141
720

.THS_

 

 

.UBRARY
Michigan State
University

 

 

 

This is to certify that the
thesis entitled

THE IMPACT OF PARENTAL MONITORING ON
DELINQUENT BEHAVIORS

presented by

BYUNG H. LEE

has been accepted towards fulﬁllment
of the requirements for the

MS. degree in Criminal Justice

 

 

 

ﬂ

Illlajor P ofessor’s Signature
/ 9/02/02” ?

Date

MSU is an Afﬁnnative ActiorVEquaI Opportunity Employer

.
.I—l-l-O-0-a_----u-I-I-I--o-o-—.—-

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DaleDue.indd

THE IMPACT OF PARENTAL MONITORING ON DELINQUENT BEHAVIORS
By

Byung Hyun Lee

A THESIS
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
MASTER OF SCIENCE

Criminal Justice

2009

ABSTRACT
THE IMPACT OF PARENTAL MONITORING ON DELINQUENT BEHAVIORS
By
Byung Hyun Lee
Previous studies of juvenile delinquency suggest that various family factors

inﬂuence youth engagement in delinquent behaviors. Although there are numerous
causes of delinquency among youth, research indicates that the role of parental
supervision and the parents' presence in the children’s daily lives decrease the probability
of delinquent behaviors. Using Hirschi’s social bond theory (1969), the purpose of this
cross-sectional study is to examine the effect of parental monitoring on adolescent
engagement in delinquency and substance use based on the National Longitudinal Survey
of Youth 1997, a nationally representative sample of adolescents. In addition, the study
investigates the aspects of family structure and peer inﬂuence and its effect on the two
outcomes. The multivariate regression results reveal that parental monitoring was not
foremost predictor of delinquency and substance use among other correlates. Instead, the
variables associated with family structure and peer inﬂuence came out to be the
signiﬁcant predictors of delinquency as well as substance use. Due to the limitations of a
cross-sectional design in establishing causal relationships, future research should conduct

a longitudinal analysis including a larger number of variables related to family context.

To my wife Jeong 800 and our son Philip

iii

ACKNOWLEDGEMENTS

Throughout the years during my master’s program, I am thankful for having
various “helpers.” I would like to convey my truthful gratitude to my committee
members: To Dr. Sheila Maxwell, chair of the thesis committee, who provides me with
the support, guidance and advice toward the completion of this study, I am very thankful
for her presence. To Dr. Vincent Hoffman, who has always offered me meaningful
advices and encouragement. To Dr. Chris Maxwell, who present his expertise and
insights for the study. The feedback and criticism I received helped me to successfully
complete my thesis.

My acknowledgement also goes to my family, especially, my wife. To my wife,
J eong 800, who always gave me positive encouragement, and emotional support for me.
Thank you for your prayers and understanding. Lastly, I acknowledge God for always
leading to a good path during tough times. I am grateful for your unconditional love for

me and my family.

iv

TABLE OF CONTENTS

LIST OF TABLES .............................................................................. vi
CHAPTER 1 - INTRODUCTION .............................................................. 1
CHAPTER 2 - A REVIEW OF THE LITERATURE ...................................... 4
I. HIRSCHI’S SOCIAL BOND THEORY .................................................... 5
II. CONCEPTUALIZATION OF PARENTAL MONITORING ........................ 11
III. PARENTAL MONITORING AND DELINQUENCY ................................ 14
CHAPTER 3 -— DATA AND METHODOLOGY ................ - ............................. 18
I. RESEARCH HYPOTHESIS ................................................................. 18
II. DATA COLLECTION ....................................................................... 19
Data ............................................................................................. 19
Sample .......................................................................................... 20
Sampling Procedure ........................................................................... 22

III. VARIABLES ................................................................................ 24
Dependent Variables .......................................................................... 24
Independent Variables ....................................................................... 27
Demographic Variables ...................................................................... 30
CHAPTER 4 - DATA ANALYSES AND FINDINGS ....................................... 32
1. METHODS OF ANALYSIS ................................................................ 32
II. FINDINGS .................................................................................... 33
Bivariate Statistics ............................................................................ 33
Variables associated with Delinquency .................................................... 33
Variables associated with Substance Use ................................................. 37
Summary of the Bivariate Relationships ................................................... 4O
Multivariate Statistics ....................................................................... 41
Multicollinearity Diagnostics ................................................................ 41
Logistic Regression Analysis (Delinquency) .............................................. 43
Logistic Regression Analysis (Substance Use) ........................................... 48
CHAPTER 5 - DISCUSSION AND CONCLUSION ......................................... 54
I. SUMMARY AND DISCUSSION .......................................................... 54
Summary ....................................................................................... 54
Study Limitations ............................................................................. 58

11. CONCLUSION AND IMPLICATIONS FOR FUTURE STUDY. . . . . 59
Conclusion ..................................................................................... 59
Implications for Future Study ............................................................... 61
APPENDIX A - MULTICOLLINEARITY ..................................................... 63
BIBLIOGRAPHY. . . . . . .. .................................................................................................. 64

LIST OF TABLES

Table 1. Descriptive Statistics and Variable Coding (NLSY 1997) ......................... 27
Table 2. Percent of Delinquency by Parental Monitoring, Parental Structure, and
Demographic Characteristics ............................................................ 33
Table 3. Percent of Substance Use by Parental Monitoring, Parental Structure, and
Demographic Characteristics ........................................................... 37
Table 4. Collinearity Diagnostics ................................................................ 42
Table 5. Logistic Regression Models of Delinquency ......................................... 45
Table 6. Logistic Regression Models of Substance Use ....................................... 51

Table 7. Z-Score of Difference between Regression Estimates for Delinquency and
Substance Use ............................................................................. 57

vi

CHAPTER I
INTRODUCTION

After the sharp rise in crime during the mid-19803 throughout the early 19903 in
the US, American society feared that the trend of violence would continue in an upward
direction (Butts & Travis, 2002). According to the Uniform Crime Reports (UCR) and
National Crime Victimization Survey (N CVS), the overall crime rate sharply increased
throughout the 19805 and early 19903. Nobody expected that America would witness a
signiﬁcant drop in crime in the mid-19905 in most categories of crime and in all
geographic parts of the country (Blumstein & Wallman, 2000; Levitt, 2004). Even though
scholars debated the causes of the sudden drop in violent crime in the US. during the
mid-19903, they came to the consensus that the national trend in violent crime was
related in some way to the descending rates of juvenile violence across the country
(Butts, 2000). Today, researchers assert that the recent decline in most categories of
violent crime was related to dropping rates of juvenile crime (Butts, 2000; Blumstein &
Wallman, 2000; Snyder & Sickmund, 2006)

This apparent downward trend in juvenile crime is partly illustrated by the decline
in juvenile arrests. As indicated by a recent research report of youth violence in America
from 1980 to 2000, of the 14 million arrests made by law enforcement agencies, 17
percent included juveniles under the age of 18 — a signiﬁcant decrease from previous
years (Butts & Travis, 2002). According to the report by Butts and Travis, between 1994
and 2000, juvenile arrests for murder were reduced by 68 percent, arrests for burglary and
robbery by juveniles were down 33 percent and 51 percent respectively, and juvenile

arrests for auto theft dropped by 42 percent (Butts & Travis, 2002). Hence, the rate of

decline in juvenile crime between the years 1994 and 2000 was conspicuous, particularly
in four main categories of the Violent Crime Index such as homicide, forcible rape,
aggravated assault and robbery (Butts & Travis, 2002). Despite the conjecture about
plausible reasons for the sharp increase of crime in the 19803 and early 19903 as well as
the steep decline in the mid to late 19903, a majority of scholars seem to agree that the
effect of ﬂuctuation in juvenile population was associated with the varying rates of crime
(Butts & Travis, 2002; Blumstein & Wallman, 2000).

According to Juvenile Offenders and Victims: 2006 National Report, although
there has been a steady increase in violent crime by youth beginning the mid-19803, there
has also been constant decline in the same category since the mid-19903 (Snyder &
Sickmund, 2006). For example, the report shows that between the years 1994 and 2002,
the number of murders by juvenile offenders fell 65 percent to its lowest level since 1984.
Moreover, the use of marijuana had declined from 1997 to 2004 for school youth in 8th,
10th and 12th grade (Snyder & Sickmund, 2006). Although these measures may not
reﬂect the true picture and magnitude of juvenile crime, it illustrates a comprehensive
depiction of the trends and levels of crime committed by adolescents across the nation in
the last decade.

Assuming that the problem posed by violent juvenile crime persists, the need for
further study of the correlates of adolescent delinquency such as family characteristics
will increase as the juvenile population grows. According to a report by the US. Census
Bureau (2001), the population of juveniles grew through 2002 and projects that it will
continue to escalate steadily throughout ﬁrst half of the 213t century. The Census Bureau

(2001) estimates that it will increase 14 percent between the years 2000 and 2025 and, by

2050, the juvenile population will be 36 percent larger than the ﬁgure in 2000. The
escalation of the juvenile population conveys implications for future juvenile crimes and
it is an issue that cannot be neglected as more adolescents reach their years of crime-
prone ages.

In seeking the causes of juvenile delinquency, research looked into parenting
behaviors and parental inﬂuence on children. Hirschi’s social bond theory has been at the
core of the criminological research and has received a high level of empirical support in
examining the relationship between social bond and delinquency. Using Hirschi’s social
control theory, the purpose of this study is to examine the effect of parental supervision
on adolescent engagement in delinquency based on the National Longitudinal Survey of
Youth 1997 (NLSY 97), a nationally representative sample of adolescents. This study
will also take into account the youth’s gender, race, family income, peer inﬂuence, and
living with parents to create a broader picture of the relationship between parental
supervision and delinquency. By understanding the inﬂuence of parental supervision on
child delinquency, scholars, practitioners and policy-makers will be able to better assess
the relationship between these two variables and, in tum, produce polices that help to

prevent delinquent behaviors and to mitigate its related risks.

CHAPTER 2

A REVIEW OF THE LITERATURE

Control theories argue that people will commit crimes if they do not develop
normative social bonds. Furthermore, individuals may still engage in criminal behaviors
due to various inﬂuences such as biological, social or psychological inﬂuences. The
control theory seeks to explain why individuals do not engage in deviant behaviors rather
than why they do. The control perspective argues that most people do not commit
criminal acts if they are strongly bonded to social institutions such as family and
“controlled” by their inﬂuence that inhibit them from criminal activities; the social bond
formed could curb deviant behaviors. However, if the individuals fail to form the strong
bond to signiﬁcant others, especially to family, they are likely to engage in criminal
activities.

Among the control theories, Hirschi’s social bond theory (1969) is one of the
most frequently investigated frameworks for empirical examination in the literature on
the parent-child relationship and delinquency (V old, Snipes, & Bernard, 2002). Since
Hirschi’s publication of Causes of Delinquency (1969), control theories have arguably led
the way for criminological research (V old, Bernard, & Snipes, 2002). While the concept
of social control can be discussed in different theoretical models, Hirschi’s concept of
social control (1969) centers around the family as the main source for reducing
delinquency.

Studies have attempted to explain delinquency by focusing on family inﬂuence
and parenting (Elliot et al., 1979; Patterson, 1982; Cernkovich & Giordano, 1987; Laub

& Sampson, 1988; Krohn, 1986; Thomberry, 1987). The dimensions of the family

process concept, such as attachment, involvement, quality of parent-child relationship,
and supervision, serve as core elements of social control theory by Hirschi (1969). They
are fundamental to social bonding interactions that take place in human relationships,
particularly in parent-child relationships. A number of studies illustrate the role of family
and parents in curbing delinquency and substance use (Barnes et al., 2006; Cernkovich &
Giordano, 1987; Fletcher et al., 2004; Jensen & Brownﬁeld, 1983; Krohn, 1974; Loeber
& Stouthamer-Loeber, 1986; McCluskey & Tovar, 2003; McCord, 1991; Oxford et al.,
2000; Rankin & Wells, 1990; Taylor, 1985; Wasserrnan et a1. 1996; Weintraub & Gold,
1991; Wells & Rankin, 1988). Attachment and involvement of parents are the two
elements of social control that are conceptually closely related to parental control,
whether it is direct or indirect control (Wells & Rankin, 1988). A prime example of
parental control or regulation included in a number of related studies is parental
supervision or monitoring (Petit et al., 2001; Wells & Rankin, 2006). The terms,
monitoring and supervision, seem to be used interchangeably in the literature. This study
speciﬁcally investigates the relationship between parental supervision and delinquent
behaviors of the adolescents. Hirschi’s social theory suggests that parental monitoring
will be predictive of adolescent outcome variables like delinquency and substance use.

As a theoretical framework, the social bond theory is discussed in greater detail below.

I. Hirschi’s Social Bond Theory
Hirschi (1969) provides theoretical discussions and empirical analyses of his
social bond theory. Like most control theorists, Hirschi’s theory also rests on the

Hobbesian supposition that human beings are not innately programmed to conform to

social norms or rules, but rather guided by primitive instincts and hence “naturally
capable of committing criminal acts” (Hirschi, 1969; Wiatrowski, Griswold, & Roberts,
1981). One of Hirschi’s ﬁmdamental propositions is that there is an inverse relationship
between social bonds and delinquency (Hirschi, 1969). Hirschi emphasized that
delinquency can be restrained by the social bonds that children develop in prevailing
social order (Hirschi, 1969; Nye, 1958; Gottfredson & Hirschi, 1990). It assumes that
delinquent behaviors occur when the individual’s bond to society is weak or broken. In
theory, Hirschi makes the claim that adolescents and young adults are likely to conform
to societal norms and parental expectations.

