University of Southern California Dissertations and Theses

Family-based risk and protective mechanisms for youth at-risk of gang joining

(USC Thesis Other)

Family-based risk and protective mechanisms for youth at-risk of gang joining

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content FAMILY-BASED RISK AND PROTECTIVE MECHANISMS
FOR YOUTH AT-RISK OF GANG JOINING
by
Monica Lin Whitlock
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
Doctor of Philosophy
SOCIOLOGY (MARRIAGE AND FAMILY THERAPY)
May 2004
Copyright 2004 Monica Lin Whitlock
UMI Number: DP71343
All rights reserved
INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.
in the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
UMI
D issertation R jblishing
UMI DP71343
Published by ProQuest LLC (2015). Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code
uesf
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 48106- 1346
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90089-1695
This dissertation, written by
V\o»4v C jO U \a)VmV\qcVS
under the direction o f dissertation committee, and
approved by all its members, has been presented to and
accepted by the Director o f Graduate and Professional
Programs, in partial fulfillment o f the requirements fo r the
degree o f
DOCTOR OF PHILOSOPHY
Director
Date May 1 4 , 2 0 0 4
Dissertation Committee
Chair
DEDICATION
In memory of my grandfather, Richard William DeGrow.
ACKNOWLEDGEMENTS
Yes, it is finally done.
It's amazing what can happen to your life in ten years. It is alarming to
realize that I have spent a decade working toward the completion of this degree. I
have moved to five different cities during this time, I birthed two babies and buried
three grandparents. I managed to keep the same husband who also completed his
doctoral degree and have learned that two ABDs and two toddlers should not be
under the same roof at once. Yellow caution tape should be placed around the
perimeter of the building.
While this work represents a significant scholarly effort, it is only the
beginning of my pursuit of the knowledge of resiliency. It represents a mere
exploration of the mechanisms that allow surprisingly positive outcomes despite
compromised beginnings. While gang joining was never an option for me, my own
resistance to other outcomes would not have been possible without the presence of
my grandfather in my life. This work is dedicated to his memory. I am thankful for
his investment in my life. He has been my biggest fan - my greatest encouragement.
He has changed the course of our family’s destination. Generations will continue to
benefit from his strength of character.
Husband. You have sacrificed greatly to assist me in the completion of this
degree. Your constant support and belief in me sustained me through this process.
iii
Thank you for your love, for your support, and for your companionship. I applaud
your success, and look forward to the long dissertation-free days ahead of us. We
survived.
To my girls, Jordan Michal and Collin Shae. I pray that the Lord will erase
all the memories of the frustrating days when I was working too hard and carrying
too many burdens. I will forever remember your attempts to understand my ‘tashun’
and your attempts to give me space to work on it. You have been beyond noble in
sharing your mother with USC.
To my parents, Steve and Christina Murillo, and Carl and Gwen Whitlock.
Thank you for your sacrificial efforts to help me complete this degree. The girls
enjoyed many special days with you. And to my extended family and dear friends,
Janis Prince, Julie Condrin, Wendy DeBoer, Kimberly Barbarick, and so many
others, thank you for providing coveted support and encouragement. This work truly
has been a joint effort.
I am thankful for the opportunity to work with numerous distinguished
faculty members from the University of Southern California.
To Malcolm Klein. I have tested the limits of mentorship by requiring your
involvement beyond your retirement from teaching. I have learned from you how to
conduct research with the most rigorous standards. I greatly respect your insistence
on excellence and I have a passion for research because of your contribution to my
life.
IV
To Cheryl Maxson. I am honored to have occupied a place in your life where
I could observe and leam from your every move. You are a brilliant grant
administrator. Thank you for your investment of time and your mentorship.
To Tim Biblarz. Your excellence in teaching provided the inspiration for my
passion for numbers. Thank you also to Gayla Margolin, Constance Ahrons,
Alexander Taylor, Irv Borstein, Marcia Lasswell, and Carlfred Broderick for
guidance in this pursuit. I am extremely grateful for my experience at USC.
My ticket has been punched. Ready, here I go.
V
TABLE OF CONTENTS
Dedication...................................................................................................................... ii
Acknowledgements..................................................................................................... iii
List of Tables.............................................................................................................. viii
List of Figures.............................................................................................................. x
Abstract........................................................................................................................... xii
Chapter 1: Introduction................................................................................................ 1
Chapter 2: Resiliency research; conceptual considerations............................ 8
Prior resiliency research. ............ 11
Resiliency and resistance....................... 19
Risk and protective confusion; the misuse of protection 21
Potential protective processes.......................................................... 24
Conceptual model............................................................. 27
Guidelines for the measurement of protection............................. 28
Chapter 3 ; Family Processes................................................................................... 31
Family resiliency.................................................................................... 32
Gangs and families................................................................................ 34
F amily structure.................................................................................... 39
Family management........................................................................... 42
Family conformity and deviance...................................................... 44
Chapter 4; Sample and measures.......................................................................... 46
Cross-sectional data................................................................................ 47
Dependent variable................................................................................ 47
Location.................................................................... :............................... 53
Sources..................................................................................................... 55
Independent measures....................................................................... 60
Family structure................................................................................ 62
F amily management ...................................................................... 62
Family conformity and deviance.................................................. 70
Correlations............................................................................................. 70
Chapter 5 ; Analytic approach..................... 77
Cumulative risk.................................................................................... 88
Cumulative protection....................................................................... 90
vi
Chapter 6: Findings................................................................................................. 92
Bivariate results ...................................................................... 92
F amily structure............................................................................ 95
F amily management................................................................... 101
Family conformity and deviance.............................................. 110
Interaction analyses............................................................. 115
Risk model............................................................................................. 128
Cumulative risk.......................................... 130
Cumulative protection....................................................................... 133
Chapter 7: Conclusion............................................................................................. 140
Family risk............................................................................................. 141
Nondistinguishing family characteristics..................................... 143
Methodological procedure............................................................... 144
Limitations............................................................................................. 147
Bibliography.............................................................................................................. 149
Appendix 1.................................................................................................................. 158
Appendix 2 .................................................................................................................. 160
Vll
LIST OF TABLES
Table 1. Prevalence o f gang affiliation......................................................................................... 49
Table 2. Six-month prevalence rates for delinquency scales................................................. 51
Table 3. Indicators o f gang presence in study areas (percent yes).......................................53
Table 4. San D iego and study area characteristics....................................................................54
Table 5. Percent o f sample belong to a gang by age............................................................... 56
Table 6. Sources o f respondents................................................................................................... 57
Table 7. Location o f interviews...................................................................................................... 59
Table 8. Variable distributions for continuous measures........................................................61
Table 9. Prevalence o f exposure to family violence items.......................................................65
Table 10. Prevalence o f fear o f violence in hom e....................................................................67
Table 11. Prevalence o f family victimization.............................................................................67
Table 12. Prevalence o f parental defiance scale items............................................................ 69
Table 13. Correlation matrix...........................................................................................................72
Table 14. Complete variable list sorted by bivariate relationship to gang joining ....79
Table 15. Model o f expected probabilities.................................................................................80
Table 16. Prevalence o f family characteristics by gang status...............................................93
Table 17. Family risk factors for gang joining; odds ratios.................................................. 94
Table 18. Potential protective factors for gang joining; odds ratios..................................95
Table 19. Family structure...............................................................................................................97
Table 20. Percent live with two parents......................................................................................98
Table 21. Percent living without fathers.....................................................................................98
Table 22. Family structure and extended families..................................................................100
viü
Table 23. Percent live in small families...................................................................................... 101
Table 24. Percent reporting high parental discipline inconsistency...................................102
Table 25. Parental educational aspirations................................................................................ 103
Table 26. Family attachment.........................................................................................................104
Table 27. Family conflict resolution styles................................................................................105
Table 28. Family cohesiveness......................................................................................................106
Table 29. Family-based self-esteem............................................................................................ 107
Table 30. Parental defiance............................................................................................................108
Table 31. Parental monitoring......................................................................................................109
Table 32. Parents know friends....................................................................................................110
Table 33. Family gang deviance................................................................................................... I l l
Table 34. Family incarceration history.......................................................................................113
Table 35. Family drug or alcohol history.................................................................................. 115
Table 36. Relationships o f risk and protective factors........................................................... 116
Table 37. Risk factors for gang joining: multivariate logistic regression parameters... 128
Table 38. Risk factors for gang joining: multivariate logistic regression statistics -
reduced m odel...........................................................................................................................129
Table 39. Regression coefficients, cumulative risk scale........................................................131
Table 40. Odds ratios for cumulative risk scale points.......................................................... 132
Table 41. Mean scores o f protection scales by gang status................................................. 134
Table 42. Regression coefficients. Protection Scale - A ....................................................... 134
Table 43. Regression coefficients. Protection Scale - B....................................................... 135
IX
LIST OF FIGURES
Figure 1. Model o f pure protection: Protection reduces probability at high risk with no
effect at low .................................................................................................................................. 30
Figure 2. Model o f pure protection: Protection reduces probability at high risk with no
effect at low .................................................................................................................................. 81
Figure 3. Protection reduces probability at low risk with no effect at high risk............... 82
Figure 4. Protective factor reduces probability at both high and low risk..........................83
Figure 5. Protective factor aggravates at high level o f risk with no effect at low .............84
Figure 6. Protective factor aggravates at low with no effect at high.................................... 85
Figure 7. Protective factor reduces risk at high, aggravates at low ....................................... 86
Figure 8. Protective factor reduces risk at low, aggravates at high....................................... 87
Figure 9. Protective factor increases probability at both high and low levels o f risk 88
Figure 10. Prevalence o f family forms, total sample.................................................................96
Figure 11. Percent live with extended family and nonfamily adults.....................................99
Figure 12. Frequency distribution o f person in family youth gets along best with....... 104
Figure 13. Percent o f youth with family members ever in gangs........................................112
Figure 14. Percent o f youth with family members ever in jaü or prison...........................114
Figure 15. Family structure (risk) by close family adult (prot.)............................................119
Figure 16. Family structure (risk) by family closeness (prot.)...............................................120
Figure 17. Family structure (risk) by family compromises (prot.).......................................121
Figure 18. Parental monitoring (risk) by punishment consistency (prot.)........................ 122
Figure 19. Parental monitoring (risk) by reliable punishment (prot.)................................ 124
Figure 20. Parents know friends (risk) by family fights (prot.)............................................ 125
Figure 21. Family prison history (risk) by punishment consistency (prot.)......................126
X
Figure 22. Family drug/alcohol history (risk) by parental educational aspirations (prot.)
....................................................................................................................................................... 127
Figure 23. Frequency distribution o f count o f risk factors, total sample.............. 130
Figure 24. Percent joined gang by number o f risk factors...................................................131
Figure 25. Parental monitoring (risk) by Protection Scale A ..............................................136
Figure 26. Family drug/alcohol history (risk) by Protection Scale - A ..............................137
Figure 27. Family Structure (risk) by Protection Scale - B ...................................................138
XI
ABSTRACT
This research addresses the misuse of the term protective factor in the
criminological literature where it has been loosely used to describe factors that are
inversely related to the risk of an unwanted outcome. The assumption of this
research is that reducing risk and increasing protection are two different efforts. The
justification for using the term protection is satisfied only when the catalytic effect of
protection can be demonstrated. This study examines protection as a distinct
conceptual and statistical identity within the context of examining the familial
contribution to risk and resiliency development among adolescent males at high risk
for joining gangs. A definition of protective factors is utilized to identify family-
based mechanisms that place youth at greater risk for joining gangs, and family-based
protective factors that reduce that risk.
The definition of protection required that each protective factor not have a
direct effect on gang joining, but to have a statistically significant interaction with a
risk factor in a logistic regression equation. Family factors are analyzed within the
categories of family structure, management and deviance. Results of the bivariate
analysis identify ten significant family-based risk factors for joining gangs. In
multivariate analyses, three of these risk factors significantly correlate with gang
joining. Living without both biological parents, low family-based self-esteem, and
family gang history increase risk for joining gangs. Each of ten risk factors and
eleven potential protective factors were paired in a series of equations. Eight of 110
pairings of risk and protective factors yielded significant interaction terms. Four of
xii
the eleven potential protective factors reduced the probability of joining at high risk,
and three protective factors reduced it at low risk. A cumulative count of risk factors
indicated that for every additional risk factor experienced, the odds of joining
increased at a rate of 3.5. The use of an analytic approach which identifies an
interaction effect is urged to justify the use of the term protective factor.
xm
CHAPTER 1: INTRODUCTION
Although research in risk and resiliency first appeared several decades ago,
there is a resurgent interest in risk and protective factors for a multitude of
destructive behaviors in children and youth. Researchers have long been concerned
with identifying those characteristics that place certain children at risk for negative
outcomes, and there is a multitude of studies that identify risk factors for such
behaviors as depression, delinquency, violence, or substance abuse. O f fairly recent
interest, however, is the concern with identifying those youth that, despite their
increased risk for negative sequelae, nevertheless avoid the outcome for which they
are at risk. These protected youth are thought to negotiate life situations in such a
way that their risk does not operate in the same way as in unprotected youth. Their
risk is mediated by these protective factors.
There is a current misuse of the term “protective factor” in the criminological
literature that has been left unaddressed. Protection is commonly operationalized as
simply the opposite of risk. A factor is said to be protective by virtue of its inverse
relationship with the outcome. However, when one considers the original meaning
of protection as a process which reduces the effect of risk, it must, conceptually, be
entirely different from risk. Rather than simply occupying the opposite end of the
continuum of risk, protection must interact with risk to reduce the latter’s effect. If
protection represents an inverse relationship with an outcome, then the term
protection is actually unnecessary, and researchers should simply refer to varying
levels of risk and its relationship to outcomes.
Rutter, a pioneer in the development and explication of conceptual and
methodological issues surrounding the study of resiliency, explains that protection
must be something other than the positive end of the continuum; protection must
have a separate meaning (1987). While it is possible that what operates as risk in one
situation, may in fact be protective in another, we must not simply rely on the
direction of a statistical relationship, or arbitrary conceptual thinking, to guide our
understanding of the nature of the variables.
IMPORTANCE OF THE STUDY
During the Presidential Address at a recent American Society of Criminology
annual meeting, Farrington (2000) suggested we move ahead in our knowledge of
etiology to clarify this risk/protection ambiguity. This clarification of the correct use
of terms is important and timely because major national and local policies are being
developed from studies currently incorrectly identifying protective factors. Titles of
articles now commonly refer to “risk and protective factors” with no regard given to
defining the terms appropriately. Few criminological studies to date identify
protection as unique from risk. (The exceptions are commonly found in mental
health and substance abuse literatures) (see Aldwin 1994, Hetherington & Blechman
1996, and Rolf, Masten, Cicchetti, Nuechterlein, & Weintraub 1990 for examples).
The extensive use of the terms risk, protection, resiliency and buffering across
disciplines, researchers, clinicians and policy makers has resulted in a lack of clarity
of the original meanings of the terms (Cowan, Cowan, & Schulz 1996).
The most recent effort produced by the Office of Juvenile Justice and
Delinquency Prevention under the direction of Loeber and Farrington, “Highlight[ed]
the link between risk and protective factors and programming” (Loeber, Farrington,
Rumsey, Kerr, & Allen-Hagen, 1998, p. 1). However, Loeber et al.’s use of the term
protection simply refers to the obverse of risk. They made no attempt at linking risk
and protective factors, but instead defined protection as those characteristics
inversely related to serious, violent and chronic offending. These research findings
respond to the Office of Juvenile Justice and Delinquency Prevention's (OJJDP)
charge to, “Construct an analysis of risk and protective factors that will enable
communities to build effective prevention and intervention programs for serious and
violent juvenile offenders” (Loeber et al., 1998, p. 1). In addition, the recommended
approach is to measure risk and protective factors adequately in order to target
successfully those youth at risk for serious, violent and chronic offending (Loeber et
al., 1998).
Another example is the Sherman, Gottfredson, MacKenzie, Eck, Reuter, &
Bushway report to the United States Congress (1997) which explicitly defined
protective factors as, “Inversely related: the lower the level of a risk factor, the
higher the level of a protective factor” (p. 2-17). The Strengthening America’s
Families Initiative, also initiated by OJJDP, also is entrenched in the protective factor
confusion. In their first bulletin of the Family Strengthening Series, Kumpfer &
Alvarado (1998) discuss the meaning of protective factors accurately. Kumpfer &
Alvarado (1998) assert, “The goal of family-focused prevention programs should be
not only to decrease risk factors, but also to increase ongoing family protective
mechanisms” (p. 3). This is exactly right; reducing risk and increasing protection
are two different efforts. However, they then describe Bry, Catalano, Kumpfer,
Lochman and Szapocznik’s (1998) five major types of family protective factors for
substance abuse which are:
Supportive parent-child relationships
Positive discipline methods
Monitoring and supervision
Families who advocate for their children
Parents who seek information and support (Kumpfer & Alvarado 1998, p. 3).
This list could have been titled a list of risk factors (as determined by a review of
research regarding family risk factors for substance abuse) if each type of factor was
phrased in the negative direction. The list would have read:
■ Unsupportive parent-child relationships
■ Poor discipline methods
■ Poor monitoring and supervision
■ Uninvolved families
■ Lack of parental involvement in seeking information and support.
Unless it can be shown that these “protective factors” do in fact mediate the effect of
nsk, they cannot correctly be considered an indication of anything other than low
risk.
The research to be reported here is particularly essential in the field of gang
research. Public and policy interests in the proliferation of gangs have generated
much needed attention to the causes and prevention of gang joining. In the last three
decades, American cities have experienced a dramatic proliferation of gang problems
(Klein 1995; Miller 2001). The most recent National Youth Gang Survey data
estimates that more than 24,500 gangs and 772,500 gang members were active in the
United States in 2000 (Egley & Aqunan 2002). Klein (1995) reported a 345 percent
increase in the number of gang-involved cities fi*om the 1960s, and all states now
report gang problems. The growth in gangs and gang members is no longer reserved
for large cities. Miller (2001) reports, “[There] was a striking increase in the growth
of gang problems in the Nation’s smaller cities, towns, and villages. The size of the
average gang city population fell from 182,000 to 34,000, an 81 percent decline”
(Miller 2001, p. x). Until recently, when the analyses of prospective delinquency
studies (which included gang items) became available, little was known regarding
factors which placed youth at greater risk for joining gangs.
This present study examines protection as a distinct conceptual and statistical
identity using the logic of Rutter (1987) (see also Aldwin 1994 and Garmezy &
Masten 1986) within the context of examining the familial contribution to risk and
resiliency development among adolescent males at high risk for joining gangs.
While practitioners and politicians quickly cite the demise of the family as a
fundamental cause of gang joining and other youth problems, criminologists
traditionally have shown less interest in family explanations. Families from which
delinquent youth come have been examined as descriptions of the environment, but
the examination of specific family processes often has been left to the field of
psychology. This present study examines families as sources of both risk and
protection for gang joining. While the role of peers has become clear in promoting
delinquency, the family, the primary socialization agent, continues to have an
influence on youthful decisions. This study addresses the effect of alternative family
structures, family management strategies, and family deviance that might explain
gang joining.
The study begins in Chapter 2 with a discussion of the meaning of resiliency,
resistance, and risk and protective factors. In this chapter, the misuse of protection is
clarified and the resiliency research is reviewed. Chapter 3 presents the research that
addresses family sources of resiliency and discusses family contributions to gang
joining within three categories: family structure, family management and family
deviance. Chapter 4 discusses the method used for the study, including a description
of the dataset, measures, and a discussion of the analytic steps. Chapter 5 presents
the findings of the risk and protective analyses in several steps. First, the bivariate
correlates between family factors and gang joining are discussed. The findings of the
statistical interactions between risk and protective factors are then presented,
followed by a discussion of the accumulation of risk and protective factor analyses.
The concluding chapter. Chapter 6 provides an overview of the study’s purpose,
method and findings, and offers suggestions for further research.
This project is exploratory. Because the empirical data regarding risk factors
for gang joining are fairly recent, and theoretical explanations of gang joining,
specifically concerning the family, are seldom addressed, the state of knowledge on
mediating effects on the gang joining process are undeveloped. This project serves
to begin the dialogue of protection from gang joining, using a definition of protection
that has not been utilized in gang research.
CHAPTER 2: RESILIENCY RESEARCH:
CONCEPTUAL CONSIDERATIONS
Speaking of a resilient child of an alcoholic, journalist Butler (1997) wrote,
“He was self-righting; he sprang back to his original shape despite the blows of
outrageous fortune” (p. 24). There is little controversy that some youth face
seemingly insurmountable levels of risk, but manage to ignore — resist - even react
to risk in such a way as to avoid getting into trouble - and even succeed masterfully.
Studies of pathology have noted that there is great variability in the way
children and youth respond to stress (Rutter 1993). In fact, it is unusual for more
than half of those with potent risk factors to develop serious disabilities or persistent
