses a continuous (vs.
dichotomous) response scale and separates gang activity from gang involvement, which
may address the lack of variation seen in the within-subjects analyses.
Finally, measuring employment with only one item, hours employed, also had
limitations. Solely examining hours employed overlooks certain processes unique to
gangs. For example, survey data shows that 48% of gang members endorse belonging to
a gang that operates legitimate businesses (Knox et al., 1995). Youth in BEP also
endorsed working for these gang-owned operations, and it is likely that youth in US also
held these type of jobs. However, the current study did not examine these employment
distinctions. In order to address this issue, multi-item employment measures are needed
that examine employment-oriented processes in more detail. For example, future
measures could include items measuring (a) co-worker characteristics (e.g., prosocial or
antisocial attitudes), (b) employment type (e.g., gang-owned, community agency-
sponsored, or conventional), and (c) job stability (e.g., consistent work opportunities,
contract-to-contract employment or potential for long-term growth in job). A measure
56

including these items would be valuable for future research as they would capture
theoretically relevant processes underlying treatment effects.
Conclusions
This dissertation finds that increases in employment predict reductions in gang
involvement. Based on within-subjects results, each hourly increase in employment
decreases the probability of being involved in gang activities by 0.59% or reduces time
spent with the gang by 3.36 hours each week. In other terms, a full-time job (i.e., a 40-
hour increase in employment) is expected to decrease the probability of being involved in
gang activities by 23.6% and to decrease time spent with the gang by 5.6 days each week.
The between-subjects results revealed that BEP leads to a small-to-medium effect on
gang involvement, supporting the continued use of BEP. Previous reviews highlight that
most gang interventions have not been evaluated (Fisher et al., 2008; Howell, 2000). The
present study uses two methods, a within- and between-subjects approach, to contribute
to the evidence-base for BEP. While the present results are important, they represent
only a first step in evaluating BEP outcomes. The long-term impact of BEP is in the
process of being evaluated (McDaniel, Smith, & Huey, 2010). This outcome study will
offer further clarification on the effectiveness of this already promising intervention.
57

Bibliography
Andrews, D. A., Zinger, I., Hoge, R. D., Bonta, J., Gendreau, P., & Cullen, F. T. (1990).
Does correctional treatment work? A clinically relevant and psychologically
informed meta-analysis, Criminology, 28, 369-404. doi: 10.1111/j.1745-
9125.1990.tb01330.x

Aryan, H. E., Jandial, R., Bennett, R. L., Masri, L. S., Lavine, S. D. & Levy, M. L.
(2005). Gunshot wounds to the head: Gang- and non-gang-related injuries and
outcomes. Brain Injury, 19, 0269-9052.

Azrin, N. H. & Besalel, V. A. (1980). Job Club Counselor's Manual. Austin, TX: Pro-ed.

Barnes, S. A., Lindborg, S. R., & Seaman, J. W. (2006). Multiple imputation techniques
in small sample clinical trials. Statistics in Medicine. 25, 233-245. doi:
10.1002/sim.2231

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in
social psychological research: Conceptual, strategic and statistical considerations.
Journal of Personality and Social Psychology, 51, 1173-1182. doi:10.1037/0022-
3514.51.6.1173

Bendixen, M., Endresen, I.M., & Olweus, D. (2006). Joining and leaving gangs:
Selection and facilitation effects on self-reported antisocial behaviour in early
adolescence. European Journal of Criminology, 3, 85-114.
doi:10.1177/1477370806059082

Bjerregaard, B. (2002). Self-definitions of gang membership and involvement in
delinquent activities. Youth and Society, 34, 31-54. doi:
10.1177/0044118X02034001002

Caldwell, L. & Altschuler, D. (2001). Adolescents leaving gangs: An analysis of risk and
protective factors, resiliency and desistance in a developmental context. Journal
of Gang Research, 8, 21-34.