According to Hirschi (1969), the concept of social bond consists of four elements
— attachment, involvement, commitment, and belief. Hirschi (1969) argued that these four
elements, independently and collectively, constituted signiﬁcant barriers to adolescent’s
delinquent behavior. The children’s attachment to parents is the most crucial among the
bonds, including peers or teachers (Hirschi, 1969). Attachment includes the emotional
bonds of the child to the signiﬁcant others such as parents, peers, and teachers. If a child
is not attached to others, he or she may be insensitive to other people’s emotions and to
norms set by others. Family environment is the foundation upon which a strong positive
attachment could be built to serve as a barrier to delinquency (Jensen & Brownﬁeld,
1983). Involvement is when a child is engaged in conventional activities with close
acquaintances, or is associated with healthy organizations or activities such as youth
groups, which not only allow children to pursue conventional goals and values but also to
conﬁrm to community values. Commitment could be illustrated in a child’s dedication to

conventional goals within the community he or she is part of. Belief refers to when a

child is able to believe in the norms and values established by community that helps the
child to refrain from delinquent choices.

The underlying assumption of Hirschi’s account reinforces the argument that
delinquent behaviors occur when a child’s bond to family or society is weak. If the bond
of affection to the family is strong, the attachment formed may be able to deter criminal
behaviors. The theory implicitly states that the stronger this bond, the higher the
likelihood the child will internalize the social norms and make rational considerations
when faced with moral dilemmas. A3 Hirschi’s social control theory has received much
scholarly attention from criminologists, there have been several empirical studies that
tested this theory; many provided empirical support and development and some contested
its supposition and questioned its validity.

In support of Hirschi’s account, the study by Krohn and Massey (1980) had a
three-fold purpose. First, using data from a sample of more than 3,000 male and female
adolescents in six communities of three Midwestern states, they examined the “overall
and relative effects” of the four elements of the social bond theory to improve upon the
previous limited inquiries that address Hirschi’s work. Krohn and Massey (1980)
incorporated Hirschi’s social bond elements of involvement and attachment to social
institutions into the concept of commitment in its operationalization of the variable.
Secondly, for their analysis, the measures were divided into major and minor forms of
delinquency as well as substance use with a total of four distinct measures, compared to
previous research that examined the effects based on scales that illustrate the degree of
seriousness for delinquent behaviors. Lastly, gender difference was investigated in

regards to the effect of social bond on deviance. Major ﬁndings show that among the

three elements, the commitment variable has a stronger correlation with delinquency than
either one of the two remaining elements. Regarding the gender difference, results show
that their model consistently explains more variance in delinquent behaviors for female
adolescents compared to the male adolescents. Despite the modiﬁcation of the
commitment variable, the study on the whole found a moderate level of support for the
social bond account by Hirschi.

Another study that examined the social bond theory is the longitudinal study
conducted by Wiatrowski et a1. (1981), who attempted to test and improve on Hirschi's
original model using data drawn from Youth in Transition Study by Bachman (1975)
with a sample of 2,212 school boys. In order to concurrently consider how the four
elements of social bond operate with respect to delinquency, Wiatrowski et al. (1981)
examined three different multivariate models of social control theory that include the four
elements of social bond for explaining delinquency: a simple, a complex, and a modiﬁed
complex model of the social bond. Limitations of the study presented by the authors were
the questionable status of causal ordering in their path models and the use of
unidirnensional measure of delinquency. Despite these limitations, the study produced
valuable results; parental attachment have a strong negative relationship with
delinquency, as predicted by the social control theory (Wiatrowski et al., 1981).
Additionally, the study (1981) provided empirical support for the claim that social bonds
are built not only in context of family but also in an educational setting such as schools.
These empirical tests of Hirschi’s social control theory strengthen the explanatory power
of certain elements of social bond, particularly attachment in predicting delinquency

among younger adolescents.

The study by Knight and Tripodi (1996) contributes to the collective support for
the social bond account. In their empirical test of Hirschi's social bond theory, Knight and
Tripodi (1996) applied Hirschi’s theoretical argument to a group of 168 severely
delinquent juveniles. They interviewed a total of 168 severely delinquent juveniles from a
regional detention center in 1993. They focused on the effect of bonding to society in
three main areas: attachment to parents, peer association, and school bonding. The key
unanticipated ﬁnding of this study is that as the juveniles commit more crimes, the
attachment to their parents increase. Knight and Tripodi (1996) provide a partial
explanation that the population of these severely delinquent youth formed a close
bonding with their parents, 70 percent of which have been incarcerated previously.
Parallel to Hirschi’s idea that strong bonds, including the four elements of attachment,
belief, involvement, and commitment, guide youth to conform to societal norms. These
severely delinquent juveniles are susceptible to conforming to a deviant subculture, partly
through bonding with their deviant parents (Knight & Tripodi, 1996). With their
statistical analysis, they conclude that "strong bonds to criminal subculture" inﬂuence
youth to be less likely to conform to conventional activities and behavior (Knight &
Tripodi, 1996).

More recent studies by illustrate that social bond theory continues to be
empirically investigated by current researchers (Junger & Marshall, 1997; Sims, 2002;
Chapple, McQuillan, & Berdahl, 2005; Rankin, 2006; Booth, Farrell, & Varano, 2008;
Ozbay & Ozcan, 2008). Among these latest studies, an inquiry by Junger and Marshall
(1997) used Hirschi’s social bond theory to test the generalizability of social bond in a

interethnic and cross-national context. The research was conducted on a sample of 788

male adolescents originally from Surinam, Morocco, Turkey, and the Netherlands, with
all of them residing in the Netherlands. The results reveal that social bond is predictive of
general delinquency as well as a variety of delinquent behaviors in each of the four ethnic
groups. Four of the most consistent social bonding variables predictive of delinquency
were beliefs, school conﬂict, involvement in unconventional activities, and virtual
supervision by family (Junger & Marshall, 1997). Findings suggest that youth involved in
delinquency are more likely to have "lower degree of direct supervision by their parents,
more conﬂicts with teachers, more supportive attitudes toward deviant activities and
more involvement in unconventional leisure activities” (Junger & Marshall, 1997, p.
212). Although there are limitations concerning validity and generalizability, the study
provides empirical support for Hirschi’s social bond theory.

In contrast to the studies above, Agnew considers that Hirschi’s theoretical
application should be reevaluated. In Agnew’s longitudinal test (1985) of social control
theory using a national sample of male adolescents, the data analysis indicates that
variables related to Hirschi’s concept of social control explains only a trivial percentage
of variation in future delinquent behavior, especially when examining more serious forms
of delinquency and among older adolescents. Agnew (1985) also pointed out that only a
few studies with limited sample have tested the empirical validity of the social bond
theory. Despite the unsettled challenges and criticisms, Hirschi’s social control theory
continues to be investigated as a theoretical framework by researchers all over the globe
(Junger-Tas, 1992; Nagasawa, Qian, & Wong, 2000; Wang, Qiao, Hong, & Zhang, 2002;

Burcu, 2003; Ozbay & Ozcan, 2008).

10

Following the criticisms of his theory’s shortcomings, Hirschi collaborated with
Gottfredson to provide an alternative explanation for delinquency. The General Theory of
Crime by Gottfredson and Hirschi (1990) support the link between social control and
self-control under the assumption that supervision by parents not only prevents
delinquent behaviors but also trains children to turn away from temptations for criminal
conduct on their own. Gottfredson and Hirschi (1990) claim that ineffective child-rearing
by parents is the main source accountable for low self-control which leads to delinquent
and irresponsible behaviors. Gottfredson and Hirschi (1990) also claim that lack of self-
control is a strong predictor of delinquent behaviors. They argue that children brought up
in unstructured family environments fail to develop the ability to control and manage
their behaviors and consequently are prone to engage in irresponsible behaviors that often
involve risks leading to commit crimes. Gottfredson and Hirschi (1990) argued that in
order to control deviant behavior and teach children about self-control, parents need to
instigate proper monitoring, acknowledge children’s deviant behavior, and utilize
adequate rewards and punishment. Children will come to realize and internalize the social
norms and rules of conduct and acquire a sense of discipline. Although some believe the
theory has a limitation in that it does not explain enough variance in criminal behavior
compared to other sociological explanations such as social learning and differential
association, several studies testing Gottfredson and Hirschi's theory have provided
empirical support concerning the relationship between self-control and crime (Evans et

al, 1997; Pratt & Cullen, 2000).

II. Conceptualization of Parental Monitoring

11

Efforts of researchers to analyze family process variables had been
multidisciplinary, and many of the studies have conceptualized and operationalized
measures of family variables such as parental monitoring in various ways (Canter, 1982;
Cernkovich & Giordano, 1987; Van Voorhis, Cullen, Mathers & Garner, 1988). Due to
the multidimensional nature of parenting and its varied conceptualization in different
disciplines, variables related to social control such as parental supervision, have only had
empirically limited tests (J ang & Smith, 1997). There is no consistent way of
conceptualizing parental supervision and many of the attempts to do so overlap in parts.
Among the studies, a list of the mostly frequently mentioned elements of the
conceptualizing parental supervision is the physical presence of a parent in overseeing the
child’s activities, direct parental control, and the parent’s awareness and knowledge about
the child (McCord, 1991; Nye, 1958; Warr, 2005; Wells and Rankin, 1998; Wright and
Cullen, 2001).

For the purpose of this study, the parental awareness and knowledge will be the
main focus because it closely resembles the measurement and conceptualization of the
parental monitoring variable in the NLSY 97 dataset. Essentially, it measured how much
the parent(s) know about the child’s well-being such as the child’s school attendance, use
of free time activities at home and at school, or feelings towards drug use. In other words,
the parental monitoring in the study is generally deﬁned as the extent of knowledge to
which parents attempt to give attention to, to track, or control their children’s activities,
their whereabouts, peers, and adaptations (Dishion & McMahon, 1998; Kerr & Stattin,

2000; Jacobson & Crockett, 2000).

12

The task of conceptualization is not simple because it may be difficult to capture
dynamic processes, such as parental control, supervision, or involvement. In particular,
the terms of supervision and monitoring seem to be used interchangeably in different
contexts. Despite the ambiguous use of these terms, studies have taken steps in
conceptualizing parental control, characterizing direct control by parents to comprise of
close monitoring, parental supervision, and family guidelines within the household and
rational sanctions for non-compliance (Patterson, 1980).

First of all, one of the main elements of the conceptualizing parental supervision
seems to be the physical presence of a parent in overseeing the child’s activities. For
example, McCord (1991) conceptualized parental supervision as the “degree to which the
boy’s activities after school were governed by an adult,” focusing on the physical
presence of an adult supervisor. In a similar case, other research characterizes direct adult
supervision as the “physical presence of at least one adult” in monitoring the activities of
the adolescents (Warr, 2005). Secondly, Wells and Rankin (1998) provide a
reconceptualized version of direct parental control, constructing from Nye’s (1958)
concept of direct control, that include the following three elements: normative regulation,
which is a process in which parents set the standards for the children’s behaviors;
monitoring which refers to supervision by the parents to conﬁrm whether children abide
by the guidelines set; and discipline which involves applying the idea of punishment and
rewards depending on the compliance to the rules. This conceptualization includes not
only the aspects of supervision but also dimensions of disciplining and regulating the

children’s behaviors. Lastly, Wright and Cullen (2001), in their recent study that consists

l3

of measuring for direct parental control, include three dimensions, which are parental
supervision, parental expectations of the child, and parental household rules.

Monitoring also consists of the parent’s awareness and knowledge about the
child’s well-being such as the child’s school attendance, use of free time activities at
home and at school, or feelings towards drug use (Snyder & Patterson, 1987).
Nonetheless, parental monitoring is generally deﬁned as the extent of knowledge to
which parents attempt to give attention to, to track, or control their children’s activities,
their whereabouts, peers, and adaptations (Dishion & McMahon, 1998; Jacobson &

Crockett, 2000; Kerr & Stattin, 2000; Snyder and Patterson, 1987).

III. Parental Monitoring and Delinquency

In the criminological literature, various elements of family inﬂuence, including
family structure, relationships and process have been examined for their impact on
hindering delinquent youth behaviors'(Gottfredson & Hirschi, 1990; Gove & Crutchﬁeld,
1982; Larzelere & Patterson, 1990; Wells & Rankin, 1988). Different dimensions of
family inﬂuence such as attachment, supervision or control were identiﬁed as signiﬁcant
determinants of delinquency among other important correlates (Barnes & Farrell, 1992;
Fletcher, Steinberg, & Williams-Wheeler, 2004; Krohn, Stern, Thomberry & J ang, 1992).
In particular, parenting variables such as attachment, involvement, and monitoring have
been found to signiﬁcantly inﬂuence delinquency (Glueck & Glueck, 1962; Larzelere &
Patterson, 1990; Wells & Rankin, 1988).

In examining the relationships between family variables and delinquency, studies

report that discipline, supervision, protection and affection tends to be absent in

14

households where delinquents reside (Gottfredson & Hirschi, 1990; Warr, 2005). For
example, in a longitudinal study conducted by Oxford et a1. (2000), adolescents from
“prosocial” family environments, structured with rules, monitoring and attachment had a
lower probability of initiating substance use, demonstrating a signiﬁcant negative effect
even while controlling for the deviant peer inﬂuence. In the well-known study by the
Gluecks (1968), they contended that three factors were signiﬁcant within the household:
parental supervision, child-parent attachment, and parental discipline. Their ﬁndings
reveal that when the parents failed to provide supervision and proper disciplining, the
children were more likely to engage in delinquent behaviors. On the whole, research
indicates that the parental supervision and the parents' involvement in the children’s daily
lives may play a signiﬁcant role in altering the outcome in their delinquent behaviors
(Larzelere & Patterson, 1990; Loeber & Stouthamer—Loeber, 1986; McCord, 1991; Nye,
1958; Taylor, 1985). Additionally, evidence reveals that higher levels of parental support
and monitoring may reduce the levels as well as the trajectories of adolescent problem
behaviors; furthermore, the effects may buffer the inﬂuence of peer deviance (Barnes,
Hoffman, Welte, Farrell & Dintcheff, 2006). Moreover, Sampson and Laub (1994), in
their reanalysis of the study by the Gluecks, found that there is a strong relationship
between parental monitoring and delinquency.

A majority of the studies support the claim that parental supervision or monitoring
is negatively associated with delinquency (Steinberg, Fletcher, & Darling, 1994; Ary et
al., 1999; Jacobson & Crockett, 2000; Pettit et al., 2001). Among the studies covering
parental supervision, it is generally perceived that if parents fail to monitor children’s

disruptive behaviors, they will have the liberty or opportunity to associate with

15

delinquent peers who in turn tempt them into criminal conducts (Patterson, Reid, &
Dishion, 1992). In a study that examined the relationship between parenting and
children’s antisocial behavior, low levels of monitoring and supervision turned out to be
one of the signiﬁcant explanations of delinquency, particularly among young male
adolescents in urban setting. (W asserman et al., 1996).

Studies view that parental supervision is a proper strategy for parents seeking to
restrain youth from engaging in delinquent conducts, in particular substance use
(Fletcher, Darling, & Steinberg, 1995, p. 270). Snyder and Patterson (1987) also conﬁrm
the positive effect of supervision, that it “fosters appropriate parental reaction to
antisocial and delinquent behaviors.” Studies also suggest that effective parenting
techniques can protect against child’s peer deviance; children whose parents exercise
control on their activities and their peer relationships commit less serious and less
frequent delinquent acts than others (Hirschi, 1969; Rankin & Wells, 1990; Dishion et al.,
1991; Dishion & Mc Mahon, 1998; J ang & Smith, 1997). Hence, monitoring seems to be
an important parenting aspect associated with positive adjustment for children. Overall,
these studies have made the clear statement that parental supervision efforts “work
protectively by keeping youths away from deviant peer contexts and out of trouble” (Kerr
& Stattin, 2003).

Concerning the effect of supervision and peers, it is perceived that if parents fail
to monitor adolescents with disruptive behaviors, they will be exposed to the inﬂuence of
delinquent peers, leading to their own engagement in antisocial or criminal behaviors
(Patterson, Reid, & Dishion, 1992). However, if parents are constant supervisors of their

children's daily activities and behaviors, it is more likely that the monitoring will help to

16

prevent the children from being inﬂuenced by deviant peers. In short, parental monitoring
is important in discouraging delinquent behaviors and averting the potential association

with peers who engage in deviant acts.

l7

CHAPTER 3

DATA AND METHODOLOGY
1. Research Hypotheses
H_1_A: Youth with strong parental monitoring are less likely to engage in delinquent
behaviors than those without it.
H_1_B: Youth with strong parental monitoring are less likely to be involved in substance
use than those without it.
H_2_A: Youth with supervision by mother are less likely to engage in delinquent
behaviors than those without it.
H_2_B: Youth with supervision by father are less likely to be involved in substance use
than those without it.
H_3_A: Youth with positive peer inﬂuence are less likely to engage in delinquent
behaviors than those without it.
H_3_B: Youth with positive peer inﬂuence are less likely to be involved in substance use
than those without it.
H_4_A: Youth with negative peer inﬂuence are more likely to engage in delinquent
behaviors than those without it.
H_4_B: Youth with negative peer inﬂuence are more likely to be involved in substance
use than those without it.
H_5_A: Youth who are living with both parents are less likely to engage in delinquent
behaviors than those do not.
H_5_B: Youth who are living with both parents are less likely to be involved in

substance use than those do not.

18

11. Data Collection

Dag

The data used for this study is called the National Longitudinal Survey of Youth
1997 (N LSY97), designed by the United States Department of Labor, consisting of seven
rounds of data collected from 1997 to 2003. The NLSY97 survey is sponsored and
directed by the US. Bureau of Labor Statistics and conducted by the National Opinion
Research Center at the University of Chicago, with assistance from the Center for Human
Resource Research at The Ohio State University (Bureau of Labor Statistics, 1997). The
dataset used for this research was downloaded from the Inter-University Consortium for
Political and Social Research (ICPSR) at the University of Michigan.

The survey was designed to create a nationally representative sample of
approximately 9,000 male and female adolescents with age range of 12 to 16 when ﬁrst
interviewed in 1997. The major purpose of the NLSY97 data was to collect extensive
data on youth labor market behavior and educational experiences. However, the actual
content of the NLSY97 is greatly broader due to the interests and concerns of Bureau of
Labor Statistics and other governmental agencies, including United States Department of
Justice (U SDOJ ), Ofﬁce of Juvenile Justice and Delinquency Prevention (01] DP) and the
National Institute of Child Health and Human Development (N ICHD). With the
sponsorship of these agencies, various set of questions such as delinquency, criminal
activity, health, family dynamics and individual behaviors are included in the self-
administered portion of the youth questionnaire.

Two core samples were drawn in this national survey which comprised of a cross-

sectional sample representing the US. youth population born in the years 1980 through

19

1984 and supplemental samples of the African-American and Hispanic population born
in those years. During the initial survey round in 1997, the cross-sectional sample
contained 6,748 respondents and is designed to be representative of the youth living in
the United States. The supplemental sample of 2,236 respondents was designed to be
collected by a procedure of oversampling the African-American and Hispanic youths
living in the United States during the initial survey round and born during the same
period as that of the cross-sectional sample. Overall, the NLSY97 sample allows
researchers to analyze the experiences and behaviors of the population as a whole and

alternate group of Hispanics and Blacks.

Sample

The speciﬁc dataset used for this study was the dataset for the year 1997 collected
in round 1. The main reason for selecting the dataset of round 1 was due to the dropping
retention rates after round 1 in 1997. After the initial survey round, some NLSY97
respondents did not respond to one or more following interviews. The list of reasons for
non-interview included deceased, not locatable, technical problem, illness of respondent,
unavailability, refusal of interview.

The round 1 survey in 1997 was conducted by using three separate questionnaires
which were the ‘Screener, Household Roster, and Non-Resident Roster questionnaire,’
the ‘Parent questionnaire,’ and the ‘Youth questionnaire.’ The ‘Screener, Household
Roster, and Non-Resident Roster questionnaire’ was distributed to a household resident
above the age of 18. The ‘Parent questionnaire’ was given to a parent or parental ﬁgure

of the youth who resides in the household. The ‘Youth questionnaire’ was handed out to

20

the youth respondent residing in the household. As a data collection tool, a computer-
assisted interview called CAPI was used to conduct the survey. The interviews were
largely conducted as in-person, and telephone interviews were conducted occasionally.

In terms of data collection, the ‘Youth questionnaire,’ as a main questionnaire,
was administered to every youth respondent for each round of the survey. The one-hour
lasting interview gathered detailed information on topics of school activities, employment
history, family background as well as ﬁnancial characteristics, social behavior, and health
issues.

The purpose of the 'Screener, household roster, and nonresident roster
questionnaire' was to collect the basic information (e. g. name, age, birth date) of each
resident in the household. When an eligible youth was selected, a two-part roster
questionnaire was given to the youth respondents. First, the household roster part
gathered information on every member residing in the youth's household. Second, the
nonresident roster part collected data on the individuals related to the youth's immediate
family who resided outside of the household; examples would be parents, siblings, or
spouse. These screening interviews were also conducted only in round 1, despite updates
on some parts of the survey in the later rounds.

The parent interviews were conducted only in round 1, distributing the 'Parent
questionnaire' to one parent of every youth for participation. The purpose of this 'Parent
questionnaire' was to gather and provide wide-ranging background information. The
parent interview contains responses to questions regarding marital history, employment
as well as information related to the family and youth. During the screening process, the

selection limited the criteria to parents who actually reside with the youth in the same

21

household. Moreover, interviewers created a list for choosing the preferred parent
respondent. For instance, the biological mother was selected ahead of the biological
father. If the youth respondent did not reside with a parental ﬁgure or resided with a
parent or guardian who was not listed, parent interview was not conducted. It should be
noted that not all youth respondents with an eligible parent will include a fully completed
parent questionnaire due to non-response. Out of a total of 8984 youth respondents, 88.4
percent (7942) of them have information for parent interview.

In additional to the questionnaires and instruments used for the survey, special
data collections in the NLSY97 study included high school surveys, the computer-
adaptive form of the Armed Services Vocational Aptitude Battery (CAT-ASVAB)
administered only in round 1, and the gathering and coding of academic transcripts. In
particular, school survey questionnaires were designed with a focus on topic of school-to-
work transition for round 1 of the study. In 1996, these questionnaires, as mail surveys,
were delivered to all public and private schools located in one of the primary sampling
units. It should be noted that school surveys in round 1 was considered a census of
schools in the primary sampling units. The school surveys requests information regarding
the characteristics of student body, programs for students, and delinquency-related

reports such as alcohol or drug use, weapons possession, and gang activity.

Sampling Procedure

In terms of sampling procedure, the NLSY97 data comprises two independent
probability samples, which are the cross-sectional sample and the oversarnple of African-

American and/or Hispanic respondents. In the process of selection, a list of housing units

22

for the cross-sectional sample in the ﬁrst phase and the oversample was drawn from two
independently selected, stratiﬁed multistage area probability samples. Its purpose was to
ensure an accurate representation of different portions of the population deﬁned by race,
income, and other demographic variables. In selecting a sample of 8,984 respondents,
interviewers screened 75,291 households in 147 primary sampling units that did not
overlap. For this study, a primary sampling unit was a either metropolitan area or single
or group of county for non-metropolitan areas.

In terms of data collection, a broad range of information in the NLSY97 had been
gathered through several different survey instruments. For the purpose of properly
identifying and screening the eligible youths for the survey, the instruments of screener,
household roster and nonresident roster questionnaire was used to collect demographic
information on household residents and key nonresident relatives during the initial survey
round. With use of several questionnaires and the computerized version of the Armed
Services Vocational Aptitude Battery (CAT-ASVAB), the survey was conducted as a
face-to-face interview in the households. The youth questionnaire, which was
administered in each round for an hour, contains topics pertaining to the youth’s
employment history, school experience, family background, health, and attitudes and
behaviors. The parent questionnaire in round one interviewed one of the youth's parents
or the caregivers about the family’s history and structure for half-hour as well as about
key features of the parent's own life for an hour. School surveys and transcripts collected
information in several rounds about schools in the sample areas and the academic records
of the youths. In addition, the survey used a computer-assisted personal interviewing

(CAPI) system, which automatically and efﬁciently guides respondents to follow certain

23

question paths depending on age of respondents and responses to preceding questions.
The main purpose of this system was to enhance the face-to-face data collection method
and to lower the probability of inconsistency of the data, both during interviews and in
the full collection period, compared to the outcomes of traditional interviews using paper
and pencil. To assure completeness of the questionnaires, the respondents were provided
cash incentives; the youth and their corresponding parents received ten dollars for their

participation in each round.

III. Variables

Dependent Variables

Two dependent variables will be analyzed in this study. First is the delinquency
score index, which examines whether or not youth is involved in delinquent incidents.
Second is substance use index, which looks into whether or not the youth has a history of
substance use. According to the youth report, both indexes were created, using
questionnaires distributed to youth aged from 12 to 16 years.

First, delinquency was measured by an index composed of a modiﬁed form of
items developed by Del Elliot (1983) to measure delinquency and criminality in the
National Youth Survey (NY S). Ten question items contain topics of running away, gang
membership, possession of gun or drugs, theft, assault and arrests, as the response were
dichotomous (0=No, 1=Yes). The delinquency score index was created by summation of
the responses from the number of delinquent/criminal acts the youths identiﬁed having
ever done through the youth report, for a possible total score of 10. According to the

youth report, higher scores indicate more incidents of delinquency.

24

It should be noted that scores on the delinquency index were attained if
respondents answered at least eight of the ten question items. Those youth who answered
at least eight of the ten questions were assigned a weighted score based on the 10-point
scale, whereas youth who answered less than eight question items were coded as missing.
For this scale, there was a trivial percentage of missing data. Out of 9007 cases, only 15
were considered missing; its mean was 1.33 and standard deviation was 1.84.

Secondly, substance use was measured by an index comprising items modiﬁed
from National Survey of Family and Households (N SFH-2) in which the youth
questionnaires were given to those aged from 10 to 17. The three question items include
topics of use of cigarette, alcohol and drugs, as the responses were dichotomous (0=No,
1=Yes). The substance use index score was created by adding the number of substances
the youth reported having ever attempted for a possible total score of 3. According to the
youth report, higher scores indicate more incidents of substance use.

It should be noted that scores on the substance use index were obtained from only
those respondents who answered all three question items, whereas respondents who
answered fewer than three items were coded as missing. There was only little missing
data for this scale; out of 8999 cases, only 23 were considered missing; its mean was 1.02
and standard deviation was 1.33.

For the two dependent variables, they were receded into dichotomous variables in
order to run a logistic regression. Originally, the range of delinquency was from O to 10.
However, due to its highly skewed distribution, the delinquency variable was coded into

a binary category. For substance use, its previous range was from O to 3, but was coded

25

into a binary variable. Table 1 illustrates how the variables were coded and percentage of

each response.

Table 1: Descriptive Statistics and Variable CodjpglNLSY 1997)

 

 

 

Variable Coding Percentage
I. Dependent Variables
a. Delinquency O=No delinquent incident 46.6
1=At least one incident of
delinquency 53.2
999=Missing 0.2
b. Substance Use 0=No substance use 46.4
1=At least one use of substance
53.3
999=Missing 0.3
11. Independent Variables
a. Monitoring by Mother 0=No Monitoring 41.2
1=Monitoring 58.8
b. Monitoring by Father 0=No Monitoring 53.0
1=Monitoring 47.0
c. Monitoring by both Mother & Father
0= No Monitoring 54.1
1=Monitoring 45.9
d. Living with Bio. Parent(s) 00=No Bio. Parents 6.5
01=Only Bio. Father 5.6
10=Only Bio. Mother 36.7
1 l=Both Bio. Parents 51.2
999=Missing 0.0
III. Demographic Variables
a. Age (Youth) 12=12 years old 19.7
l3=13 years old 20.1
14=14 years old 20.5
15:15 years old 20.9
l6=16 years old 18.8
b. Gender (Youth) 0=Male 51.2
1=Female 48.8
c. Race (Youth) 0=Non-White 40.9
1=White 58.2

(Table 1 continued)

26

999=Missing 0.9

d. Parent Education Range = 0-7
Mean = 2.54
Standard Deviation= 1.37

e. Household Income Range = 0-246,474
Mean = 44084.68
Median = 37750.00
Standard Deviation= 36267.39

f. Positive Peers Range = 4-20
Mean = 12.1 1
Standard Deviation= 2.89

g. Delinquent(Negative) Peers Range = 5-25
Mean = 1 1.09

Standard Deviation= 4.77
(Table 1 continued)

Independent Variables

Study by Martens (1997) found that parental knowledge of children’s
whereabouts has a signiﬁcant effect on delinquency and antisocial behavior for youth
both boys and girls. The variable ‘parental monitoring’ in this study is composed of four
questions regarding the parents’ (both father and mother) awareness and knowledge of
children whereabouts. In a study that examined the relationship between parenting and
children’s antisocial behavior, low levels of monitoring and supervision turned out to be
one of the signiﬁcant explanations of delinquency, particularly among young male
adolescents in urban setting. (Wasserman et al., 1996).

In support of social control theory (Hirschi, 1969), parental monitoring is
analyzed as a major variable of this study. Parental monitoring, either by residential or
non-residential parents, is measured by source of items that comprise standard questions

used widely by the recognized researchers of the family contexts (Hetherington, Cox, &

27

Cox, 1982; Maccoby & Mnookin, 1992). The four question items include “How much
does he/she know about your close friends, that is, who they are?”, “How much does
he/she know about your close ﬁiends' parents, that is, who they are?”, “How much does
he/she know about who you are with when you are not at home?”, and “How much does
she know about who your teachers are and what you are doing in school?” These
responses were measured on a 5-point scale, from 0 to 4. The report consisted of youth
aged from 12 to 14 years.

In terms of scale created for measuring the parental monitoring, a scale was
created for each of the four possible parental ﬁgures. Those ﬁgures are residential
mother, residential father, non-residential biological mother, and non-residential
biological father. The responses to the four items were added up so that the scores would
range from 0 to 16 points. According to the youth reports, it is indicated that higher
scores account for greater parental monitoring. In effect, the parental monitoring scale
illustrates the awareness of the youth's activities by parental ﬁgures

It is noted that respondents who answered three of the four items were given a
weighted score based on the 16-point scale, but respondents who answered less than three
items were coded as missing. However, detailed codebook shows that there was a very
small amount of missing data. In examining the reliability of the scale, results of
Cronbach’s alpha for these scales were measured and considered satisfactory (Bureau of
Labor Statistics, 1997). Research suggests that Cronbach’s alpha is the preferential
measure of reliability or internal consistency (Carmine & Zeller, 1985).

For the monitoring variables, there are three separate variables. First would be the ’

monitoring by mother. Originally, the monitoring by mother was divided into residential

28

and non-residential mother. However, they were added and recoded into a combined
variable to include for the all types of mother. Second would be the monitoring by father.
It was also recoded as a newly created including residential and non-residential.
Furthermore, among the independent variables, another important variable would be the
family structure variable, speciﬁcally focusing on whether or not the youth is living with
a biological parent(s). This variable was also added up and created to comprise four
categories to represent the residence of mother and/or father in the youth’s household. It
is a important consideration because it provides not only the percentage of youth who are
living with their biological parent(s) but also a exploration point of how a family
structure could be associated with a family function variable such parental monitoring.
Regarding the farme structure variable (‘Living with biological parent(s)’), in the
round 1 dataset (for 1997) there were originally two dichotomous variables that indicated
whether the youth was ‘living with a biological mother’ or ‘living with a biological
father’ measured as either ‘yes’ or ‘no.’ However, the two variables were added up and
recoded into a four category variable that conveyed the four types of family structure.
Those categories are ‘living with both biological parents,’ ‘living with only biological
mother,’ ‘living with only biological father,’ ‘living with no biological parent.’ The ﬁrst
category of ‘living with both biological parents’ was selected as a reference category
because of its considerable ﬁ'equency. This family structure variable was considered as
the “main” variables along with the three monitoring variables in order to examine the
association between the family structure and family function aspects. The two different
sorts of family variables may provide further insights or explanations when analyzing the

correlates of delinquency and substance use. For example, the effect of monitoring by a

29

parent within an “intact” family environment would be different from that within a “non-

intact” family household.

Demographic Variables

Main demographic characteristics of the sample will be controlled for the purpose
of eliminating spuriousness during the data analysis. Those variables are age, gender,
race, parent education, household income, and peer inﬂuence. Age was partly limited that
its range was from 12 to 16. Gender and race were also included as essential variables.
Parent education and household income were selected to control for the socioeconomic
status of the parent because it may likely have an impact on the monitoring skill of a
parent. In terms of coding, level of education for father and mother were added together
to create a newly combined variable representing level of parent education as a whole.
Additionally, two types of peer inﬂuence were included as peer inﬂuence, which is
considered one of the highly recognized correlates of delinquency and substance use in
the criminological literature. Speciﬁcally, those are the positive and delinquent (negative)
peer inﬂuence. For the positive peer inﬂuence, 4 peer characteristics were added up and
recoded into a new variable; its range is 4 to 20. For the negative peer inﬂuence, 5 peer
characteristics were added up and recoded into a new variable; its range is 5 to 25.

In regards to the age-crime relationship, age is certainly correlated with crime.
The age-crime distribution describing the increasing criminal involvement to a peak in
the years of adolescence and aﬁerwards decreasing with age is well-known in the
criminological literature (Hirschi & Gottfredson, 1983). This age-crime relationship was

claimed to be invariant across historical periods, geographic locations, and crime types

30

(Hirschi & Gottfredson, 1983). Critics pointed out the overstatement of the age-crime
relationship and the invariance of age effect, but the impact of age in crime is still
considered one of the most important correlates associated with crime (Steffensmeier et
al., 1989). Although this cross-sectional study is supported by Hirschi and Gottfredson
(1983), longitudinal research would allow one to examine the causation by establishing
the order of factors (V old, Bernard, & Snipes, 2002).

Regarding the relationship between gender and delinquent behavior, research
reveal that girls tend to be more emotionally attached to and controlled by their parents
compared to boys; this may explain that girls have lower rates of delinquency (Giordano
& Cernkovich, 1997; Svensson, 2004). Also, in the criminological literature, it is
perceived that males are more aggressive and more often engage in delinquent or
criminal activities, compared to females. Peer inﬂuence has been frequently mentioned in
the literature as one of the signiﬁcant predictors of youth delinquency (W arr, 1993;

Krohn, 1994).

31

CHAPTER 4
DATA ANALYSES AND FINDINGS
1. Methods of Analysis 1

Three analytic strategies were used to explore the impact of the independent
variables on the pair of dependent variables. Firstly, univariate statistics will be presented
for all independent and dependent variables, including demographic variables. The
descriptive statistics illustrates the percentage and variable coding for each variable in the
whole sample. It also indicates that the missing cases for all variables were recoded as
‘999’ in the SPSS program as ‘user-deﬁned.’

Secondly, a bivariate analysis will be conducted to examine the relationship
between the independent and dependent variables through the display of contingency
tables. Chi-square tests of independence are generally used to investigate the relationship
between two discrete variables that comprise binary or multinomial distributions
(Bachman & Paternoster, 2004). Chi-square analysis will allow testing for the
independence of a selected set of independent variables and the two dependent variables.
Moreover, T-tests will be used to test for continuous variables such as income.

Lastly, multivariate statistics will presented by using the logistic regression
technique to analyze the dataset and to examine the relationship between parental
monitoring and delinquency. Logistic regression will be used because the two dependent
variables are binary. The study will analyze three models through the logistic regression.
The ﬁrst basic model will examine only the monitoring variables and family structure
variable such as living with one’s biological parent(s). The second model will include the

two types of peer inﬂuence (positive and negative peer inﬂuence) resting on the ﬁrst

32

model. The third model will use the full list of variables, including all demographic
variables. All three model are conducted for the two dependent variables (delinquency
and substance use). Prior to running the logistic regression, the study will also check for
the possible multicollinearity issues with the use of collinearity diagnostics available

through the SPSS program.

11. Findings

Bivariate Statistics

Bivariate analysis was performed to examine the relationship between two
selected variables. The relationship was tested whether it was signiﬁcant or not. For the
chi-square analysis, the measure of association was either Phi or Cramer’s V values to
illustrate the associations of all the independent to the dependent variables (Bachman &
Paternoster, 2004). Tables 2A and 2B illustrates the summary statistics of bivariate
analysis of the independent variables by the two dependent variables. Most of the
bivariate relationships are fairly consistent with the hypothesized statements supported by
the criminological theories, except the examples between monitoring by father and

delinquency / substance use.

Variables associated with Delinquency

Table 2A shows considerable support for the social bond theory. Youth with no
monitoring were more likely to engage in delinquent acts, compared to youth who were
monitored. However, this was only true for youth who were monitored by only mother

and by both parents. The monitoring by only father did not yield to be signiﬁcant.

33

Overall, the family structure variable, living with biological parent(s), came out to be
signiﬁcant. Presumably, youth who resided with both parents were less likely to be
delinquent, compared to 3 other family structures. But, it is interesting to view that youth
residing with only biological mother were most likely to be delinquent among all 4 cases.
The well-known age-crime curve supports the ﬁndings that youth who were 16 years old
were most likely to be delinquent as youth crime peaks when reaching the age of 16. In
terms of gender, male adolescents were more likely to engage in delinquent activities
than female counterparts. Race did not yield a signiﬁcant outcome. For the level of parent
education, it was generally true that the higher the parents’ education, the less likely the
youth’s delinquency. With regards to the two types of peer inﬂuence, the more positive
peer inﬂuence youth receives, the less likely the youth will become delinquent.
Moreover, the more negative peer inﬂuence youth receives, the more likely the youth will
lured into delinquent acts. All of these signiﬁcant relationships are consistent with the

criminological literature.

Table 2. Percent of Delinquency by Parental Monitoring, Parental Structure, and
Demographic Characteristics

 

Variables Number Percent of
of Cases Delinquency
All Cases 8984 53.2
Parental Monitoring
Monitoring by Only Mother No Monitoring 3688 58.9
Monitoring 5281 49.4
Chi square 77.8l3***
(Cramer’s V) (.093)
Monitoring by Only Father No Monitoring 4753 58.3
Monitoring 4216 47.7
Chi square 1.022
(Cramer’s V) (.107)
Monitoring by Mother & Father No Monitoring 4846 58.3
Monitoring 4123 47.5

 

34

 

Chi square 1.046***

(Cramer’s V) (.108)
Living with Biological Parent(s) in Living w/ No 578 60.6
Household Biological Parent

Living w/ Only 499 64.7

Biological Mother

Living w/ Only 3209 57.6

Biological Father

Living w/ Both 4602 48.1

Biological Parents

Chi square l.138***

(Cramer’s V) (.113)
Demographic Variables
Age (of Youth) 12 years old 1766 44.7

13 years old 1804 47.8

14 years old 1836 55.9

15 years old 1872 57.8

16 years old 1691 60.4

Chi square 1.292***

(Cramer’s V) (.120)
Gender (of Youth) Male 4588 62.6

Female 43 81 43 .6

Chi square 3.228***

(Cramer’s V) (.190)
Race Non-white 3663 53.8

White 5226 53.1

Chi square .411

(Cramer’s V) (.007)
Parent Education Level 0 189 45.0
(Mother & Father Combined)

0.5 199 47.2

1 1 123 5 1 .3

1.5 448 54.9

2 900 55.8

2.5 740 54.9

3 1340 52.0

3 .5 5 1 8 54.8

4 425 50.1

4.5 238 53.8

5 246 52.8

5.5 81 55.6

6 87 54.0

6.5 14 21.4

7 45 53.3

iTable 2 continued)

 

35

 

Chi square 21 .747***

(Cramer’s V) (.057)
Income Mean (SD) for 45263.6 (37315.7)
Delinquency
Mean (SD) for Non- 43089.8 (35154.6)
Delinquency
T-test 2.831"
Positive Peers 4 33 54.5
5 66 59.1
6 141 62.4
7 250 69.2
8 410 64.4
9 660 62.4
10 919 58.5
1 1 1110 56.2
12 1225 54.3
13 1 125 51.9
14 921 47.0
15 713 45.7
16 501 43.9
17 297 44.8
18 158 40.5
19 75 41.3
20 35 25.7
Chi square 1.715***
(Cramer’s V) (.141)
Negative (Delinquent) Peers 5 1156 31.2
6 738 43.2
7 687 46.3
8 576 45.0
9 584 53.1
10 530 54.2
11 561 56.7
12 547 54.3
13 567 61.2
14 505 59.8
15 453 66.4
16 380 65.0
17 325 72.9
18 273 74.4
19 ' 207 70.5
20 192 74.0
21 1 15 77.4
22 76 69.7
23 50 82.0
24 41 82.9

(T able 2 continued)

36

 

25 20 75.0
Chi square 5903*"

(Cramer’s V) (.262)

** p<.05 ***p<.01
(Table 2 continued)

Variables associated with Substance Use

Table 2B illustrates notably similar results to table 2A. In general, there is
considerable difference between youth with no monitoring and at least some monitoring
in predicting delinquency. Among the three monitoring variables, monitoring by only
mother and by both parents came out to be signiﬁcant. Regarding the family structure
variable, youth who lived with both biological parents were less likely to be delinquent,
compared to 3 other family structures. Among the four categories, the youth residing with
only biological mother were most likely to be delinquent among all 4 cases. The
percentage of delinquency increases signiﬁcantly as age ranges from 12 to 16.
Contrasting to delinquency, the percentages of substance use by male and female youth
were relatively close. Contrary to table 2A, race came out to be signiﬁcant in predicting
substance use. White youth were slightly more likely to use substance, compared to non-
white counterparts. Even though the level of parent education became higher, it was not
clear whether substance use by youth declined. In terms of peer inﬂuence, there was an
inverse relationship between positive peer inﬂuence and substance use. For negative peer
inﬂuence, as negative peer inﬂuence increases, the more likely the youth will engage in

substance use.

Table 3. Percent of Substance Use by Parental Monitoring, Parental Structure, and
Demographic Characteristics

 

Variables Number Percent of

37

 

of Cases

Substance Use

All Cases 8984 53.3
Parental Monitoring
Monitoring by Only Mother No Monitoring 3685 68.3
Monitoring 5276 43.1
Chi square 5559*"
(Cramer’s V) (.249)
Monitoring by Only Father No Monitoring 4748 63.9
Monitoring 4213 41 .7
Chi square 4.438
(Cramer’s V) (.223)
Monitoring by Mother & Father No Monitoring 4841 63.6
Monitoring 4120 41.6
Chi square 4349*“
(Cramer’s V) (.220)
Living with Biological Parent(3) Living w/ No 577 56.7
in Household Biological Parent
Living w/ Only 499 64.1
Biological Mother
Living w/ Only 3288 58.5
Biological Father
Living w/ Both 4597 48.3
Biological Parents
Chi square 1.075***
(Cramer’s V) (.1 10)
Demographic Variables
Age (of Youth) 12 years old 1765 31.4
13 years old 1802 41.3
14 years old 1834 56.2
15 years old 1872 65.1
16 years old 1688 73.7
Chi square 8384*“
(Cramer’s V) (.306)
Gender (of Youth) Male 4585 55.5
Female 4376 51.4
Chi square 15.072***
(Cramer’s V) (.041)
Race Non-white 3660 49.3
White . 5221 56.7
Chi square 46.946***
(Cramer’s V) (.073)
(Table 3 continued)

 

38

 

Parent Education Level
(Mother & Father Combined)

Income

Positive Peers

Negative (Delinquent) Peers

(Table 3 continued)

0

0.5
l
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7

Chi square

(Cramer’s V)

Mean (SD) for
Delinquency
Mean (SD) for
Non-Delinquency

T-test

\OOO\IO\UIL

10
ll
12
13
14
15
16
17
l8
19
20

Chi square

(Cramer’s V)

5
6
7
8

39

33
66
141
250
410
660
919
1108
1225
1124
919
713
501
296
158
75
35

1155
733
687
576

188 45.2
199 43.7
1123 51.9
447 55.3
900 57.7
740 55.9
1338 52.8
517 55.9
425 49.4
238 54.2
246 47.2
81 54.3
87 55.2
13 53.8
45 48.9
31 .247***
(.069)
44113.6 (36039.9)

44051.1 (36557.2)

2.831M

63.6
57.6
68.8
65.6
61.2
65.0
59.8
57.4
56.3
53.7
47.2
46.4
38.3
41.9
30.4
37.3
20.0
2.492***
(.170)

23.7
34.2
41.6
46.0

 

9 584 52.7

 

10 530 55.5
11 561 59.9
12 548 60.4
13 567 63.5
14 505 69.5
15 451 70.3
16 380 74.7
17 325 74.2
18 273 80.2
19 207 77.3
20 192 77.6
21 115 81.7
22 74 79.7
23 50 82.0
24 41 70.7
25 20 65.0
Chi square 1.094***
(Cramer’s V) (.357)
an: p<.05 ***p<.01
(Table 3 continued)

Summm of the Bivariate Relationships

The ﬁndings of bivariate relationships illustrate that there is a signiﬁcant
relationship between majority of the independent variables and delinquency / substance
use. For delinquency and substance use, it is shown that there is a signiﬁcant relationship
between the 2 monitoring variables and delinquency / substance use. However, the
monitoring by only father variable did not yield to be signiﬁcant for both dependent
variables. It would be logical to ﬁgure out the amount of time youth ranged from 12 and
16 years spent with the mother and father. Mothers may tend to spend more time with
their adolescents, compared to their fathers. The issue of parent-child communication
could also be a factor for the outcome of father’s monitoring. After entering the

interaction of monitoring by mother and father, the result came out to be signiﬁcant. One

40

can assume that the monitoring by mother had an ample effect on the outcome’s
signiﬁcance.

On a bivariate level of analysis, 9 and 10 out of the 11 variables examined were
signiﬁcantly associated with delinquency and substance use, respectively. Notably, tables
2A and 2B show that percentage of delinquency and substance use by youth vary with
demographic characteristics of the youth and their parents and the peer inﬂuences. As
predicted in core criminological literature, the bivariate analysis indicates that many of
the demographic variables such as age, gender, parent education, and income were
related to delinquency and substance use on a national context. While many of the
bivariate relationships are coherent with the crime theories, a multivariate analysis would
be required to identify which variables have a signiﬁcant impact on the two dependent

variables.

Multivariate Statistics

The bivariate results in tables 2A and 2B illustrated signiﬁcant relationships
between the monitoring / family structure / demographic and the two dependent
variables. Yet, these ﬁndings reveal signiﬁcant associations between no more than two
variables. In order to further explore more complex interrelationships between variables
of interest, this study exploited the multivariate logistic regression. The logistic
regression allows researchers to quantify the effect of all selected independent variables
on the dependent variable of interest (Bachman & Paternoster, 2004). The quantitative
outcomes for delinquency and substance use are illustrated in tables 4 and 5, respectively.

For the purpose of logistic regression, the two dependent variables (delinquency and

41

substance use) were recoded into a set of dichotomous responses. In short, the outcomes

were recoded as ‘0’ for ‘no’ and ‘1’ for ‘yes.’

Multicollinearitv Diagnostics

In order to check for multicollinearity between the variables, the OLS regression
was performed along with the collinearity diagnostics to assess whether collinearity is
problematic. Checking for tolerance and variance inﬂation factors (VIF) is a common
way of detecting multicollinearity in statistical analysis. Also, another form of
collinearity diagnostics — correlation matrix — was obtained for all independent variables.

Initially, the correlation matrix was examined to check for multicollinearity (See
Appendix A). The correlation matrix allows researchers to view how variables are related
to each other, particularly between two variables. A high correlation may indicate
multicollinearity issue, but it would be necessary to use a more advanced diagnostics
method.

Secondly, the tolerance and variance inﬂation factor (V IF) were taken as
measures of collinearity. In general, low values of tolerance and high measures of
variance inﬂation factor (VIF) display signs of collinearity that help determine whether to
include a variable or not in the model.

Based on the obtained collinearity diagnostics in Table 3 below, the variance
inﬂation factor (VIF) measures are close to zero and not higher than value of 4.0.
However, few variables have tolerance measures that may indicate potential collinearity

problem. Table 3 demonstrates that the tolerance measures for all 11 variables came out

42

to be smaller than 0.10. Moreover, the variance inﬂation factor (V IF) measures provide

some assurance with multicollinearity.

Table 4. Collinearity Diagnostics

 

 

Independent Variables B Tolerance VIF

Constant .301

Monitoring by Mother -.106 0.259 3.865
Monitoring by Father -.188 0.869 1.151
Monitoring by Both Mother & Father .068 0.716 1.396
Living with Biological Parent(3) .806*** 0.836 1.196
Positive Peers -.048*** 0.903 1.108
Negative (Delinquent) Peers .127*** 0.720 1.389
Age (of Youth) .021 0.252 3.965
Male (compare to Female) -.873*** 0.982 1.018
White (compare to Non-White) .230*** 0.869 1.150
Parent Education Level

(Mother & Father Combined) '027 08” 1'144
Income -.367 0.828 1.207

 

a: II: *p<.0 1
Note: B value refers to unstandardized coefﬁcients.

Qgistic Regression Analysis (Delinquengp)

Table 5 presents three models predicting youth delinquency. Table 5 illustrates
the results of each variable entered into the three models. Model I examined solely the
effects of monitoring variables and family structure variables on delinquent behavior by
youth. Model 2 displays the effect of monitoring / family structure (living with biological
parent) variables on delinquency while controlling for peer inﬂuence variables. Finally,
Model 3 shows the effect of monitoring / family structure variables on youth delinquency
while controlling for peer inﬂuence and demographic variables.

For Model 1, the logistic regression coefﬁcients of all variables yielded to be
statistically signiﬁcant predictors of youth delinquency, except for the interaction term of

monitoring by both mother and father. The main independent variables, monitoring by

43

mother and by father, came out to be signiﬁcant in predicting delinquency. Even though
the interaction terms (monitoring by both parents) had been statistically signiﬁcant in the
bivariate analysis, it did not yield to be signiﬁcant in Table 5. Furthermore, the logistic
regression coefﬁcients reﬂecting the effects of monitoring by mother and monitoring by
father came out to be expected — a negative relationship. This is in accordance with the
hypothesized inverse relationship between parental monitoring and delinquency; the
higher the degree of monitoring, the lower the likelihood of delinquency.

Concerning the family structure variable, a signiﬁcant relationship between the
family structure and delinquency is established. A reference group (youth living with
both biological parents) is presented along with the three dummy variables that represent
the youth’s living arrangements (see Table 5). The results reveal that all three living
arrangements were signiﬁcantly related to the adolescent’s delinquency. As displayed in
Table 5, the odds ratio produced for living with no parent, living with only biological
mother, and living with only biological father were 1.548, 2.036, and 1.392, respectively.
These values of odds ratio are higher than that of the reference group. In turn, the
comparison of the odds ratio indicates that the youth residing in these three types of
household structure were more likely to become delinquent, compared to those living
with both biological parents. Among the three types of family structure, youth living with
only his or her biological mother had the highest odds ratio (2.03 6). Additionally, the
effect of the three types of family structure on youth delinquency was all positive.
Overall, model 1 predicts about 3 percent of the variation in delinquency, as indicated by

the Nagelkerke Pseudo R-square value.

44

For Model 2, three of the monitoring variables came out to be insigniﬁcant,
contrary to model 1. The three monitoring variables are not predictive of the odds of
delinquency. However, the peer inﬂuence variables were signiﬁcantly related to
delinquency. It can be suggested that the peer inﬂuence variables have a stronger effect
on delinquency, compared to monitoring by parents (W arr, 1993; Gifford-Smith, Dodge,
Dishion & McCord, 2005). The regression coefﬁcients for the two peer inﬂuence
variables indicate a negative relationship for the positive peer variable and a positive case
for the delinquent peer inﬂuence. This corresponds to the currently hypothesized
relationship between peer inﬂuence and delinquency; the more positive peers the youth
associates with, the less likely one becomes delinquent. The more negative peers the
youth forms relationships, the more likely one becomes delinquent.

Similar to Model 1, the three dummy variables of family structure came out to be
signiﬁcant. The results reveal that all three living arrangements were signiﬁcantly related
to the adolescent’s delinquency. These variables are predictive of the variation in
delinquency as in model 1 as well as in model 2. As displayed in Table 5, the odds ratio
produced for living with no parent, living with only biological mother, and living with
only biological father were 1.367, 1.842, and 1.190, respectively. Since the values of odds
ratio are higher than that of the reference group, the odds of delinquency among youth
living with no parent, only biological mother, and only biological father were higher,
compared to those living with both biological parents. Similar to Model 1, among the
three types of family structure, youth residing with only biological mother had the highest

odds ratio (1.842). Moreover, the effect of the three types of family structure on youth

45

delinquency was all positive. Overall, model 2 predicts about 10 percent of the variation

in delinquency, as indicated by the Nagelkerke Pseudo R-square value.

Table 5. Logistic Regression Models of Delinquency

 

 

 

 

Model 1 Model 2 Model 3
Independent Variables B Odds B Odds B Odds
Ratio Ratio Ratio

Family Variables
Monitoring by Mother -.237*** .789 .146 1.157 .265 1.303
Monitoring by Father -.580*** .560 -.l48 .863 -.238 .788
Monitoring by Both Mother .417 1.518 .025 1.025 .019 1.019
and Father
Living with Biological
Parent(s)

(Living w/ No Parent) .437*** 1.548 .312*** 1.367 .450*** 1.568

(Living 8” Bio. Mother) .711*** 2.036 .611*** 1.842 .811*** 2.251

(Living w/ Bio. Father) .331*** 1.392 .174*** 1.190 .274*** 1.315

( Ref. Category: Living

w/ Both Bio. Parents )
Peers Variables
Positive Peers -.055*** .946 -.O41*** .959
Negative (Delinquent) Peers .102*** 1.107 .127*** 1.135
Demographic Variables
Age (of Youth) .004 1.004
Male (compare to Female) -.968*** .380
White (compare to Non-White) .l67*** 1.182
Parent Education Level .053*** 1.054
(Mother & Father Combined)
Household Income .000 1.000
Constant .167*** -.429*** -.752
-2 Log Likelihood 12203.627 10975.426 7743.829
Model Chi Square 190.812*** 644.160*** 833.305***
Pseudo R Square (Nagelkerke) .028 .098 .168
** p<.05 ***p<.01

A “full” Model 3 is presented with all thirteen variables included. Similar to

model 2, three monitoring variables came out to be insigniﬁcant. Hence, the parental

46

monitoring variables are not predictive of the variation in delinquency when other
demographic variables are controlled. On the other hand, the two types of peer inﬂuence
appear to be signiﬁcant like those of Model 2. Peer inﬂuence variables continue to be
strong predictors of likelihood of delinquency among youth, providing statistical support
for the current hypotheses concerning the relationship between peers and delinquency.
Furthermore, the three dummy variables of family structure came out to be signiﬁcant.
These three family structure variables appear to be predictive of the variation in
delinquency in all three models. The odds ratio produced for living with no parent, living
with only biological mother, and living with only biological father were 1.568, 2.251, and
1.315, respectively. Because the values of odds ratio are higher than that of the reference
group, the odds of delinquency among youth living with no parent, only biological
mother, and only biological father were higher than those living with both biological
parents. A3 in all three models, among the three types of living arrangements, youth
residing with only biological mother had the highest odds ratio (2.251). Interestingly, the
family structure variables displayed dominance over the family function variable such as
parental monitoring.

Additionally, several demographic variables -— gender, race, and parent education
— seem to be signiﬁcantly related to youth delinquency. Concerning gender issue, Table 5
indicates that the odds of delinquency are 0.380 times higher for males than females. For
race, abnormal result shows that the odds of delinquency are 1.182 times higher for
whites, compared to non-whites. For parent education, an unusual ﬁnding illustrates that
the odds of delinquency are higher for youth whose parents’ education level was high

than those whose parents’ education level was low. In contrast, age and income of the

47

youth came out to be statistically insigniﬁcant. In summary, youth who have the
following characteristics are signiﬁcantly more likely to engage in delinquent activities --
- the youth who are males, who are living with no biological parent or single biological
parent, who are whites, whose peers are less delinquent and more positive, and whose
parents’ education level are high. Among these characteristics, family structure of ‘living
with only biological mother’ proved to be the strongest predictor (2.251) of youth
delinquency. Overall, model 3 predicts about 17 percent of the variation in delinquency,
as indicated by the Nagelkerke Pseudo R-square value. The Nagelkerke Pseudo R-square
value of explaining delinquency appears to increase as it moves from model 1 (3 percent)

to model 3, as the model becomes “fuller.”

Logistic Regression Analysis (Substance Use)

Table 6 presents three logistic regression models predicting substance use. Model
1 shows solely the effects of monitoring variables and family structure variables on
substance use. Model 2 displays the effect of monitoring / family structure variables
while controlling for peer inﬂuence variables on substances use by youth. Finally, Model
3 shows the effect of monitoring / family structure variables on substance use while
controlling for peer inﬂuence and other demographic variables.

For model 1 (Table 6) of substance use, the logistic regression coefﬁcients of all
variables yielded to be statistically signiﬁcant predictors of substance use. These results
did not lose the signiﬁcance when compared to the bivariate analysis. The signiﬁcant
ﬁndings provide statistical support for the current hypothesized relationship between the

parental monitoring and substance use. All three of the monitoring variables came out to

48

be signiﬁcant predictors of substance use. For example, the odds of using substance by
youth were 0.364 times higher for the youth who were monitored by the mother than
those who were not monitored. Also, the odds of substance use were 0.262 times higher for

the youth who were monitored by the father than those who were not. lastly, the odds of
substance use were 3.554 times higher for youth monitored by both parents compared to those

who were not monitored. Three of the family structure variables also came out to be
signiﬁcant. The ﬁndings show that all three living arrangements were signiﬁcantly
related to substance use. As displayed in Table 6, the odds ratio produced for living with
no parent, living with only biological mother, and living with only biological father were
1.259, 2.035, and 1.472, respectively. Since the values of odds ratio are higher than that
of the reference group, the odds of substance use among youth living with no parent, only
biological mother, and only biological father were higher, compared to those living with
both biological parents. Among the three types of family structure, youth residing with
only biological mother had the highest odds ratio (2.035). The effect of the three types of
family structure on youth delinquency formed positive relationships.

Reviewing the logistic regression coefﬁcients of each variable in Model 1, one
can observe that the regression coefﬁcients for monitoring by mother and monitoring by
father came out to be negative. These negative directions are in conformity with the
hypothesized inverse relationship between parental monitoring and delinquency; the
higher the degree of monitoring by mother or father, the less likely the juveniles
participate in substance use. However, the variable of monitoring by both mother and
father came out to be positively associated with substance use. Overall, Model 1 predicts
about 10 percent of the variation in substance use, as indicated by the Nagelkerke Pseudo

R-square value.

49

For Model 2 of substance use, three of the monitoring variables came out to be
signiﬁcant, unlike Model 1 of delinquency (Table 6). Thus, the three monitoring
variables are predictive of the odds of substance use, even when controlling for the peer
inﬂuence variables. The regression coefﬁcients for the monitoring by mother and
monitoring by father were negative, which conﬁrms that inverse relationship between
parental monitoring and substance use. Regarding the logistic regression coefﬁcients in
Model 2, the regression coefﬁcient values for monitoring by mother and monitoring by
father came out to be negative. Similar to Model 1 of substance use, these negative
associations are in line with the hypothesized inverse relationship between parental
monitoring and delinquency. However, the variable of monitoring by both mother and
father came out to be positively related to substance use. Although this is an odd,
interesting result, this occurred in the logistic regression model 1 of substance use as well
(Table 6).

Among the three family structure variables, only the arrangement of ‘living with
no parent’ yielded to be insigniﬁcant, while the two other single-parent structure
variables came out to be signiﬁcantly related to substance use by young adolescents. This
may be interpreted that even though monitoring by parent(s) occurs, the living
arrangements of a biological parent may be important in preventing activities of
substance use. As shown in Table 6, the odds ratio produced for living with only
biological mother and only biological father were 1.745 and 1.239, respectively. Since
the values of odds ratio are higher than that of the reference group, the odds of substance
use among youth living with only biological mother and only biological father were

higher, compared to those living with both biological parents. Similar to Model 1, among

50

the three types of family structure, youth residing with only biological mother had the
highest odds ratio (1.745). Moreover, the effect of the two types of family structure on
substance use by youth came out to be both positive.

Notably, the two types of peer inﬂuence variable were also signiﬁcantly related to
substance use. Similar to Model 2 of delinquency (Table 5), the regression coefﬁcients
for the two peer inﬂuence variables indicate a negative relationship for the positive peer
variable and a positive association for the delinquent peer inﬂuence, which supports the
hypothesized relationship between peer inﬂuence and substance use. Overall, model 2
predicts about 18 percent of the variation in delinquency, as indicated by the Nagelkerke

Pseudo R-square value.

Table 6. Logistic Regression Models of Substance Use

 

 

 

Model 1 Model 2 Model 3
Independent Variables B Odds B Odds B Odds
Ratio Ratio Ratio

Family Variables
Monitoring by Mother -1.011*** .364 -.631*** .532 .103 1.108
Monitoring by Father -1.338*** .262 -.854*** .426 -.155 .857
Monitoring by Both Mother 1.268*** 3.554 .870*** 2.387 .060 1.061
and Father
Living with Biological
Parent(s)

( Living w/ No Parent ) .230*** 1.259 .091 1.095 .174*** 1.190

( Living w/ Bio. Mother) .711*** 2.035 .557*** 1.745 602*" 1.825

(Living w/ Bio. Father) .387*** 1.472 .214*** 1.239 394*" 1.483

( Ref. Category: Living

w/ Both Bio. Parents )

Peers Variables
Positive Peers -.053*** .949 -.054*** .947
Negative (Delinquent) Peers .125 *** 1.133 .131 *** 1.140
Demographic Variables
Age (of Youth) .277*** 1.319
Male (compare to Female) -.294*** .745

 

51

White (compare to Non-

 

White) .608*** 1.837
Parent Education Level .014 1.014
(Mother & Father Combined)
Household Income .000 1.000
Constant .600*** -.280*** -4.954***
-2 Log Likelihood 11694.646 10350.393 7394.952
Model Chi Square 648.868*** 1248.379*** 1161 .425***
Pseudo R Square (Nagelkerke) .098 .184 .228

** p<.05 ***p<.01

(Table 6 continued)

For a “full” model 3 of substance use, it included all thirteen variables, presented
in Table 6. Contrary to models 1 and 2 of substance use, the three monitoring variables
came out to be insigniﬁcant. Hence, the parental monitoring variables are not predictive
of the variation in substance use when other demographic variables are controlled.
However, the three variables for family structure appear to be signiﬁcant. This presents a
similar picture as the logistic regression Model 3 of delinquency (Table 5), in which none
of the three monitoring variables yielded signiﬁcance but all of the family structure
variables came out to be signiﬁcant. It is logical to question why the monitoring variables
were insigniﬁcant and the family structure variables were signiﬁcant in predicting the
variation in substance use or delinquency when controlling for all other variables.

Concerning the family structure characteristics, the results reveal that all three
living arrangements were signiﬁcantly related to substance use. These variables are
predictive of the variation in substance use in all three models. Table 6 shows that the
odds ratio produced for living with no parent, living with only biological mother, and
living with only biological father were 1.190, 1.825, and 1.483, respectively. Given that

the values of odds ratio are higher than that of the reference group, the odds of substance

52

use among youth living with no parent, only biological mother, and only biological father
were higher, compared to those living with both biological parents. Among the three
types of family structure, youth residing with only biological mother had the highest odds
ratio (1.825). The effect of the three types of family structure on substance use by youth
yielded positive relationships.

Similar to Model 2 of substance use (Table 6), the two types (positive and
negative) of peer inﬂuence continue to be strong predictors of likelihood of substance
use. These signiﬁcant ﬁndings provide statistical support for the current hypotheses
concerning the relationship between peers and substance use. In addition, several
demographic variables such as age, gender, and race came out to be signiﬁcantly related
to substance use. Concerning age of youth, Table 5 illustrates that the odds of substance
use are 1.319 times higher for older youth than younger ones. For gender, the result
reveals that the odds of substance are 0.745 times higher for males, compared to females.
The negative regression coefﬁcients also support that male youth are more likely to
engage in substance use than female counterpart. For race, the ﬁnding shows that the
odds of substance use are 1.837 times higher for whites than non-whites. However, parent
education and income yielded to be statistically insigniﬁcant.

In summary, youth who are more likely to engage in substance displayed the
following characteristics --- youth who are living with no biological parent or single
biological parent, who are older, who are males, who are whites, and whose peers are less
delinquent and more positive. Among all the variables, the family structure of ‘living
with only biological mother’ and the race of youth proved to be the two strongest

predictors of substance use by young juveniles.

53

Overall, logistic regression Model 3 predicts about 23 percent of the variation in
substance use, indicated by the Nagelkerke Pseudo R-square value. As with models for
delinquency, the Nagelkerke Pseudo R-square values of explaining substance use also

seem to increase as it progresses from model 1 (10 percent) to model 3.

54

CHAPTER 5
DISCUSSION AND CONCLUSION
1. Summary and Discussion

Summa_ry

After reviewing all of the logistic regression models and tables, two general
statements can be made regarding the issues of delinquency and substance use by young
adolescents. First would be the outcome of delinquent behaviors. Youth who have the
following characteristics are more likely to become delinquent than those who have the
contrary --- those who are males, who are living with no biological parent or single
biological parent, who are whites, whose peers are less positive and more delinquent, and
whose parents’ education level are high. Second would be the outcome of substance use.
Youth who have the following qualities are more likely to take part in substance use ---
those who are males, who are older, who are whites, who are living with no biological
parent or single biological parent, and whose peers are less positive and more delinquent.
Overall, youth who are males and whites, who are living with no biological or single
biological parent, and whose peers are less and more delinquent, are more likely to
engage in not only delinquent behaviors but also use of substance. However, the two
outcomes did not overlap on the issues on age and parent education.

For the “full” model of delinquency, ‘living with biological mother’ proved to be
the strongest predictor (2.251) of youth delinquency among all the variables. For
substance use, two strongest predictors of substance use were evident among all the
variables — ‘living with biological mother’ (1.825) and race (1.837). In terms of

explanatory power, the logistic regression models for substance use seem to be higher

55

than those three models of delinquency. Logistic regression model 3 predicting substance
use appears to yield the highest value of Nagelkerke Pseudo R-square (23%).

Overall, for the logistic models of delinquency, family structure characteristics
seem to have a dominant effect over other important variables such as parental
monitoring and peer inﬂuence variables. For models of substance use, family structure
and peer inﬂuence seem to be more signiﬁcant cases, compared to those of parental
monitoring. In addition, the two monitoring variables (by only single parent) yielded a
negative relationship in predicting substance use. However, the interaction term
(monitoring by both mother and father) came to form a positive, signiﬁcant relationship
in predicting substance use by youth. This particular effect leads to mixed explanations.

By examining the ﬁndings of both logistic models, it demonstrates that other
important variables such as peers and family structure may have a stronger effect on
delinquency and substance use. Although studies indicate the importance of parental
monitoring in reducing levels of delinquent behaviors, an ecological perspective advises
against making unwarranted conclusions concerning the comprehensiveness of the
parenting variable, as there are different spheres of family-delinquency link such as peer
groups and community (Bronfenbrenner, 1979; Jacobson & Crockett, 2000). Studies
reveal that family structures such as single parent household are found to be signiﬁcant
predictors of risks for substance use and delinquent behaviors (Demuth & Brown, 2004;
Dishion & McMahon, 1998; Van Voorhis et al., 1988). Furthermore, although it is
debatable, studies show that peer may have an equal or stronger effect as a predictor of

delinquent behaviors compared to parents (Barnes et al., 2006; Elliott & Menard, 1996;

56

Gifford-Smith et al., 2005; Snyder, Dishion, & Patterson, 1986; Sabatelli & Anderson,
1991; Warr, 1993).

Several important ﬁndings were obtained ﬁrom this study. Notably, the impact of
parental monitoring was not foremost predictor of delinquency and substance use among
other correlates. Instead, the family structure variables came out to be the signiﬁcant
predictors of delinquency as well as substance use among the main independent
variables. Moreover, the variables associated with peer inﬂuence came out to be the
signiﬁcant predictors of both dependent variables in four of the logistic regression
models. While controlling for peer inﬂuence and all other demographic factors, the
ﬁndings did not support the view of social bond theory, particularly parental monitoring
as applicable predictor of delinquency and substance. Another noteworthy ﬁnding was
concerned with the family structure -- whether the youth was living with a no biological
parent, a biological mother, a biological father, or both biological parents. Studies support
that the claim that parental support and control may be lacking in the single-parent
households compared to the households with both parents (Barnes & Farrell, 1992; Gove
& Crutchﬁeld, 1982).

In order to examine the invariance of the explanatory variables within the
regression framework, one of the frequently employed methods would be to “determine
the signiﬁcance of the difference between two regression coefﬁcients within two
independent samples” (Paternoster et al., 1998, p.860). In other words, the purpose of this
statistical test is to compare the regression coefﬁcients of the two models and examine
the effect of the explanatory variables are invariant across the two outcomes (delinquency

and substance use) in the study. In this case, the null hypothesis would be that there is no

57

difference between the two coefﬁcients in the population. Below is the table for test of
invariance of the independent variables in both models of delinquency and substance use.
It should be noted that only the independent variables that yield to be signiﬁcant
predictors of delinquency were chosen for this statistical test for the equality of
regression coefﬁcients. In Table 7, a z-score above the absolute value of 1.96, whether
positive or negative, indicates a signiﬁcant difference between the two regression
coefﬁcients. Overall, all of the variables came out to be signiﬁcant except the variable of
‘living w/ father.’ Hence, for the six signiﬁcant variables, the null hypothesis that no
difference exists between the two regression coefﬁcients can be rejected. This proves that
the two regression coefﬁcients are not equal. In other words, the impact of the signiﬁcant
variables is greater for outcome of delinquency than for that of substance use, or vice

versa.

Table 7. Z—Score of Difference Between Regression Coefficients for Delinquency and
Substance Use

 

 

. Delinquency Substance Use

B SE B SE Z'Sc‘m’
Living w/ No Parent .592 .089 .225 .053 354"
Living w/ Mother .808 .099 .346 .059 4.00**
Living w/ Father .270 .051 .183 .031 1.45
Gender -.873 .042 -. 136 .025 1507’”
Race .230 .045 .330 .027 -3.70**
Positive Peer -.048 .008 -.O30 .005 190*
Negative Peer .127 .005 .075 .003 4.84**

 

NOTE: Unstandardized regression coefﬁcients and z-scores were statistically signiﬁcant at
*p < .05, "p < .01. All of the 7 independent variables are signiﬁcant itself in the models (Please
tables 5 & 6).

58

Study Limitations

There are several limitations of this study that should be mentioned. First of all,
the study did not contain other signiﬁcant family context variables such parental
attachment, parental discipline, parental affection towards the children, or parental
conﬂict. Previous studies examined a variety of variables related to family context
(Barnes, Hoffman, Welte, Farrell, & Dintcheff, 2006; Barnes and Farrell, 1992; Loeber
and Stouthamer—Loeber, 1986; Rankin and Wells, 1990).

Secondly, the data analysis conducted in the study is based on a cross-sectional
data. Thus, the signiﬁcant ﬁndings in this study do not provide causal relationships
between the independent and dependent variables. Although the main independent
variables — parental monitoring and family structure — came out to be signiﬁcant
predictors of delinquency and substance use, it is unclear whether these variables directly
predict delinquency. Moreover, a cross-sectional study that uncovered the effect of
parental monitoring may not be helpful if the parent is not able to maintain the function
of monitoring. In other words, there may be other variables that moderate or interact with
the current parenting and family variables in predicting delinquency and substance use.
Stouthamer-Loeber and Loeber (1988) suggest that different sets of factors contribute to
the different levels of delinquency, rather than the case where a single set of causal
factors predict delinquency.

For future research, it may be useful to review the longitudinal study by Klein et
al. (1997), which examined the predictors of early adolescence delinquency with a
sample 132 Caucasian families. The study included a number of family variables and a

type of maternal parenting skill as its main independent variables. Its ﬁndings indicate

59

that maternal communication and problem-solving skills and family variables both
‘independently and interactively predict’ delinquency during early adolescence (Klein et
al., 1997). Due to the limitations of a cross-sectional design in establishing causal
relationships, future research should conduct a longitudinal analysis including a larger
number of variables related to family context such as those related to family process and
structure. This would help in better capturing the dimensions of the speciﬁc relationships
between the parent and the child within the different family households.

Thirdly, even though the analysis is based on a dataset that is nationally
representative, the main methodical issue was the highly skewed distribution and large
percentage of missing variables. For the independent variables, the zero values comprised
about 40 to 50 percent for the monitoring variables. Moreover, the signiﬁcance of
predictors may be due to the nature of large secondary dataset; those predictors may be
not signiﬁcant if analyzed in a smaller sample.

Lastly, the study utilized the logistic regression technique by recoding the
dependent variables into dichotomous ones. Although the logistic regression allows one
to examine the simple effect of parental monitoring on delinquency, it does not convey
the types and degree of delinquent behaviors. Future research should consider examining
the relationship between the different types and degree of delinquency by family function

and structure aspects.

11. Conclusion and Implications for Future

Conclusion

The main purpose of this study was to examine the relationship between parental

60

monitoring and delinquency / substance use. In terms of directionality, the study
examined whether monitoring by parents can explain the delinquent behaviors and
substance use by young adolescents. Total of eleven variables that have been veriﬁed to
be strong predictors of delinquency were included in the study analysis. The monitoring
variables were supported by empirical evidence testing Hirschi’s social bond theory.
Irnportantly, the study contains the investigation of an aspect of family structure. A
family structure variable, speciﬁcally a component of living arrangements with youth’s
biological parent(s), was included in the equation as a major independent variable. By
including the family structure variable, the study was able to examine the independent
contributions of family function and family structure variables to adolescent delinquency
and substance use, while controlling for other demographic factors. The family structure
variables were considered as the major variables along with the three monitoring
variables in order to examine the association between the family structure and family
function aspects. However, the proportions of the contribution of each type of family
variables are unknown. For example, the effect of monitoring by a parent within an
“intact” family environment would be different from that within a “non—intact” family
household. Furthermore, by including the peer variables as another signiﬁcant competing
explanation for delinquency, the models were able to provide further insights or
explanations when analyzing the correlates of delinquency and substance use (Warr,
2005)

This study also demonstrates the importance of the interaction term by
incorporating newly created monitoring variable in the statistical models to examine the

effects of interaction between the two monitoring variables, rather than comparing

61

monitoring by only mother and only father. Although the results do not indicate the
signiﬁcance or the magnitude of an interaction term, but the attempt may be worthwhile
in investigating the impact of monitoring by both mother and father on the dependent

variables.

Implications for Future Study

The methodological technique used in the current study is suitable for a model
with dependent variables that are dichotomous. However, the dependent variables ---
delinquency and substance use --- are initially measured as variables with a limited range
which can be found in the NLSY 97 dataset. Furthermore, the two dependent variables
are considered count data where they represents the number of counts received;
delinquency ranging from 0 to 10 and substance use ranging from 0 to 3. Hence, the
Poisson regression model using count data may be statistically more adequate for the type
of current dependent variables (Cameron & Trivedi, 1998). Future study should examine
the effect of family contextual variables (family process, family function, & family
structure) as well as peer inﬂuence on these two dependent variables by using the Poisson
count data model. This will not only increase the rigor of the methodology but also
generate sophisticated models for explaining the effects of parental factors on
delinquency and substance use.

The ﬁndings of the study regarding the main variables are mixed. Albeit the
reasonable speculation that family structure and peer variables had a dominant inﬂuence
over the monitoring variables, the reasons for the insigniﬁcance of monitoring variables

are still uncertain. Presumably, the weak performance of the monitoring variables on both

62

outcomes may be attributed to the other contextual inﬂuence and social dynamics that
may be occurring in the household. Although some of the results for parental monitoring
lead to unclear understanding of the family dynamics, it may be constructive to
investigate the impact of other family or parenting variables that could account for
insigniﬁcant results such as marital quality, inter-parental conﬂict, mother-adolescent and
father-adolescent relationship, parental communication skills, and family stressors
(Burcu, 2003; Barnes et al., 2006; Bahr, 1979; Cernkovich & Giordano, 1987; Dishion, et
al., 1991; Larzelere & Patterson, 1990; Loeber & Stouthamer-Loeber, 1986). While it
would be difﬁcult to include a comprehensive set of components related to the family
dynamics, future study should attempt to include a wide range of variables that may be
accountable for the variation in delinquent behaviors as well as substance use.

Regarding the explanatory power of the models for delinquency and substance
use, the results reveal that family structure (living arrangements), peer , and several
demographic (gender and race) variables all contribute to the explanation of the
relationship between parental inﬂuence and child delinquency. However, only small
percentage of the effect of parental monitoring is explained. It seems that mechanisms
through which family function (monitoring) and family structure (living arrangements)

affect delinquent behaviors are complex.

63

Appendix A. Correlation Matrix CN=89841

 

 

Variables 1 2 3 4 5 6 7 8 9 10 11
Monitoring By
Both Parents (1)
M . I ﬁt.

omtonng By .79 1.00
Mother (2)
M 't . B it! 0“

0m 0mg y .98 .77 1.00
Father (3)
Family Structure ... ... ...

.11 .07 .07 1.00
(4)
Age (5) —.67”’-.85"'-.68"‘-.02 1.00
Gender (6) -.03‘ -.01 -03" .02 .01 1.00
Race (7) .11‘” .02 .11‘" .09’” -.01 -.01 1.00
Parent Education m .. ... ... . m
(8) .07 .03 .07 .06 -.03 -.01 .25 1.00
Income (9) .11“" -.01 .11‘” .08'” .02 -.01 .23‘” .29"’1.00
POSitive Peer 15“. 14... 15... 05“! 17". 05“. O3. 11!!! 14ttt1 00
Inﬂuence (10) . . . . . . . . . .
Negative Peer it. 0‘. I”. ﬁt. Oil 0.. ﬁt. #0. ttt it!
-.37 -.39 -.38 -.07 .47 .10 -.11 -.07 -.10 -.27 1.00

Inﬂuence (11)

 

64

BIBLIOGRAPHY

Agnew, R. (1985). Social control theory and delinquency: A longitudinal study.
Criminology, 23, 47-61

Bachman, R. and Paternoster, R. (1997). Statistical methods for criminology and criminal
justice. New York: McGraw-Hill.

Bahr, S. (1979). Family determinants and effects of deviance. In W. Burr, R. Hill, F. Nye,
& I. Reiss (Eds), Contemporary theories about the family: Research-based theories (Vol.
1). New York: Free Press.

Barnes, G. M. and Farrell, M. (1992). Parental support and control as predictors of
adolescent drinking, delinquency, and related problem behaviors. Journal of Marriage
and the Family, 54, 763-776.

Barnes, G. M., Hoffman, J. H., Welte, J, W., Farrell, M. P., & Dintcheff, B. A. (2006).
Effects of parental monitoring and peer deviance on substance use and delinquency.
Journal of Marriage and Family, 68, 1084-1104.

Bowlby, J. H. (Ed3.) (1969-1982). Attachment, Separation and Loss. (3 volumes). New
York: Basic Books.

Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA:
Harvard University Press.

Burcu, E. (2003). Alcohol and drug use in young apprentices: Effect of social control in
the family. Journal of Youth and Adolescence, 32, 291-299.

Bureau of Labor Statistics, US. Department of Labor. 2002. National Longitudinal
Survey of Youth 1997 cohort, 1997—2001 (rounds 1—5) [machine-readable data ﬁle].
Chicago, IL: National Opinion Research Center, University of Chicago [producer].
Columbus, OH: Center for Human Resource Research, Ohio State University
[distributor].

Butts, J. (2000). Youth crime drop. Washington, DC: Urban Institute. Justice Policy
Center.

Bureau of Labor Statistics, US. Department of Labor. National Longitudinal Survey of
Youth 1997 cohort, 1997-2003 (rounds 1-7) [computer ﬁle]. Produced by the National
Opinion Research Center, the University of Chicago and distributed by the Center for
Human Resource Research, The Ohio State University. Columbus, OH: 2005.

65

Butts, J. & Travis, J. (2002). The Rise and Fall of American Youth Violence: 1980 to
2000. Research Report, Urban Institute, 2100 M Street, N.W., 5th Floor, Washington, DC
20037. Tel: 202-833-7200; Fax: 202-728-0232; Web site: http://www.urban.org.

Cameron, A. and Trivedi, P. (1998). Regression Analysis of Count Data. New York:
Cambridge University Press.

Canter, R. J. (1982). Family correlates of male and female delinquency. Criminology, 20,
149-167

Cave, D. (April 12, 2008). Eight Teenagers Charged in Internet Beating Have Their Day
on the Web. In New York Times. Retrieved April 13, 2008, from
http:// www.nytimes.com/ 2008/ 04/ l 2/us/ 1 2ﬂorida.htrnl?pagewanted=print

Cernkovich, S. A & Giordano, P. C. (1987). Family relationships and delinquency,
Criminology, 25, 295-321

Chang, J., Kazdin, C., Hagan, K. & Ibanga, 1. (April 10, 2008). Mean Girls' Fights
Haven't Increased Teen Violence: Bullies Use Internet as a Way of Winning Bragging
Rights for Exploits. In ABC News. Retrieved April 13, 2008, from
http://abcnews. go.com/GMA/story?id=4626462&page=1 &GMA=true

Chubb, N. H., and Fertman, C. I. (1992). Adolescents' perceptions of belonging in their
families. Families in Society: The Journal of Contemporary Human Services, 73, 387-
393

Conger, R. (1976). Social control and social learning models of delinquent behavior: A
synthesis. Criminology, 14, 17-40

Demuth, S. & Brown, S. (2004). Family structure, family processes, and adolescent
delinquency: The signiﬁcance of parental absence versus parental gender. Journal of
Research in Crime and Delinquency, 41, 58-81.

Dishion, T. & McMahon, R. (1998). Parental monitoring and the prevention of child and
adolescent problem behavior: A conceptual and empirical formulation. Clinical Child
and Family Psychology Review, I, 61-75.

Dishion, T.J., Patterson, G.R., Stoolrniller, M. & Skinner, ML. (1991). Family, school,
and behavioral antecedents to early adolescent involvement with antisocial peers.
Developmental Psychology 27, 172-180.

Duncan, D. F. (1978). Attitudes toward parents and delinquency in suburban adolescent
males. Adolescence, 13, 365-369

Elliott, D. NATIONAL YOUTH SURVEY [UNITED STATES]: WAVE VI, (1983)
[Computer ﬁle]. ICPSR version. Boulder, CO: University of Colorado, Behavioral

66

Research Institute [producer], 1992. Ann Arbor, MI: Inter-university Consortium for
Political and Social Research [distributor], 1994.

Elliott, D., & Menard, S. (1996). Delinquent ﬁiends and delinquent behavior: Temporal
and developmental patterns. In Hawkins, J. D. (ed.), Delinquency and Crime: Current
Theories. Cambridge University Press, Cambridge, England.

Evans, D. T., Cullen, F. S., Burton, V. S., Dunaway, G. R. & Benson, M. L. (1997). The
Social Consequences of Self-Control: Testing the General Theory of Crime. Criminology.
5, 475-501.

Fletcher, A. C., Steinberg, L., & Williams-Wheeler, M. (2004). Parental inﬂuences on
adolescent problem behavior: Revisiting Stattin and Kerr. Child Development, 75, 781-
796.

Garbarino, J. (1992). Children and Families in the Social Environment. (2nd ed.). New
York: Aldine De Gruyter.

Gifford-Smith, M., Dodge, K. A., Dishion, T. J ., & McCord, J. (2005). Peer inﬂuences in
children and adolescents crossing the bridge from developmental to intervention science.
Journal of Abnormal Child Psychology, 33, 255—265.

Glueck S. and Glueck E. (1962). Family Environment and Delinquency. Boston:
Houghton Mifﬂin Co.

Glueck, S. and Glueck, E. (1950). Unraveling juvenile delinquency. New York: The
Commonwealth Fund.

Gottﬁ'edson, M. and T. Hirschi. (1990). A General Theory of Crime. Stanford: Stanford
Univ. Press.

Gove, W. and Crutchﬁeld, R. (1982). The Family and Juvenile Delinquency. Sociological
Quarterly. 23, 301-319.

Hagan, J. (1991). Destiny and drift: Subcultural preferences, status attainments, and the
risks and rewards of youth. American Sociological Review, 56, 567-582.

Hagestad, G. (1981). Problems and promises in the social psychology of intergenerational
relations. In R. Fogel, E. Hatﬁeld, S. Kiesler & J. March (Eds), Stability and change in
the family. New York: Academic Press

Hindelang, M. J. (1973). Causes of delinquency: A partial replication and extension.
Social Problems, 20, 471-487.

Hirschi, T. (1969). Causes of delinquency. Berkeley, CA: University of California Press.

67

Hirschi, T. and Gottfredson, M. (1983). Age and Explanation of Crime. American
Journal of Sociology, 89, 552-584.

Hirschi, T. and Gottfredson, M. (2001). Self-control theory. In R. Paternoster and R.
Bachman (eds.) Explaining criminals and crime, 81—96.

Hirschi, T., & Gottfredson, M. (2006). Social control and self-control theory. In S. Henry
and M. M. Lanier (Eds.), Essential criminology reader (pp. 111-118). Boulder, CO:
Westview Press.

Jacobson, K. C., & Crockett, L. J. (2000). Parental monitoring and adolescent adjustment:
An ecological perspective. Journal of Research on Adolescence, 10, 65-97.

Jensen, S. F., and Brownﬁeld, D. (1983). Parents and drugs: Specifying the consequences
of. attachment. Criminology, 21, 543-554

Junger-Tas, J. (1992). An empirical test of control theory. Journal of Quantitative
Criminology, 8, 9-28.

Kerr, M., & Stattin, H. (2003). Parenting of adolescents: Action or reaction? Chapter 9
(pp. 121-151) in AC. Crouter & A. Booth (Eds.) Children's inﬂuence on family

dynamics: The neglected side of family relationships. Mahwah, NJ: Lawrence Erlbaum.

Krohn, M. D. (1974). An investigation of the effect of parental and peer associations on
marijuana use: An empirical test of differential association theory. In Riedel, M. and
Thomberry, T. P. (eds), Crime and Delinquency: Dimensions of Deviance (pp. 75—87).
Praeger, NY.

Krohn, M. D., Stern, S. B., Thomberry, T. P. and Jang, S. J. (1992). The measurement of
family process variables: An examination of adolescent and parent perceptions of family
life on delinquent behavior. Journal of Quantitative Criminology. 8, 287-315.

Kumpfer, K. L. & Alvarado, R. (2003). Farrrily-Strengthening Approaches for the
Prevention of Youth Problem Behaviors. American Psychologist. 58, 457-65.

Larzelere, R. E., and Patterson, GR. (1990). Parental management: Mediator of effect
of socioeconomic status on early delinquency. Criminology, 28, 301-323.

Laub, J. H. and Sampson, R. J. (1988). Unraveling Families and Delinquency: A
Reanalysis of the Gluecks' Data. Criminology, 26, 355-380

Loeber, R. and Farrington, D. P. (2000). Young children who commit crime:

Epidemiology, developmental origins, risk factors, early interventions, and policy
implications, Development and Psychopathology, 12, 737-762.

68

Loeber, R. and Stouthamer-Loeber, M. (1986). Family Factors as Correlates and
Predictors of Juvenile Conduct Problems and Delinquency, Crime and Justice, 7, 29-149

Loury, G. (1987). The family as context for delinquency prevention: Demographic trend
and political realities. In J. Q. Wilson & G. C. Loury (Eds.), From children to citizens
(Vol. 3, pp. 3-26). New York: Springer-Verlag.

Matsueda, R. L. and Heimer, K. (1987). Race, family structure, and delinquency: A test
of differential association and social control theories. American Sociology Review, 52,
826-840

McCluskey, C.P. and Tovar, S. (2003). Family Processes and Delinquency: The
Consistency of Relationships by Ethnicity and Gender. Journal of Ethnicity in Criminal
Justice, 1, 37-62.

McCord, J. (1991). Family relationships, juvenile delinquency, and adult criminality.
Criminology, 29, 397-417.

McCord, J. & Tremblay, RE. (1992). (Eds.), Preventing Antisocial Behavior:
Interventions from Birth through Adolescence. New York: Guilford Press.

McCord, J. (1994). "Family Socialization and Antisocial Behavior: Searching for Causal
Relationships in Longitudinal Research." In E. G. M. Weitekarnp & H-J Kemer (Eds),

Cross-National Longitudinal Research on Human Development and Criminal Behavior
(pp. 177-188). Dordrecht, Netherlands: Kluwer.

Miller, B. C. (1993). Families, science, values: Alternative views of parenting effects and
adolescent pregnancy. Journal of Marriage and the Family, 55, 7-21.

Nagasawa, R., Qian, Z., & Wong, P. (2000). Social Control Theory as a Theory of
Conformity. Sociological Perspectives, 43, 581—603.

Newcombe, M. D. & Bentler, P. M. (1988). Consequences of Adolescent Drug Use:
Impact on the lives of young adults. Newbury Park: Sage.

Nye, F. I. (1958). F arnily Relationships and Delinquent Behavior. New York: Wiley

Oxford, M. L., Harachi, T. W., Catalano, R. F., & Abbott, R. D. (2000). Preadolescent
predictors of substance initiation: a test of both the direct and mediated effect of family

social control factors on deviant peer associations and substance initiation. American
Journal of Drug and Alcohol Abuse, 27, 599-616.

Ozbay, O. and Ozcan, Y. (2008). A Test of Hirschi's Social Bonding Theory: A

Comparison of Male and Female Delinquency. International Journal of Offender Therapy
and Comparative Criminology, 52, 134-157.

69

Paternoster, R., Brarne, R., Mazerolle, P. & Piquero, A. (1998). Using the correct
statistical test for the equality of regression coefﬁcients. Criminology, 36, 859-867.

Patterson, G. R. (1980). Children who steal. In T. Hirschi & R. Gottfredson (Eds),
Understanding crime (pp. 73-90). Beverly Hills, CA: Sage Publications.

Patterson, G. R. & Stouthamer-Loeber, M. (1984). The correlation of family management
practices and delinquency. Child Development, 55, 1299-1307

Pratt, T. C., & Cullen, F. T. (2000). The empirical status of Gottfredson and Hirschi’s
general theory of crime: A meta-analysis. Criminology, 38, 931-964

Rankin, J. H. and Wells, L. E. (1990). The Effect of Parental Attachments and Direct
Controls on Delinquency. Journal of Research in Crime and Delinquency, 27, 140-165

Reiss, A. J. (1951). Delinquency as the failure of personal and social controls. American
Sociological Review, I 6, 196-207.

Robbins, M. S., & Szapocznik, J. (2000). Brief Strategic Family Therapy. Bulletin.
Washington DC: US Department of Justice, Office of Justice Programs, Ofﬁce of
Juvenile Justice and Delinquency Prevention.

Sampson, R. & Laub, J. (1990). Stability and change in crime and deviance over the life
course: The salience of adult social bonds. American Sociological Review, 55, 609-627.

Sampson, R. & Laub, J. H. (1993). Crime in the Making: Pathways and Turning Points
Through Life. Cambridge, MA: Harvard University Press.

Sampson, R. & Laub, J. H. (1994). Urban poverty and the family context of delinquency:
A new look at structure and process in a classic study. Child Development, 94, 523-540.

Sabatelli, R. M. and Anderson, S. A. (1991). Family system dynamics, peer relationships,
and adolescents' psychological adjustment. Family Relations, 40, 363-369.

Smith, C. A. & Stern, S. B. (1997). Delinquency and antisocial behavior: A review of
family processes and intervention research. Social Service Review, 71, 382-420.

Snyder, Howard N., and Sickmund, Melissa. 2006. Juvenile Offenders and Victims: 2006
National Report. Washington, DC: US. Department of Justice, Office of Justice
Programs, Ofﬁce of Juvenile Justice and Delinquency Prevention.

Snyder, J., Dishion, T.J., & Patterson, CR. (1986). Determinants and consequences of

associating with deviant peers during preadolescence and adolescence. Journal of Early
Adolescence, 6, 29-43

70

Snyder, J., & Patterson, G. (1987). Family Interaction and Delinquent Behavior (pp. 216-
243). In H. Quay Handbook of Juvenile Delinquency. New York, NY: Interscience
Publications.

Stattin, H. and Kerr, M. (2000). Parental monitoring: A reinterpretation. Child
Development, 71 , 1072-1085

Steffensmeier, D. (1989). On the Causes of “White-Collar Crime”: An Assessment of
Hirschi and Gottfredson’s Claims, Criminology, 27, 345—58.

Stern, S. B. and Smith, C. A. (1995) Family processes and delinquency in an ecological
context. Social Service Review, 69,703-731.

Svensson, R. (2004). Shame as a Consequence of the Parent-Child Relationship A Study

of Gender Differences in Juvenile Delinquency. European Journal of Criminology, 1,
477-504.

Taylor, L. S. (1985). The family as an adaptable and enduring social unit: A case for
assessing delinquent behavior from a family systems perspective. Today’s Delinquent, 4,
69-97.

Thomberry, T. P., Huizinga D. and Loeber, R. (1995). The prevention of serious
delinquency and violence: Implications from the Program of Research on the Causes and
Correlates of Delinquency. pp. 213-237 in James C. Howell, Barry Krisberg, J. David
Hawkins and John J. Wilson (eds.), Sourcebook on Serious, Violent, and Chronic
Juvenile Offenders. Thousand Oaks, CA:

Trojanowicz, R. C., Morash, M., & Schram, P. J. (2001) Juvenile delinquency: concepts
and control (6th ed.). Upper Saddle River, NJ: Prentice Hall.

US. Census Bureau. (2001). Population projections of the United States by age, sex,
race, Hispanic origin, and nativity: 1999 to 2100. Revised January 19, 2001.
<www.census.gov/population/ www/projections/natdet.html>.

Van Voorhis, P, Cullen, F., Mathers, R., & Garner, C. (1988). The Impact of Family
Structure and Quality on Delinquency: A Comparative Assessment of Structural and
Functional Factors. Criminology, 26, 235-261.

Vila, B. (1994). A general paradigm for understanding criminal behavior: Extending
evolutionary ecological theory. Criminology, 32, 311-359.

Vitaro, F ., Brendgen, M. & Tremblay, R. E. (2000). Inﬂuence of Deviant Friends on

Delinquency: Searching for Moderator Variables. Journal of Abnormal Child
Psychology, 28(4)

71

Vold, G., Bernard, T., and Snipes, J. (2002). Theoretical Criminology, New York: Oxford
University Press.

Wang, G. T., Qiao, H., Hong, S., & Zhang, J. (2002). Adolescent social bond, self-
control, and deviant behavior in China. International Journal of Contemporary Sociology,
39, 52—68.

Warr, M. (1993). Parents, peers, and delinquency. Social Forces, 72, 247-264

Warr, M. (2005). Making Delinquent Friends: Adult Supervision and Children’s
Afﬁliations. Criminology, 43, 77-105.

Wasserrnan, G. A., Miller, L., Pinner, E., and Jaramillo, B. S. (1996). Parenting
predictors of early conduct problems in urban, high-risk boys. Journal of the American
Academy of Child and Adolescent Psychiatry, 35, 1227—1236.

Weintraub, K. J. & Gold, M. (1991). Monitoring and delinquency. Criminal Behavior
and Mental Health, 1, 268-281.

Wells, E. L. and Rankin, J. H. (1986). The broken homes and juvenile delinquency:
Analytic issues. Journal of Research in Crime and Delinquency, 23, 68-93.

Wells, E. L. and Rankin, J. H. (1988). Direct parental controls and delinquency.
Criminology, 26, 263-285.

Wright, J. P. and Cullen, F. T. (2001). Parental efﬁcacy and delinquent behavior: Do
control and support matter. Criminology, 39, 677-706.

72

 
   

A

“11111111111111llllllllﬂlllilllllf