problems (Rutter 1985; Wemer & Smith 1992). It is important to recognize this
diversity of response to risk (Hetherington & Blechman 1996). Understanding how
those who are exposed to tragic circumstances escape unscathed brings hope for
prevention (Rutter 1993). Rutter (1993) asserts:
If only we knew what it was that enabled people to “escape” damage from
serious adverse experiences, we would have the means at our disposal to
enhance everyone’s resistance to stress and adversity (p. 626).
Note that the purpose of prevention here is not to reduce risk, but to enhance
resiliency. Risk must precede resilience; it is not possible to be resilient unless one
has been exposed to risk. It is possible that without the experience of risk, resilient
characteristics would not develop. It may only be because o f the risk that the
development of resilient characteristics can occur. In other words, exposure to risk
can lead to the development of greater personal strength. It is one of the purposes of
this project to develop guidelines for enhancing resistance to gang joining among
urban youth.
But does resiliency simply represent a list of characteristics of successful
children? Can the debate surrounding distinguishing protection from risk be settled
by taking our plentiful lists of risk factors and dividing them by their positive or
negative relationships with an outcome? Certainly such a list would be minimally
informative, but it would be unhelpful in the attempt to quantify resiliency or
protective processes.
To understand the importance of distinguishing protection from risk, one
must return to the incipient meaning of protection offered by the pioneers of
resiliency research. Psychologists have advanced the understanding of children’s
ability to cope with stress in such a way that many cope, succeed and even thrive
despite increased odds of psychological disorder. Studies of resiliency, coping, stress
management, and competency development have been examined in the fields of risk
and coping research, child development, and developmental psychopathology
(Masten, Morison, Pellegrini, & Tellegen 1990; Sameroff & Seifer 1990; Wemer &
Smith 1992). Rutter, Giller, & Hagel (1998) point out, however, that much of what
is defined in studies as resiliency is in fact, “No more than variations in level of risk”
(p. 207). Much of the investigation of resiliency that has used the term protective
factor has been conducted without thought about the way those factors are associated
with reduced probabilities (Richters & Weintraub 1990).
Rutter (1993) asserts that resilience is not the avoidance of risk. He gives the
example of immunity to infections which comes from controlled exposures to the
pathogen, not through its avoidance. Discovering the processes by which children
manage their risk is something very different than simply measuring a variation in
the level of risk exposure. Protection must interact with risk in order to develop a
level of invulnerability or immunity to risk. Obviously this process is not the same
as the measurement of the opposite of risk. This present study enters this field of
investigation by extending the dialogue of exploring the continuities and
discontinuities in the development of delinquent outcomes within criminology by
asking what is different about families that increases youths’ level of invulnerability
to gang joining.
With the availability of large longitudinal, representative, ethnically diverse
datasets in the field of criminology, criminologists must approach the study of
resistance utilizing the accumulated findings from social science disciplines. The
investigation of resiliency must be conducted with the rigorous standards developed
for decades in the psychological developmental literature.
10
PRIOR RESILIENCY RESEARCH
Perhaps the most cited reference on protective factors is the landmark
publication edited by Garmezy and Rutter (1983) entitled Stress, Coping, and
Development in Childhood. Garmezy (1984) reviews the studies of resiliency and
describes three categories of factors that distinguish competent from incompetent
children. They are: 1) pre-existing personality characteristics of the child; 2) family
cohesion and an absence of discord; and 3) the “availability of external support
systems that encourage and reinforce a child’s coping efforts” (Rutter 1987, p. 316).
These factors had been shown to be associated with beneficial outcomes for children
(Garmezy 1984). While these factors may suggest likely protective processes, by
themselves, they do not represent anything more than variations in risk (Rutter 1987).
As listed, they have “limited prevention implications’ (Rutter 1987, p. 317) because
their mediating effect has not been demonstrated. It is unclear how these factors are
causally related to the outcome (Richters & Weintraub 1990). Descriptively, these
correlates are interesting, and are associated with lower rates of the various
outcomes, but it is unclear whether protective processes explain the reduced
probabilities. Analytic strategies must focus on how to elicit these protective
processes (Rutter 1987).
Rutter et al. (1998) describe several approaches that have been used in the
study of resiliency. One approach involves identifying individuals who have endured
stressful risk experiences but have managed their lives successfully. First, this
11
method of research identifies an at-risk group and then extracts the survivors for
comparison to the unsuccessful group. The purpose is to explain the success of the
survivors. Along this line, Garmezy and colleagues (Garmezy, Masten & Tellegen
1984; Masten et al. 1990) conducted a comparison in Project Competence where
they examined characteristics distinguishing high-risk resilient youth firom low-risk
youth. This method of examining resiliency, by examining high risk resilient youth
(defined by some arbitrary criteria like social competence, the absence of a criminal
record or psychiatric diagnosis) does not address the problem of distinguishing risk
from protection.
The first descriptions of youth that resisted the odds of psychological
disorders attributed the success of resilient youth to the individual character traits that
enabled them to succeed. Individual ability to respond constructively to stress was
believed to explain their successful mastery over their risk (Anthony 1974; Garmezy
1983; Radke-Yarrow & Sherman 1990; Rutter 1987); Resilience, by definition,
seemed to be located within the individual; it referred to the “individual variations in
response to risk” (Rutter 1987, p. 317). Radke-Yarrow & Sherman (1990) described
Anthony’s (1974) pioneering discussion of invulnerability in which he asked whether
resilience was a natural immunity or an “inborn higher threshold of sensitivity to
stress” (Radke-Yarrow & Sherman 1990. p. 99). However, Rutter (1987, 1993)
explains that resiliency does not in fact reside in the individual. An individual may
react to one life stressor resiliently, but in different circumstances, at a different point
12
in life, the risk negotiation may not be successful (Rutter 1987). Resilience does not
simply represent an individual characteristic of inner strength and stoicism. Rather,
it is:
an interactive and systemic phenomenon, the product of a complex
relationship of inner strengths and outer help throughout a person’s lifespan.
It is the outward and visible sign of a web of relationships and experiences
that teach people mastery, doggedness, love, moral courage and hope (Rutter
et al. 1998, p .25-26).
Radke-Yarrow & Sherman (1990) explained the success of their identified
survivors (children of depressed parents) as the survivors’ ability to minimize the
psychological and physiological costs of their risk and utilize inborn character
strengths to cope with tremendous amounts of stress. Being a child of a depressed
parent - inheriting risk -made their sample of children what they termed
‘constitutionally vulnerable’ (Radke-Yarrow & Sherman 1990). They were at risk of
negative outcomes because of parental modeling of psychopathology. Among these
children of depressed parents (n=20), Radke-Yarrow & Sherman (1990), identified
four resilient children from whom they gathered case histories. These four children
were considered successful ‘copers’ because they were happy with themselves,
physically healthy, behaving ‘masterfully’, and were “learning to be a positive
contributor to ones’ immediate society” (p. 100). They had “no psychiatric
diagnoses.. .[were] performing at grade level in school, relate[d] well to peers and
adult authorities in school and at home, and ha[d] a positive self-concept” (p. 100).
13
They found three protective qualities among this group: 1) above average IQ;
2) a quality that elicits positive responses from other persons (see also Wemer &
Smith 1992); and 3) “parental match” where some psychological or physical inborn
quality in the child matches a need in the parent. There was a “clear conception
[among these children] that there is something good and special about himself or
herself’ (Radke-Yarrow & Sherman, p. 112). Radke-Yarrow & Sherman (1990)
write, “These children receive the maximum social-emotional resources that their
families are capable of providing” (p.112-113). Each held a special place in their
families. Their styles of interacting with their families similarly operated among
extra-familial relationships where these children garnered rewarding relationships
with other adults. “The children are all alert and curious, and they impress observers
as having a zest for living. All have winning smiles and are attractive, charming and
socially engaging. All have above average intelligence” (Radke-Yarrow & Sherman
1990, p. 114).
Radke-Yarrow & Sherman then examined how effective the protective
factors were in predicting survivors. They found that 1 of 4 of the children with the
protective factors was not a survivor, and only 4 of 20 children who did not have the
protective factors were considered survivors. The assumption of this method of
research is that if the study sample is known to be at risk, then the characteristics of
resilient youth are considered protective. Case studies like this are contributive in
generating qualities of protection that should then be tested for the mediating effect.
14
Wemer & Smith are well known for their contribution to the resiliency
literature from their research in Kauai with 505 children and their parents whom they
followed from prenatal experiences to adulthood. Wemer & Smith (1992) defined
risk as those children experiencing chronic poverty, having mothers with little
education, moderate to severe perinatal complications, developmental delays or
irregularities, genetic abnormalities, and parental psychopathology. Their homes
were troubled by discord, desertion, or divorce, or marred by parental alcoholism or
mental illness. A combination of these risk factors placed children at greater risk for
major mental health problems. Two out of three who had four or more cumulative
risk factors before the age of two subsequently developed serious leaming and/or
behavior problems by age 10 or had a record of delinquencies, mental health
problems, or pregnancies by age 18 (Wemer & Smith 1992). One-third of the high
risk children had developed into a competent, confident and caring young adult by
age eighteen.
Wemer & Smith then divided the children in the study into three groups
defined by their success in adulthood: low-risk, high-risk resilient and high-risk
nonresilient. The high-risk resilient group was assessed on the basis of school and/or
work achievement, relationship with spouse or mate, relationship with their children,
relationship with parents, in-laws and siblings, relationship with peers, and the
degree of overall satisfaction an individual expressed with his or her present state in
life. A criminal record, a record of spouse or child abuse or delinquent child support.
15
and a record of chronic substance abuse, and/or psychosomatic or psychiatric
disorders were considered signs of unsuccessful adaptation to adult life (Wemer &
Smith 1992). Wemer & Smith’s research design was advantageous because it
allowed for the examination of protective factors that occurred subsequent to the
onset of risk. They found that, “As disadvantages and the cumulative number of
stressful life events increased, more protective factors in the children and their care
giving environment were needed to counterbalance the negative factors and to ensure
positive developmental outcomes” (Wemer & Smith 1992, p. 2). This makes sense;
the more risk one encounters, the greater the need for protection. While Wemer &
Smith did not utilize an analytic strategy that identified statistical interaction terms,
they identified factors that were associated with competence among a high-risk
sample. Wemer & Smith’s list of risk and protective factors is a list of variables that
are either positively or negatively associated with competence.
Wemer & Smith’s work clearly fits within a systemic framework wherein
children are viewed as dynamic organisms that interact with the environment; this
exchange of experiences and relationships contributes to each child’s level of
functioning. This level of functioning can be viewed on a resiliency-vulnerability
continuum. Each child has a place on this continuum resulting from his biological
and psychological foundation. Each micro process of risk and protective
experiences, then, alters one’s level of resilience. Wemer & Smith note that
vulnerability is different from risk factors, which she defined as “biological or
16
psychosocial hazards that increase the likelihood of a negative developmental
outcome in a group of people” (Wemer & Smith 1992, p. 3). Vulnerability refers to
an “individual’s susceptibility to a disorder.” For example, Wemer & Smith
identified some youth that were vulnerable, even in the absence of biological stress
or financial constraints. Some at-risk youth seem to elicit negative responses fi*om
their environments. They note, “The bases of resistance are both environmental and
constitutional, and the degree of resistance varies over time and according to life’s
circumstances (Wemer & Smith 1992, p. 4).
In their longitudinal investigation of the lives of children of schizophrenic
women, Sameroff & Seifer (1990) avoided the confusion of what to label various
factors by examining “a set of variables as they impact on competence in children,
without introducing conceptual difficulties associated with categorizing all measures
into one of two categories (risk or protective)” (p. 63). Sameroff & Seifer (1990)
eschew the conceptual debate, and highlight the complexity in differentiating risk
from protection. They claim to be “more concerned with identifying the relevant
systems that impact the development of competence in individual children, whether
or not they develop in the context of one or another type of family” (Sameroff &
Seifer, p. 63). They explain that if researchers adopt a medical model wherein risk is
a constitutional deficit, then protection is clearly defined as anything other than that
constitutional deficit. If, however risk is not so clearly delineated, and may in fact
reside in multiple factors and domains of life, the definition of risk becomes more
17
difficult (Sameroff & Seifer 1990, p. 62). Sameroff & Seifer (1990) suggest an
alternative strategy of defining risk in the selection of the sample. The risk factors
are chosen a priori, and subjects are selected based on these eligibility criteria. An
example of this for gang research would be defining the at-risk sample based on
neighborhood criteria of gang presence in the neighborhood. The apparent problem
with this strategy is whether to include with this gang-presence risk factor all of the
correlated risk factors that co-occur with gang presence, such as low SES, exposure
to violence, and drug availability.
Like others (Radke-Yarrow & Sherman 1990; Wemer & Smith 1992), Pianta
et al. (1990) assessed levels of risk based on how individuals scored on various a
priori defined risk factors. They used one composite score of risk and divided the
children at the median. The high risk - or top half - of the children were then
divided into competent and incompetent groups. They derived competence groups
using hierarchical cluster analyses of the Child Behavior Checklist -Teacher report.
They then examined the differences between means on the various factors. The
significant factors were considered protective factors. Again, this does not avoid the
obverse problem except that the protective factors occurred at a later point in time. It
seems important to address how risk is assumed in the selection of the sample
(generalized risk), and how risk for specific outcomes is measured (O’Dougherty &
Wright 1990).
18
RESILIENCY AND RESISTANCE
The term resilience has been used in the literature to represent cumulative
masterful experiences in negotiating risk (Anthony 1974; Radke-Yarrow & Sherman
1990). It has been used to describe an overall positive functioning in which a youth
has avoided many negative outcomes; it is developed through successful responses
to risks. Thus, resilient youth may not be resistant to all risk outcomes (Rutter 1993;
Radke-Yarrow & Sherman 1990).
Rutter (1993) describes resistance as the ability to avoid a specific outcome.
Resistance represents specific combinations of risk and protective factors for
particular outcomes. Wemer & Smith (1992) write, “Whereas resilience is a
characteristic that varies from person to person, protective factors or mechanisms are
more specific and more narrowly defined. Protective factors modify (ameliorate,
buffer) a person’s reaction to a situation that in ordinary circumstances leads to
maladaptive outcomes” (p. 5). Thus, protective processes enable resistance by
mediating the effect of risk for specific outcomes, and repeated successful masteries
over risk (resistances) result in resiliency.
Rutter (1993) maintains that one must examine the various ways that people
may suffer and attempt to accommodate that diversity in measures of resilience).
Studies that focus on only one dimension of functioning are measuring resistance to a
specific risk, not resilience. This caution is especially relevant for criminological
19
research where multi-dimensional outcome research is conducted by only a very few
(see Elliott, Huizinga & Ageton 1985; Elliott, Huizinga & Menard 1989).
Resiliency can mask other negative symptoms that result from risk
experiences. Youth may experience a series of protective processes that enable them
to avoid certain risks, thus appearing successful and adaptive by most measures;
however, there may exist other complications such as depression that results from
their risk experience. Hauser cautions, “Protective factors do not necessarily lead to
resilience” (1999, p. 4). If the severity of the risk is too great, protective factors may
not provide sufficient buffers (Hauser 1999). Some children are more resistant than
others (Rutter 1993). Only with comprehensive examinations of total functioning
can resiliency adequately be measured (Radke-Yarrow & Sherman 1990).
Radke-Yarrow & Sherman (1990) provide a description of the costs of
survival that appeared among their survivor sample. They insist that children who
survive are not invulnerable. In fact they pay considerable costs for their survival. In
a case study of one of their survivors, they explain that it was clear that the child
expended an enormous amount of emotional energy to win love and affection from
others, but that the payoff was adequate for this expenditure. They suggest it may be
that resilient children are pushed to early maturity and are unable to demand that
their needs are met within the family (Radke-Yarrow & Sherman 1990).. Radke-
Yarrow & Sherman note that Anthony (1974) asks if it is “useful to imagine coping
20
behavior as drawing on a reservoir of resources that can eventually be ‘used u p / with
the costs appearing later in development” (Radke-Yarrow & Sherman 1990, p. 99)
Rutter (1987) suggests that there has been a shill in the locus from risk
variables to a negotiation of risk situations. This process whereby the individual
responds to risk in such a way that a positive outcome is produced may be difficult to
measure. How is it that one youth can witness a violent event, and yet not choose to
participate in the violence, while a second youth may witness the violence and decide
the excitement and rush of the street life is appealing? What is the iterative process
that explains how youth exposed to the same risk, even the same commitment to
delinquent peers, can yield such opposing outcomes? The current research is
concerned with identifying factors that reduce the predicted effect of the risk by
examining one possible outcome of residing in urban gang prevalent areas - gang
joining. This is a study of resistance.
RISK AND PROTECTIVE CONFUSION: THE MISUSE OF PROTECTION
Making the distinction between a risk and a protective factor is more than
simple rhetoric. “Important answers often are obscured by the ways we frame our
questions” (Richters & Weintraub 1990, p. 67); framing protection as the opposite of
risk is misleading because it neglects the unique mediating effect of protection.
If the researcher’s intention is to describe factors such as school achievement
as either positively or negatively associated with a delinquent outcome at a point in
21
time, then the research path many have taken in the discussion of risk and protective
factors can be continued. That is, many researchers have a priori conceptualized
certain variables as either risk or protective and tested them as such. The
implications of the findings from this method are that increasing or decreasing the
level of the variable will either increase or decrease the outcome. Richters &
Weintraub (1990) note that factors that reduce the probability of negative outcomes
“are not necessarily protective factors in any meaningful (scientifically productive)
sense of the word” since they indicate only the absence of causal factors (p.79).
Inversely related factors do not demonstrate protection. Richters & Weintraub
(1990) further describe their concerns below:
Importantly, our concern here is not with the protective factors concept itself
but rather with the practice of automatically invoking it as a quasi explanation
for virtually any factor associated with reduced rates of negative outcomes
among high-risk offspring. Garmezy rendered an invaluable service to
psychology by emphasizing the need to consider protective or buffering
factors in any meaningful analysis of the influences of childhood stress.
Progress beyond Garmezy’s insights to an empirically based understanding of
the processes through which protective factors work, however, will require a
prior understanding of the processes through which stressors themselves exert
an influence. Moreover, a necessary (though not sufficient) condition for
invoking ‘protective factors’ as an explanation for positive outcomes in high-
risk offspring samples should be a demonstration of the proximal stressors to
which the offspring are being subjected and from which they are being
protected, (p. 80).
Neglecting this distinction obfuscates any potentially instructive discussion of
reducing the effect of risk (Kumpfer 1994). Reducing the exposure to risk, however,
is not the same as protecting those at nsk. It may not be possible to reduce risk, as in
reducing one’s exposure to violence, but, once the youth is exposed to violence, there
22
may be points of intervention which may protect the youth from the risk already
experienced.
So the resolution to the terminology confusion is quite simple; the protective
factor must protect from something Jenkins & Smith 1990; Richters & Weintraub
1990; Rutter 1987). To justify the use of the term protection, we must ask what
youth are being protected from and demonstrate its effect (Richters & Weintraub
1990; Wemer & Smith 1992). Rutter (1981) explains, “The notion...is of factors
which are largely inert on their own but, to use a chemical analogy, serve as catalysts
when combined with acute stressors of some type.. .When they tend to reduce the
effect of stressors, they are usually termed protective factors” (p. 340-341). The
protective factor must statistically interact with the risk factor to reduce the
probability of the outcome; “ .. .the vulnerability or protective effect is evident only
in combination with the risk variable” (Rutter 1987, p. 317). This distinction is more
than merely semantic; “[it has] and will continue to have important consequences for
our ability to refine our risk classifications and progress toward a process-level
understanding of the origins of deviance in affected high-risk offspring” (Richters &
Weintraub 1990, p. 90). It appears that this caution asserted over a decade ago has
not been integrated into etiological criminological research.
Rutter (1993) defines protective processes as, “Features that, while not
directly promoting good outcomes, enhance resistance to psychosocial adversities
and hazards of various kinds” (p. 630). Notice in this definition that protective
23
processes do not have a direct effect with an outcome. Protective mechanisms
operate indirectly to reduce the effect of risk mechanisms. In Rutter’s (1991) critique
of Jenkins & Smith’s (1990) work on children from disharmonious homes, he
emphasizes that a protective factor must statistically interact with risk, while it may
or may not have a main effect on the outcome (Rutter 1991, p. 151). He explains:
Instead of searching for broadly based protective factors we need to focus on
protective mechanisms and processes. That is, we need to ask why and how
some individuals manage to maintain high self-esteem and self-efficacy in
spite of facing the same adversities that lead other people to give up and lose
hope. How is it that some people have confidants to whom they can turn?
What has happened to enable them to have social supports that they can use
effectively at moments of crisis? Is it chance, the spin of the roulette wheel
of life, or did prior circumstances, occurrences, or actions serve to bring
about this desirable state of affairs? The search is not for broadly defined
protective factors but, rather, for the developmental and situational
mechanisms involved in protective processes (Rutter 1987, p. 317).
POTENTIAL PROTECTIVE PROCESSES
To move beyond the empirical search for protective factors and mechanisms,
it is important to begin to determine patterns of mechanisms that indeed reduce the
impact of risk. Rutter (1993) developed a series of guidelines from the available
literature, describing probable characteristics of protective mechanisms. They
included the following:
1) those that reduce sensitivity to risk (such as by previous successful
coping with challenges);
24
2) reduction of risk impact (as by parental supervision or monitoring, a
positive peer group, avoidance of drawing children into parental conflict, and the
child’s own distancing from a deviant parent);
3) reduction of negative chain reactions (e.g., successful handling of
family conflict, effective social problem-solving strategies, avoidance of damaging
coping strategies such as drugs or alcohol);
4) increasing positive chain reactions (such as through the eliciting of
supportive responses from other people);
5) promotion of self-esteem and self-efficacy (as through secure and
supportive personal relationships, responsibility and success in task
accomplishments, and successful coping with manageable stresses);
6) neutralizing or compensatory positive experiences that directly
counter the risk effect;
7) opening up of positive opportunities (as through educational and
career opportunities, broadening of marital choice, and change of home
environment);
8) positive cognitive processing of negative experience (acceptance
rather than denial or distortion, focus on positive aspects, and incorporation into
personal schemata).
25
Rutter et al. (1998) note that these protective concepts are preliminary hypotheses
derived from the limited resiliency research available. These guidelines provide
direction to researchers for distinguishing protective from risk mechanisms.
In addition, Rutter et al. (1998) identified qualities that promote resistance for
antisocial behavior that emerged from the available research. They include:
1) lack of genetic vulnerability;
2) higher IQ;
3) temperamental and other personality features;
4) the maintenance of a stable, v^arm, harmonious relations v^ith at least one
parent - in the context of overall family discord;
5) parental supervision;
6) good experiences at school;
7) a prosocial peer group;
8) experiences that open up ne^v opportunities through academic
advancement or change in peer group or altered social circumstances may
provide a turning point;
9) an attitude of mind that involves a sense of self-efficacy, a positive
approach to planning, and social problem solving may be protective, (p.
210-211).
Clearly what can be seen from the literature is that investigations of protective factors
must take into account individual characteristics, family interactions, and the broader
social system (Pianta et al. 1990). Radke-Yarrow & Sherman point out that the
challenge is to understand how factors such as intelligence, sex, age and family
socioeconomic status can “deflect risks and reflect processes that mediate the
experiences of the individual” (1990, p. 101). Wemer & Smith (1992) summarize:
26
Despite the heterogeneity of the risk conditions studies, and despite
conceptual and methodological differences in the assessment of quality of
adult adaptation, one can begin to discern a common core of individual
dispositions and sources of support which ameliorate or buffer a person’s
response to both constitutional risk factors (such as parental
psychopathology) or stressful life events (economic hardship, divorce,
breakdown of parenting). These [Wemer & Smith’s] findings have been
replicated in different historical eras, across different generations, and in
different metropolitan areas of the US and Europe, with both Black and
Caucasian populations (p. 13).
CONCEPTUAL MODEL
The purpose of this research is threefold:
1) to examine risk processes in families that differentiate gang from nongang
youth;
2) to examine distinct protective processes in families that reduce the impact
of risk on gang joining; and
3) to address the ambiguity in the use of the terms risk and protective factors
by advancing the measurement of protective mechanisms.
The primary question conceming this research is this: Given the same risk status,
what protective processes within the family function to reduce the effect of risk so
that a high risk youth has odds of joining a gang approaching those of a low risk
youth, despite their different levels of risk?
27
GUIDELINES FOR THE MEASUREMENT OF PROTECTION
The following guidelines for the measurement of protection are adapted from
Rutter’s (1987) conceptualizations of risk and vulnerability processes. These