Centers for Disease Control and Prevention, National Center for Injury Prevention and
Control (2007). Web-based Injury Statistics Query and Reporting System
(WISQARS). Retrieved from http://www.cdc.gov/ncipc/wisqars/default.htm

Cloward, R. & Ohlin, L. (1960). Delinquency and opportunity. New York: Free Press.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159. doi:
10.1037/0033-2909.112.1.155
58

Curry, G. D., & Spergel, I. (1992). Gang involvement and delinquency among hispanic
and african american adolescent males. Journal of Research in Crime and
Delinquency, 29, 273-291. doi:10.1177/0022427892029003002

Curry, G. D., & Thomas, R.W. (1992). Community organization and gang policy
response. Journal of Quantitative Criminology, 8, 357-374.
doi:10.1007/BF01093640

Decker, S. H., & Lauritsen, J. L. (2002). Leaving the gang. In C. R. Huff (Ed.), Gangs in
America III (pp. 103-122), Thousand Oaks, CA: Sage Publications, Inc.

Egley, A. (2002). National youth gang survey trends from 1996-2000. Juvenile Justice
Bulletin, 3.

Elgar, F., Waschbusch, D.A., McGrath, P.J., Stewart, S.H., & Curtis, L.J. (2004).
Temporal relations in daily-reported maternal mood and disruptive child behavior.
Journal of Abnormal Child Psychology, 32, 237-247.
doi:10.1023/B:JACP.0000026138.95860.81

Erdfelder, E., Faul, F., & Buchner, A. (1996). GPower: A general power analysis
program. Behavior Research Methods, Instruments, & Computers, 28, 1-11.

Esbensen, F. A. & Huizinga, D. (1993). Gangs, drugs, and delinquency in a survey of
urban youth. Criminology, 31, 565-589. doi:10.1111/j.1745-9125.1993.tb01142.x

Esbensen, F. A., Winfree, L. T., He, N., & Taylor, T. J. (2001). Youth gangs and
definitional issues: When is a gang a gang, and why does it matter?. Crime &
Delinquency, 4, 105-130. doi:10.1177/1477370805048626

Fagan, J. (1990). Social processes of delinquency and drug use among urban gangs. In
C.R. Huff (Ed.) Gangs in America. (pp. 183-219). Newbury Park, CA: Sage
Publications.

Fairchild, A. J. & Mackinnon, D. P. (2009). A general model for testing mediation and
moderation effects. Prevention Science, 10, 87-99. doi: 10.1007/s11121-008-
0109-6

Fisher, H., Montgomery, P., & Gardner, F. (2008). Opportunities provision for preventing
youth gang involvement for children and young people (7-16). Cochrane
Database of Systematic Reviews, 2, 1-32.

59

Frazier, P., Tix, A., & Barron, K. (2004). Testing moderator and mediator effects in
counseling psychology research. Journal of Counseling Psychology, 51, 115-134.
doi: 10.1037/0022-0167.51.1.115

Gatti, U., Temblay, R.E., Vitaro, F., & McDuff, P. (2005). Youth gangs, delinquency and
drug use: A test of the selection, facilitation, and enhancement hypotheses.
Journal of Child Psychology and Psychiatry, 46, 1178-1190. doi:10.1111/j.1469-
7610.2005.00423.x

Goldstein, A.P. (1991). Delinquent gangs: A psychological perspective. Champaign, IL:
Research Press.

Graham, J. W. & Schafer, J. L. (1999). On the performance of multiple imputation for
multivariate data with small sample size. In R. H. Hoyle (Ed.), Statistical
strategies for small sample research. Newbury Park: Sage.

Hagedorn, J. M. (1988). People and folks: Gangs, crime and the underclass in a rustbelt
city. Chicago, IL: Lakeview Press.

Herrmann, D. S., McWhirter, J. J., Sipsas-Herrmann, A. (1997). The relationship between
dimensional self-concepts and juvenile gang involvement: implications for
prevention, intervention, and court referred diversion programs. Behavioral
sciences and the law, 15, 181-194. doi:10.1002/(SICI)1099-
0798(199721)15:2<181::AID-BSL268>3.0.CO;2-O

Hirschi, T. (1969). Causes of delinquency. Berkeley and Los Angeles: University of
California Press.