guidelines serve as hypotheses of the manner in which protection is expected to
function.
1) Protective factors are distinct from risk factors.
Conceptually, protective factors represent distinct processes, and statistically
this distinctness can be captured through the measurement of statistical interactions
between risk and protective factors.
2) Protective factors do not have a direct relationship with the outcome.
Since a protective factor is a response to risk, it cannot have a direct effect on
an outcome. A risk factor directly increases the probability of negative outcomes. A
protective factor interacts with a risk factor to reduce the likelihood of the occurrence
of the event (in this case, gang joining).
3) Protection is a process that is risk specific.
Since protection is a process that reduces the effect of risk, it is risk specific.
That is, protection responds to - or kicks in - when exposed to certain types of risk;
i.e., it is a form of resistance rather than of resiliency. Protective processes may have
28
no effect on certain forms of risks. Protective processes are micro-processes.
Accumulated protective experiences contribute to an overall level of resiliency. So
to characterize resilient individuals, one must identify the processes that reduced the
risk effects.
4) Protection reduces the effect of risk only at higher levels of risk.
Protection will have no effect at low risk where it is not needed; protective
factors are only necessary buffers at higher levels of risk. In other words, protective
factors respond to risk such that in a no-risk sample, the protective factor does not
make a difference in prediction to gang membership. On the other hand, with a high-
risk sub sample, the protective factor occurs in response to the risk experience, and
together, the risk and protective combination protect the youth from joining a gang.
This means that a correctly termed protective factor statistically interacts with risk
with the result that the high risk youth who is ‘protected’ displays an outcome similar
to the lower-risk youth. This relationship is displayed in Figure 1 where gang joining
decreases as protection increases for those at high risk, while gang joining for those
at low risk is unaffected by protection.
29
Figure 1. Model of pure protection: Protection reduces probability at high risk
with no effect at low.
High R isk Low Risk
I
I
I
High Protection Low Protection
30
CHAPTER 3: FAMILY PROCESSES
This project is exploratory. It involves an application of a method of
distinguishing risk from protection among youth at-risk for joining gangs.
Theoretical propositions conceming the operation of family stmcture, management
and norms on gang joining can be developed after the definitional and
methodological considerations are overcome. It is not possible to develop
hypotheses conceming the relationship of family factors until it can be established
that protection is tmly different from risk and is operating in reaction to the risk.
Until this is accomplished, policy will continue to be built upon a misunderstanding
of the meaning of protection (see Catalano & Hawkins 1996 and Loeber et al. 1998)
and research will continue to be ineffective at locating the family processes which
actually intermpt the gang-joining process. This is a terribly understudied area;
much information is missing regarding a clear explanation of the role family
stmcture, process and socialization play in the evolution of a gang member. This
project attempts to address this gap in the literature by examining relevant family
factors that show considerable promise.
This chapter begins with a presentation of resiliency principles developed
from the family resiliency research. While this body of research does not address
gang joining, it addresses potential family sources of resiliency which can inform the
study of gang families. In the remainder of the chapter, the research which examines
31
gang members’ families is examined within the three areas of influence: family
structure, family management and family norms.
It is helpful to consider individual behavior within the framework of systems
theory. The application of systems theory to the family suggests that, just as in a
cybernetic system, the family functions as a whole so that when something is wrong
with the system, symptoms of the inefficiency emerge in and among family members
(see Broderick 1993). The symptom of a faulty system might evidence itself among
different family members and among different dyads at various times in the family
life course. From this perspective, gang membership can be considered a result of
community, family, peer, and individual system inefficiencies that result in a youth’s
decision to join a gang. It is this systemic assumption that allows us to examine gang
joining as an effect of disordered families.
FAMILY RESILIENCY
Resiliency researchers have observed families in order to discern the familial
predictors of resiliency among youth. While some studies simply observed low risk
families, or did not examine the differential impact of family processes on siblings
(Kumpfer & Alvarado 1998), an exception is the Family Research Consortium on
Risk and Resilience, which, while not addressing the etiology of gang joining,
examined the family processes that “support healthy or problematic development”
(Cox & Brooks-Gunn 1999, p. 1). One of the Consortium’s products, an edited
32
volume entitled Conflict and Cohesion in Families, addressed the multi-level
systemic impacts on the family system. Similarly, Sameroff & Seifer (1990) describe
their work as examining the “social and cultural pressures that serve to promote the
maintenance of strong family relationships through development and the means by
which these strengths are manifest in child competence” (p. 62-63). Clearly, an
individual’s behavior cannot be evaluated without an examination of the many layers
of influences on development.
Hauser, Vieyra, & Jacobson (1989), in their review of the literature of
families and resiliency, describe the direct and indirect influences of the family on
individuals. The direct influences include three facets of the family; 1) family
structure and home environment (e.g., number of siblings, and household
composition); 2) parental attitude, support and discipline; and 3) family
communication and interaction, which they refer to as micro-processes (Hauser et al.
1989). The indirect effects of the family include the way in which the family system
influences personal characteristics such as self-esteem, personality, and temperament
(Hauser et al. 1989). It seems likely that these family influences would affect peer-
selection and gang joining as well.
In addition, Richters & Weintraub (1990) emphasize the importance of
examining the effects of family processes on siblings. It cannot be assumed that
family experiences are shared among siblings; the experiences of parental
monitoring and affection may differ for each child within the family. Rutter (1993)
33
gives the example of family discord that can affect each child in the home differently.
He explains that protection may come in the ability of one child to avoid being
scapegoated or brought into the family discord. Protection may also come by way of
holding a prized place within the family, and by bringing out positive interactions
from parents (Radke-Yarrow & Sherman 1990). Thus it is important to recognize
the individual experiences of family process variables; “Nonshared environmental
influences tend to have a greater effect than shared ones” (Rutter 1993, p. 627). In
fact, Richters & Weintraub (1990) assert, “Neglecting to look at individual
differences within families may explain some of the weak associations found
between family factors and negative outcomes in prior research” (p. 82). While the
examination of sibling pairs is not undertaken in this study, it is important to
recognize that family characteristics affect children differently.
Perhaps family protective mechanisms that have been identified in the
resiliency literature with various outcomes also can be found in this present research
with the dependent variable, gang joining. This present research adds to this family
resiliency literature by examining family sources of resiliency for youth at risk of
joining gangs.
GANGS AND FAMILIES
While family research and gang investigations are seldom joined, examining
family process is an obvious foundation from which to begin to understand the
34
development of gang membership. Neglecting to examine the family relationships
that may fundamentally affect the progression of gang-related behavior provides only
a limited picture of the gang-joining process.
Traditionally, criminological theories viewed the family as an important
context in which delinquency emerges. Rutter et al. 1998 explain, “Many of the
early theories about the causes of crime took as their starting point an assumption
that most delinquents came from a socially disadvantaged background...The findings
across studies of diverse samples are reasonably consistent in showing that much of
the risk for antisocial behavior associated with poverty and social disadvantage is
moderated by the adverse effects of prolonged economic (and associated) stresses on
family functioning” (p. 199, 201). Often, however, proximal factors such as
delinquent peers and poor school achievement are given more explanatory attention
than the possible root causes. Rutter et al. 1998 draw attention to the difficulty in
isolating the effect of specific risk mechanisms from constellations of risk processes.
Recently, there has been a modest swing of interest among criminologists
toward the re-examination of the family processes that contribute to developing
delinquent youth (although not necessarily gang members). This resurgent interest in
understanding the role of families in delinquency, and in identifying efforts toward
strengthening families, is evidenced in the work of the Office of Juvenile Justice and
Delinquency Prevention (OJJDP) which commissioned a series of studies that
describe successful strategies to improve family functioning (see Kumpfer &
35
Alvarado 1998). In addition, in 1993, OJJDP instituted a Comprehensive Strategy
fo r Serious, Violent, and Chronic Juvenile Offenders (Wilson & Howell 1993) which
developed programs nationwide to strengthen the family and other social institutions
in an attempt to address the needs of serious, violent and chronic juvenile offenders
(Loeber et. al 1998). Families are an obvious link to youth behavior, but little
attention has been given to the ways that the same family can produce both a
‘struggler’ and a ‘survivor.’ The institution of the family warrants this attention.
Surprisingly few gang researchers address family contributions to gang
joining. It is striking that major texts of gang research do not discuss the families
from which gang members emerge. For example, Klein’s (1995) seminal work The
American Street Gang and Covey, Menard & Franzese’s (1997) textbook
Gangs have no reference to families in the indexes. Spergel’s (1995) text. The Youth
Gang Problem devotes three pages to a discussion of the way that families may
predispose youth to gang joining. While Decker and Van Winkle (1996) have an
entire chapter devoted to families, only a few pages of the chapter address possible
family causes. Most of the chapter addresses the impact of gang joining on families.
Thomberry, Krohn, Lizotte, Smith & Tobin (2003) tested seven family-based risk
factors for gang joining, none of which significantly associated with gang
membership (for males only). Thus, only parental attachment to child was included
in a model of causal processes associated with gang membership. Family deviance
36
was not examined. The role of the family as the foundation from which deviant
networks develop is often presumed, but seldom explicated.
Despite its neglect in the literature, the impact families can have on youth
development and decisions to join a gang can be profound. Weak, aggrieved
families may provide the shaky foundation ftom which youth easily slip into the
attractiveness of gang affiliation. This premise was found in the early works of gang
researchers (Cohen 1955; Thrasher 1927). For example, Cohen (1955) explains,
“The child’s relationships to his family may in various ways facilitate or interpose
barriers to such [gang] participation” (p. 153). It is this very statement asserted in
1955 to which this research now returns.
Gang researchers have stressed the importance of recognizing the economic
and neighborhood contexts ftom which urban families at risk attempt to raise their
children (Decker and Van Winkle 1996; Moore 1991; Thomberry et al. 2003; Vigil
2002). Decker and Van Winkle (1996) note, “Within the context of declining
neighborhoods, families - generally headed by poor single women - often stood
alone in their efforts to control the behavior of their children” (p. 230-231).
Certainly, neighborhood constraints chisel away at family strengths.
Urban researchers have offered propositions to explain the process by which
economic hardships may lead to family difficulties. While it is beyond the scope of
this study to address the effects of neighborhood stress, the study youth reside in
areas characterized by hardship and disorder. Thus, a brief discussion of the effect of
37
neighborhood context on families reminds the reader of the exogenous influences on
the family’s ability to parent successfully.
Vigil (1988) conducted an in-depth ethnographic analysis of East Los
Angeles gangs, in which he concluded that the process of multiple marginality,
where families are economically and socially marginalized, leaves their youth to the
socialization of the streets. Vigil (2002) explains how economic and racial
discrimination have left families with fathers who were absent or “tenuously
connected” with their children. Long-term unemployment in the communities which
resulted from discrimination and overcrowding in Los Angeles resulted in families
headed by working women who spent considerable time away from home. Moore
(1991) cautions that strain associated with poverty is not sufficient to explain gang
joining. Moore’s (1991) gang and nongang families did not differ demographically;
“It is when we turn to the emotional climate of the family during childhood that
problems begin to be evident” (p. 89). Wilson (1987), while not examining gang
families in particular, concluded that economic hardships lead to the dissolution of
marriages; when families are pressed and stressed financially, Wilson proposes,
there is little time to supervise and engage children in positive activities.
Most gang research has assumed that youth approached attraction to join
gangs from even playing fields in terms of their families. The processes within the
family that either encourage or prevent a youth from joining a gang have been under
investigated, and no researchers have examined the processes that foster resiliency.
38
Instead, the available research examines the characteristics that are either positively
or negatively related to gang joining. Family factors such as socio-economic status
or family structure are treated as control variables, while process variables are
seldom examined. The studies that do predict to gang membership are reviewed
below within the categories of family structure, family management and family
norms.
FAMIL Y STR UCTURE
Female-headed households have been targeted politically as representing the
demise of the family, and youth workers quickly cite the absent father as a major
contributing factor to the demise of youth. For example, interviews with personnel
from the community-based agencies that participated in this present research
revealed that many community residents and agency personnel believe that father
absenteeism was the reason youth became involved in gangs. Whether confirmed
with research or not, family structure is quickly cited as one of the primary reasons
youth seek out the camaraderie of a gang.
The early assumptions about the influence of fathers on families were that
their roles were critical (Tennyson 1967). Neglecting to address the effect that a
father’s absence may have on a child relegates fathers to a very small influence on
the lives of children. Since Miller’s (1958) observation that gangs were more
prevalent among communities consisting of large portions of female-headed families.
39
there has been a supposition that family structure is related to gang joining. Cohen
(1955) noted, “youth from ‘broken’ homes or from homes where both parents work
receive less supervision over their associations” (p. 153). In Vigil’s (1988) analysis
of East Los Angeles families, most of his Mexican gang informants were from single
parents; fathers were absent by death or divorce. Decker & Van Winkle’s (1996)
interviews with African American gang members revealed that most gang members
did not live with both parents, and a quarter of those studied lived away from both
parents.
In constrast, Tennyson (1967) observed African American and Caucasian
boys in Chicago and found little difference in family structure among gang and
lower-class nongang boys within race. Lyon (1991) also found that family structure,
when examined as one or two parent homes, did not distinguish gang and nongang
families. Results from the Pittsburgh youth study did not conclude that gang
members were more likely to come from single-parent families (Lahey, Gordon,
Loeber, Stouthamer-Loeber, & Farrington 1999). Similarly, Curry & Spergel (1992)
did not find that family structure explained gang-related activities.
Recent findings from two of the large high-risk prospective studies that were
designed to address the causes and correlates of youthful offending indicated that
living with intact families did distinguish gang members from nongang youth
(Esbensen, Huizinga & Weiher 1993 in Denver; Thomberry 1998 in Rochester). In
their analysis of the Denver Youth Survey, Esbensen et al. (1993) found that gang
40
members differed from nongang nonoffenders in terms of the structure of their
families. They did not, however, differ from nongang street offenders (40% of
nongang nonoffenders, 44% of street offenders and 47% of gang members were from
single parent homes.) Esbensen et al. (1993) note that, “Gang members were less
likely to live in intact families. Sixteen percent of gang members lived in intact
families and 35% of nonoffenders were from intact families. Twenty-three percent
of gang members lived in “other” arrangements while 9% of nonoffenders lived in
comparable arrangements.” Battin-Pearson, Guo, Hill, Abbott & Hawkins (1998)
found that having one parent or one parent plus one other adult in the household
predicted transient gang members v. nonmembers. “Youth with no parents in the
household were four times as likely to become multiple-year gang members” (Battin
et al. 1998).
Larson & Myeroff (1967) cautioned long ago that family structure alone is an
ambiguous measure reflecting other independent processes. Tennyson (1967)
rightfully notes that parental attachment should be considered an intervening factor
between family structure and gang joining. The intervening effect of family
management strategies on gang joining for various family structures should also be
considered. In addition, measures of family structure may be deceiving in that they
hide the existence of the conflictual, strife-ridden families that remain intact (Larson
& Myeroff 1967). Conflict in families can be more harmful to children than the
effects of separate spousal living arrangements (Ahrons & Rodgers 1989).
41
FAMIL Y MAN A GEMENT
As Thrasher’s first work on gangs indicated, “The most important agency in
directing the sparetime activities of the boy is the family” (Thrasher 1927, p. 65).
Thrasher (1927) considered lack of parental supervision as a contributor to gang
joining. Females, he said, were less involved in gangs because they received greater
parental supervision and held increased responsibilities for care of the home. It is
this form of supervision over activities that he asserted could prevent boys from
joining gangs.
Lack of supervision distinguished gang and nongang families (Lahey et ah
1999; Loeber et al. 1998; Smith, Lizotte, Thomberry & Krohn 1997; Thomberry
1998 - for females). Results from the Pittsburgh Youth Study (Lahey et al. 1999)
found lack of parental supervision predicted early gang joining, but less supervision
was associated with lower risk of serious gang entry in late adolescence. Walker-
Bames & Mason (2001) found that parenting practices predicted gang activity even
after considering the influence of a negative peer group. They found that for black
youth, parenting played a particularly clear role. “Higher levels of behavioral control
and lower levels of lax and psychological control were related to decreases in gang
involvement for Blacks, with higher levels of behavioral control and lower levels of
lax control also related to decreases in gang delinquency” (Walker-Bames & Mason
2001, p. 1826).
42
Many ethnographic examinations of gang members’ families suggest that
poor family management strategies encourage gang joining. For example, Moore
(1991) found that the emotional climate of the family (as expressed between parents,
and between parents and children), and violence (including incest) were
characteristic of many East Los Angeles gang families, although female gang
members reported more family problems than male gang members. Moore, however,
did not include a nongang comparison group, so it is impossible to discern how
different the gang members’ families were from other families in the area. Vigil
(1988) examined several East Los Angeles Latino housing project gangs and their
families and found that parental practices and family relations differentiated gang and
nongang families.
The findings from the quantitative studies show similar patterns. Low
attachment to mother (Hill et al. 1999; Thomberry 1998 - for males), and family
management problems (Thomberry 1998), including poor parent-child
communication and serious marital discord (Loeber et al. 1998) were found to
increase the likelihood of gang joining. Gang families also were characterized by
harsh and erratic discipline practices and maltreatment or neglect (Loeber et al.
1998). Adler, Ovando & Hocevar (1984) studied 30 Mexican American gang
members and found they spent less time with their families in social activities and
that their families expressed fewer positive feelings toward one another.
43
FAMILY CONFORMITY AND DEVIANCE
Few would doubt the tremendous continuance of deviance across generations.
Youth who are exposed to nonconforming behaviors within their families - their
primary place of valuation and identity development - are likely at greater odds of
choosing deviant activities (Rowe & Farrington 1997; Sampson 1992). This appears
to be especially true for gang joining in Los Angeles where generations of families
have affiliated with gangs (Klein 1995; Moore 1978, 1991; Vigil 1988, 2002).
Vigil (2002) describes the experiences of Mexican families in Los Angeles where the
father is “traditionally the head.. .and his stable presence brings a great deal of social
control and security to the family (p.39).” Fathers who are absent by desertion,
death, or involvement in the gang subculture are “absent and unable to provide the
ongoing direction and supervision [their] children need.. .a father’s life and behavior
nevertheless influence his offspring by providing a symbolic source of antisocial
identification for them to emulate” (Vigil 2002, p.39).
It is not uncommon for youth to say they joined a gang because a family
member (e.g., brother, cousin, or uncle) was in a gang (Vigil 2002). This
relationship was less evident among the St. Louis gangs studied by Decker and Van
Winkle (1996). Most gang members’ parents were not involved in gangs, although
45% of their gang members had a brother in a gang (p. 233). Curry & Spergel (1992)
also noted that the presence of a gang member in the family was the best predictor of
gang activity.
44
Other measures of family deviance have been found to correlate with gang
joining. Parental substance abuse (Loeber et al. 1998) and antisocial or criminal
behavior increase the likelihood of gang joining greatly (Loeber et al. 1998; Moore
1991; Curry & Spergel 1992). In the Seattle longitudinal study, parental attitudes
favorable to violence increased the likelihood of gang joining (Hill et al. 1999).
Moore’s exgang members reported families which experienced heavy alcohol use,
drug addiction, and arrests (Moore 1991). Esbensen et al. (1993) found that gang
members differed from non-gang members in family normlessness, but gang
members reported similar levels of normlessness to nongang street offenders.
Rutter & Quinton (1984) reviewed the research in search of possible
explanations of the processes by which parental psychopathology and offspring
maladjustment are linked. They identified several causal mechanisms: 1) genetic
transmission; 2) direct exposure to parent’s symptoms; 3) indirect effects such as
marital discord. Thus, it is particularly the latter two mechanisms through which
children are socialized into family norms and values that maintain gang affiliation
across generations.
45
CHAPTER 4: SAMPLE AND MEASUREMENT
Data for this study were taken from a youth interview project conducted at the
University of Southern California’s Social Science Research Institute under the
direction of Drs. Cheryl Maxson and Malcolm Klein. The California Wellness
Foundation and the National Institute of Justice funded the study to investigate
factors associated with joining and resistance to gang membership. This study
examined individual, family, school, peer and neighborhood factors related to gang
joining among African American males and females in San Diego, California (see
Maxson, Whitlock & Klein 1998). This study presents the findings from the San
Diego study on gang resistance within the family domain (African American males
only).
Females were excluded from this analysis for two reasons. First, the small
sample size (26 gang members and 30 nongang members) reduces the likelihood of
finding statistical interactions among the variables. Second, girls’ experiences in
families were expected to differ from boys; girls are socialized differently regarding
appropriate roles both inside and outside the family (Belknap 2001; Chesney-Lind,
Shelden, & Joe 1996). The few studies that examine correlates of female gang
joining have found several variables, especially among family processes, to predict
female gang involvement uniquely (Esbensen & Deschenes 1998; Esbensen,
Deschenes, & Winfree 1999; Maxson & Whitlock 2002; Thomberry 1998).
46
CROSS-SECTIONAL DATA
This is a cross-sectional dataset. While there are clear advantages to the use
of longitudinal data for testing risk and protective factors (e.g., the examination of
protection which occurs subsequent to the onset of risk, and an examination of
changes in risk and protection over time), cross-sectional data designs are useful for
generating promising hypotheses (Richters & Weintraub 1990; Rutter 1994).
Considering the costs associated with longitudinal research in terms of money, labor,
and time, it is reasonable to develop models for the operationalization of protection
with a cross-sectional dataset that subsequently can be tested within a longitudinal
model. This present test of a model of distinguishing risk from protection is
especially useful since data on the characteristics of gang resistance are undeveloped.
DEPENDENT VARIABLE
The dependent variable used in this analysis is measured by the use of self-
reported gang membership. The investigators chose not to use law-enforcement data
or school records to identify gang members because 13-15 year old youth often
escape labeling by law enforcement or are mistakenly over-identified by school
administrative personnel. Thus, the self-nomination technique of defining a gang
was utilized to allow youth to assign themselves to a gang.
There is a long history of divergence over the characteristics that constitute a
gang (Klein 1971; 1995; Miller et al. 1961; Short & Strodtbeck 1965). One of the
47
techniques commonly used in response to the debate is the self report technique. The
assumption behind this technique of gang identification is that there is shared
meaning regarding what a gang is. The use of this method has been found to be a
robust measure; Esbensen, Winfree, He & Taylor (2001) found that, ’’the simple
question, ‘Have you ever been a gang member?’ was understood by the respondents
in such a manner that one can surmise that there exists a shared understanding of
what this term means, not only by former and current gang members, but also by
nongang youth.”
Previous commentary and research concerning the reliability of self reported
gang membership provide robust findings of differences between gang and nongang
youth attitudes and behavior using this technique (Bjerregaard 2002; Bjerregaard &
Smith 1993; Curry & Decker 1998; Curry 2000; Esbensen & Huizinga 1993; Klein
1995; Thomberry, Krohn, Lizotte, Chard-Wierschem 1993; Thomberry et al. 2003).
The use of the self-report method is supported by studies which show that admitted
gang members commit serious and violent offenses at a rate several times higher than
nongang youth (Battin et al. 1998; Bjerregaard 2002; Bjerregaard & Smith 1993;
Esbensen & Huizinga 1993; Hill et al. 1999; Howell 1998; Huff 1998; Thomberry
1998; Thomberry & Burch 1997; Thomberry et al. 1993). This increased
involvement in delinquency is apparent during periods of gang membership, and
retums to previous levels after leaving the gang (Thomberry 1998). Gang joining
increases the risk of a multitude of negative outcomes (e.g., school failure and
48
strained law enforcement relationships); once involved, the expected negative life
effects can be demonstrated.
In this study, youth were asked, “Have you ever been a member of a youth
gang?” and were given the response categories, “1) No, and you don’t want to be; 2)
No, but you’d like to be; 3) Yes, you once were but no longer are; or, 4) Yes, you are
now.” Table 1 provides the frequency distribution of the question from which the
dichotomous gang member (yes=l; no=0) was formed. The gang member category
included youth that reported they were currently involved in a gang, or had
previously been a member of a gang, but did not currently consider themselves to be
a member (responses #3 and #4). Seventy-seven youth claimed a current or former
gang affiliation. Eighty-eight youth reported no gang affiliation.
Table 1. Prevalence of gang affiliation.
Have you ever been a member o f a youth gang? Percent n
No, and you don’t want to be 48.5 80
No, but you’d like to be 4.8 8
Yes, you once were but you no longer are 17.0 28
Yes, you are now 29.7 49
Total 100.0 165
49
Notice that the dependent variable represents a category of youth who have
ever joined a gang. This study addresses risk factors for gang joining, regardless of
the length of gang involvement. It is expected that gang youth, regardless of their
current status, had experienced the risk factors that caused them to join a gang.
These youth are different than nongang youth because at some point they ascribed
themselves as gang members, even if they no longer claim membership. Although a
worthwhile exploration, it is beyond the scope of this study to examine factors
associated with stable vs. transient gang membership. In this study, exgang youth are
not significantly older than current gang members. The mean age of gang members
is 14.25 (sd=.84) and the mean age of exgang members is 14.43 (sd=.68) p=.31.
These exgang youth report hanging out with the gang at a mean age of 10.96,
becoming a full member at an average age of 11.82, and leaving the gang at an
average age of 13.19. Thus the average duration of gang membership was long
enough to suggest that exgang members were not simply flirting with joining, but
actually experienced approximately one year of active membership.
As a test of the self nomination method used in this study, gang and nongang
boys’ involvement in delinquency was examined to determine whether the sampled
gang members were more delinquent than nongang youth. As expected, gang
members were more likely to have committed a variety of delinquent offenses,
including violence, a scale of general delinquency, drug use and drug sales in the
prior six month period (i.e., six-month prevalence) (see Table 2). (See Appendix 1
50
for scale compositions.) For every type of offense, the six-month prevalence for
gang members was greater than for nongang involved youth. (Appendix 2 lists the
six-month prevalence for individual self-reported delinquent behaviors.)
Table 2. Six-month prevalence rates for delinquency scales^
Gang
(percent)
Nongang
(percent)
Vioknt Offending 79% 44%***
General Delinquency 90% 44%***
Drug sales (including marijuana) 38% 6%***
AlcohoFliquor use 58% 32%***
Drug use (including marijuana) 58% 23%***
Minor offending 91% 72%**
**p<.01; ***p<.001
While gang members report greater delinquent involvement, notice the high
prevalence of offending among the nongang youth noted in Table 2. While the
nongang youth had not joined a gang, they were still very active in offending. Gang
membership, however, markedly increased that involvement. These data confirm the
gang designation for this sample and demonstrate the level of risk for the nongang
sample.
51
The test of gang and nongang differences in measures of gang awareness in
neighborhoods (reported in Table 3 below) provides an additional test of the
reliability of the grouping of the sample and further demonstrates the distinctiveness
of these two groups of youth. In addition, this examination demonstrates the
appropriateness of the sampling of neighborhoods where gang joining was a viable
option for youth. (A description of the neighborhood selection follows.)
Youth who reside in these neighborhoods report significant gang presence
and importance in their neighborhoods. Gang members, of course, report greater
gang presence than nongang youth; however, nongang members face considerable
exposure to gang activity in their neighborhoods. In fact, more than two thirds of
nongang youth reported a lot of talk about gangs, and a lot of gang activity around
their neighborhoods. Thus, one can be confident that these nongang youth
represented a group of youth at significant risk for joining gangs. They have been
exposed to neighborhood risk, but did not choose to join a gang. There were no
differences between gang and nongang youth in receiving warnings about gangs or in
reported pressure to join gangs. The majority of nongang youth reported there was
no pressure to join gangs. So while gang joining is an accessible option to these
youth, the pressure to join does not explain the attraction of gang joining. Table 3
displays these various measures of gang presence in the neighborhood.
52
Table 3. Indicators of gang presence in study areas (percent yes).
Gang
(percent/n)
Nongaiig
(percent/n)
Clii-squaie
Have adults in your home warned you about gangs
in your neighborhood?
88% (68) 96% (84) 2.89
Is there a lot o f talk about gangs around your
neighborhood?
94% (72) 69% (61) 15.37***
Is there a lot o f gang activity around here? 87% (67) 69% (61) 7.39**
Are there gang rivalries close by? 79% (61) 59% (51) 7.50**
Is there pressure on neighborhood kids to join
gangs around here?
56% (43) 45% (39) 1.98
Among the kids in the neighborhood, how
important is it to be a member o f a gang?
(percent somewhat or very important)
79% (61) 52% (46) 13.08***
Are any o f the people living on your street
members o f a gang?
87% (67) 67% (58) 9.33**
**p<.01; ***p<.001
LOCATION
San Diego was chosen as the site for this investigation because it is one of the
largest gang cities in California. San Diego has a long history of gangs, dating from
1974, with 86 active youth gangs and 5,118 gang members (NYGC 2002). Two
areas within the city of San Diego were chosen for this sample because they
represented the highest gang activity areas in the city. These two areas were
determined by examining maps of police field interviews with African American
53
male gang youth. Informed by these maps, interviews were conducted with key
community based program directors and police officers to confirm those areas in San
Diego that represented the highest areas of gang activity for African American males.
Census data were used (e.g., percent below poverty level, median household income,
percent unemployed, and number of African American male youth) to identify
economically homogeneous census areas that contained a sufficient population of
young African American males. As expected, these areas were more economically
disadvantaged than most areas in the city of San Diego. Table 4 displays selected
characteristics of these two areas.
Table 4. San Diego and study area characteristics.
High Gang Area M oderate Gang
Area
San Diego city
Percent below poverty level 38% 29% 13%
Median household income $16,980 $18,982 $33,686
Percent unemployed 14% 11% 6%
Population 74,490 66,725 1,110,549
Source: 1990 U.S. Census.
54
SOURCES OF RESPONDENTS
The study design called for equal numbers of gang and nongang youth in
order to assess factors associated with resistance to gang joining. A total of 165
African American males ages 13-15 were interviewed. This age group was chosen
by the investigators to represent the peak ages of joining gangs, and was designed to
allow the assessment of factors associated with resistance to gang joining at the age
when gang joining usually becomes an option. These ages are derived from the body
of gang research which identifies ages of gang members that range from young
adolescence to early adulthood (Bjerregaard & Smith 1993; Esbensen et al. 1993;
Klein 1995.) In this sample, age was modestly correlated with gang membership,
although the magnitude of the differences was small, reflecting a difference of only a
few months. (The average age of gang members is 14.36 (sd=.74) and the average
age of nongang youth is 14.14 (sd=.82) p=.03). An examination of the crosstab of
age and gang status indicates that the trend of the data is as expected, where fewer 13
year old boys claimed gang membership (see Table 5), although the chi-square was
not significant.
55
Table 5. Percent of sample belong to a gang by age.
Gang Nongang
(percent/n) (percent/n)
13 years old 33 3 (12) 66.7 (24)
14 years old 47.2 (25) 52.8 (28)
15 years old 52.6(40) 47.4 (36)
X ^ = 3.663; d .f= 2 ; p=.16
Seventy-seven youth were gang members and eighty-eight were nongang
members. Structured face to face interviews were conducted by trained interviewers.
Employees of SAND AG, a governmental research unit in San Diego, conducted most
interviews. Interviewers were trained in procedures regarding contacting youth and
administering the interview.
Youth were obtained from a variety of sources, including community
agencies (e.g., YMCA, neighborhood recreation and social organizations), street
contacts and “snowball” sampling, and probation school referrals. The inclusion of a
large proportion of street contacts was important to ensure that the youth did not
represent only those youth that had already been identified and served by helping
agencies. This large group of street contacts included youth that had not yet been
filtered through the informal and formal justice and community networks. It was
necessary to employ a convenience sampling strategy in order to ensure an adequate
base rate of gang membership. This strategy has been employed in other high-risk
research where, for example, children of schizophrenics are followed. Choosing this
56
‘high risk’ population is propitious for securing enough cell numbers to satisfy
statistical requirements (Richters & Weintraub 1990). Prospective probability
samples are simply hard to come by.
Table 6 displays the various sources of respondents. Note there was some
variation in the sources from which gang and nongang youth were drawn, but none of
these differences were statistically significant. Over half of the sample came from
community agencies such as youth recreation and counseling organizations. In the
process of selecting appropriate sites within the city, numerous youth-serving
agencies were interviewed in order to provide a description of the types of services
available to youth in these neighborhoods. Many of these agencies offered space for
conducting interviews and forwarded names of potentially eligible youth to the
research staff.
Table 6. Sources of respondents.
Gang
(percent/n)
Nongang
(percent/n)
Total
(percent/n)
Agencies 53% (41) 51% (45) 52% (86)
Street contacts/snowballs 34% (26) 42% (37) 38% (63)
Alternative schools 13% (10) 5% (4) 9% (14)
Juvenile Hall/Probation - 2% (2) 1% (2)
X = 5.97; d .f= 3; p = .ll
57
Street contacts filled in most of the remainder of the sample. (Ninety percent
of the total sample was derived from both agency and street contacts). The process
of contacting youth on the streets required extensive training in which interviewers
were instructed on methods of approaching youth on streets, in parks, at fast-food
restaurants and neighborhood markets, and, in general, ‘hanging out’ in public
places. The street contact category includes youth who were referred to the study
from other youth (snowballs), and youth referred from neighborhood informants that
would gather groups together to meet with interviewers. Other youth were obtained
from a number of alternative educational schools run by the San Diego County
Probation Department which enrolled boys referred by court order, and several from
contacts from juvenile hall.
The ability to develop a sample of neighborhood agency and street contacts
was facilitated greatly by permission to conduct these interviews without parental
consent. The United States Department of Health and Human Services and the
University of Southern California’s Institutional Review Board approved the research
request to avoid parental contact. This approach was pursued because the researchers
anticipated that the very youth needed would be less compliant, and more difficult to
reach through conventional methods. The researchers needed to be able to interview
gang members ‘on-the-spot’, and did not want to cause alarm among parents
unnecessarily who may mistakenly (or not) assume their child was gang-affiliated
because of the nature of the study. Thus, for most of the youth, there was no contact
58
with parents. For those parents that were met, the study was explained briefly and
the interview was never refused.
The interviews took place between January 1995 and July 1996. Most of the
interviews took place at one of several agencies that agreed to provide private office
space in their facilities. The second most common location was in an open setting
such as a park or field where the interview could not be overheard. Table 7 lists the
interview locations by gang status. In each case, the interviewer ensured the privacy
of the setting.
Table 7. Location of interviews.
Gang
(percent/n)
Nongang
(percent/n)
Total
(percent/n)
Home or other residence
1%(1)
2%(2) 2%(3)
School 12%(9) 5%(4) 8%(13)
Agency 53%(41) 52%(46) 53%(87)
Fast food restaurant 3%(2) 7%(6) 5%(8)
Park, street or open setting 21%(16) 31%(27) 26%(43)
Other 10%(8) 3%(3) 7%(11)
= 8.937; d .f= 5 ; p = .ll
59
INDEPENDENT MEASURES
I grouped the variables that were tested for their relationship with gang
joining into the three categories discussed in Chapter 3: 1) family structure, 2)
family management strategies, and 3) family conformity and deviance. Each
category contained multiple measures of one or more constructs. These measures
were not combined into latent constructs in order to preserve the detail in the
measurement of the independent variables, thus allowing risk-specific protective
factors maximum opportunity to emerge. The complete list of variables and their
distributions are presented in Table 8 and are described in detail below. These items
were designed by the research staff, and drawn from the Denver Youth Study
(Esbensen & Huizinga 1993). The exceptions are the family attachment scale (from
Rochester Youth Development Study) (Thomberry et al. 1993), the family self
esteem scale (Hare 1977) and the parental defiance scale (Friedman, Mann &
Friedman 1975).
For each variable, percentage distributions and scale reliabilities were
examined. Scales with reliabilities less than .6 were deconstmcted and the individual
items were evaluated for possible inclusion. Four of twenty-one variables were
nominal and nine items were skewed. (Skewed items are marked with an asterisk in
Table 8). In order to utilize a common metric for the interpretation of odds ratios for
all variables, the continuous variables were divided at the median. Thus, for each
odds ratio reported, the value represents the increase in odds associated with the
60
presence or absence of a risk or protective factor. While it often is preferable to
preserve the continuous nature of variables, this method of cutting at medians or
quartiles is common in risk factor approaches which utilize logistic regression (Hill
et al. 1999; Farrington 1998; Thomberry et al. 2003).
Table 8. Variable distributions for continuous measures.
Variable Description range mean sd
Fam ily Structure
Lives without both biological parents na
Family size (number o f siblings)* 0-11 3.78 2.14
F am ily M anagem ent
Talk parents out o f punishment 1-3 1.89 .67
(l=alm ost never..,3=often)
Parents punish and not punish for the same thing 1-3 1.87 .62
(l=alm ost never...3=oflen)
Parents disagree about punishment 1-3 1.80 .74
(l=alm ost never...3=oflen)
Difference in punishment depending on parents’ mood 1-3 1.99 .78
(l=alm ost never...3=often)
Parental educational aspirations (3=graduate from high 3-6 5.24 .76
school...6=more than college)*
Family attachment (l=negative...4=positive)* 1-4 3.52 .39
Family talks out conflict (l=almost never...4=often)* 1-4 3.07 .98
Family compromises (l=alm ost never...4=often)* 1-4 2.83 .95
Family threatens (l=alm ost never.. .4=0 ften) * 1-4 1.72 .96
Family physically fights (l=alm ost never...4=often)* 1-4 1.96 1.04
Family yells (l=alm ost never.. .4=often)* 1-4 3.02 .97
Family cohesiveness (l=not close at all...4=very close) 1-4 3.11 .61
Family self-esteem (l=low...4=high) 1-4 3.10 .42
Shout at mother (l=almost never...3=0ften)* 1-3 1.48 .64
Parental monitoring (l=low...3=high) 1-3 2.39 .37
Parents know friends (l=none...4=some) 1-4 2.79 .83
Fam ily Conformity and Deviance
Family member ever in a gang na
Family member ever in prison or jail na
Family member with drug or alcohol use na
* skewed distribution
61
FAMIL Y STR UCTURE
Family structure is measured by asking, “Which adults, that is people over
age 18, do you live with now?” Youth were given the following choices: mother,
father, stepfather, stepmother, mother’s boyfriend, father’s girlfriend, other relative
and other nonrelative adult. For the risk/protection analysis, family structure was
measured as a dichotomous variable where l=both parents and O=other.
A measure of family size was included in the category of family structure to
assess whether small families were associated with lower risk of gang joining. Youth
were asked, “How many brothers and sisters, including step and half brothers and
sisters, do you have?” Youth reported a range of 0 - 11 siblings, with a mean of 3.78
(s.d. = 2.14). Because of the skewed distribution of this measure, the variable was
dichotomized at the median (3) so that a score of one represented 3 or fewer siblings
and a score of zero represented four or more.
FAMILY MANAGEMENT
A measure of consistency of discipline was designed as a 4-item scale but
because of a poor alpha (alpha =.31), the items were deconstructed into individual
items representing their separate constructs. The questions that were asked are listed
below. The response choices for the first three items were often, sometimes, or
almost never.
62
1. “If your parents had planned some punishment for you, how often can you
talk them out of it?”
2. “How often do your parents punish you for something and at other times
not punish you for the same thing?”
3. “How often do your parents disagree about how to discipline you?”
4. “Do your parents give you a different kind of punishment depending on
whether they are in a good or bad mood?” Would you say there is no
difference, some difference, or a big difference?
Parental educational aspirations were measured using a single item which
asked, “How far do you think your parents would like you to go in school?” The
response choices were: 8 ^^ grade or less; 9^^-l 1* grade; graduate from high
school/GED; some college; graduate from college; or more than college. Because
of the skewed distribution of this measure (only nine percent of the sample believed
their parents wanted them to earn less than a college degree) the variable was cut at
the median so that 1 = parents want child to earn more than a college degree, and 0 =
parents want their child to graduate from college or less. More will be said about the
boys’ perceived parental high expectations for educational achievement in the next
chapter.
Close family attachment was measured with an 11-item scale in which boys
were asked to think about the adult in their family that they got along with best, and
then answer a series o f questions about that relationship. If the boy could not choose
63
which family member, the interviewers were instructed to use mother. The response
categories were often, sometimes, rarely or never. Relevant items were reverse
coded, and a mean score of the items was created (alpha = .77). The items were
(How often do):
You get along well with (adult)
You feel that you can really trust (adult)
Y our (adult) does not understand you
Your (adult) is too demanding
You really enjoy your (adult)
You have a lot o f respect for your (adult)
Y our (adult) interferes with your activities
You think your (adult) is terrific
Y ou feel very angry towards your (adult)
You feel violent towards your (adult)
You feel proud o f your (adult).
Because of the skewed distribution of this measure (most youth reported a positive
attachment), the variable was cut at the median so that 1= close family attachment
and 0= less close family attachment.
Family conflict management style was measured as a six-item scale reflecting
both positive and negative strategies. Youth were asked: “How often have you seen
a family member handle a serious disagreement by: a) talking it out;
b) compromising; c) walking away^; d) threatening someone; e) yelling; and
f) physically fighting.” The response categories were often, sometimes, rarely, or
never. The negative items were recoded and the resulting alpha for this scale was
1 This item was thrown out o f the analysis because the construct validity o f the item was questionable
(e.g., is walking away an avoidance strategy or is it an appropriate response to conflict?)
64
.55. Because of the low alpha, the scale was not used, but individual items were
retained for further analyses.
A five-item measure of exposure to violence in the home (alpha=.42) was
originally included in the family management category. Because of the low alpha for
this scale, the individual items were evaluated for possible inclusion in the
subsequent analyses. The questions asked if any of the following yes/no questions
occurred during the last few years. The items included:
During the last few years have:
Grown ups in your home hit each other?
Grown ups in your home threatened to stab or shoot each other?
Grown ups in your home yelled at each other?
You actually seen somebody in your home get shot or stabbed?
You seen a gun in your home?
Table 9 displays the prevalence for these items.
Table 9. Prevalence of exposure to family violence items.
percent n
Grown-ups hit each other 11.5% 19
Threaten to stab/shoot each other 2.4% 4
Grown-ups yelled at each other 71.5% 118
Somebody in home shot/stab 8.5% 14
Seen a gun in home 40.0% 66
65
Notice that the boys in this sample rarely reported observing physical
violence or threats of shooting or stabbing in their homes. (A separate single item
not included in this scale asked whether the boys’ parents physically fought in the
past year. Only 2% (n=4) boys responded yes.) Yelling in the home, was quite
common; however, this item was not retained in the analysis since measures of
family conflict resolution (including yelling) were already included. The item
regarding gun presence in home was dropped because it may be that gun presence in
the home is an effect of gang joining, rather than a risk factor. While the threat and
occurrence of violence were rare, it is unsettling that even 9% of boys reported
witnessing that someone in their home had been stabbed or shot. Although it is
beyond the scope of this study to assess the effect of witnessing such an event in
one’s home, it is possible that the boys’ development would be altered by witnessing
such violence. The three items recording hitting, threats and shootings or stabbings
were dropped from subsequent analyses because they occurred so infrequently.
Additionally, youth were asked, “Have you ever been afraid that someone
would hurt you at home?” Ten percent of boys reported that they were sometimes or
often afraid (n=17). (See Table 10.) Six percent of boys (n=10) reported that they
had been physically or sexually abused in their homes (see Table 11). Since there
was little variation on these items, they were dropped from subsequent analyses.
66
Table 10. Prevalence of fear of violence in home.
Have you ever been afraid that someone would
hurt you at home? Would you say you ’ re:
percent n
Never afraid 89.7 148
Sometimes afraid 7.9 13
Often afraid 2.4 4
Table 11. Prevalence of family victimization.
percent n
Have you ever been beaten or physically abused by an adult
in your family? 5.5 9
Have you ever been sexually assaulted, molested or raped? .6 1
A scale measuring family cohesiveness (alpha =.66) included the following
three items:
1. “How much fun would you say you and your family have together?
Would you say, a great deal, pretty much, a little, or none at all?
2. How much time do you spend playing, talking or doing things with your
family? Would you say, a great deal, pretty much, a little, none at all?
3. Compared to most families, would you say yours was very close to each
other, somewhat close, not very close or not close at all?”
A mean score was created for this scale.
67
Family-based self-esteem was measured with a 10-item scale (alpha = .83).
The items were:
My parents are proud o f the type o f person that I am.
No one pays much attention to me at home.
My parents feel that I can be depended on.
I often feel that if they could, my parents would trade me for another child.
My parents try to understand me.
My parents expect too much from me.
I am an important person in my family.
I often feel unwanted at home.
My parents believe that I will be a success in the future.
I often wish that I was bom into another family.
The response categories were strongly agree, agree, disagree and strongly disagree.
A mean score for this scale was created with the negatively worded statements
reverse coded.
Parental defiance was measured as a 4-item scale, asking “How often have
you shouted at your mother? Cursed at your mother? Struck your mother?” and
“Struck your father?” The response categories were, “often, sometimes,” and
“almost never.” The scale did not scale well (alpha = .41) because of low prevalence
on these items (see Table 12). Most youth in this sample were respectful to their
parents in that they did not shout, curse or strike their parents. No youth reported
striking their mothers. Thus, in the risk/protective analysis, only the shouting at
mother item was retained since it was the only item with an adequate distribution.
68
Table 12. Prevalence of parental defiance scale items.
Som etmes/often : percent n
Shouted at mother 40.0 66
Cursed at mother 9.7 16
Struck mother 0 0
Struck father 7.9 13
A 7-item scale measuring parental monitoring (alpha = .66) included the
following items. “How often do:
1. your parents talk with you about what you actually did during the day?
2. your parents talk with you about how things are going in school?
3. you leave a note for your parents or call them about where you are going if they are not
at home?
4. your parents know who you are with when you are away from home?
5. you know how to get in touch with your parents if they are not at home?
6. your parents find time to listen to you when you want to talk to them?
7. your parents know where you are when you’re not at home or at school?
The response categories were often, sometimes and almost never. A mean score for
all items was created.
To measure parental involvement, youth were asked, “How many of your
friends do your parents know?” Response choices were none of them, most of them,
some of them, or all of them.
69
FAMILY CONFORMITY AND DEVIANCE
Family gang membership was measured by asking, “Has anyone in your
family ever been a gang member?” If yes, the interviewer recorded up to three
family relationships. For the purpose of the risk/protective analysis, the variable was
coded as a dichotomous variable (yes=l; no = 0).
Family prison history was measured by asking, “Has anyone in your family
spent time in prison, or jail or somewhere like that?” If yes, the interviewer recorded
up to three family relationships. As above, the risk/protective factor was coded as a
dichotomous variable (I=yes; 0=no).
Family alcohol or drug history was measured by asking two questions: “Has
anyone that you’ve ever regularly lived with used alcohol a lot?” and “Has anyone
that you’ve ever regularly lived with used illegal drugs a lot?” If the respondent
indicated a yes response to either question, the variable was coded as yes (I=yes to
either; 0=no to both).
CORRELATIONS
The correlation matrix (Table 13) displays the Pearson product-moment
correlation coefficients for each pair of variables from the complete variable list
(refer to Table 8). While some correlations are statistically significant, none are so
highly correlated as to require their removal from subsequent analyses. The
70
statistically significant pairs are discussed below. (All factors are coded so that a
score of one represents the risk end of the variable.)
71
21
o\|
s
o
o
o
o
\0
o
I'i
■ v O
C N
OO ON CO CO CN I C N J
O C O T — I T — I O T — I
m
o
CO
O ON C N j
I O O
^ NO ^ CN CO o
o o o o o o o
* ON CO LO CN
0 ^ 1 o f— I T - M T - M
o
o
ool o
o
o In CO o o
o o o o
CN
OO
CN ON
CN T —(
IN OO ON OO
o o o o
tN | O r-( IN
o o o o
* (N T —( T f NO CN In T —( T —( 1 —I LO
I D O O O O O O O O O O
nOI ON O NO CO IN OO CO
O T —( O T —( O O r— I
CN ID ON nO
O O O o
lO l
s
In NO o
CO GO ,- 4
* C N t-i CNCNOO^OCNO^OOCN
NOo O O O O O t—i O O O O t-i
■^1 o ^
CO CO
^ CN O LO O
T I ^ 0 T-H ^ 0 T —I
col
T^LO*CNCONOCN^rH
T—I t—inO o O O O O O
m
CO CO ^
CO T —I CO
NO
o In NO
t -h o O
CN! O ^ CN
0 0^— 1
O CO
T —I o
O C O C O ^ O O t —I ^ C N C N L O C N
t -h O O O O O O O O O O
CN In
o o o
o
4| CO GO UO NM
o o o o OO o
m CN
T —I O g So
IN
CO
t -h OO O
o o ^
ON
ON ON
CO CO
IN IN
CO CO
s
I
s
I
’q3
a
Î
W ) o >
I &
II
0>
o o C O O
.1 C N I I col ^ 1 m l NOI INI ool o i 2 1 ; i : | 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 7 2
"O
0
G
1
1
g
"G
top cd CG
. a G G
V h
0 f-i ^4
o> 0>
2
0 > >
1
i
o>
V h
0>
V h
G o> o>
1
0
e
G
fi h
G 0 0> 0>
a
13
a a
1
è -
1 1 I i
y-J CG pH pp
INI 001 ON| o |
cnI
s
1 1
C C
o
tu o
a
1 3
X
Î
b J O
CNI
(N i
O
O
(Ni
O r-H
O O
S
o
o
OO
o
o
o
e n
e n
GO
o
o
co
e n I N
T —I O
IN IN
CN o
O T —I
CN
m |
§ §
IN LO NO NO
g
COj O
O
o
o
IN
O
I §
ON CN
GO O
OO
e n
o
IN CN IN
o o o
IN
2
CNI o OO LO
o o o
e n
o
ON
o
GO
GO
LO
CN
GO
I
G
§
I
I
H
O ) o O
o O c o c o
^ 1 (NI eni Tfi LOI N O I INI 001 oni 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 1 2 1 n I 2 1
O
G
• S
&
g
•G
td G h
.a G G
O >
.■ y > d ) d )
g o > >
I
s
d ) d )
J - t
G d ) d )
1
o rO rO
G
a a
G O d ) d )
a
1 3
a a
a J
C L h
1
1 1 i i
P h pp pp
O N l G O I ) N | 0 0 |
cnI
II
( d
0
bJO
1
rG
d )
I
to û
73
The individual measures of family conflict management strategies - which
came from a deconstructed scale with a poor alpha - are expectedly correlated.
Family threatening is correlated with family compromising (r= .15*), family fighting
(r= .30***), and family yelling (r= .16*). In addition to correlating with threatening,
yelling among family members to resolve conflict is correlated with family
compromising (r= .16*) and talking out conflict (r= .21**). Family fighting is also
correlated with talking out conflict (r= .16*). Talking out conflict and compromising
are correlated (r= .19**). Two of the conflict management measures correlated with
other family characteristics. The family fighting measure is correlated with
consistency of punishment (r= .19*) and family threatening is correlated with
parental educational aspirations (r= -.15*).
Self-esteem in the home is correlated with seven of the available twenty-one
variables: talking parents out of punishment (r= .15*); family attachment (r= .18*);
family cohesiveness (r= .27***); parental monitoring (r= .23**); and three
measures of family conflict management strategies - talking out conflict (r= .19*),
compromising (r= .24**), and family fighting (r= .23**). The numerous
correlations with home-based self-esteem hint to a hypothesis that the executive
functioning of the parents may produce boys with various levels of home-based self
esteem, which may then influence gang participation.
Parental monitoring is correlated with family yelling (r= .18*), family
cohesiveness (r= .28***), family attachment (r= .22*), family structure (r= .19**),
74
and the single item asking whether parents knew all of their son’s friends (r=.31***).
Since this last item is correlated with the parental monitoring scale, it is predictable
that this single item also is correlated with family attachment (r= .19**) and self-
esteem in the home (r= .17*), in addition to the monitoring scale.
Note that family structure is correlated with monitoring which perhaps
reflects the difficulty of single parent households to monitor activities. Family