Hirschi, T., & Gottfredson, M. (1983). Age and the explanation of crime. American
Journal of Sociology, 91, 22-27. doi:10.1086/228243

Hoeppner, B. B., Goodwin, M. S., Velicer, W. F., Mooney, M. E., & Hatsukami, D. K.
(2008). Detecting longitudinal patterns of daily smoking following drastic
cigarette reduction. Addictive Behavior, 33, 623-639.
doi:10.1016/j.addbeh.2007.11.005

Houston, J. (1996). What works: The search for excellence in gang intervention
programs. The Journal of Gang Research, 3, 1–16.

Howell, J. C. & Egley, A. E. (2005). Moving risk factors into developmental theories of
gang membership, Youth Violence and Juvenile Justice, 3, 334-354. doi:
10.1177/1541204005278679

60

Howell, J. C. (1998). Promising programs for youth gang violence prevention and
intervention. In R. Loeber & D. P. Farrington (Eds.), Serious and violent juvenile
offenders: Risk factors and successful interventions (pp. 284-312). Thousand
Oaks, CA: Sage Publications.

Howell, J. C. (2000). Youth Gang Programs and Strategies. Summary. Washington, DC:
U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice
and Delinquency Prevention.

Howell, J. C. (2007). Menacing or mimicking? Realities of youth gangs. The Juvenile
and Family Court Journal, 58, 9-20. doi:10.1111/j.1755-6988.2007.tb00137.x

Hoyle, R. H., & Kenny, D. A. (1999). Statistical power and tests of mediation. In R. H.
Hoyle (Ed.), Statistical strategies for small sample research. Newbury Park:
Sage.

Huff, C. R. (1990). Denial, overreaction, and misidentification: A postscript on public
policy. In C.R. Huff (Ed.) Gangs in America. Newbury Park, CA: Sage
Publications.

Huff, C. R. (1998). Comparing the criminal behavior of youth gangs and at-risk youths.
National Institute of Justice.

Hutson, H. R., Anglin, D., Kyriacou, D. N., Hart, J., & Spears, K. (1995). The epidemic
of gang-related homicides in Los Angeles County from 1979 through 1994. The
Journal of the American Medical Association, 274, 1031-1036.
doi:10.1001/jama.274.13.1031

Hutson, H.R., Anglin D., & Pratts, M.J. (1994). Adolescents and children injured or
killed in drive-by shootings in Los Angeles. New England Journal of Medicine,
330, 324-327. doi:10.1056/NEJM199402033300506

Johansson, P. & Hoglend, P. (2007). Identifying mechanisms of change in
psychotherapy: Mediators of treatment outcome. Clinical Psychology and
Psychotherapy, 14, 1-9. doi: 10.1002/cpp.514

Johnson, D. R. (1995). Alternative methods for the quantitative analysis of panel data in
family research: Pooled time-series models. Journal of Marriage and the Family,
57, 1065-1077. doi:10.2307/353423

Klein, M. W. (1971). Street gangs and street workers. Englewood Cliffs: Prentice-Hall,
Inc.

61

Klein, M. W. (1995). The American street gang: Its nature prevalence, and control. New
York: Oxford University Press.

Klein, M. W., & Maxson, C. L. (2006). Street gang patterns and policies. New York:
Oxford University Press.

Knox, G. W., Tromanhaouser, E. D., Houston, J. G., Martin, B., Morris, T. F., McCurrie,
J. L. , Laskey, J. L. , Papachristos, D., Feinberg, J., & Waxman, C. (1995). The
Economics of Gang Life: A Task Force Report. Chicago State University:
National Gang Crime Research Center.

MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the
indirect effect: Distribution of the product and resampling methods. Multivariate
Behavioral Research, 39, 99-128. doi:10.1207/s15327906mbr3901_4

Maxson, C. L., Whitlock, M. & Klein, M. (1998). Vulnerability to street gang
membership: Implications for prevention. Social Service Review, 72, 70-90.

McDaniel, D., Bridgelal, I., & Huey, S. (2010). A meta-analysis on the effectiveness of
employment-related programs for delinquent youth. Unpublished manuscript.

McDaniel, D., Smith, C., & Huey, S. (2010). [Employment and gang involvement
outcomes in BEP evaluation]. Unpublished raw data.

Mills, C. M., & Walter, T. L. (1977). A behavioral employment intervention program for
reducing juvenile delinquency. Behavioral Therapy, 8, 270-272.
doi:10.1016/S0005-7894(77)80280-3

Moher, D., Schulz, K., & Altman, D. (2001). The CONSORT statement: revised
recommendation for improving the quality of reports of parallel-group
randomised trials, Lancot, 357, 1191-1194. doi: 10.1016/S0140-6736(00)04337-3

Moffitt, T. E., (1993). Adolescence-limited and life-course-persistent antisocial behavior:
A developmental taxonomy. Psychological Review, 100, 674-701. doi:
10.1037/0012-1649.39.2.222

Moloney, M., Mackenzie, K., Hunt, G., & Joe-Laidler, K. (2009). The path and promise
of fatherhood for gang members. British Journal of Criminology, 49, 305-325.
doi:10.1093/bjc/azp003

62

Moore, K. J., Osgood, D. W., Larzelere, R. E., & Chamberlain, P. (1994). Use of pooled
time series in the study of naturally occurring clinical events and problem
behavior in foster care setting. Journal of Consulting and Clinical Psychology, 62,
718-728. doi: 0022-006X/94/S3.00

Pager, D. (2007). Marked: Race, crime, and finding work in an era of mass
incarceration. Chicago: University of Chicago Press.

Pager, D., Western, B., & Bonikowski, B. (2008). Race at work: A field experiment of
discrimination in low-wage labor markets. Princeton University.

Paternoster, R., Bushway, S., Brame, R., & Apel, R. (2003). The effect of teenage
employment on delinquency and problem behaviors. Social Forces, 82, 297-335.
doi:10.1353/sof.2003.0104

Peterson, D., Taylor, T.J., & Esbenson, F. (2004). Gang membership and violent
victimization. Justice Quarterly, 21, 793-815. doi:10.1080/07418820400095991

Pillen, M. B., & Hoewing-Roberson, R. C. (1992). Determining youth gang membership:
Development of a self-report instrument. Bloomington, IL: Chestnut Health
Systems.

Pizarro, J. M., & McGloin, J. M. (2006). Explaining gang homicides in Newark, New
Jersey: Collective behavior or social disorganization? Journal of Criminal Justice,
34, 195-207. doi:10.1016/j.jcrimjus.2006.01.002

Preacher, K. J. & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect
effects in simple mediation models. Behavior Research Methods, Instruments,
and Computers, 36, 717-731.

Preacher, K. J. & Hayes, A. F. (2008). Asymptotic and resampling strategies for
assessing and comparing indirect effects in multiple mediator models. Behavior
Research Methods, 40, 879-891. doi: 10.3758/BRM.40.3.879

Quantitative Micro Software. (2009). Eviews (Version 7.0) [Computer Software]. Irvine,
CA: Quantitative Micro Software.

Sánchez-Jankowski, M. (1991). Islands in the street: Gangs and American urban society.
Berkeley, CA: University of California Press.

Schafer, J.L. (1997) Analysis of Incomplete Multivariate Data. Chapman & Hall,
London.

63

Seals, R. A. (2009). Are gangs a substitute for legitimated employment? Investigating the
impact of labor market effects on gang affiliation. KYKLOS, 62, 407-425.
doi:10.1111/j.1467-6435.2009.00443.x

Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental
studies: New procedures and recommendations. Psychological Methods, 7, 422-
445. doi:10.1037/1082-989X.7.4.422

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural
equation models. In S. Leinhardt (Ed.), Sociological Methodology 1982 (pp. 290-
312). Washington DC: American Sociological Association.

Soliday, E., Moore, K. J., & Lande, M. B. (2002). Daily reports and pooled time series
analysis: Pediatric psychology applications. Journal of Pediatric Psychology, 27,
67-76. doi:10.1093/jpepsy/27.1.67

Spano, R., Freilich, J. D., & Bolland, J. (2008). Gang membership, gun carrying, and
employment: Applying routine activities theory to explain violent victimization
among inner city, minority youth living in extreme poverty. Justice Quarterly,
25, 381-410. doi:10.1080/07418820802024911

Spergel, I. A. (1990). Youth gangs: Continuity and change. Crime and Justice, 12, 174-
275. doi: 10.1086/449166

Spergel, I.A., Curry, G.D., Chance, R., Kane, C., Ross, R., Alexander, A., Simmons, E.,
& Oh, S. 1994. Gang Suppression and Intervention: Problem and Response.
Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office
of Juvenile Justice and Delinquency Prevention.