structure also is correlated with measures of family threatening (r= .18*) and the item
measuring parents disagreement about punishment (r= -.18*).
As expected, a correlation between family attachment and family
cohesiveness was found (r= .21**). Family attachment relationship also is correlated
with the measure of talking parents out of punishment (r= -.16*).
The measures of family deviance predictably covary. Family gang
membership is correlated with family drug/alcohol abuse (r= .19**) and family
prison history (r= .43***). Family prison history and alcohol/drug abuse in the
family also are correlated (r= .34***). Family prison history is modestly correlated
with three family conflict management strategies (talking out conflict (r= .16*),
fighting (r= -.15*) and yelling (r= .17*)). Prison history, drug/alcohol abuse, and
gang membership are correlated with family self-esteem (r= .16*; r= .16*; and r=
.15*, respectively). The three deviance measures also are correlated with parental
monitoring (r= .17*; r= .17*; r= .30***, respectively). In addition, family gang
75
membership is correlated with talking parents out of punishment (r= .16*) and
parents knowing all friends (r= .28***).
For the purpose of the analysis at hand, the relatively low correlations
between the variables did not indicate much overlap; thus, all items were retained in
the subsequent analyses. The low correlations between family items is surprising as
it suggests that there is no underlying family construct relating to gang membership.
Family styles or profiles may not be easy to categorize. This finding supports the
importance of measuring numerous aspects of family life, as they are expected to
affect risk for gang joining differentially.
76
CHAPTER 5: ANALYTIC APPROACH
The method for this research draws heavily from the work of Rutter (1987)
and Aldwin (1994) who specify a statistical model for examining interaction effects.
This model of testing for interaction effects has been advanced in psychological
developmental research where the interest has been on identifying coping strategies
that mediate or buffer the effect of stress (Aldwin 1994). Aldwin points out that the
examination of protective factors (what she calls coping) using a direct effect or
correlational model is too simple to understand the complexities of the relationships
involved. Examining only direct effects may account for the small effect sizes in
such studies. In this present study, the interactional method is used to begin to
distinguish risk and protective factors for gang joining, since no conceptual literature
indicates which variables should operate in which direction.
The first stage of analysis involved testing the list of potential correlates with
gang joining in a series of bivariate logistic regression equations (O’Dougherty &
Wright 1990). Gang membership (l=yes; 0=no) was regressed on each independent
variable to determine which variables had direct effects on gang joining.
All other variables that did not have a direct effect on gang joining were
considered potential protective factors; “potential” here means only that there is no
direct effect - a statistical, not yet a conceptual issue. These potential protective
factors, by definition, did not have a direct relationship with gang joining, but were
77
retained in the analyses since they may have a mediating effect on risk. To test this,
each potential protective factor was entered into a multiple logistic regression
equation as an interaction term with the risk factor. This logistic regression equation
included the following terms: 1) the risk factor; 2) the potential protective factor;
and 3) the multiplicative interaction term of the risk factor and the potential
protective factor. A total of 110 such equations were run (10 risk factors by 11
potential protective factors). These variables are listed in Table 14 below.
78
Table 14. Complete variable list sorted by bivariate relationship to gang
joining.
Lives without both biological parents
J
Sometimes/rarely/never talk out conflict
Often/sometimes/rarely threaten
Often yell
Low family self-esteem
Low parental monitoring
Parents know none/some/most friends
Small family size
F am ily Functioning
Talk parents out o f punishment (sometimes or
almost never)
Parents punish and not punish for the same
thing (sometimes or almost never)
Parents disagree about punishment (sometimes
or almost never)
Some/no difference in punishment depending
on parents’ mood
B elief that parents want them to earn more
than a college degree
Close family attachment
Often compromise
Never or rarely physically fight
High family cohesiveness
Almost never shout at mother
Fam ily Conformity and Deviance
Family member ever in a gang
Family member ever in prison or jail
Family member with drug or alcohol use
Each regression equation was then evaluated to determine if the model chi-
square was significant and if the p for the interaction term was significant. A
significant interaction term indicated the protective factor interacted with risk, but it
79
did not indicate in which direction (see Rutter 1983). That is, it was possible for the
protective factor to increase risk rather than buffer it.
To determine the direction of the interaction, the cell probabilities were
plotted to reflect the effect of risk on gang joining in the presence or the absence of
the protective factor. The hypothesized effect of a protective factor on risk is
indicated in Table 15 below, where, under conditions of high risk and high
protection, the probability of gang joining would be lower than the probability
associated with a high risk status without protection.
Table 15. Model of expected probabilities.
Low Protection High Protection
Low risk LOW LOW
High risk HIGH LOW
Thus, the expected model of protection indicates that under conditions of high risk,
the presence of the protective factor reduces the probability of gang joining; under
conditions of low risk, the presence of the protective factor has no effect on the
probability of joining gangs. This relationship is charted in Figure 2 below.
80
Figure 2. Model of pure protection: Protection reduces probability at high risk
with no effect at low.
g Low Protection
□ H ig h Protection
H igh Risk Low Risk
Alternately, the relationship between risk and protection may take other forms. The
exhaustive possible relationships are charted below.
Figure 3 indicates that the potential protective factor may reduce risk at the
low end of risk rather than at the hypothesized high end of risk. In this example, the
potential protective factor enhances a low risk status. The protective factor is
inactive at highest risk, but mediates risk at lower levels of risk.
81
Protection reduces probability at low risk with no effect at high risk. Figure 3
Low Protection
□ H igh Protection
^ 0.7
am #
_
%
# 0 #
High Low
The protective factor may reduce the effect of risk at any level (see Figure 4),
indicating that the protective factor has a universal impact on risk, regardless o f the
level of risk.
82
Protective factor reduces probability at both high and low risk. Figure 4
Low Protection
□ H igh Protection =
c 0.7
H igh Low
Figure 5 indicates the relationship where a potential protective factor may
actually aggravate the effect of risk. In this case, the potential protective factor is
operating as a risk mechanism that increases risk only when in combination with the
risk factor.
83
Protective factor aggravates at high level of risk with no effect at low.
Low rru tfd iim
□ H igh Protection Ij
Figure 5
__
High Low
Figure 6 describes the relationship where the potential protective factor
aggravates risk at low risk instead of high.
84
Protective factor aggravates at low with no effect at high. Figure 6
^ 0.7
S 0.5
'
Low Protection
□ H igh Protection
H igh Low
Figure 7 demonstrates the relationship between risk and the potential
protective factor when the potential protective factor reduces the risk of gang joining
at the high end of risk, but increases the risk of gang joining at the lower level of risk.
85
Figure 7. Protective factor reduces risk at high, aggravates at low.
C 0.7
WÊ Low Protection |
□ H igh Protection [
H igh Low
Figure 8 demonstrates the relationship between risk and the potential
protective factor when the potential protective factor increases the risk of gang
joining at the high end of risk, but reduces the risk of gang joining at the lower level
of risk.
86
Figure 8. Protective factor reduces risk at low, aggravates at high.
0.5
MM
Ml
■ L o w Protection
□ High Protection
H igh Risk Low Risk
Figure 9 shows a diagram of the potential protective factor operating as a risk
mechanism at both high and low levels of risk.
87
Figure 9. Protective factor increases probability at both high and low levels of
risk.
B Low Protection
□ H igh Protection
_
High Low
These figures represent the eight logical - but not likely — relationships
between two variables. Only Figure 2 fits the Rutter model, and it is an empirical
question whether any of the others appear.
CUMULATIVE RISK
The variable-by-variable analyses of risk and protective factors described
above allow for a broad view of the patterns of relationships that exist. Ideally,
protective factors that buffer the effect of risk can be found among multiple risk
factors. In addition, certain risk factors may emerge as particularly amenable to
88
family protective buffers, and particular protective factors may emerge as frequent
risk reducers. While this examination has benefits, examining cumulative effects of
both risk and protection seems more promising for guiding those in service delivery
to reduce risk and enhance protection. Risk and protection are likely to come from
multiple sources.
Research has demonstrated that accumulation of risk from several sources is a
more powerful predictor of delinquent outcomes (DeWit & Silverman 1995; Elliott
et. al 1985; Loeber et al. 1998; O'Dougherty & Wright 1990; Rolf et al. 1990)
Hetherington & Blechman (1996) note that while some stressors occur singly and are
handled before other stressors occur, it is more likely that one stressor would trigger
subsequent stressors.
Radke-Yarrow & Sherman (1990) question whether risk operates as a
continuous measure, where each additional risk factor contributes to maladaptive
behavior, or if there is a vulnerability threshold at which point there is a “massive
breakdown in the child’s adaptive behavior” (1990, p. 98). While the composition of
the risk experiences for each youth may differ, the measure of accumulation of risk
can account for intensity of risk arising from multiple sources. Thus, examining the
cumulative effect of risk on gang joining may more effectively capture the nature of
risk in urban families.
In this study, all ten risk factors were entered into a multivariate logistic
regression equation in order to determine the strongest correlates of gang joining
89
while controlling for the others. This model was trimmed so that only those
variables with significant contributions to the model were included in an
accumulation.
Then, the risk factors from the reduced model were combined into a single
item count of risk factors. Gang membership was regressed on this scale, and the
model fit for the risk scale equation was examined. The relationship between
number of risk factors and prevalence of gang joining was also examined. The
analytic plan was to search for protective factors among the potential factors, and
then to enter each protective factor into a regression equation with the cumulative
risk scale to determine whether a single protective factor could mediate the effect of
cumulative family risk on gang joining.
CUMULATIVE PROTECTION
It seems unlikely that a single protective factor would have the strength to
buffer the impact of multiple sources of risk (Rutter et al. 1998). Measuring the
cumulative effect of protection, however raises an additional question of how to
combine variables which are known not to relate directly to gang joining. Is there a
point of critical mass wherein the accumulation of protection finally exerts enough
strength to buffer risk?
In this study, protection was accumulated by creating a cumulative scale
which included only those protective factors which demonstrated an individual
90
protective effect with a risk factor in the risk scale. Gang membership was regressed
on this protection scale and parameters were examined. In addition, the protective
scale was entered as an interaction term with each risk factor and with the risk scale.
These findings are described in the next chapter.
91
CHAPTER 6: FINDINGS
BIVARIATE REGRESSIONS
All variables first were examined individually for their ability to differentiate
gang and nongang youth. Table 16 displays prevalence for each of the items by gang
status. O f the twenty-one family variables examined, ten significantly differentiated
gang and nongang youth and eleven did not. All risk factors correlated with gang
joining in the predicted direction.
92
Table 16. Prevalence of family characteristics by gang status.
Variable Description Gang
(n=77)
(percent/n)
Nongang
(n=88)
(percent/n)
Fam ily Structure
Lives without both biological parents 89.6% (69) 75.0% (66)**
Small family (0-3 sibs) 51.9% (40) 60.2% (53)
Fam ily Functioning
Talk parents out o f punishment (sometimes or 79.2% (61) 86.4% (76)
almost never)
Parents punish and not punish for the same thing 84.0% (63) 8&6?4(78)
(sometimes or almost never)
Parents disagree about punishment (sometimes or 82.2% (60) 80.2% (69)
almost never)
Some/no difference in punishment depending on 68.8% (53) 72.7% (64)
parents’ mood
B elief that parents want them to earn more than a 39.5% (30) 37.5% (33)
college degree
Close family attachment 42.1% (32) 54.5% (48)
Sometimes, rarely, or never talk out conflict 64.9% (50) 52.3% (46)*
Often compromise 19.5% (15) 29.5% (26)
Often, sometimes or rarely threaten 49.4% (38) 36.4% (32)*
Never or rarely physically fight 62.3% (48) 70.5% (62)
Often yell 45.5% (35) 31.8% (28)*
High family cohesiveness 48.1% (37) 50.0% (44)
Low family self-esteem 66.2% (51) 44.3% (39)**
Almost never shout at mother 58.4% (45) 61.4% (54)
Low parental monitoring 64.9% (50) 44.3% (39)**
Parents do not know all friends 84.4% (65) 72.7% (64)*
Fam ily Conformity and Deviance
Family member ever in a gang 88.3% (68) 52.3% (46)***
Family member ever in prison or jail 77.9% (60) 52.3% (46)***
Family member with drug or alcohol use 48.1% (37) 31.8% (28)*
*p<.05; **p<.01; ***p<.001
Gang membership was regressed on each of the twenty-one variables using
logistic regression. Those that were statistically significant predictors were called
risk factors because they directly increased the probability of gang joining. Table 17
93
displays the regression coefficients and odds ratios for each bivariate equation. Most
of the odds were doubled when the risk factor was present. Family structure, family
prison history and family gang history were particularly strong risk factors.
Table 17. Family risk factors for gang joining: odds ratios.
Beta SE Odds
Fam ily Structure
Lives without both biological parents 1.06 .45 2.88**
Fam ily Functioning
Sometimes/rarely/never talk out conflict .53 .32 1.70*
Often/sometime s/rarely threaten .53 .32 1.71*
Often yell .58 .32 1.79*
Low family self-esteem .90 .32 2.46**
Low parental monitoring .84 .32 2.33**
Parents know none/some/most friends .71 .40 2.03*
F am ily Conformity and Deviance
Family member ever in a gang 1.93 .41 6.90***
Family member ever in prison or jail 1.17 .35 3.22***
Family member with drug or alcohol use .68 .32 1.98*
*p<.05; **p<.01; ***p<.001
Table 18 displays the regression coefficients and odds ratios for the eleven
variables that did not directly increase the probability of gang joining. Each of these
risk and potential protective factors is discussed in detail below.
94
Table 18. Potential protective factors for gang joining: odds ratios.
Beta SE Odds
Fam ily Structure
Small family size -.34 .32 .71
Fam ily Functioning
Talk parents out o f punishment (sometimes or almost never) -.51 .42 .60
Parents punish and not punish for the same thing (sometimes or -.40 .46 .67
almost never)
Parents disagree about punishment (sometimes or almost never) -.13 .41 .88
Some/no difference in punishment depending on parents’ mood -.19 .34 .83
B elief that parents want them to earn more than a college degree .08 .32 1.09
Close family attachment -.50 .32 .61
Often compromise -.55 .37 .58
Never or rarely physically fight -.37 .33 .69
Family cohesiveness -.08 .31 .93
Almost never shout at mother -.12 .32 .89
*p<.05; **p<.01; ***p<.001
FAMIL Y STR UCTURE
The majority o f the youth in this sample (61%; see Figure 10 below) live in
single parent households. Two parent households (a category of stepfamilies and
natural families) are the second most common family form at 29%, followed by a
small percentage of youth living with neither parent (10%).
95
Figure 10. Prevalence of family forms, total sample.
Both parents
Step parents
Single Parent
N either parents
0% 20% 40% 60% 80% 100%
Gang members are less likely than nongang youth to come from homes where
both mother and father are living (see Table 19). Nine out of 10 gang members, and
3 out of 4 nongang youth indicated they lived without both parents. Stated in terms
of odds, a youth that does not reside with both parents has an almost 3 times greater
odds of joining a gang than a youth that lives in an intact family. (Odds ratio
displayed in Table 17.) This intact family category was the only family form that
distinguished gang and nongang youth.
96
Table 19. Fam ily structure.
Variable Description Gang
(n=77)
(percent/n)
Nongang
(n=88)
(percent/n)
Total sample
(percent/n)
Odds
Ratio
Doth parents 10.4% (8) 25.0% (22)** 18.2% (30) .35**
Mother and stepfather 11.7% (9) 8.0% (7) 9.7% (16) 1.53
Father and stepmother 1.3% (1) 1.1% (1) 1.2% (2) 1.15
Mother-headed 55.8% (43) 53.4% (47) 54.5% (90) 1.10
Father headed homes 5.6% (5) 6.5% (5) 6.0% (10) 1.78
Neither parent 13.0% (10) 8.0% (7) 10.3% (17) 1.73
*p<.05; **p<.01; ***p<.001
I combined stepfamilies with natural families to form a category of two-
parent households to see if the distinction between gang and nongang youth
remained. If there is an expected innoculative effect of residing in natural families
then including stepfamilies with natural families would dilute the distinction. This
category o f two-parent households did not distinguish gang and nongang youth (see
Table 20). Thus, the only family form that marked gang youth as distinct was their
likelihood to live in families that have experienced some sort of parental
reorganization. Only youth from intact families, defined conservatively, have lower
odds of joining gangs.
97
Table 20. Percent live with two parents.
Gang Nongang Total sample Odds
(n=77) (n=88) (percent/n) Ratio
(percent/n) (percent/n)
Two parents (natural or step) 23.4% (18) 34.1% (30) 29.1% (48) .59
Chi-square = 2.29; df=l; p = .07
The absence o f the boys’ fathers, thus, appears to increase the risk of joining
gangs. While three fourths of the boys in this sample reported the absence of their
fathers (see Table 21), gang members were more likely than nongang youth to report
absent fathers. The results, as expected, are quite similar to those found when
examining the effect of natural families. A youth whose father is absent has almost
two times greater odds of joining a gang than a youth whose father is present. On the
other hand, almost 70% of boys who did not live with their fathers also did not join
gangs. While a father’s absence increases the risk of joining gangs, it alone does not
ensure gang joining. Many youth at greater risk of gang joining because of an absent
father, nevertheless resist.
Table 21. Percent living without fathers.
Gang Nongang Total sample Odds
(n=77) (n=88) Ratio
(percent/n) (percent/n) (percent/n)
Father’s absence 80.5% (62) 69.3% (61) 74.5% (123) 1.83*
Chi-square = 2.72; df=l; p = .05
98
Considering the high prevalence of single parent households within this
sample, I asked whether youth’s families include extended family and nonfamily
adults (other than parents) that may provide assistance in raising these boys. These
additional family members may provide increased opportunities for adult
relationships and mentoring. Twenty-two percent of youth live with at least one
extended family adult, not including adult siblings, 26% live with adult siblings, and
only 6% live with nonfamily adults (see Figure 11 below.) (These groups are not
mutually exclusive.)
Figure 11. Percent live with extended family and nonfamily adults.
Live with
adult
siblings
Live with
nonrelatives
Live with
mm
0% 20% 40% 60% 80% 100%
These percentages vary by family structure (as shown in Table 22), but the small
sample sizes preclude an analysis of statistical differences. There were no
gang/nongang differences for these family characteristics (data not displayed). This
99
picture of gang and nongang families who reside in high-risk neighborhoods is not
the picture one might expect; there is no evidence of large families which include
“outsider” adults that presume a lack of control.
Table 22. Family structure and extended families.
Family structure Sample
percent
(percent/n)
Percent live
with extended
family adults
(other than
sibs)
(percent/n)
Percent live
with adult sibs
(percent/n)
Percent live
with
nonrelative
adults
(percent/n)
Mother and father 18.2% (30) 6.7% (2) 33.3% (10) -
Mother-headed 54.5% (90) 18.9% (17) 22.2% (20) 2.2% (2)
Mother and stepfather 9.7% (16) 12.5% (2) 18.9% (3) 6.3% (1)
Father and stepmother 1.2% (2) 100% (2) -
—
Father-headed 6.0% (10) 30% (3) 20% (2) -
Neither parent 10% (17) 58.8% (10) 41.2% (7) 17.6% (3)
*p<.05; **p<.01; ***p<.001
These boys report that they have from 0-1 1 siblings (median = 3), some of
whom they may not currently live with. They report an average o f 5 persons (x =
4.75; sd=2.01) including themselves living in their homes. The size o f the boys’
families did not distinguish gang from nongang youth. While nongang boys were
slightly more likely to have smaller families, this difference was not statistically
100
significant (see Table 23). Natural families are the least likely to have extra family
adults living in the home with them.
Table 23. Percent live in small families.
Gang Nongang Total sample
(n=77) (n=88)
(percent/n) (percent/n)
(percent/n)
Small family size (0-3 siblings) 51.9% (40) 60.2% (53) 56.4% (83)
Chi-square = 1.15; df=l; p = .2 9
FAMIL Y M AN A GEMENT
Six of the sixteen family management items were risk factors for joining
gangs. Measures of conflict resolution strategies, parental monitoring, and family-
based self-esteem distinguish gang from nongang youth. Multivariate analyses will
reveal the persistence of these risk factors. Each family management factor is
described below.
Gang members reported that the discipline they received from their parents
was slightly more inconsistent than that reported by nongang youth, although this
difference was not statistically significant. Most youth, regardless of gang
membership, reported that their parents were consistent in punishment delivery.
101
Thus measures of parental inconsistency did not increase a boys’ risk of joining
gangs (see Table 24).
Table 24. Percent reporting high parental discipline inconsistency.
Variable Description Gang
(n=77)
(percent/n)
Nongang
(n=88)
(percent/n)
Total sample
(percent/n)
Odds
Ratio
Often talk parents out o f punishment 20.8% (16) 13.6% (12) 17.0% (28) 1.66
Parents often punish and not punish
for the same thing
16.0% (12) 11.4% (10) 13.5% (22) 1.49
Parents often disagree about
punishment
17.8% (13) 19.8% (17) 18.9% (30) .88
Percent reporting big difference in
punishment depending on
parents’ mood
31.2% (24) 27.3% (24) 29.1% (48) 1.21
*p<.05; **p<.01; ***p<.001
Youth were asked how far they thought their parents wanted them to go in
school. Most youth (91.4%) (data not shown) believed that their parents wanted
them to earn at least a college degree. Approximately 40% of youth (see Table 25)
thought that their parents wanted them to go to graduate school. It is striking to note
that, even in areas where educational opportunities may seem blocked, perceptions of
the importance of education remain strong to these families — so strong that post
university education is the aspiration. Tliere was little variation in this measure.
102
Thus, as expected, gang and nongang youth did not differ in their reports of parental
expectations.
Table 25. Parental educational aspirations.
Variable Description Gang Nongang Total sample Odds
(n=77) (n=88) (percent/n) Ratio
(percent/n) (percent/n)
Percent believe parents want them to
earn more than a college degree
39.5% (30) 37.5% (33) 38.4% (63) 1.09
*p<.05; **p<.01; ***p<.001
Youth reported on the quality of the relationship of the adult in their family
that they felt they got along with best^. The most frequent family member chosen
was mother, followed by a sibling, father, and grandparent. A few youth chose aunt,
uncle, cousin, or others (see Figure 12). Nongang youth were slightly more likely to
choose the father than gang youth (7% and 15% respectively; data not shown),
presumably because nongang youth were more likely to reside with their fathers.
^ If the respondent could not choose a family member, he was instructed by the interviewer
to answer the questions about his relationship with his mother.
103
Figure 12. Frequency distribution of person in family youth gets along best
with.
mother
father
21% sibling
grandparent
aunt
uncle
other I I '/''
0% 20% 40% 60% 80% 100%
Nongang youth felt greater attachment to their closest family adult than did
gang involved youth, although this difference was not
statistically significant (see Table 26). Most youth regardless of gang affiliation felt
close to someone in their family. The mean score for the entire sample was 3.52
(sd=.39) on a scale from 1 to 4 with 4 representing the most positive value. Because
of the skewed distribution on this scale, the measure was dichotomized at the
median.
Table 26. Family attachment.
Variable Description Gang Nongang Total sample Odds
(n=77) (n=88) (percent/n) Ratio
(percent/n) (percent/n)
Percent with close family attachment 42.1% (32) 54.5% (48) 48.8% (80) .61
*'p<.05; **p<.01; ***p<.001
104
Youth who do not participate in gangs report increased exposure to positive
conflict resolution strategies within their families than do gang-involved youth (see
Table 27.) Nongang families are more likely to talk out conflict, compromise, and
are less likely to yell, threaten and physically fight than are gang families. Notice
that walking away from conflict is more common among gang families than nongang
families. This reversal may be explained if one interprets walking away from
conflict to be a passive and unconstructive response to conflict which only delays
conflict resolution.
Table 27. Family conflict resolution styles.
Variable Description Gang
(n=77)
(percent/n)
Nongang
(n=88)
(percent/n)
Total sample
(percent/n)
Odds
Ratio
Often walk away from conflict 26.0% (20) 12.5% (11) 18.8% (31) 2.46*
Often talk out conflict 35.1% (27) 47.7% (42) 41.8% (69) .59*
Often compromise 19.5% (15) 29.5% (26) 24.8% (41) .58
Never threaten 50.6% (39) 63.6% (56) 57.6% (95) .59*
Never or rarely physically fight 62.3% (48) 70.5% (62) 66.7% (110) .69
Never, rarely or sometimes yell 54.5% (42) 68.2% (60) 61.8% (102) .56*
*p<.05; **p<.01; ***p<.001
Most youth in this sample felt that their families were close, spent time
together, and had fim (see Table 28). Nongang youth were slightly more likely to
105
report spending more time with their families; however, the overall measure of
family cohesiveness did not distinguish gang and nongang youth.
Table 28, Family cohesiveness.
Variable Description Gang Nongang Total
(n=77) (n=88) sample
(mean, sd) (mean, sd) (mean, sd)
Cohesiveness scale score 3.04 (.64) 3.17 (.59) 3.11 (.61)
Scale items:
Fun in family 2.92 (.89) 3.10 (.84) 3.02 (.87)
Time spent with family 2.77 (.78) 2.98 (.80)* 2.88 (.79)
Closeness 3.43 (.80) 3.42 (.66) 3.42 (.73)
(4 = a great deal/very close; 3 = pretty much/somewhat close; 2 = a little/not very close; 1 = none at
all/not close at all.) *p<.05; **p<.01; ***p<.001
Gang-involved youth reported lower self-esteem within their families than
did nongang youth (see Table 29). On average, youth reported generally positive self
esteem within their families, although gang-involved youth reported poorer family
esteem.
106
Table 29. Family-based self-esteem
Variable Description Gang
(n=77)
(mean, sd)
Nongang
(n=88)
(mean, sd)
Total
sample
(mean, sd)
Mean self-esteem 2.96 (.34) 3.22 (.46)*** 3.10 (.42)
Scale items:
My parents are proud o f the type o f person that I
am
2.74 (.70) 3.27 (.62)*** 3.02 (.71)
No one pays attention to me at home 2.03 (.65) 1.80 (.61)** 1.90 (.64)
My parents feel that I can be depended on. 3.06 (.57) 3.23 (.62)* 3.15 (.60)
I often feel that if they could, my parents would
trade me for another child.
1.78 (.74) 1.48 (.59)** 1.62 (.68)
My parents try to understand me. 3.09 (.54) 3.24 (.55)* 3.17 (.55)
My parents expect too much from me. 2.58 (.83) 2.31 (.82)** 2.44 (.84)
I am an important person in my family. 3.13 (.50) 3.30 (.66)* 3.22 (.60)
I often feel unwanted at home. 2.12 (.73) 1.88 (.83)* 1.99 (.79)
My parents believe that I will be a success in the
future.
3.09 (.57) 3.31 (.63)** 3.21 (.61)
I often wish that I was bom into another family. 2.03 (.74) 1.69 (.61)*** 1.85 (.69)
(4 = strongly agree; 3 = agree; 2 = disagree; 1 = strongly disagree) *p<.05; **p<.01 ; ***p<.001
The majority of boys in this sample did not express defiance toward their
mothers. Gang-involved boys did not express anger more frequently than nongang
youth, as shown in Table 30.
107
Table 30. Parental defiance.
Variable Description Gang Nongang Total sample Odds
(n=77) (n=88) Ratio
(percent/n) (percent/n) (percent/n)
Almost never shout at mother 58 4% (45) 61.4% (54) 60.0% (99) .89
*p<.05; **p<.01; ***p<.001
Gang-involved youth report less parental monitoring than nongang youth.
Behaviors such as leaving notes and tracking activities appear to reduce the
likelihood of gang joining (see Table 31).
108
Table 31. Parental monitoring.
Variable Description Gang
(n=77)
(mean, sd)
Nongang
(n=88)
(mean, sd)
Total
sample
(mean, sd)
Mean monitoring 2.30 (.38) 2.47 (.36)** 2.39 (.37)
Scale items:
How often do:
Your parents talk with you about what you
actually did during the day?
2.17 (.68) 2.20 (.71) 2.19 (.69)
Y our parents talk with you about how things are
going at school?
2.53 (.60) 2.63 (.55) 2.58 (.57)
You leave a note for your parents or call them
about where you are going if they are not at
home?
2.08 (.82) 2.31 (.72)* 2.20 (.77)
Your parents know who you are with when you
are away from home?
2.03 (.71) 2.36 (.68)*** 2.21 (.71)
Y ou know how to get in touch with your parents
if they are not at home?
2.69 (.54) 2.63 (.59) 2.65 (.57)
Do your parents find time to listen to you when
you want to talk to them?
2.58 (.57) 2.68 (.47) 2.64 (.52)
Do your parents know where you are when
you’re not at home or at school?
2.05 (.69) 2.48 (.61)*** 2.28 (.68)