Spergel, I. A. (1995). The youth gang problem. New York, NY: Oxford University Press.

Spergel, I. A. (2007). Reducing youth gang violence: The Little Village Gang project in
Chicago. New York: AltaMira Press.

Staff, J., & Uggen, C. (2003). The fruit of good work: Job quality and adolescent
deviance. Journal of Research in Crime and Delinquency, 40, 263-290.
doi:10.1177/0022427803253799

StataCorp. (2009). Stata Statistical Software (Release 11) [Computer Software]. College
Station, TX: StataCorp LP.

Steinhauer, J. (2009). California joblessness reaches 70-year high. New York Times, p.
A12.
64

Steketee, G. & Chambless, D. L. (1992). Methodological issues in prediction of treatment
outcome. Clinical Psychology Review, 12, 387-400. doi:10.1016/0272-
7358(92)90123-P

Stinchcomb, J. B. (2002). Promising (and not so promising) gang prevention and
intervention strategies: A comprehensive literature review. Journal of Gang
Research, 10, 27-45.

Tabachnik, B. G., & Fidell, L. S. (2007). Using multivariate statistics, 5th ed. Boston:
Allyn and Bacon.

Thornberry, T. P. (1998). Membership in youth gangs and involvement in serious and
violent offending. In R. Loeber and D. P. Farrington (Eds.), Serious and violent
juvenile offenders: Risk factors and successful interventions, Thousand Oaks, CA:
Sage Publications.

Thornberry, T. P., (Ed.). (1997). Developmental theories of crime and delinquency. New
Brunswick, NJ: Transaction Publishers.

Thornberry, T.P., Huizinga, D., & Loeber, R. (2004). The causes and correlates studies:
Findings and policy implications. Juvenile Justice, 9, 3-19.

Thornberry, T.P., Marvin D.K., Lizotte, A.J., Smith, C.A. and Tobin, K. (2003). Gangs
and delinquency in developmental perspective. New York: Cambridge University
Press. doi:10.1017/CBO9780511499517.

Tita, G., & Abrahamse, A. (2004). Gang homicide in LA, 1981–2001. At the local level:
Perspectives on violence prevention, No. 3. Sacramento, California: California
Attorney General’s Office.

Uggen, C. & Staff, J. (2003). Work as a turning point for criminal offenders. Corrections
Management Quarterly, 5, 1-16.

Vigil, J. D. (1988). Barrio gangs: Street life and identity in Southern California. Austin,
TX: University of Texas Press.

Walker-Barnes, C. J. & Mason, C. A. (2004). Delinquency and substance use among
gang-involved youth: The moderating role of parenting practices, American
Journal of Community Psychology, 34, 235-250. doi:10.1007/s10464-004-7417-1

Walter, T. L., & Mills, C. M. (1980). A behavioral-employment intervention program for
reducing juvenile delinquency. In R. R. Ross & P. Gendreau (Eds.), Effective
correctional treatment (pp. 187-200). Toronto: Butterworths.
65

Wells, L. E. & Weisheit, R. A. (2001). Gang problems in nonmetropolitan areas: A
longitudinal assessment. Justice Quarterly, 18, 791-823.
doi:10.1080/07418820100095101

West, S. G. & Hepworth, J. T. (1991). Statistical issues in the study of temporal data:
Daily experiences. Journal of Personality, 59, 609-662.

Willman, M. T. & Snortum, J. R. (1982). A police program for employment of youth
gang members. International Journal of Offender Therapy and Comparative
Criminology, 26, 207-214. doi:10.1177/0306624X8202600302.

Winfree, L. T., Backstrom, T. V., & Mays, G. L. (1994). Social learning theory, self-
reported delinquency, and youth gangs. Youth and Society, 26, 147-177.
doi:10.1177/0044118X94026002001.