(3 = often; 2 = sometimes; 1 = almost never) *p<.05; **p<.01; ***p<.001
In addition, when parents know their son’s friends, they are less likely to join
gangs. While very few youth indicated that their parents did not know any of their
friends (4%), nongang youth were more likely to report that their parents knew all of
their friends. (See Table 32.)
109
Table 32. Parents know friends.
Variable Description Gang Nongang Total sample
(n=77) (n=88)
(percent/n) (percent/n)
(percent/n)
Parents do not know all friends 84.4% (65) 72.7% (64)* 78.2% (100)
"p<05
FAMILY CONFORMITY AND DEVIANCE
Gang members more frequently reported family members that had been or
currently participated in gangs than did youth who did not join gangs (see Table 33).
Eighty-eight percent of gang members reported that someone in their family had ever
been in a gang compared to 52% of nongang youth. Most gang members were not
the only family member currently involved in gangs, as almost 75% of gang
members reported that other family members were also in a gang. Nearly one out of
five gang members reported that someone in their family encouraged them to join a
gang. For boys who did not join gangs, only 25% reported family members currently
involved in gangs, and only 3% had family members that encouraged them to be a
part of a gang. Note, however, that over half of the boys who did not join gangs
reported a family gang history. Thus, while gang members clearly evidence
increased exposure to gang lifestyles within their families, many youth who are
similarly exposed nevertheless chose not to be involved. It is precisely the protective
1 1 0
processes that interrupt this familial transmission of gang membership that this
research attempts to uncover.
Table 33. Family gang deviance.
Variable Description Gang
(n=77)
(percent/n)
Nongang
(n=88)
(percent/n)
Total sample
(percent/n)
Odds
Ratio
Family member ever in a gang 88.3% (68) 52.3% (46)*** 69.1% (114) 6.90***
Mother/Father ever in a gang 20.8% (16) 1.1% (1)*** 10.3% (17) 22.81**
Brother/Sister ever in a gang 36.4% (28) 17.0% (15)** 26.1% (43) 2.78**
Other family ever in a gang 63.6% (49) 43.2% (38)** 52.7% (87) 2.30**
Any family currently in gang 72.7% (56) 25.0% (22)*** 47.3% (78) 8.00***
Family member encourage to join 18.2% (14) 3.4% (3)*** 10.3% (17) 6.30**
*p<.05; **p<.01; ***p<.001
Most commonly, the family members with gang involvement were cousins,
followed by uncles, then brothers, and fathers (see Figure 13.) Few youth reported
involvement by mother, sisters or aunts. As expected, the number of family
members involved in gangs was greater for gang-involved youth than for nongang
youth. Gang members reported an average of 3 family members ever in gangs
(x=3.49; sd=5.64) compared to an average of 1 (x=1.33; sd=3.44) for nongang youth
(p=.002).
Ill
Figure 13. Percent of youth with family members ever in gangs.
m other I 1 " < i
father
brother
uncle
aunt I I'll
cousin
other ■ 2%
100%
The examination of the odds ratios that describe the relationship between
family gang history and gang joining is even more striking. If a youth reported that
someone in his family had ever been in a gang, he had almost 7 times greater odds of
joining a gang than a youth without that history. His father’s participation in gangs
was particularly compelling, increasing a boy’s odds of joining a gang to nearly 23
times greater odds than youth whose fathers were not in gangs. Clearly, gang joining
tends to run in families. Families with extended traditions of gang participation
place their children at great risk of joining.
Illegal behaviors within the family that resulted in incarceration also are
familiar to both gang and nongang youth. While family criminal histories were
common to the majority of sampled youth, gang members were substantially more
1 1 2
likely than nongang youth to report that someone in their family had been in prison
or jail (see Table 34). A youth with a family criminal history had a three times
greater chance of joining a gang than a youth without that histoty.
Table 34. Family incarceration history.
Variable Description Gang Nongang Total sample Odds
(n=77) (n=88) Ratio
(percent/n) (percent/n)
(percent/n)
Family member ever in prison
or jail
77.9% (60) 52.3% (46) 64.2% (106)*** 3.22***
*p<.05; **p<.01; ***p<.001
The most common family member with incarceration history was a father or uncle,
followed by a cousin or a brother (data not shown). It is striking that one of four of
these sampled boys’ fathers had spent time in prison or jail. These boys are clearly
exposed to illegal behaviors (that resulted in incarceration) directly from their
families. Their ability to choose nondelinquent paths is hindered by such exposure.
113
Figure 14. Percent of youth with family members ever in jail or prison.
m other H i 5%
father
- T ^ I
brother 1
uncle 24Vv
other 4%
0% 20% 40% 60% 80% 100%
Gang members reported that someone in their family that they regularly lived
with abused alcohol or drugs more often than did nongang youth. A youth with
exposure to familial drug or alcohol abuse had almost two times greater odds of
joining a gang, compared to youth without such exposure. Alcohol misuse was more
prevalent than drug abuse (35% and 18% respectively for the overall sample), but the
presence of either in a youth’s home distinguished youth who joined gangs from
those who did not.
114
Table 35. Family drug or alcohol history.
Variable Description Gang
(n=77)
(percent/n)
Nongang
(n=88)
(percent/n)
Total sample
(percent/n)
Odds
Ratio
Family member with drug or
alcohol use
48.1% (37) 31.8% (28) 39.4% (65)* 1.98*
Family alcohol use 42.9% (33) 28.4% (25) 35.2% (58)* 1.89*
Family drug use 29.9% (23) 8.0% (7) 18.3% (30)*** 4.87***
*p<.05; **p<.01; ***p<.001
INTERACTION ANALYSES
Each potential protective factor was tested for its ability to interact with risk
and reduce the likelihood of gang joining. Out of 110 logistic regression equations in
which each risk factor, potential protective factor, and the interaction terms were
entered, only eight of the 110 interaction terms were statistically significant (p<.05).
The results of these tests are charted in Table 36.
O f the eight equations where the interaction terms were statistically
significant, seven potential protective factors reduced risk (four at high, and three at
low). These relationships are described in detail below. Eight interactions represent
at least five percent of the possible interactions, thus these results are slightly beyond
those expected by chance.
115
PQ
CM 0 3
o o |
O Q
voi
I
I
A
§
I
O
SA
I -
C A
a
Î
VO
o t
H
T fi
m l
C N I
I
I
I
1
I
I
O
I
C/2
• S
I
I
O
I
I
I
I
I
I
I
I
O h
.S
g
a >
I
O
& Û
I
I
a
•â
O Q
0-, .S O n
:g
I 1
116
Five of the eight risk factors responded to protection from at least one source.
These are family structure, parental monitoring, knowing friends, prison history and
drug/alcohol history. O f these five, only family structure responded to three sources
of protection. The remaining five risk factors were unresponsive to any family
source of protection. These unresponsive risk factors were measures of conflict
resolution (talking out, threatening and yelling), self-esteem in the home, and family
gang history.
Seven of the eleven potential protective factors reduced risk (at either high or
low levels of risk) associated with at least one risk factor. These protective factors
were punishment consistency, punishment changes with moods, parental educational
aspirations, close family attachment, compromising in conflict, physically fighting,
and family cohesiveness. One protective factor interacted with two risk factors, and
in both instances protection operated in the same direction. Punishment consistency
reduced the odds of gang joining associated with low parental monitoring and family
prison history.
The picture of risk and protective relationships that is shown in Table 36
appears somewhat random. There is no particular protective factor that is set apart as
a frequent risk reducer, nor does a risk factor emerge as particularly responsive to
family strengths. Each pair of relationships is now discussed in detail below.
Figure 15 displays the relationship between family structure and family
attachment. Here we see that for youth that do not live with both biological parents,
117
having a close family adult relationship does not reduce the probability of joining a
gang, but for boys who do live with both biological parents, having a close family
adult relationship reduces the likelihood of joining a gang. In other words, having a
close family adult mentor is only protective when a boy lived with both parents. This
interaction seems counterintuitive; when a boy is at less risk for joining, having the
protective factor is particularly helpful. A possible interpretation may be that the risk
associated with alternative family structure is so strong that it is unaffected by the
presence of a positive adult relationship, whereas when youth have both parents
available, but are still at risk from other unmeasured sources, the close family adult
reduces risk. This is an example of a protective factor that augments the low risk
status.
118
Figure 15. Family structure (risk) by close family adult (prot.).
Lack of close fam ily adult relationship (0)
□ H ave a close fam ily adult relationship (1)
s 0.5
0.2 4
n=66 n=68
Live without both parents (1)
n=22 n=8
Live with both parents (0)
Family structure
Family attachment
Family structure*family attachment
Constant
Model chi-square
.43
-2.31*
2. 02*
-.25
12.35; p=.006
se
.56
1.15
1.20
.50
Figure 16 displays an interaction with the same family structure risk factor in
which the protective factor again augments the low risk status. Notice the similarity
in Figures 15 and 16. The similarity in the depictions may support the earlier
supposition that risk associated with alternative family structure is less receptive to
family buffers. Figure 16 indicates that for youth who live with both biological
parents, having a close family reduces the probability of gang joining, but for youth
119
living in alternative living arrangements, having a close family does not reduce risk
of gang joining.
Figure 16. Family structure (risk) by family closeness (prot.).
50 0.6
V 0.4
n=66 n=69
Live without both parents (1)
- ■ Lack of fam ily clo sen ess (0)
1 □ C lose fam ily (1)
.40
k .13
n=22 n=8
Live with both parents (0)
Family structure
Family cohesiveness
Family structure*family cohesiveness
Constant
Model chi-square
.38
-1.47
1. 62*
-.41
9.14; p=.03
.58
.92
.99
.53
Figure 17 models true protection where the presence of the protective factor
reduces the likelihood of gang joining at high risk with no reduction at low risk.
While the difference between the percents at low risk is sizeable, the difference is not
1 2 0
statistically significant. Thus, youth who live without both parents are at less risk for
joining gangs when their family compromises during conflict. A family that
compromises does not reduce risk for youth who live with both biological parents.
Figure 17. Family structure (risk) by family compromises (prot ).
■ Fam ily som etim es, rarely or n ever com p rom ises (0)
□ Fam ily often com p rom ises (1)
.57
n=66 n=69
L ive w ith o u t b oth parents (1)
n=22 n=8
Live w ith b oth parents (0)
Family structure
Family compromise
Family structure*family compromise
Constant
Model chi-square
1 .6 2 * *
1 .3 4
- 2 . 2 7 *
- 1 . 3 4
1 3 .4 5 7 ; p = . 0 0 4
se
.5 4
.9 6
1 .0 4
.5 0
The interaction between parental monitoring and punishment consistency also
models true protection (see Figure 18). For youth who report low parental
1 2 1
monitoring, consistent punishment reduces the likelihood of gang joining. For youth
with high parental monitoring, punishment consistency did not reduce the likelihood
of joining. Thus, if parents poorly monitor, they can reduce some of the risk caused
by poor monitoring by consistently applying punishment.
Figure 18. Parental monitoring (risk) by punishment consistency (prot.).
P u nish m en t is often incon sisten t (0)
0.9 -
□ P u n ish m en t is som etim es, rarely or n ever
incon sisten t (1)
a, 0.6
o 0.5 -
a 0.4
n=39 n=48
L ow parental m on itorin g (1)
n=49 n=27
H igh parental m on itorin g (0)
b se
Parental monitoring 2.49** 1.03
Punishment consistency .45 .72
Parental monitoring*punishment consistency -1.90* 1.09
Constant -.98 .68
Model chi-square 10.69; p=.01
1 2 2
Figure 19 demonstrates true protection, although the graphing of the expected
probabilities is somewhat misleading. For youth with low parental monitoring,
reliable punishment decreases risk for gang joining. For youth with high parental
monitoring, reliable punishment has no (statistically significant) effect. Statistically,
there is no aggravating effect of protection at low risk since the logit (.83) is
nonsignificant.
123
Figure 19. Parental monitoring (risk) by reliable punishment (prot.).
1 T
0.9
0.8
■ Punishment often changes with
mood (0)
□ Punishment sometimes, rarely or
never changes with m oods (1)
0.5 -
0.1 -
n=39 n=50
Low parental monitoring (1)
n=49 n=27
High parental monitoring (0)
Parental monitoring
Reliable punishment
Parental monitoring*reliable punishment
Constant
Model chi-square
I.95**
.83
-1.49*
-1.25
II.07; p=.01
.69
.63
.78*
.57
The following protective factor reduces risk at low, but has no effect at high
(see Figure 20). Parental lack of knowledge of friends places youth at higher risk of
gang joining. When parents are more knowledgeable of friends, lack of fighting
greatly enhances the low risk status - and reduces the likelihood of gang joining.
124
Figure 20. Parents know friends (risk) by family fights (prot.).
Parents som etim es or often fight (0)
□ Parents rarely or n ever fight (1)
*0 0.6
H 0.4 -
n=64 n=65
Parents do n o t k n o w m ost, so m e or
n on e o f friends (1)
n=24 n=12
Parents k n o w all friends (0)
P a r e n t s k n o w f r i e n d s
F a m i l y f i g h t i n g
P a r e n t s k n o w f i i e n d s * f a m i l y f i g h t i n g
C o n s t a n t
M o d e l c h i - s q u a r e
- .4 1
- 2.20* *
2. 22**
.4 1
1 1 . 7 7 ; p = . 0 0 8
.6 2
.8 2
.9 0
.5 3
Figure 21 demonstrates the effect of punishment consistency on reducing the
risk associated with family prison history - a case of true protection. When youth
report that someone in their family has spent time in jail or prison, punishment
consistency reduces the likelihood of gang joining which is increased by the deviant
modeling available to the youth. At low risk, punishment consistency has no
reduction effect.
125
Figure 21. Family prison history (risk) by punishment consistency (prot.).
w 0.6
o 0.5 -
■ P u n ish m en t is often incon sisten t (0)
□ P u n ish m en t is som etim es, rarely or n ever
in co n sisten t (1)
n=46 n=58
Fam ily prison h istory (1)
' .31 -
.13
n=42 n=17
N o fam ily prison h istory (0)
Family prison history
Punishment consistency
Family prison history*punishment consistency
Constant
Model chi-square
3.24**
1.16
-2.37*
-1.95
16.31; p=.001
se
1.25
1.11
1.30
1.07
Figure 22 displays true protection where high parental educational aspirations
reduce risk associated with family drug or alcohol history.
126
Figure 22. Family drug/alcohol history (risk) by parental educational
aspirations (prot.)
Parents w an t college degree or less (0)
□ Parents w an t m ore than college degree for son
0.6 - —
n=28 n=37
Fam ily w ith d ru g/alcohol u se (1)
n=60 n=39
N o fam ily d ru g/alcohol u se (0)
Family drug/alcohol use
Parental educational aspirations
Family drug/alcohol use*educational aspirations
Constant
Model chi-square
1.15**
.57
-1.14*
-.67
7.90; p=.05
se
.42
.42
.67
.28
These findings point to the need to examine sources of protection fi’ om other
domains of influence such as individual strengths, peer influences, school variables
and neighborhood contexts. It appears that, in most cases, family strengths alone are
unable to compensate for family weaknesses. In the next section, multivariate
models of risk and protection are examined in order to identify the most
parsimonious and promising points of intervention.
127
RISKMODEL
All ten risk factors were entered into a regression model to identify the most
prominent family factors that explain gang joining (see Table 37). Tluee lisk factors
significantly explain gang joining in this sample. They are family structure, family-
based self-esteem and family gang history.
Table 37. Risk factors for gang joining: multivariate logistic regression
parameters.
Variable b S.E. Exp (b) tolerance
Family structure 1.00* .51 2.72 .87
Family talking .11 .37 1.11 .90
Family threatening .16 .37 1.17 .92
Family yelling .18 .38 1.19 .90
Family-based self-esteem .74* .37 2.11 .87
Parental monitoring .11 .39 1.11 .78
Parents know friends -.05 .49 .96 .84
Family gang history 1.60*** .47 4.94 .73
Family prison history .31 .42 1.36 .71
Family alcohol / drug history .30 .39 1.34 .84
Constant -3.08 .68 .05
-2 log likelihood 187.87
Model chi-square 40.134***
*p<.05; ***p<.001
Not much explanatory power is lost by dropping the nonsignificant items
from the model (see Table 38). Thus it appears that youth who live without both
biological parents, who have low self-esteem in their families, and whose family
members have been members of gangs, are at particular risk for joining gangs.
128
Family conflict resolution strategies, parental monitoring and family prison and drug
history lose their significance in this multivariate model.
Table 38. Risk factors for gang joining: multivariate logistic regression
statistics - reduced model.
Variable B S.E. Exp (b)
Family structure 1.12** .49 3.06
Family-based self-esteem ^6** .36 2.37
Family gang history 1.82*** .42 6.15
Constant -Z86 .61 .06
-2 log likelihood 190.51
Model chi-square 37.49***
Kp<01; ***p<.001
Estimated probabilities calculated from the regression coefficients show that
seventy two percent of boys ((1.12*1) + (.86*1) + (1.82*1) - 2.86 = .94; exp .94 =
2.56; (2.56/(1+2.56) = .72) with all three risk factors — with low self-esteem, a family
member in a gang, and not living with both parents - are likely to join gangs.
Twenty-nine percent of boys living without both parents, with low family self-
esteem, and without a family member in a gang are likely to join gangs. Family
history of gang joining continues to increase the odds of gang joining greatly, even
beyond the contribution of family self-esteem and structure.
129
CUMULA LIVE RISK
For the 3-item risk model, a count of the number of risk factors was
calculated for each youth, figure 23 displays the distribution of the count of risk
factors for the total sample. All but four nongang respondents had at least one family
risk factor. One-third of the sample had all three risk factors.
Figure 23. Frequency distribution of count of risk factors, total sample.
8 0 - -
N o n e 1 r i s k f a c t o r 2 r i s k f a c t o r s
Number of risk factors
3 r i s k f a c t o r s
As expected, gang members reported more sources of risk (mean = 2.44;
sd=.62) than nongang youth (mean = 1.72; sd =.84) (p<.001). H alf of gang members
experienced all three risk factors, compared to only 21% for nongang youth (data
tabled in Figure 24.).
130
Figure 24. Percent joined gang by number of risk factors.
G an g
N uiigdiig
100 1
90
60
60
40
N o n e
(n=4)
1 risk factor
(n=40)
2 risk factors
(n=64)
3 risk factors
(n=57)
When gang joining is regressed on the cumulative risk score, the model fit is
significant. The odds ratio indicates that, for every additional risk factor
experienced, the odds increase at a rate of 3.5.
Table 39. Regression coefficients, cumulative risk scale.
Variable B S.E. Exp (b)
Risk Scale 1.26*** .24 3.52
Constant -2.77 .55 .06
-2 log likelihood
Model chi-square
193.538
34.37***
"p<.001
131
The odds of gang joining associated with each point on the risk scale (a value
of 0, 1, 2, or 3) were calculated from the regression coefficients. Results are
displayed in Table 40. Notice how the odds of gang joining increase exponentially
with each additional risk factor. (The change in odds reflects the 3.5 increase (e.g.,
.22 X 3.5=.77).) For each additional risk factor experienced, the odds of gang joining
increases at a rate of 3.5. The dramatic effect of cumulative risk is demonstrated by a
comparison of the odds of joining for a boy with no risk factors compared to the odds
for a boy with all three. A boy with three risk factors has 45 times greater odds of
joining a gang than a boy with no risk factors (2.72 divided by .06). Thus, family
risk, while not explaining the complete picture of risk for gang joining, clearly
elevates risk for gang joining.
Table 40. Odds ratios for cumulative risk scale points.
Variable Odds
No risk factors .06
1 risk factor .22
2 risk factors .77
3 risk factors 2.72
Each o f the eleven protective factors was tested for its ability to reduce the
risk associated with cumulative family risk by regressing gang joining on the
cumulative risk scale, the individual protective factor and the multiplicative
interaction term for both items. None of the interaction terms were statistically
132
significant. Thus no single protective factor was effective at reducing the risk of
gang joining associated with multiple family disadvantages.
CUMULATIVE PROTECTION
Accumulating a scale of protection that consists of a number of factors that
are indirectly related to the outcome may seem counterintuitive. However,
assembling such a list allows for the examination of the cumulative buffering effect
of protection. While perhaps unlikely, it is probable that multiple sources of
protection chip away at risk, and, in combination with each other, finally exert a
protective influence. Simply stated, perhaps there is strength in numbers.
Recall that seven protective factors buffered risk (refer to Table 36). Four
protective factors reduced risk at high risk and three protective factors reduced risk at
low risk. Combining all seven factors into a cumulative protection scale may
produce a scale with no effect since protection has been shown to be dependent on
risk level; the protective effects may cancel each other out. To avoid this, two
cumulative protection scales were created. Protection Scale - A is composed of four
true protective factors, where protection reduces risk only at high levels of risk.
Protection Scale - B is composed of three items that reduce risk at low levels of risk,
but not at high. None of the items in either scale aggravated risk. Table 41 displays
the mean scores for the protection scales by gang status.
133
Table 41. Mean scores of protection scales by gang status.
Variable Description Gang Nongang Total
(n=77) (n=88) sample
(mean, sd) (mean, sd) (mean, sd)
Protection Scale - A 2.09 (.92) 2.28 (.93) 2.19 (.93)
Protection Scale - B 1.52 (.98) 1.75 (.85) 1.64 (.92)
Turning now to the question concerning whether accumulated protection has
the ability to reduce risk, I first examined the regression parameters for the equations
in which gang joining was regressed on the cumulative protection scales (separately).
The logit for Protection Scale — A was not significant (see Table 42). For Protection
Scale — B (see Table 43), the logit is significant, although the model chi-square is not
significant. Note that the protection scales, in their cumulative forms, now have
direct (though inverse) effects on gang joining.
Table 42. Regression coefficients, Protection Scale - A.
Variable B S.E. Exp (b)
Protection Scale - A -.23 .17 .80
Constant .36 .40 1.44
-2 log likelihood 226.22
Model chi-square 1.79
134
Table 43. Regression coefficients, Protection Scale - B.
Variable B S.E. Exp (b)
Protection Scale - B .2 8 * .17 .76
Constant .32 .32 1.38
-2 log likelihood 225.38
Model chi-square 2.91
*p<.05
Next, gang joining was regressed on the risk factor, the cumulative protection
scale (A and B in separate runs) and the multiplicative interaction term of both items
to determine whether the cumulative protection scale reduced the probability of gang
joining associated with specific risks. Two risk factors responded to Protection Scale
— A and one risk factor responded to Protection Scale - B. The pairs of variables
with significant interaction terms are charted in Figures 25 through 27 below.
Parental monitoring and family drug history were mediated by Protection Scale — A
and family structure was mediated by protection scale - B.
135
Figure 25. Parental monitoring (risk) by Protection Scale - A.
■ () pri)tocti\o tjc to rs
E l 1 p ro to c ti\o t.ictor
□ 2 prDloctivi.'
O 1 p roti’ctivo
□ 4 protective t.ictors
O 0.5 -
M 0.4 -
0.2 -
Low parental monitoring (1) High parental monitoring (0)
b s e
P a r e n t a l m o n i t o r i n g 2 .27 * * .8 8
P r o t e c t i o n s c a l e - A .1 7 .2 6
P a r e n t a l m o n i t o r i n g * P r o t e c t i o n s c a l e — A - . 6 6 * .3 6
C o n s t a n t - . 9 8 .6 6
M o d e l c h i - s q u a r e 1 1 .5 8 ; p = .0 1
136
Figure 26. Family drug/alcohol history (risk) by Protection Scale - A.
0.9
0.8 -f
■ O p rotective tJctors
□ 1 protective tjc to r
□ 2 p ro te ttiv e tJctors
□ 3 p rotective t.litors
□ 4 protêt.tivo t.ictors
60 0 .6 -
m
m
Fam ily w ith drug/alcohol histo ry (1) N o fam ily drug/alcohol histo ry (0)
b s e
F a m i l y a l c o h o l / d r u g h i s t o r y 2 . 0 9 * * .8 8
P r o t e c t i o n s c a l e - A .0 7 .2 3
F a m i l y a l c o h o l / d r u g h i s t o r y * P r o t e c t i o n s c a l e — A - .6 7 * .3 7
C o n s t a n t - .5 7 .5 6
M o d e l c h i - s q u a r e 9 .0 7 ; p = . 0 3
137
Figure 27. Family Structure (risk) by Protection Scale - B.
N o protection
□ protection f " "j
“ 0.6
X 0.4
Live without both parents (1) Live with both parents (0)
b s e
F a m i l y s t r u c t u r e - .3 8 .9 0
P r o t e c t i o n s c a l e - B - 1 . 2 9 * .6 0
F a m i l y s t r u c t u r e * P r o t e c t i o n s c a l e - B 1 .1 3 * .6 3
C o n s t a n t .6 9 .8 2
M o d e l c h i - s q u a r e 1 3 .2 2 ; p = . 0 0 4
As expected, the cumulative protection scale A which is composed o f true protective
factors displays a true protective relationship where protection reduces risk at high
and not at low (see Figures 25 and 26). Protection Scale - B reduces risk at low, also
as expected. While some interactive effects were found, cumulative protection did
138
not convey a substantial increase in the number of risk factors that responded to
protection.
139
CHAPTER?: CONCLUSION
This project was designed to identify distinct risk and protective factors for
youth at risk of gang joining, using a statistical method that allowed for the
demonstration of the interactive reduction effect of protection. While an
examination of characteristics of gang and nongang youth provides an adequate
model for differentiating these groups, it does not address the primary concern of this
project which is to identify interactive processes wherein the probability of joining is
reduced in the presence of a protective factor. This project asks, under what
conditions is risk for gang joining mediated by family factors.
Logistic regression equations were run with a dichotomous measure of gang
joining as the dependent variable. Bivariate regressions with each family factor and
gang joining were examined. Those factors with direct effects were called risk
factors. The remaining factors were called potential protective factors and were
tested in a subsequent analytic step for their ability to reduce risk. Each protective
factor was then entered into a multivariate regression equation as a multiplicative
interaction term with each risk factor. For each equation with a significant
interaction term, estimated probabilities were examined to assess the pattern of
relationships among the variables. Finally risk and protection were accumulated and
tested within the regression model.
140
FAMILYRISK
The results of this project offer several promising contributions to the
literature on family risk and protection for gang joining. First, this research
demonstrates ten family-based risk factors for gang joining. Three of the ten factors
remained significant in the multivariate model. These are described first.
Family risk for gang joining comes from three main sources. First, family
structure emerges as an important correlate of gang joining. While there is some
variation in previous research regarding whether family structure has an impact on
offending and gang joining, this present research confirms that families that
experience reorganization increase risk for joining gangs. In this study, the presence
o f a boys’ father was critical in reducing risk for gang joining for African American
middle school aged boys. While some studies indicate that step-parents are able to
compensate for the loss of a parent, this research does not support that. Only living
with both mother and father reduced the risk o f joining gangs among this at-risk
sample. This relationship held true even after considering the effects o f family
management practices such as family cohesiveness and parental monitoring. Family
reorganization increases risk for gang joining.
The second main source of family risk is low family-based self-esteem. Boys
who feel important and connected to their families are less likely to join gangs. This
measure of self-esteem is likely an effect of parental practices; some families are
more effective at creating a sense of belonging and acceptance for their children than
141
others. The hope for reducing risk perhaps lies in working with families and youth to
increase home-based self-esteem with the effect of reducing risk for gang joining.
I he third and strongest family source of risk for joining gangs is family
history of gang involvement. The odds of joining a gang are greatly increased (7
times greater) when a boy reports a family member is or has been involved in gangs.
Thus, reducing youth gang involvement appears to require extended family
interventions. To address efforts for gang intervention at a boy without including his
cousins, uncles, and of course parents, may be unproductive.
Sutherland’s (1939) theory of differential association explains the process of
learning criminality that occurs as a result of the frequency and types of associations
that are favorable toward law breaking. His theory is typically used to explain how
peer associations may lead to delinquency; however, it is a likely extension of his
work to describe how family relationships provide the primary set of definitions
favorable for gang joining. In families where gang joining is common among its
members, the family provides an environment that defines gang joining as acceptable
- or tolerable. This gang family provides models of gang behavior for imitation, and
reinforces behavior consistent with gang involvement. Sutherland might suggest that
these family relationships have added significance because of the familial bond; in
other words, those bonds may be more influential. While peer and family influences
on delinquency often have been examined as separate domains of influence, this
current research suggests that it is important to address family and peer networks
142
together when considering factors associated with gang joining. Gang networks will
likely include family members.