Winton, R. (2005, May 13). L.A. home turf for hundreds of neighborhood criminal
groups. Los Angeles Times, p. B2.

Wright, J. P. & Cullen, F. T. (2004). Employment, peers, and life-course transitions.
Justice Quarterly, 21, 183-206. doi:10.1080/07418820400095781.

Wright, J., Cullen, F., & Williams, N. (2002). The embeddedness of adolescent
employment and participation in delinquency: a life course perspective. Western
Criminology Review, 4, 1-19.
66
Appendix A: Session Data for BEP Youth
Figure A1

BEP Session Data for Youth 1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 1
Gang Activity
Days With Gang
Hours Employed

67
Figure A2

BEP Session Data for Youth 2
1
4
7
10
13
16
19
22
25
28
31
34
37
40
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 2
Gang Activity
Days With Gang
Hours Employed

68
Figure A3

BEP Session Data for Youth 3
1
3
5
7
9
11
13
15
17
19
21
23
0
5
10
15
20
25
30
35
40
45
50
55
Units
Session
BEP Youth 3
Gang Activity
Days With Gang
Hours Employed

69
Figure A4

BEP Session Data for Youth 4
1
3
5
7
9
11
13
15
17
19
21
23
25
27
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 4
Gang Activity
Days With Gang
Gang Membership
Hours Employed

70
Figure A5

BEP Session Data for Youth 5
1
4
7
10
13
16
19
22
25
28
33
36
0
10
20
30
40
Units
Session
BEP Youth 5
Gang Activity
Days With Gang
Gang Membership
Hours Employed

71

Figure A6

BEP Session Data for Youth 6
1
3
5
7
9
11
13
15
17
19
22
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 6
Gang Activity
Days With Gang
Gang Membership
Hours Employed

72
Figure A7

BEP Session Data for Youth 7
1
4
7
10
13
17
20
23
26
29
32
35
38
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 7
Gang Activity
Days With Gang
Gang Membership
Hours Employed

73
Figure A8

BEP Session Data for Youth 8
1
6
11
15
19
25
32
36
40
45
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 8
Gang Activity
Days With Gang
Gang Membership
Hours Employed

74
Figure A9

BEP Session Data for Youth 9
1
5
9
13
17
21
25
29
33
37
41
0
10
20
30
40
Units
Session
BEP Youth 9
Gang Activity
Days With Gang
Gang Membership
Hours Employed

75

Figure A10

BEP Session Data for Youth 10
1
3
5
7
9
11
13
15
17
19
21
23
0
5
10
15
20
25
Units
Session
BEP Youth 10
Gang Activity
Days With Gang
Gang Membership
Hours Employed

76

Figure A11

BEP Session Data for Youth 11
1
3
5
7
9
11
13
15
17
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 11
Gang Activity
Days With Gang
Gang Membership
Hours Employed

77
Figure A12

BEP Session Data for Youth 12
1
3
5
8
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 12
Gang Activity
Days With Gang
Gang Membership
Hours Employed

78
Figure A13

BEP Session Data for Youth 13
1
2
3
4
5
6
7
8
9
0
5
10
15
20
25
30
35
40
Units
Session
BEP Youth 13
Gang Activity
Days With Gang
Hours Employed

79
Appendix B: Example of Full Model
Table B1

Full Model for Gang Activity
w

 
LPM FGLS

LPM FGLS

LPM FGLS
Coefficent -0.0059241 -0.0043432 ID 4 Coefficent 0.3616288 0.2816433 Age 17 Coefficent 0.0094179 0.0280737
Hours Employed
t -4.81 -4.37 t 4.04 2.42 t 0.12 0.51
Coefficent
0.085151 0.0719955 ID 5
Coefficent
0.3245836 0.2004947 Age 18
Coefficent
-0.1332724 -0.0965452
February
t 0.98 1.02 t 3.68 2.69 t -1.36 -0.84
Coefficent
0.1140235 0.1079314 ID 7
Coefficent
1.221081 1.156144 Age 19
Coefficent
0.043094 0.0159187
March
t 1.3 1.32 t 16.24 19.77 t 0.4 0.14
Coefficent
0.158652 0.171473 ID 10
Coefficent
0.6800939 0.6302332 Intercept
Coefficent
-0.2761312 -0.2063822
April
t 2 1.69 t 5.97 2.98 t -2.1 -1.58
Coefficent
0.1605742 0.159092 ID 11
Coefficent
0.9450914 0.8481671
R
2

0.8274 0.6807
May
t 2.04 1.54 t 11.51 9.89 Durbin-Watson Statistic 1.253 1.905
Coefficent
0.3079259 0.23175 ID 13
Coefficent
1.292264 1.191626
ρ
0.4037621
June
t 3.95 1.96 t 13.52 14.14
Coefficent
0.419541 0.330847 ID 15
Coefficent
1.1489
1.081472
July
t 5.55 2.74 t 10.95 8.64
Coefficent
0.3010972 0.2917228 ID 19
Coefficent
0.3355876 0.2594748
August
t 4.02 2.43 t 4.01 1.95
Coefficent
0.1601338 0.1223592 ID 23
Coefficent
1.252161 1.170261
September
t 2.11 1.08 t 16.26 21.41
Coefficent
0.0846898 0.0591986 ID 25
Coefficent
0.0644218 0.0364178
October
t 1.13 0.56 t 0.87 0.72
Coefficent
0.0416683 0.0245558 ID 26
Coefficent
1.102621 1.038265
November
t 0.5 0.25 t 10.68 8.54
Coefficent
-0.0962187 -0.058561 ID 27
Coefficent
1.326406 1.245473
December
t -1.17 -0.63 t 11.59 17.31

Note. LPM = Linear Probability Model. FGLS =
Feasible Generalized Least Squares. ID = Individual
Differences. t = t value. ρ = Degrees of Freedom.

Asset Metadata

Creator McDaniel, Dawn Delfín (author)

Core Title Reducing gang involvement through employment: a pilot intervention

Contributor Electronically uploaded by the author (provenance)

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 07/20/2010

Defense Date 04/05/2010

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Behavioral Employment Program,Department of Probation,employment,gang-involved,Gangs,intervention,jobs,juvenile offenders,Los Angeles,OAI-PMH Harvest,probationers,program,Treatment,Work,Youth

Place Name California (states), Los Angeles (counties)

Language English

Advisor Huey, Stanley J., Jr. (committee chair), Farver, Jo Ann M. (committee member), Margolin, Gayla (committee member), Schneiderman, Janet U. (committee member), Schwartz, David (committee member)

Creator Email dawn.mcdaniel@gmail.com,ddmcdani@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3198

Unique identifier UC1304954

Identifier etd-McDaniel-3870 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-350149 (legacy record id),usctheses-m3198 (legacy record id)

Legacy Identifier etd-McDaniel-3870.pdf

Dmrecord 350149

Document Type Dissertation

Rights McDaniel, Dawn Delfin

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email uscdl@usc.edu

Abstract (if available)

Abstract Gang-related activity exacts a heavy toll in terms of morbidity and mortality, law enforcement and corrections resources, and quality of life. Previous intervention efforts to reduce gang involvement have had negligible effects, and current knowledge about the processes that redirect youth away from gangs is limited. This dissertation attempts to contribute to our understanding of gang-related treatment process by examining the relationship between employment and gang involvement within the context of a pilot employment-focused intervention for gang-involved youth. A sample of 27 gang-involved, juvenile offenders were recruited from the Los Angeles Department of Probation and randomly assigned to either the Behavioral Employment Program (BEP) or usual probation services (US). Both within-subjects analyses (i.e., pooled time series) and between-subjects analyses (i.e., multiple regression) examined the impact of BEP-initiated employment on gang involvement. Increased employment was significantly related to reductions in gang involvement for within- and between-subjects analyses. Although no association between treatment condition and increased employment was found, BEP led to marginally significant reductions in gang involvement at 6-months post-entry. Employment programs for gang-involved youth are common in community settings, although empirical support for such programs is rare. Results from this study offer initial support for BEP and the continued use of employment as a strategy to reduce gang involvement. Confirmation of these promising findings will require future research with larger samples.