Seven other risk factors distinguished gang fiom nongang youth but lost their
significance in the multivariate model. Nevertheless, they are important
characteristics of gang youth which demonstrate points of intervention for youth.
These items reflected three measures of conflict resolution styles, two measures of
parental monitoring, and two measures of family deviance (prison history and
drug/alcohol abuse). While parental monitoring was unable to reduce gang joining
when considering the powerful impact of family gang history, family structure and
family self-esteem, it does appear that monitoring may prevent some youth from
joining gangs. Further analyses of the effect of monitoring are needed.
NONDISTINGUISHING FAMILY CHARACTERISTICS
A second contribution of this research is the identification of factors that do
not distinguish gang and nongang youth. Since few studies have identified variables
associated with gang joining, this project included some demonstrated risk factors
drawn from the extensive delinquency literature. Half of the proposed factors (11 out
of 21) did not differentiate gang and nongang youth. These results indicate that large
families, parental inconsistency in punishment, parental lack of emphasis on
education, low family attachment, negative conflict resolution styles, low
cohesiveness, and parental defiance do not increase risk of gang joining.
143
Family risk factors for gang joining are a subset of those for engaging more
generally in delinquent activities, such that predictions for gang joining are more
difficult. The items that were used in Lhis reseaich were drawn from the extensive
delinquency literature regarding risk factors for delinquency (Elliott et al. 1985;
Rutter & Giller 1984; West & Farrington 1977; Wolfgang, Figlio & Sellin 1972).
Since gang joining represents only a small portion and a select group of those who
engage in delinquency, factors that are known to correlate with delinquency may not
explain gang joining very well. Data from the Rochester Youth Development Study
found 25 of 40 tested causal factors to predict to gang joining. Seattle Social
Development Project (Hill et al. 1999), on the other hand, found 21 of 25 tested
factors predicted gang joining. Comparisons of risk factors to delinquency and gang
joining from the same datasets are needed to address this question.
METHODOLOGICAL PROCEDURE
This research stresses the need to examine interaction effects for the purpose
of demonstrating the buffering effect of protective factors. This definitional use can
be demonstrated in a logistic regression model by estimating probabilities. Since
such protective effects can be demonstrated, one can conclude that protection is not
simply the absence of risk. In addition, this research demonstrates that protection is
specific to type of risk and to level of risk. Some “protective factors,” as revealed by
the statistical interactions, may heighten or aggravate risk. Thus one needs to look
144
closely at the nature of the interaction patterns. These findings point to the need to
identify correctly types of risk experiences, and to direct interventions carefully.
Many risk factors identified in this study were unaffected by any individual
protective influence or even by a cumulative protection scale. Thus applying
protective-sounding interventions toward families may be a misuse of resources if
risk is not correctly diagnosed.
While no single protective factor emerged as particularly effective against
family risk, several protective factors reduced risk associated with particular risk
factors. A few possible family buffers were identified. These were consistency of
punishment, family closeness, family cohesiveness, family compromising during
conflict, and refraining from physically fighting. Thus, it appears that these family
processes can protect a youth from the heightened risk associated with certain risk
factors.
None of the protective factors were able to reduce the effect of cumulative
family risk. Hetherington & Blechman (1996, p. vii) note that while moderate levels
of risk may activate coping and adaptive responses, with high levels of cumulative
stress there may be little evidence of resilience. The results of this study certainly
show the strength of cumulative risk.
While the purposes of this project were to test a new methodology and to
address issues concerning the family, it is expected that other domains will influence
youth’s decision to join a gang. Perhaps additional protective factors can be
145
identified by including individual, peer, school, and neighborhood protective factors.
Indeed, this research asked the same family that produced the gang member to also
protect. Certainly, examining the external resources outside the family that are
available to youth in efforts toward resistance may provide additional understanding
of the process of resistance.
Rutter’s insistence on demonstrating the buffering effect of protection in
order to invoke the term protection seems essential. This research provided a test of
this method, using a restrictive definition of protection where protection could not
have a main effect on gang joining. Thus, using this definition made protection
difficult to find. Perhaps limiting protection to only those mechanisms that have an
indirect effect prevents the emergence of findings wherein “protection” decreases
risk directly, and decreases risk also. This mechanism, however, would require a
different name. Rutter and others have suggested moving away from using the term
factor and instead using the term mechanism to describe such effects since a factor
may operate differently under different conditions. That is, a mechanism may
operate as a protective mechanism or a vulnerability mechanism depending on the
conditions. It remains essential, however, to demonstrate the conditions under which
protection “kicks in” its effect.
146
LIMITATIONS
This research was conducted with African American boys in one southern
Caiitbmia city with a relatively small sample size. Future research should examine
protection with additional populations. This small cross-sectional dataset allowed
for a test of the relationships for this group of youth, and allowed for the
demonstration of techniques available for use in testing for interactive effects.
However, the generalizability of these findings must be confirmed.
Cross-sectional data preclude a determination of the causal relationships
between variables (Richters & Weintraub 1990, p. 68; Rutter 1994). Resiliency
scholars (Rutter et al.(1998; 1993, p. 626; O ’Dougherty & Wright, 1990, p. 123)
recommend that resiliency is measured at multiple points in time to allow for the
assessment of protective processes that occur after the onset of risk.
In addition, when a single image is taken of risk and the outcome, it neglects
the real life process by which people move in and out of risk statuses (Cowan et al.
1996). This is especially true for gang joining where the joining process is just that,
a process, and may not be marked by a single joining event.
Future research which continues to examine protection as an interactive
process is critical to the development of models of continuities and discontinuities of
deviant outcomes. Risk and protection must be measured comprehensively and
cumulatively, accounting for chronicity, multiplicity and or severity of risk
147
(O’Dougherty & Wright 1990). Hypotheses regarding the relationships of risk and
protective factors must now be tested.
Broderick (1993) suggests that there is an individual chaiacteiistic that, in
interaction with one’s environment, produces what he terms a “transitional character”
(p. 241). This transitional character, “is one who, in the course of a single
generation, changes the course of a lineage...who grows up in an abusive,
emotionally destructive environment, [but] somehow finds a way to metabolize the
poison and refuses to pass it on to [the next generation]" (Broderick, 1992, p. 18).
Research must address these unexpected outcomes, where thriving is an option
despite a disadvantaged family history. This project presents a move in this
direction. It is important to understand why some youth do not join gangs when we
expect them to - and why some do join when our expectations preclude gang
involvement. While our ability to predict and explain the causal paths that lead to
negative outcomes has increased, most theories have not adequately explained the
exceptions - the predictions to competence that occur, despite a difficult family
history. The challenge for future research is to identify processes that promote
success and unexpected outcomes of competence.
148
BIBLIOGRAPHY
Adler, P., Ovando, C., & Hocevar, D. 1984. “Familiar Correlates of Gang
Membership; An Exploratory Study of Mexican-American Youth.” Hispanic
Journal o f Behavioral Sciences. 6(1): 65-76.
Ahrons, C.R. & Rodgers, R.H. 1989. Divorced Families: A Multidisciplinary
Developmental View. New York: W.W. Norton.
Akers, R.L. 1977. Deviant Behavior: A Social Learning Approach. 2^^ edition.
Belmont: Wadsworth.
Aldwin, C.M. 1994. Stress, Coping and Development: An Integrative Perspective.
New York: Guilford Press.
Anthony, E.J. 1974. “The Syndrome of the Psychologically Invulnerable Child.” In
E.J. Anthony & C. Koupemik (Eds.) The Child in His Family. Voll. Ill:
Children at Psychatric Risk. New York: Wiley.
Battin-Pearson, S., Guo, J., Hill, K.G., Abbott, R.D., Hawkins, J.D. 1998. “Early
Predictors of Sustained Adolescent Gang Membership.”
Bjerregaard, B. 2002. “Self-definitions of Gang Membership and Involvement in
Delinquent Activities.” Youth and Society. 31-54.
Bjerregaard, B. & Smith, C. 1993. “Gender Differences in Gang Participation,
Delinquency, and Substance Use.” Journal of Quantitative Criminology. 9:
329-355.
Broderick, C.B. 1993. Understanding Family Process: Basics o f Family Systems
Theory. Newbury Park, CA: Sage.
Bry, B.H., Catalano, R.F., Kumpfer, K.L., Lochman, J.E., & Szapocznik, J. 1998.
“Scientific Findings From Family Prevention Intervention Research.” In R.S.
Ashery, E.B. Robertson, & K.L. Kumpfer (eds.). Drug Abuse Prevention
Through Family Interventions. NIDA Research Monograph 177. US
Department of Health and Human Services. National Institutes of Health.
Butler, K. 1997. “The Anatomy of Resilience.” The Family Therapy Networker.
21(2): 22-31. March/April.
149
Catalano, R.F. & Hawkins, J.D. 1996. “The Social Development Model: A Theory
of Antisocial Behavior.” In J.D. Hawkins (Ed.), Delinquency and Crime:
Current Theories {pp. 149-197). New York: Cambridge University Press.
Chesney-Lind, M., Shelden, R.G., & Joe, K.A. 1996. “Girls, Delinquency, and
Gang Membershi
Oaks, CA: Sage.
Gang Membership.” In C.R. Huff (Ed.), Gangs in America 2”^ ed. Thousand
Cohen, A.K. 1955. Delinquent Boys: The Culture o f the Gang. Glencoe, X L : Free
Press.
Covey, H.C., Menard, S., & Franzese, R.J. 1997. Juvenile Gangs. 2"^ Edition.
Springfield: Thomas.
Cowan, P.A., Cowan, C.P., & Schulz, M.S. 1996. “Thinking About Risk and
Resilience in Families.” In E.M. Hetherington & E.A. Blechman (Eds.) Stress,
Coping and Resiliency in Children and Families. New Jersey: Lawrence
Erlbaum Associates.
Cox, M.J. & Brooks-Gunn, J. 1999. “Studying Conflict and Cohesion in Families:
An Overview.” In M.J. Cox & J. Brooks-Gunn (Eds.) Conflict and Cohesion in
Families: Causes and Consequences. New Jersey: Lawrence Erlbaum
Associates.
Curry, G.D. 2000. “Self-reported Gang Involvement and Officially Recorded
Delinquency.” Criminology 38: 1253-1274.
Curry, G.D. & Decker, S.H. 1998. Confronting Gangs: Crime and Community.
Los Angeles: Roxbury.
Curry, G.D. & Spergel, LA. 1992. “Gang Involvement and Delinquency Among
Hispanic and African-American Adolescent Males.” Journal o f Research in
Crime and Delinquency 29:273-291.
Decker, S.H. & Van Winkle, B. 1996. Life in the Gang: Family, Friends, and
Violence. Cambridge.
DeWit, D.J. & Silverman, G. 1995. “The Construction of Risk and Protective
Factor Indices for Adolescent Alcohol and Other Drug Use.” Journal o f Drug
Issues. 25(4):837-864.
Egley, Jr., A. & Aijunan, M. 2002. “Highlights of the 2000 National Youth Gang
Survey.” OJJDP Fact Sheet.
150
Elliott, D.S., Huizinga, D., & Ageton, S. 1985. Explaining Delinquency and Drug
Abuse. Beverly Hills, CA: Sage.
Elliott, D.S., Huizinga,D., & Menard, S. 1989. Multiple Problem Youth:
Delinquency, Substance Use and Mental Health Problems. New York, NY :
Springer-Verlag.
Esbensen, F.-A. & Deschenes, E.P. 1998. “A Multi-Site Examination of Gang
Membership: Does Gender Matter?” Criminology 36(4): 799-827.
Esbensen, F.-A., Deschenes, E.P., & Winffee, L.T. 1999. “Differences Between
Gang Girls and Gang Boys: Results From a Multi-site Survey. Youth and
S o c i e t y 27-53.
Esbensen, F. & Huizinga, D. 1993. “Gangs, Drugs, and Delinquency in a Survey of
Urban Youth.” Criminology 31: 565-589.
Esbensen, F., Huizinga, D., & Weiher, A.W. 1993. “Gang and Non-gang Youth:
Differences in Explanatory Factors.” Journal o f Contemporary Criminal Justice.
9:94-116.
Esbensen, F., Winffee, 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 47(1):105-130.
Farrington, D P. 2000. “Explaining and Preventing Crime: The Globalization of
Knowledge - The American Society of Criminology 1999 Presidential Address.”
Criminology. 38(1): 1-24.
. 1998. “Predictors, Causes, and Correlates of Male Youth Violence.” In
M.Tonry & M.H. Moore’s Crime and Justice Vol. 24. Chicago: University of
Chicago Press.
Friedman, C.J., Mann, F., & Friedman, A S. 1975. “A Profile of Juvenile Street
Gang Members. Adolescence. 10(40): 563-607.
Garmezy, N. 1984. “Stress-Resistant Children: The Search for Protective Factors.’
In J.E. Stevenson (Ed ), Aspects o f Current Child Psychiatry Research {Journal
o f Child Psychology and Psychiatry book supplement No. 4). Oxford:
Pergamon Press.
151
. 1983. “Stressors of Childhood.” In N. Garmezy & M. Rutter (Eds.) Stress,
Coping, and Development in Children. New York: McGraw-Hill.
Garmezy, N. & Masten, A.S. 1986. “Stress, Competence, and Resilience: Common
Frontiers for Therapist and Psychopathologist.” Behavior Therapy. 17:500-521.
Garmezy, N., Masten, A., Tellegen, A. 1984. “The Study of Stress and Competence
in Children: A Building Block for Developmental Psychopathology.” Child
Development. 55(1): 97-111.
Garmezy, N. & Rutter, M. 1983. Stress, Coping and Development in Children.
New York: McGraw-Hill.
Hare, B.R. 1977. “Hare Area Specific School, Peer, and Home Self Esteem Scale.“
In K. Corcoran & J. Fischer’s Measures fo r Clinical Practice: A Sourcebook,
volume 1. 3^^ edition. New York: Free Press.
Hauser, S.T. 1999. “Understanding Resilient Outcomes: Adolescent Lives Across
Time and Generations.” Journal o f Research on Adolescence. 9(1): 1-24.
Hauser, S T., Vieyra, M.A.B, & Jacobson, A.M. 1989. “Family Aspects of
Vulnerability and Resilience in Adolescence: A Theoretical Perspective.” In T.
F. Dugan & R. Coles (Eds.) The Child in our Times: Studies in the Development
o f Resiliency. Philadelphia, PA: Brunner/Mazel.
Hetherington, E.M. & Blechman, E.A.. (Eds.) 1996. Stress, Coping and Resiliency
in Children and Families. Manhwah, NJ: Erlbaum.
Hill, K.G., Howell, J.C., Hawkins, J.D., & Battin-Pearson, S.R. 1999. “Childhood
Risk Factors for Adolescent Gang Membership: Results from the Seattle Social
Development Project.” Journal o f Research in Crime and Delinquency. 36(3):
300-322.
Hindelang, M.J., Hirschi, T., & Weis, J.G. 1981. Measuring Delinquency. Beverly
Hills, CA: Sage.
Howell, J.C. 1998. “YouthGangs: An Overview.” OJJDP Juvenile Justice Bulletin.
Huff, C.R. 1998. “Criminal Behavior of Gang Members and At-Risk Youths.”
National Institute of Justice Research Preview.
152
Jenkins, J.M. & Smith, M.A. 1990. “Factors Protecting Children Living in
Disharmonious Homes; Maternal Reports.” Journal o f the American Academy
o f Child and Adolescent Psychiatry. 29(1): 60-69.
Klein, M.W. 2001. “Recommendations for the Operational Defiiiiliuii of Street
Gangs prepared by the Eurogang Working Group on
Conceptualizations/Definitions.”
. 1995. The American Street Gang: Its Nature, Prevalence, and Control.
New York: Oxford.
. 1971. Street Gangs and Street Workers. Englewood Cliffs, NH: Prentice-
Hall.
Kumpfer, K.L. 1994. “Family Strengthening in Preventing Delinquency: A
Literature Review.” Office of Juvenile Justice and Delinquency Prevention.
September.
Kumpfer, K.L. & Alvarado, R. 1998. “Effective Family Strengthening
Interventions.” Juvenile Justice Bulletin. Washington, DC: U.S. Department of
Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency
Prevention.
Lahey, B.B., Gordon, R.A., Loeber, R., Stouthamer-Loeber, M., & Farrington, D P.
1999. “Boys Who Join Gangs: A Prospective Study of Predictors of First Gang
Entry.” Journal o f Abnormal Child Psychology. 27(4):261-276.
Larson, W. R. & Myerhoff, B.G. 1967. “Family Integration and Police Contact.” In
M.W.Klein’s (Ed.^ Juvenile Gangs in Context: Theory, Research and Action.
Englewood Cliffs, NJ: Prentice-Hall.
Loeber, R., Farrington, D.P., Rumsey, E., Kerr, C.A., & Allen-Hagen, B. 1998
(May). “Serious and Violent Juvenile Offenders.” Juvenile Justice Bulletin.
Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office
of Juvenile Justice and Delinquency Prevention.
Lyon, J. 1991. Family and Friends: A Comparison of Delinquent Juvenile Gang
Members and Nongang Members. Dissertation. United States International
University.
153
Masten, A.S., Morison, P., Pellegrini, D., & Tellegen, A. 1990. “Competence Under
Stress: Risk and Protective Factors.” In Rolf, J., Masten, A.S., Cicchetti, D.,
Nuechterlein, K.H., Weintraub, S. (Eds.) Risk and Protective Factors in the
Development o f Psychopathology. Cambridge: New York.
Maxson, C. & Whitlock, M.L. 2002. “Joining the Gang: Gender Differences in
Risk Factors for Gang Membership.” In C.R. Huff (Ed.) Gangs in America 1 1 1 .
Thousand Oaks, CA: Sage.
Maxson, C.L., Whitlock, M.L., & Klein, M.W. 1998. “Vulnerability to Street Gang
Membership: Implications for Practice.” Social Service Review 12{\)\ 70-91.
Miller, W.B. 2001. The Growth o f Youth Gang Problems in the United States:
1970-98. Report to the Office of Juvenile Justice and Delinquency Prevention.
. 1958. “Lower Class Culture as a Generating Millieu of Gang Delinquency.”
Journal o f Social Issues 14: 5-19.
Miller, W.B., Geertz, H., & Cutter, H.S.G. 1961. “Aggression in a Boys’ Street-
comer Group.” Psychiatry 24: 283-298.
Moore, J.W. 1991. Going Down to the Barrio: Homeboys and Homegirls in
Change. Philadelphia: Temple University Press.
NYGC. 2002. San Diego Police Department Responses to 2001-2002 National
Youth Gang Survey. National Youth Gang Center.
O ’Dougherty, M. & Wright, F.S. 1990. “Children Bom at Medical Risk: Factors
Affecting Vulnerability and Resilience.” In Rolf, J., Masten, A.S., Cicchetti, D.,
Nuechterlein, K.H., Weintraub, S. (Eds.), Risk and Protective Factors in the
Development o f Psychopathology. Cambridge: New York.
Pianta, R.C., Egeland, B., & Sroufe, L A. 1990. “Maternal Stress and Children’s
Development: Prediction of School Outcomes and Identification of Protective
Factors.” In Rolf, J., Masten, A.S., Cicchetti, D., Nuechterlein, K.H., Weintraub,
S. (Eds.) Risk and Protective Factors in the Development o f Psychopathology.
Cambridge: New York.
Radke-Yarrow, M. & Sherman, T. 1990. “Hard Growing: Children Who Survive.”
In Rolf, J., Masten, A.S., Cicchetti, D., Nuechterlein, K.H., Weintraub, S. (Eds.),
Risk and Protective Factors in the Development o f Psychopathology.
Cambridge: New York.
154
Richters, J., & Weintraub, S. 1990. “Beyond Diathesis: Toward an Understanding
of High-Risk Environments.” In Rolf, J., Masten, A.S., Cicchetti, D.,
Nuechterlein, K.H., Weintraub, S. (Eds.), Risk and Protective Factors in the
Development o f Psychopathology. Cambridge: New York.
Rolf, J., Masten, A.S., Cicchetti, D., Nuechterlein, K.H., Weintraub, S. (Eds.) 1990.
Risk and Protective Factors in the Development o f Psychopathology.
Cambridge : New York.
Rowe, D C., & Farrington, D.P. 1997. “The Familial Transmission of Criminal
Convictions.” Criminology 35(1): 177-201.
Rutter, M. 1994. “Beyond Longitudinal Data: Causes, Consequences, Changes, and
Continuity.” Journal o f Consulting and Clinical Psychology. 62(5): 928-940.
. 1993. “Resilience: Some Conceptual Considerations.” q/"
Adolescent Health. 14:626-631.
. 1991. “Protective Factors: Independent or Interactive?”
American Academy o f Child and Adolescent Psychiatry (?). 30(1): 151-152.
. 1987. “Psychosocial Resilience and Protective Mechanisms.” American
Journal o f Orthopsychiatry. 57 (3): 316-331.
. 1985. “Resilience in the Face of Adversity: Protective Factors and
Resistance to Psychiatric Disorder.” British Journal o f Psychiatry. 147: 598-
611.
. 1981. “Stress, Coping and Development: Some Issues and Some
Questions.” Journal o f Child Psychol... Psychiatry. 22:323-356.
Rutter, M. & Giller, H. 1984. Juvenile Delinquency: Trends and Perspectives.
New York: Guilford Press.
Rutter, M., Giller, H., & Hagell, A. 1998. Antisocial Behavior by Young People.
Cambridge: New York.
Rutter, M. & Quinton, D. 1984. “Long-term Follow-up of Women Institutionalized
in Childhood: Factors Promoting Good Functioning in Adult Life.” British
Journal o f Developmental Psychology 18:225-234.
155
Sameroff, A.J. & Seifer, R. 1990. In (Eds.) Rolf, J., Masten, A.S., Cicchetti, D.,
Nuechterlein, K.H., Weintraub, S. Risk and Protective Factors in the
Development o f Psychopathology. Cambridge: New York. Pp. 52-66.
Sampson. R.J. 1992. “Family Management and Child Development.” In J. McCord
(Ed.), Facts, Frameworks, and Forecasts. Vol. 3 o f Advances in Criminological
Theory. New Brunswick, NJ: Transactions.
Sherman, L.W., Gottffedson, D., MacKenzie, D., Eck, J., Renter, P., & Bushway, S.
1997. “Preventing Crime: What Works, What Doesn’t, W hat’s Promising: A
Report to the United States Congress.” Prepared for the National Institute of
Justice.
Short, J.F. Jr. & Strodtbeck, F.L. 1965. Group Process and Gang Delinquency.
Chicago: University of Chicago Press.
Smith, C. A., Lizotte, A.J., Thomberry, T.P., Krohn, M.D. 1997. “Resilience to
Delinquency.” The Prevention Researcher. 4: 4-7.
Spergel, LA. 1995. The Youth Gang Problem: A Community Approach. New
York: Oxford.
Sutherland, E.H. 1939. Principles o f Criminology. Philadelphia: Lippincott.
Tennyson, R.A. 1967. “Family Structure and Delinquent Behavior.” In M.W.
Klein’s (Ed.) Juvenile Gangs in Context: Theory, Research and Action.
Englewood Cliffs, NJ: Prentice-Hall.
Thrasher, F. 1927. The Gang. Chicago: University of Chicago Press.
Thomberry, T.P. 1998. “Membership in Youth Gangs and Involvement in Serious
and Violent Offending.” In R. Loeber & D.P. Farrington (Eds.) Serious and
Violent Offenders: Risk Factors and Successful Interventions. Thousand Oaks,
CA: Sage.
Thomberry, T.P. & Burch II, J.H. 1997. “Gang Members and Delinquent Behavior.”
OJJDP Juvenile Justice Bulletin.
Thomberry, T.P., Krohn, M.D., Lizotte, A.J., & Chard-Wierschem, D. 1993. “The
Role of Juvenile Gangs in Facilitating Delinquent Behavior. Journal o f Research
in Crime and Delinquency 30: 55-87.
156
Thomberry, T.P., Krohn, M.D., Lizotte, A.J., Smith, C.A., & Tobin, K. 2003.
Gangs and Delinquency in Developmental Perspective. Cambridge: Cambridge
University Press.
Vigil, J.D. 2002. A Rainbow o f Gangs: Street Cultures in the Mega-City. Austin:
University of Texas Press.
1988. Barrio Gangs: Street Life and Idenliiy in Suulhern California. Austin,
TX: University of Texas Press.
Walker-Bames, C.J. & Mason, C.A. 2001. “Ethnic Differences in the Effect of
Parenting on Gang Involvement and Gang Delinquency: A Longitudinal,
Hierarchical Linear Modeling Perspective.” Child Development. 72(6): 1814-
1831.
Wemer, E.E. & Smith, R.S. 1992. Overcoming the Odds: High Risk Children from
Birth to Adulthood. Ithaca, NY : Comell University Press.
Wilson, J.J. & Howell, J.C. 1993. Comprehensive Strategy fo r Serious, Violent, and
Chronic Juvenile Offenders. Program Summary. Washington, DC: U.S.
Department of Justice, Office of Justice Programs, Office of Juvenile Justice and
Delinquency Prevention.
West, D.J. & Farrington, D.P. 1977. The Delinquent Way o f Life. London:
Heinemann Educational.
Wilson, W.J. 1987. The Tmly Disadvantaged. Chicago: University of Chicago
Press.
Wolfgang, M.E., Figlio, R.M., & Sellin, T. 1972. Delinquency in a Birth Cohort.
Chicago: University of Chicago Press.
157
Appendix 1. Composition of self-reported delinquency scales.
ALCOHOL/LIQUOR USE
In the last six months, have you:
1. Drunk beer or wine without your parent’s permission?
2. Drunk hard liquor without your parent’s permission?
DRUG SALES
In the last six months, have you:
1. Sold marijuana?
2. Sold crack or rock?
3. Sold hard drugs such as heroin, cocaine, LSD or acid?
DRUG USE
In the last six months, have you:
1. Used marijuana?
2. Used acid, LSD, psychedelics or hallucinogens?
3. Used cocaine or coke, other than crack?
4. Used crack or rock?
5. Used heroin?
6. Used angel dust or PCP?
7. Used tranquilizers?
8. Used downers or barbiturates?
9. Used uppers, speed or amphetamines?
GENERAL DELINQUENCY
In the last six months, have you:
1. Run away from home?
2. Carried a hidden weapon?
3. Been loud or rowdy in a public place where somebody complained and got you in
trouble?
4. Begged for money or things from strangers?
5. Gone into or tried to go into a building to steal or damage something?
6. Tried to steal or actually stolen money or things worth $5 or less?
7. How about between $5 and $50?
8. How about between $50 and $100?
9. How about over $ 100?
10. Tried to buy or sell things that were stolen?
11. Taken a car or motorcycle for a ride without the owner’s permission?
12. Stolen or tried to steal a car or other motor vehicle.
13. Been involved in a gang fight?
14. Been paid for having sexual relations with someone?
158
15. Physically hurt or threatened to hurt someone to get them to have sex with you?
16. Sold marijuana?
17. Sold hard drugs such as heroin, cocaine, LSD or acid?
18. Used a weapon or force to make someone give you money or things?
MINOR OFFENDING
In the last six months, have you:
1. Tried to steal or actually stolen money or things worth $5 or less?
2. Taken a car or motorcycle for a ride without the owner’s permission?
3. Been loud or rowdy in a public place where somebody complained and got you in
trouble?
4. Begged for money or things from strangers?
5. Run away from home?
6. Hit someone with the idea o f hurting them (other than what you have already
mentioned)?
7. Skipped class without an excuse?
8. Lied about your age to get into some place or to buy something (for example,
lying about your age to get into a movie or to buy alcohol?
9. Hitchhiked a ride with a stranger?
10. Made obscene telephone calls such as calling someone and saying dirty things?
11. Been drunk in a public place?
12. Avoided paying for things, like a movie, taking bus rides, or anything else?
VIOLENT OFFENDING
In the last six months, have you:
1. Set fire on purpose or tried to set fire to a house, building, or car, knowing
someone was inside?
2. Attacked someone with a weapon or with the idea of seriously hurting or killing
them?
3. Been involved in a gang fight?
4. Hit someone with the idea of hurting them (other than what you have already
mentioned)?
5. Used a weapon or force to make someone give you money or things?
6. Physically hurt or threatened to hurt someone to get them to have sex with you?
7. Thrown objects such as bottles or rocks at people?
159
Appendix 2. Six-month prevalence for delinquency items (percent yes).
Gang Nongang
In the last six months, have you: (Fercent yes) (Fercent yes)
Run away from home? 10% (8) 3% (3)
Skipped class without an excuse? 62% (48) 46% (40)*
Lied about your age to get into some place or to buy something
(for example, lying about your age to get into a movie or to buy
alcohol)?
39% (30) 24% (21)*
Hitchhiked a ride with a stranger? 5% (4) 1% (1)
Carried a hidden weapon? 62% (48) 18% (16)***
Been loud or rowdy in a public place where somebody
complained and got you in trouble?
52% (40) 24% (21)***
Begged for money or things from strangers? 4% (3) 2% (2)
Made obscene telephone calls, such as calling someone and
saying dirty things?
14% (11) 10% (9)
Been drunk in a public place? 35% (27) 5% (4)***
Damaged, destroyed or marked up someone else’s property on
purpose?
27% (21) 11% (10)**
Set fire on purpose or tried to set fire to a house, building, or
car, knowing someone was inside?
4% (3) 1% (1)
Avoided paying for things, like a movie, taking bus rides, or
anything else?
43% (33) 26% (23)*
Gone into or tried to go into a building to steal or damage
something?
23% (18) 9% (8)*
Tried to steal or actually stolen money or things worth $5 or
less?
26% (20) 15% (13)
How about between $5 and $50? 31% (24) 8% (7)***
How about between $50 and $100? 21% (16) 6% (5)**
How about over $100? 23% (18) 6% (5)***
160
Gang Nongang
In the last six months, have you: (Percent yes) (Percent yes)
Shoplifted or taken something from a store on purpose
(including anything you have already told me about)?
35% (27) 19% (17)*
Stolen someone’s purse or wallet or picked someone’s pocket? 12% (9) 7% (6)
Stolen something that did not belong to you? 43% (33) 16% (14)***
Tried to buy or sell things that were stolen? 36% (28) 13% (11)***
Taken a car or motorcycle for a ride without the owner’s
permission?
23% (18) 3% (3)***
Stolen or tried to steal a car or other motor vehicle? 20% (15) 2% (2)***
Forged a check or used fake money to pay for something? 5% (4)
__ *
Used or tried to use a credit card, bank card, or automatic teller
card without permission?
9% (7) 2% (2)*
Tried to cheat someone by selling them something that was not
what you said it was or that was worthless?
20% (15) 8% (7)*
Attacked someone with a weapon or with the idea o f seriously
hurting or killing them?
33% (25) 1% (1)***
Thrown objects such as bottles or rocks at people? 38% (29) 22% (19)*
Been involved in a gang fight? 60% (46) 7% (6)***
Hit someone with the idea o f hurting them (other than what you
have already mentioned)?
40% (31) 25% (22)*
Used a weapon or force to make someone give you money or
things?
17% (13) 7% (6)*
Been paid for having sexual relations with someone? 5% (4)
__ *
Physically hurt or threatened to hurt someone to get them to
have sex with you?
— —
Had or tried to have sexual relations with someone against their
will (other than what you have already mentioned)?
1% (1) 3% (3)
Sold marijuana? 36% (28) 6% (5)***
Sold crack or rock? 10% (8)
161
Gang Nongang
In the last six months, have you: (Percent yes) (Percent yes)
Sold hard drugs such as heroin, cocaine, LSD or acid? 4% (3)
—
Drunk beer or wine without your parent’s permission? 55% (42) 32% (28)**
Drunk hard liquor without your parent’s permission? 31% (24) 6% (5)***
Used marijuana? 57% (44) 24% (21)***
Used acid, LSD, psychedelics or hallucinogens? 1% (1) -
Used cocaine or coke, other than crack? - —
Used crack or rock? — —
Used heroin? —
—
Used angel dust or PCP? 8% (6)
Used tranquilizers? -
—
Used downers or barbiturates? — —
Used uppers, speed or amphetamines? 1%(1) —
*p<.05; **p<.01; ***p<.001
162

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Family -based risk and protective mechanisms for youth at -risk of gang joining

PDF

The sociolinguistics of tagging and Chicano gang graffiti

PDF

A structural and psychoanalytic analysis of the symbolic ritual process in three sets of Russian folk tales

PDF

William Lawes: His life and works

PDF

Effects of parenting style and ethnic identity on European American and Asian Indian adolescents' academic competence and self esteem

PDF

Psychologists' perceptions of older clients: The effect of age, gender, knowledge, and experience

PDF

Factors that effect the changes in degree aspirations of African American and Latino community college students

PDF

Fate in the novels of Zola and Couperus; a comparison with the Greek concept of fate

PDF

Crisis preparation in technical organizations: A study using a multi-dimensional approach

PDF

An experimental analysis of source credibility and message discrepancy under differential levels of ego-involvement in political television commercials

PDF

A study of idealism in the Nobel Prize literature

PDF

Bulimia nervosa: A response to trauma associated with parental alcoholism and sexual abuse in college sorority women

PDF

Reaffirmation and transformation of gender in popular film: A feminist approach to mythic rhetoric

PDF

Math and reading test anxiety among three populations: Mexicans, Hispanic immigrants, and Anglo-Saxon children

PDF

Era of limits; Edmund G. Brown, Jr.s' approach to policy decisions during his governorship 1975-1982

PDF

The reputation of Cicero among the English Deists (1696 - 1776)

PDF

Life goals of retirement home residents and applicants in selected organizational climates

PDF

The life and Mantuan masses of Francesco Rovigo (1541/42-1597)

PDF

Clinical social work practice with children and youth: A content analysis of social work practice literature 1947-1973

PDF

The social meaning of women's literacy in nineteenth-century America

Asset Metadata

Creator Whitlock, Monica Lin (author)

Core Title Family-based risk and protective mechanisms for youth at-risk of gang joining

Degree Doctor of Philosophy

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

Tag OAI-PMH Harvest,Social Sciences

Format application/pdf (imt)

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c37-46284

Unique identifier UC11633247

Identifier DP71343.pdf (filename),usctheses-c37-46284 (legacy record id)

Legacy Identifier DP71343.pdf

Dmrecord 46284

Document Type Dissertation

Format application/pdf (imt)

Rights Whitlock, Monica Lin

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA