University of Southern California Dissertations and Theses

The relationship between pubertal timing and delinquent behavior in maltreated male and female adolescents

(USC Thesis Other)

The relationship between pubertal timing and delinquent behavior in maltreated male and female adolescents

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

THE RELATIONSHIP BETWEEN PUBERTAL TIMING AND DELINQUENT
BEHAVIOR IN MALTREATED MALE AND FEMALE ADOLESCENTS

by

Sonya Negriff

______________________________________________________________

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

May 2007

Copyright 2007 Sonya Negriff
ii
Table of Contents

List of Tables…………………………………………………………………… iii

List of Figures……………………………………..…………………………… iv

Abstract…………………………………………….…………………………… vi

Chapter 1: Background and Significance……………………….……………… 1
Puberty as a Significant Developmental Event……………………………… 1
The Relationship between Pubertal Timing and Delinquency…………….… 2
Child Maltreatment…………………………………………………..……… 6
Child Maltreatment and Delinquency…………………………….….……… 8
Child Maltreatment and Pubertal Development…………………….…......... 10

Chapter 2: Methodological and Measurement Issues………………..……….… 13
The Measurement of Pubertal Development…………………...……………. 13
Classifying Pubertal Timing …………………….………………..………… 17
Cross-sectional Versus Longitudinal Designs…………………………..…… 18
The Measurement of Delinquency……………………………………...…… 19

Chapter 3: The Current Study……………………………………………...…… 22

Chapter 4: Research Design and Methods……………………………….…...… 25
Participants…………………………………………………………...……… 25
Procedures…...…………………………………………………..………...… 28
Measures……...…………………………………………………...………… 29

Chapter 5: Results………………………...…….……………….……………… 34
Preliminary Analyses…………………………………………….…………… 34
Missing Data Imputation………………………………………….……… 34
Tests for Normality……………………….…………………….………… 34
Psychometric Analyses…………………………………………………… 35
Bivariate Relationships…………………………………………………… 41
Substantive Analyses…………………………………………….…………… 41
Specific Aim 1……………………………………………….…………… 41
Specific Aim 2……………………………………………….…………… 50
Specific Aim 3……………………………………………….…………… 55

Chapter 6: Discussion…………………………………………...……………… 66

References………………………………………………………...…………..… 75
iii
List of Tables

Table 1. Attrition Information……...………………………...…………….….. 26

Table 2. Sample Characteristics at Time 1 and Time 2………….…...….……. 28

Table 3. Descriptive Statistics for Tanner Stages,
Pubertal Development Scale, and BMI………………………….…… 30

Table 4. Means and Standard Deviations by Age and Gender for Tanner
Stages, Pubertal Development Scale, and BMI………………….…… 31

Table 5. Items and Scales of the Adolescent Delinquency Questionnaire…..… 36

Table 6. ADQ Factor Intercorrelations and Internal Consistency
Reliabilities............................................................................................ 36

Table 7. Descriptive Statistics for Delinquency Scales………...……………… 37

Table 8. Factor Loadings for Four Factor Model of Pubertal Timing……..….. 40

Table 9. Factor Loadings for Three Factor Model of Pubertal Timing…..……. 40

Table 10. Correlations between Time 1 and Time 2 Variables……………....... 42

Table 11. Fit Statistics for Models Testing the Relationship between
Pubertal Timing and Delinquency………….....……………………. 44

Table 12. Measurement Model Estimates for Models Testing the
Relationship between Pubertal Timing and Delinquency ……...….. 45

Table 13. Structural Model Estimates for Models Testing the Relationship
between Pubertal Timing and Delinquency……………..…………. 46

Table 14. Fit Statistics for Models Testing Quadratic Relationship between
Pubertal Timing and Delinquency…...………….………...……...… 51

Table 15. Parameter Estimates for Models Testing Quadratic Relationship
between Pubertal Timing and Delinquency….………………..…… 52
iv

List of Tables cont.

Table 16. Fit Statistics for Models Testing the Interaction between
Pubertal Timing and Maltreatment………………..….….………… 57

Table 17. Parameter Estimates for Multiple Group Models between
Maltreatment and Comparison Groups………………………..…… 58

Table 18. Parameter Estimates for Multiple Group Models between
Gender with Interaction Term………………………………...…..… 62

v

List of Figures

Figure 1. Testing the linear relationship between the pubertal timing
and delinquency…………………………………………………………. 43

Figure 2. Testing cross-lag relationships between pubertal timing
and delinquency………………………………………………………….. 44

Figure 3. Testing a quadratic relationship between pubertal timing
and delinquency……………………………………………………..…… 51

Figure 4. Model testing the interaction between pubertal timing
and maltreatment………………………………………………………….. 61

Figure 5. Interaction effect of Time 1 pubertal timing x maltreatment status on
Time 1 delinquency……………………………………………….……… 64
vi
Abstract

There is significant disparity in the existing literature on pubertal development
as to whether early, late, or both early and late timing are related to delinquent
behavior, and whether this association varies based on gender. The purpose of this
study was to examine the relationship between pubertal timing and delinquent
behavior in a sample of maltreated and comparison adolescents. The sample for the
present study was drawn from a larger research project examining the effects of
maltreatment on adolescent development. The design was cross-sequential with the
age of subjects ranging from 9-13 years at the initial assessment with subsequent
assessments one year apart. Data for the present study was taken from the first and
second assessments. The first aim of this study was to examine the relationship
between pubertal timing and delinquency cross-sectionally and longitudinally for the
total sample, and then to test for gender differences using multiple group models. The
second aim examined the possibility of a nonlinear relationship between pubertal
timing and delinquency. That is, whether early, late, or early and late timing
contributed to higher levels of delinquency. The third aim was to examine the
interaction of pubertal timing and maltreatment to determine whether the experience
of maltreatment along with off-time pubertal development creates an increased risk for
delinquent behavior, and whether this relationship differs between genders. The results
showed that for both males and females earlier pubertal development relative to peers
was related to higher levels of delinquency. Multiple group models showed that
maltreatment group did not moderate the relationship between pubertal timing and
vii
delinquency when examining both genders together. However, there was evidence
of an interaction effect for females. The interpretation of this interaction was that for
maltreated females earlier pubertal timing was related to higher delinquency, whereas
for comparison females later pubertal timing was related to higher delinquency. These
findings provide evidence of the need to examine contextual factors that may affect
the amplification and direction of these relationships. Future research should continue
to elucidate moderators as well as mediators of the relationship between pubertal
development and deleterious outcomes.

1
Chapter 1: Background and Significance
Puberty as a Significant Developmental Event
Adolescence is a period of major biological, cognitive, and social-emotional
changes. Adolescents cope with these various transitions in different ways; some
negotiate the passages easily whereas others develop serious behavioral and
psychological problems. A variety of factors influence the ease or difficulty of the
transition into and through adolescence for any given individual. These factors include
gender, ethnicity, early childhood trauma, poor parent-child relations, and
neighborhood disadvantage (Ge, Brody, Conger, Simons, & McBride-Murray, 2002;
Ge, Conger, & Elder, 2001).
Puberty is a universal event that is both biologically and socially determined. It
marks the entrance into sexual maturity and is associated with specific somatic
changes such as the growth of body hair (underarm hair and pubic hair), change in
body mass, a height spurt, the development of breasts and onset of menstruation for
females, and the growth of the penis and testicles for males. The onset of puberty is
characterized by the beginning of breast development for females and testicular
enlargement for males. In the United States onset of female breast development occurs
typically begin between the ages of 8 and 13 years, and males’ testicular growth
between the ages of 9 to 13.5 years (Fechner, 2003). The effects of early pubertal
maturation versus on-time versus late maturation have been examined extensively.
Early, late, and off-time (early and late) pubertal maturation have been linked to a
number of deleterious outcomes such as delinquency, depression, anxiety, and

2
aggression (Hayward, 2003). However the results from these studies are
inconclusive as to which timing classification (early timing, late timing, or both early
and late timing) is most related to delinquency.
The Relationship between Pubertal Timing and Delinquency
In the literature addressing the relationship between pubertal development and
psychosocial adjustment there is a great deal of focus on delinquency. It is well
documented that during adolescence delinquent behavior increases (Herrera &
McCloskey, 2001). This increase has spurred a substantial body of research
investigating the relationship between puberty and externalizing problems such as
delinquency, substance use, and behavior problems (Graber, Lewisohn, Seeley, &
Brooks-Gunn, 1997; Dorn, Susman, & Ponirakis, 1999; Caspi, Lynam, Moffitt, &
Silva, 1993; Ge et al., 2002; Dick, Rose, Pulkkinen, Kaprio, 2001). Externalizing
behavior such as delinquency or substance abuse can be transitory or lead to persistent
problems. Due to variation in the permanence of these behaviors, it is essential to
follow adolescents through their development in order to determine if these
externalizing behaviors are curtailed by the entrance into adulthood or transform into
serious criminal behavior.
Several theories have been proposed to explain the association between
pubertal development and delinquent behavior. In early-maturing adolescents, a
mismatch of external appearance and cognitive development may lead to norm
violating behaviors and delinquency at an earlier age than for those individuals who
mature later. Those adolescents who appear more mature may have demands or

3
expectations placed on them that are not appropriate given their actual
developmental level. Early maturers may be drawn into risky behavior such as
delinquency by older adolescents who judge the individual by their external
appearance of maturity, not by their age. This proposition is supported by evidence
that indicates that the relationship between early pubertal maturation for girls and
delinquency is mediated by relationships with male peers (Caspi et al., 1993; Haynie,
2003). These findings suggest that male peers draw early maturing girls into
committing delinquent acts. There may be similar mechanisms that have not yet been
elucidated that operate the link between early maturation and delinquency in boys as
well.
A second explanation proposes that when an adolescent does not mature at the
same age as his or her peers it creates added stresses that may increase delinquent
behaviors as a coping mechanism. Therefore off-time (developing earlier or later than
peers) development in itself may increase risk for delinquency. Additionally, the
experience of multiple stressors such as off-time pubertal development along with
child maltreatment may further increase the risk. Delinquency itself has been found to
be more prevalent in maltreated adolescents; with the added stress of off-time pubertal
development it may amplify the risk for delinquency. Some researchers have posited
that maltreatment contributes to earlier pubertal development which subsequently
leads to higher levels of delinquency. However, there is limited empirical evidence to
support this proposition as of yet.

4
Overall, there is a great deal of research that points to the negative impact of
early maturation in regards to externalizing problems. Multiple studies have found
early maturation to be associated with the highest levels of delinquency, bullying,
truancy, disruptive behavior, and violent behavior (Ge et al., 2002; Kaltiala-Heino,
Marttunen, Rantanen, & Rimpala, 2003; Graber et al., 1997; Obeidallah, Brennan,
Brooks-Gunn, & Earls, 2004; Cota-Robles, Neiss, & Rowe, 2002; Flannery, Rowe, &
Gulley, 1993). Contrary to the bulk of the research, there are findings that show late
maturation (rather than early maturation) is associated with more behavior problems
(Dorn, Susman, et al., 1999). There are a limited number of studies that have
examined the outcome of delinquency specifically, instead many choose to examine
behavior problems, externalizing behavior, or more specific aspects of delinquent or
problem behavior. Of the studies that have examined aspects of delinquency, the
majority have employed only single gender samples. This limits what can be inferred
from those results when comparing to studies in which both genders were examined.
With all female samples, two separate studies have found early maturation to
be associated with the highest levels of delinquency (Caspi et al., 1993; Haynie, 2003)
and one found an effect for early pubertal timing only when those early maturers also
experienced neighborhood disadvantage (Obeidallah et al., 2004). However, Caspi and
colleagues (1993) found delinquency to be equal in early and on-time maturing girls at
age 15 (both of which were higher than late maturing girls). This warrants interest
because generally, on-time pubertal development is considered the least detrimental
and is thought not to present the same risks as early pubertal development. In the

5
studies with all male samples, one found early maturation to be linked to higher
levels of violent and non-violent delinquency (Cota-Robles et al., 2002), and another
found both early and late timing to be associated with the highest levels of delinquent
behavior (Williams & Dunlop, 1999). Results from studies that included both genders
found bullying, truancy, and delinquency to be higher in both early maturing boys and
girls compared to on-time or late maturing individuals (Kaltiala-Heino et al., 2003;
Flannery et al., 1993).
Therefore, although the majority of the studies find that early timing confers
the highest risk for delinquent behavior, there are results that conflict with this
inference. Considering the evidence, there is not consistent support either within or
between genders for only one type of pubertal timing as a risk factor for delinquency.
In addition, the research on females is more extensive than that on males, which calls
for clarification of the gender differences in the associations between pubertal timing
and delinquency. It may be that pubertal timing affects males and females differently
and one explanation cannot cover both cases. For example, females may be affected
by early timing, whereas for males it may be both early and late timing, or vice versa.
These relationships need to be clarified, and should not be assumed to be equal across
genders.
A limitation of the extant research is that ethnically diverse samples have not
been included in most studies. Inclusion of ethnic minorities will lead to a more
complete understanding of the impact of pubertal development across ethnicities. It is
important to identify whether there are differences in the relationship between pubertal

6
timing and delinquent behavior linked to ethnicity. Evidence points to a relationship
between pubertal timing and delinquency for African American, Hispanic, and White
boys. Additionally, findings indicate early maturing African American boys have
higher levels of delinquency than White boys (Cota-Robles et al., 2002). This
demonstrates that other ethnicities (than White) may be at higher risk when they begin
pubertal development early. Furthermore, there is evidence that indicates African
American girls on average begin to mature earlier than White girls; this may put them
at risk for a number of problems beyond those that other ethnicities experience
(Chumlea, Schubert, Roche, Kulin, Lee, Himes, et al., 2003).
Some findings point to early development being the most detrimental for the
development of delinquent behaviors, whereas other findings also point to late
development as a risk factor (Ge et al., 2002; Kaltiala-Heino et al., 2003; Williams &
Dunlop, 1999). Still other results only indicate a relationship when contextual factors
are examined (Obeidallah et al., 2004). Due to the many studies that have aimed to
uncover how puberty operates with environment to affect delinquent behaviors, it
becomes apparent that the relationship is complex and multi-faceted. Much more work
needs to be done to fully understand the interplay of puberty and psychosocial
adjustment. This means examination of ethnically diverse samples, the examination of
gender differences, and the inclusion of moderator or mediator variables.
Child Maltreatment
Child maltreatment is a pervasive social problem associated with a host of
maladaptive outcomes. Data from incidence studies conducted by the National Center

7
on Child Abuse and Neglect (NCCAN) indicate an alarmingly high number of
children in the US are victims of abuse and neglect. According to NCCAN (1996),
using the most stringent standard the annual incidence of child maltreatment was over
one and a half million children, rising to twice that number when a less stringent
standard was used. Maltreatment has been found to be associated with a variety of
maladaptive outcomes. It is well established that maltreatment in the early years has a
negative impact on developmental adaptation as defined by the salient developmental
issues of each age including cognitive, social, and mental health problems (Cicchetti
& Toth, 2000). There is substantial documentation of the impaired cognitive abilities
and poor academic achievement of maltreated youth (Kaplan, Pelcovitz, and Labruna,
1999).
Because of the consistent evidence that maltreatment adversely affects
development, it is likely that an individual with the experience of maltreatment may
experience more adjustment difficulties associated with off-time pubertal development
than an individual without this experience. The implication stemming from the
previous research is that maltreatment experience may moderate the relationship
known to exist between off-time pubertal development and maladaptive outcomes;
although it appears that this specific relationship has not yet been examined. Sexual
abuse has been found to be linked with early pubertal development (Brown, Cohen,
Chen, Smailes, & Johnson, 2004; Zabin, Emerson, & Rowland, 2005), but the
examination of maltreatment as a moderator of the relationship between pubertal
timing and adjustment difficulties has not been investigated. Based on the cumulative

8
stress model, it would be expected that the experience of multiple stressors
(maltreatment and off-time puberty) would confer more risk than if any one stressor is
experienced alone.
In addition to the actual experience of maltreatment, the events surrounding a
report of maltreatment to authorities, such as the involvement of the department of
social services, may exacerbate the experience of maltreatment itself. The removal of
a child from their family of origin, placement of the child with strangers in a foster
home, and movement from one foster home to another are among some of the stressful
events that a child might experience upon a report of maltreatment to authorities. Such
events may be added stressors associated with the experience of maltreatment that
may affect a child’s ability to cope, particularly during puberty. Early or off-time
pubertal maturation has been suggested to be stressful for some individuals; the
experience of maltreatment and consequences surrounding this experience might be
especially detrimental. There is a significant gap in the extant research as to the
relationship between pubertal timing and delinquent behavior in adolescents with the
experience of maltreatment.
Child Maltreatment and Delinquency
Child maltreatment has been documented as a major risk factor for
delinquency and violent behavior by a number of studies. For example, individuals
with substantiated abuse in childhood are more likely to have both self-reported and
official delinquency. Additionally, more extensive maltreatment has been found to
predict higher rates of delinquency (Smith & Thornberry, 1995). Developmental

9
periods during which maltreatment occurs have also been investigated as risk
factors for delinquency. Researchers found that maltreatment that begins in childhood
and continues into adolescence, and maltreatment that occurs in adolescence are
significantly related to official and self-reported delinquency (Ireland, Smith, &
Thornberry, 2002). Over numerous studies, more adolescents with a history of
maltreatment have been found to engage in delinquent behavior than those without a
history of maltreatment (Bolton, Reich, & Gutierres, 1977; McCord, 1983; Zingraff,
Leiter, Myers, & Johnson, 1993). However, these findings are from studies which
primarily used official reports of delinquency. Self-report measures have seen limited
use in the investigation of child maltreatment and delinquency. The limited findings
indicate the same trend as found with official delinquency measures; maltreatment is
correlated with delinquent behavior (Doerner, 1987; Smith & Thornberry, 1995). The
relationship between child maltreatment and delinquency has been demonstrated for
all types of abuse and neglect, and evidence suggests that this relation is particularly
strong for neglect (Brown, 1984; Weatherburn & Lind, 1998). Although there are clear
associations between child maltreatment and delinquency, there is limited
investigation into moderators or mediators that could help identify the conditions and
mechanisms by which maltreatment affects delinquent behavior. Alternatively,
maltreatment may act as a moderator or mediator between other variables, such as
pubertal development and delinquency.
Overall, child maltreatment is strongly supported as a risk factor for
delinquency and violent behavior. As reported previously, the timing of pubertal

10
development is a factor that has also been found to be related to increased
delinquency. However, there is limited research that has probed the relationship
between pubertal development and experience of child maltreatment. As both these
variables are related to increased levels of delinquency it seems apparent that the
interaction between pubertal timing and child maltreatment should be examined.
In the literature on the effects of pubertal timing there is an overwhelming
prevalence of research into these relations in females, leaving a large gap as to the
relations in males. In the research on delinquency there is an overwhelming interest in
male populations, leaving large gaps in regard to females. It is evident that both
genders should be included to clarify the links between child maltreatment, pubertal
timing, and delinquency.
Child Maltreatment and Pubertal Development
Although there is substantial research examining the relationship between
pubertal maturation and delinquency, there is far less that has attempted to examine
possible variables that might moderate or mediate the relationships that have been
reported. The relationship between pubertal timing and delinquent behavior is not as
straight-forward as some studies might imply. In addition to variables such as gender
and ethnicity, there are a number of possible factors that might affect the strength of
this relationship. It is essential to elucidate these contextual influences in order to fully
understand the complex relationship between puberty and behavior. In particular the
relationship between pubertal development and maltreatment is not well elucidated.

11
Some researchers have hypothesized that early environmental stress, such as
childhood maltreatment, triggers early pubertal development. One such theory comes
from Belsky, Steinberg, and Draper (1991) which proposes that disruptions in the
family environment prior to puberty will lead to earlier pubertal development. This
theory contrasts a more traditional view that pubertal development is primarily
biologically determined rather than environmentally determined, and serves as a risk
factor for developing adolescents. This biological view is the basis for most of the
research reviewed previously in this paper. There are findings that link sexual abuse to
early pubertal development; however the direction of the relationship is equivocal.
One study examining abused and neglected females found that age of menarche was
significantly earlier for those girls who had experienced two or more episodes of
sexual abuse (Brown et al., 2004). Additional support for a relationship between
sexual abuse and early pubertal development comes from a retrospective study of
sexual abuse and age at menarche (Zabin et al., 2005). Menarche was experienced 5
months earlier for victims of sexual abuse than non-victims. Although this difference
was significant, discrepancies in self-report of age at menarche have been found to be
as large as 18 months (Dorn, Nottelmann, Susman, Inoff-Germain, & Chrousos,
1999). Although evidence indicates that sexual abuse and pubertal timing are linked in
females, there is no research that has investigated this association in males. Nor is
there evidence that other forms of maltreatment are associated with early pubertal
development.

12
There is an absence of studies that examine maltreatment and pubertal
timing in relation to delinquency. In light of research that ties both maltreatment and
pubertal timing to high levels of delinquency, the next step appears to be examination
of the interaction of these two variables. Investigation of the relationships between
biology and environment will provide a clearer understanding of how pubertal
development and maltreatment interact to affect behavior and adjustment. In addition,
identifying conditions which may heighten risk for delinquency or violent behavior
may serve to inform interventions targeting these behaviors.

13
Chapter 2: Methodological and Measurement Issues
The Measurement of Pubertal Development
Within the literature on pubertal development there is a great deal of
methodological inconsistency. More specifically, there is inconsistency in the
approaches that have been used to measure pubertal stage and pubertal timing. One
reason for this variation in measurement is that there is not a universally established
measure to assess pubertal stage, nor is there a universally established method to
assign pubertal timing.
Foremost, it is imperative to differentiate and define pubertal stage and
pubertal timing because they provide different types of information regarding the
relationships between pubertal development and adjustment difficulties. Pubertal stage
refers the level of pubertal development at a certain point in time, whereas pubertal
timing refers to whether an individual’s pubertal development occurs earlier than, later
than, or at the same time as most of his or her peers. The effects of pubertal stage and
pubertal timing will often be confounded in cross-sectional studies limited to one age
or one grade because pubertal timing is constructed from pubertal stage. Due to the
measurement of puberty at only one time point, it cannot be determined whether the
relationships found are due to a certain stage of maturity or to attaining that stage of
maturity at an earlier or later age than peers.
The most widely used methods for operationalizing pubertal development are
through the measurement of secondary sexual characteristics, such as breast
development, genital growth, or pubic hair growth, or through the measurement of

14
other markers of puberty, such as height spurt or increase in body mass. Age at
menarche has also been used extensively for females, but lacks a comparable marker
for males. Because a goal of the present research is to examine gender differences only
those measures which can be considered comparable for males and females will be
discussed. There are two specific measures that have been used extensively to rate the
physical changes associated with puberty, the Pubertal Development Scale (PDS;
Petersen, Crockett, Richards, & Boxer, 1988) and the Tanner stages (Marshall &
Tanner, 1969, 1970). Both of these measures are considered to yield comparable
assessments of pubertal stage for males and females.
The Pubertal Development Scale
The PDS assesses pubertal development in five different areas of physical
development (Petersen, et al., 1988). For females, ratings are completed for height
spurt, breast growth, body hair growth (underarm and pubic hair), skin changes, and
menarcheal status. For males, ratings are on height spurt, body hair growth (underarm
and pubic hair), skin changes, facial hair growth, and voice change. The benefit to this
measure is that that authors have attempted to create a comparable measure of pubertal
stage for both males and females. However, there is some disagreement as to the
reliability and validity. Although it has been shown to have adequate reliability and
validity by Petersen, et al. (1988), Brooks-Gunn and colleagues (1987) found that it
has poor validity when compared with physician ratings and concluded that it should
only be used as a rough estimate of pubertal stage when ratings from the Tanner stage
drawings are not available.

15
The Tanner Stages
Research conducted by Marshall and Tanner (1969, 1970) indicates five
stages of pubertal development for breast growth, testes and penis growth, and pubic
hair growth. Based on these findings, schematic drawings were developed depicting
the five stages of development to obtain a rating of pubertal stage (Morris & Udry,
1980). Using the drawings, the rater is asked to select which picture the target
individual is closest to in terms of their pubertal development. For females, drawings
of pubic hair growth and breast growth are used, whereas for males drawings of pubic
hair growth and penis and testes growth are used. This measure has been shown to
have high reliability and validity and has been used extensively (Marshall & Tanner,
1969, 1970).
Although the Tanner stages have been used in numerous studies, the
implementation has not been universal across studies. Specifically, some studies use
only the rating of breast growth and genital growth to assign pubertal stage, whereas
others use an average of the two ratings. Furthermore, while most ratings are
completed via child or adolescent self-report, parent and physician ratings are also
common. Inconsistency in the source of the rating makes it difficult to compare results
from different studies. According to Dorn, Susman, and Ponirakis (1999) the source of
the rating (self-report, parent-report, or physical exam) has differential predictive
value for different outcome variables. Specifically, the relationship between pubertal
timing and behavior problems was more likely to be significant when the rating of

16
pubertal stage was obtained by physical exam. Therefore it is essential to take into
account the source of the rating when interpreting research findings.
It is apparent from studies on self, parent, and physician ratings that the most
accurate ratings of Tanner stages are done by health professionals (Duke, Litt, &
Gross, 1980; Brooks-Gunn, Warren, Rosso, & Gargiulo, 1987; Dorn, Susman,
Nottelmann, Inoff-Germain, & Chrousos, 1990; Finklestein D’Arcangelo, Susman,
Chinchilli, Kunselman, Schwab, et al., 1999; Taylor, Whincup, Hindmarsh, Lampe,
Odoki, & Cook, 2001). However, the use health professionals is not always feasible
and therefore many researchers turn to self and parent-ratings as alternatives. While
physical examination by a health professional is the most precise, self-ratings have
been found to provide adequate reliability with a correlation of .82 between physician
Tanner ratings and self ratings. However, adolescents’ self-ratings are not devoid of
inadequacies. Adolescents of both genders have been found to underestimate their
pubertal stage in comparison to a physician rating. However, there is also evidence
that boys in early puberty tend to overestimate their stage of pubertal development
(Dorn et al., 1990). Both self and parent ratings have been found to be less accurate at
later stages of development compared to physicians’ ratings. Overall, parent-ratings
are considered the least accurate measure of pubertal stage, especially in the case of
parents reporting on sons (Brooks-Gunn et al., 1987; Dorn et al., 1990). With
limitations in all sources of ratings, researchers should be cognizant of these
shortcomings when conducting and interpreting research on pubertal development.

17
Puberty is a complex progression of multiple hormonal and physical
changes; therefore to adequately capture the spectrum of associated changes multiple
measures of these various developments should be employed. It is apparent that
obstacles arise when attempting to measure pubertal development using any one
pubertal change. Therefore, a goal of the present study was to investigate the use of
multiple measures of pubertal development in order to determine a combination which
may give a more complete representation of pubertal development.
Classifying Pubertal Timing
To reiterate, pubertal timing refers to whether an individual’s pubertal
development occurs earlier than, later than, or at the same time as most of his or her
peers. Pubertal timing has been operationalized as both a continuous and categorical
variable. For both the Tanner stages and the PDS, categorical and continuous methods
have been employed to obtain a measure of pubertal timing. Categorical pubertal
timing is the most widely used, however across studies diverse methods have been
implemented to assign timing groups. Pubertal timing groups have been created by
classifying the individuals one standard deviation above or below the sample mean as
the early or late timing groups respectively (e.g. Ge et al., 2003; Williams & Dunlop,
1999), or by using the extreme 20-30% of the sample distribution as the early and late
timing groups (e.g. Caspi et al., 1993; Obeidallah et al., 2004). Alternatively, national
norms have been used to determine age cut-offs to define the early timing group (onset
of puberty at 11 years or younger) and the late timing group (onset of puberty at 15
years or older) (e.g. Kaltiala-Heino et al, 2003; Angold, Costello, & Worthman, 1998).

18
Continuous pubertal timing has seen more limited use and has been constructed by
regressing pubertal stage on age, separated by gender (e.g. Dorn, Susman, et al., 1999)
or by standardizing pubertal stage scores by age group and gender (e.g. Ge et al.,
2001). Clearly there is an assortment of methods for classifying pubertal timing, but
not one that has been universally established.
There are strengths and limitations of both categorical and continuous
measures. With small sample sizes a categorical approach may not be sensitive enough
to adequately detect effects, whereas with larger sample sizes categories provide a
clear way to distinguish and compare timing groups. Because categorical timing
groups are much more prevalent in the existing research, the use of a continuous
timing variable may hinder the comparison of findings. However, the conversion of a
continuous variable using cut points to generate artificial groups often results in a loss
of information, unequal cell sizes, and less power to detect interaction effects (Frazier,
Tix, & Barron, 2004; Tabachnick & Fidell, 2001).
Cross-sectional versus Longitudinal Designs
As mentioned previously, pubertal stage and pubertal timing are confounded in
cross-sectional studies limited to one age. In such studies the effects of attaining a
specific stage of puberty are confounded with the effects of entering that stage at a
certain age. In order to examine pubertal timing effects cross-sectionally the sample
needs to include individuals of varying ages, particularly age ranges that will cover
many stages of pubertal development. Pubertal timing effects can then be investigated
by comparing individuals of different ages but of similar pubertal stage.

19
The present study employed a cross-sequential design. Each subject was
assessed at two occasions separated by approximately a one-year interval. The data at
each time-point is cross-sectional, consisting of 4 different ages, while the data across
time-points is longitudinal. Thus, the design of the present study provided a basis for
effectively examining the effects of pubertal timing as well as developmental
trajectories associated with their pubertal development and experience of
maltreatment.
The Measurement of Delinquency
Traditionally, measures of delinquency have been obtained from records
produced and maintained by the criminal justice system. However many researchers
have identified problems with reliance on official reports. For example, some offences
are never discovered, some if known about are not recorded, the offence categories are
very broad (e.g., robbery), and there is an over-representation of more serious
offences. In an effort to correct for some of these issues self-report delinquency
measures were developed as a more direct measure of problem behavior. The use of
official records versus self-report measures has been the focus of considerable debate.
A criticism of self-report measures is that they have low reliability and poor
validity, however more recent evidence shows they have adequate reliability and
validity across both ethnicity and gender (Huizinga & Elliot, 1986). Concerns also
arise regarding the truthfulness of reported behaviors, but evidence shows that
deliberate falsification of self-reports is a relatively rare event (Elliott, Huizinga, &
Ageton, 1985). Inaccuracy of reports on behaviors committed several years or more in

20
the past has been remedied by curtailed reporting periods (less than 12 months) and
the use of bounding techniques (Loftus & Marburger, 1983). The reporting of trivial
events has also been of some concern. These trivial events would not be considered
delinquent acts if reported to the police, but they are over-represented in self-report
measures and reported in higher frequency than more serious offences (Elliot &
Huizinga, 1983). Adding up frequency scores to create a “total delinquency” measure
gives greater weight to items with greater frequency, but these items are usually less
serious or minor (or trivial). As a result, offense specific measures, such as serious
assault, minor assault, serious property, minor property, public disorder, status
offenses are often used (D. Huizinga, personal communication, March 21, 2006).
Additionally, differentiating not only serious offences from non-serious but also
delinquent acts that are aggressive or violent from those that are nonviolent may be
useful in the exploration of the underlying mechanisms of delinquency as well as
controlling for possible confounds.
Many of the studies that trace the origins of delinquency have focused on
males rather than females. There is evidence that the pathways to delinquency and
violent behavior may be different between genders with females showing higher levels
related to abuse and neglect in the home (Rivera & Widom, 1990). In terms of gender
differences in delinquency, male adolescents have been found to be 1.5 times more
likely to commit a delinquent act than female adolescents (Davalos, Chavez, &
Guardiola, 2005). However, female prevalence may be underestimated due to reliance
on arrest records for prevalence rates. Female delinquency has largely been neglected

21
in the empirical literature due to the belief that males are more frequent
perpetrators. However, self-report techniques have indicated that the frequencies and
types of delinquent acts of females are closely related to those reported by males
(Gomme, Morton, & West, 1984). Therefore there is a necessity to further explore
delinquency by self-report for both males and females as the frequency of these
behaviors may have been underestimated previously due to reliance on arrest records
from police reports.
Studies indicate that the majority of boys who have severe and chronic
problems with delinquency in adolescence have behavioral precursors in childhood
whereas for most seriously delinquent girls these behaviors first emerge in
adolescence (Cohen, Cohen, Kasen, Velez, & Johnson, 1993). However, less serious
problem behavior has been found to appear at similar ages for both boys and girls
(Chamberlain & Moore, 2002). That is, the high male to female ratio of problem
behaviors diminishes as individuals move from childhood to adolescence.
Additionally, while male antisocial adolescents are more likely to engage in criminal
activities as adults, females are more at risk for an assortment of poor outcomes in
adulthood such as internalizing disorders and early pregnancy (Robbins & Price,
1991). Although the delinquency literature focuses more on the investigation of males,
it is apparent that associated risk factors also need to be examined for females. And as
females are less likely to enter into the juvenile justice system, self-report measures
may capture their patterns of delinquency more accurately.

22
Chapter 3: The Current Study
The entrance into puberty is associated with multiple physical, social, and
psychological changes. These changes and the stresses accompanying them can create
vulnerability for maladaptive behavior. Delinquency during adolescence is considered
a developmentally normative behavior that often diminishes in adulthood, but in some
cases it may indicate serious behavioral problems with long-term consequences.
Overall, the empirical literature supports the notion that early pubertal timing for both
genders is associated with increased delinquent behavior. However, alternate evidence
suggests an association between early timing and delinquency only for females, and
between early and late development and delinquent behavior for males. The research
on these relationships is dominated by the use of female samples, leaving a large gap
with respect to males. Therefore an aim of this study was to clarify the relationships
between pubertal timing and delinquent behavior for each gender. Specifically, if in
fact early pubertal timing alone is associated with delinquency, or if both early and
late timing are risk factors. This question needs to be clarified prior to investigation of
more complex relationships.
Similarly to early pubertal development, maltreatment has also been found to
increase the risk for delinquency and violent behavior. This association warrants
investigation into whether both early pubertal development and experience of
maltreatment interact to create a double vulnerability above and beyond the risk
conferred by each factor alone. A large void exists in the extant literature as to the
relationship between pubertal development, maltreatment, and delinquency. It seems

23
pertinent to examine this combination of variables in order to determine the
conditions which may substantially increase risk for delinquent behavior. The plan of
this study was to examine the interaction between maltreatment and pubertal timing to
determine whether maltreatment moderates the relationship between pubertal timing
and delinquency and if this interaction differs for males and females.
These questions were examined in an ethnically diverse sample of male and
female adolescents from a large urban area. Approximately two-thirds of the sample
has substantiated reports of maltreatment as determined by the Department of Children
and Family Services. The varied ethnic composition and even split between genders
provides an opportunity to supplement and expand the existing research to more
varied populations and clearly distinguish gender differences. There is a paucity of
ethnically diverse samples in the pubertal development literature and although ethnic
differences were not be examined per se, this diversity fulfilled significant gaps in
existing research.
The specific aims of the proposed study are:
1. To examine the relationship between pubertal timing and delinquency both cross-
sectionally and longitudinally across two time points and to determine whether
these relationships are different for males and females.
2. To determine whether early, late, or off-time (early and late) pubertal timing is
most highly related to delinquency and whether this association differs between
genders.

24
3. To determine whether maltreatment moderates the relationship between
pubertal timing and delinquency, and if the interaction effect differs between male
and female adolescents.

25
Chapter 4: Research Design and Methods
Participants
The participants for this study were 454 adolescents aged 9-13 enrolled in the
Young Adolescent Project (YAP), an ongoing NICHD funded longitudinal study on
the effects of maltreatment on adolescent development. The present study used data
from Time 1 and Time 2 of the larger YAP study. The first and second assessments
took place approximately 1 year apart. In order to maximize the data available for
analyses, participants who only participated at Time 1 as well as those who
participated at both assessments were included. Slight attrition occurred between Time
1 and Time 2 assessments, approximately 86% of the initial sample have completed
the second interview. Of the 62 participants who were seen at Time 1 but not at Time
2, 19 have refused to continue in the study, 12 have moved out of the country, and 31
are avoidant or difficult to schedule. The participants who only completed Time 1
were compared with those who completed both Time 1 and Time2 on various
characteristics (see Table 1). Tests of mean differences on these variables indicate that
the groups only differ significantly on their group status (maltreated versus
comparison) with 82.5 % of the maltreated group versus 94% of the comparison group
completing both time points.
Recruitment
The maltreated sample was recruited from active cases in the Los Angeles
County Department of Children and Family Services (LACDCFS). The inclusion
criteria were: (1) a new substantiated referral to LACDCFS in the preceding month for

26
any type of maltreatment; (2) child age of 9-12 years; (3) child identified as Latino,
African-American, or Caucasian (non-Latino); (4) child residing at the time of referral
to LACDCFS in one of the 10 zip codes in urban Los Angeles County areas.
Table 1.
Attrition Information

Only Completed
Time 1

Completed
Time 1 and 2
Total
N % N % N %
Group
Maltreated 53 17.5 250 82.5 303 100
Comparison 9 6 142 94 151 100

Gender
Male 36 14.9 206 85.1 242 100
Female 26 12.3 186 87.7 212 100

Ethnicity
African American 25 14.6 146 85.4 171 100
White 7 14 43 86 50 100
Latino 24 13.6 153 86.4 177 100
Biracial 6 10.7 50 89.3 56 100
Note. Only group difference was for maltreatment versus comparison ( t(452) =
3.41, p<.01)

With LADCFS and University Park Institutional Review Board (UPIRB) approval
potential participants were contacted via postcard and asked to indicate their
willingness to participate. Contact via mail was followed up by a phone call. A total of
303 maltreated children (and their primary caretakers) were enrolled in the YAP study.
According to information abstracted from case records, the children in the maltreated
group have experienced a variety of maltreatment experiences; most had multiple
referrals as well as multiple maltreatment types. The majority of the maltreatment

27
sample experienced neglect in some form, about half of the sample experienced
physical abuse and/or emotional abuse, and approximately one fourth experienced
sexual abuse. Overall this demonstrates diversity in maltreatment experience within
the maltreatment group.
The comparison group (N=151) was recruited using names from school lists
of children aged 9-12 years residing in the same 10 zip codes as the maltreated
sample. Caretakers of potential participants were sent a postcard and asked to indicate
their interest in participating which was followed up by a phone call.
Demographics
At Time 1 and Time 2 participants were compared on a variety of demographic
information. The data indicate that upon inclusion in the study both the maltreatment
and comparison groups were relatively similar in age, gender, and ethnic distributions,
but different in terms of living arrangements (Table 2). This is not unexpected as many
of the maltreated adolescents have been removed from biological parents and live in
foster homes. At Time 2 the demographic information changed slightly from Time 1,
but as mentioned, the Time 1 and Time 2 samples did not differ significantly on any
demographic variable. The difference in the proportions of maltreated children’s living
arrangement between Time 1 and Time 2 is indicative of attrition of the children living
in foster care. That is, it appears that the children in the maltreated group living with a
biological parent returned for Time 2 whereas the children living in foster care or with
a relative were less likely to return. Neighborhood characteristics of the maltreated and
comparison groups were examined for significant differences using the year 2000 US

28
Census. Of the 72 characteristics compared, nine significant differences emerged,
but none that were likely to produce an effect because of a relationship with another
variable. These results are indicative of very similar neighborhood environments for
both groups.
Table 2.
Sample Characteristics at Time 1 and Time 2
Demographic Variable Group
Maltreated Comparison
Time 1 Time 2 Time 1 Time 2
N 303 250 151 142
Age (std deviation) 10.84 (1.15) 12.02 (1.21) 11.11 (1.15) 12.28 (1.26)
Gender (%)
Male
Female

50
50

48
52

60
40

60
40
Ethnicity (%)
African American
Latino
White
Mixed Biracial

40
35
12
13

40
36
11
13

32
47
10
11

32
45
11
12
Living Arrangement (%)
With Parent
Foster Care/Extended
Family

52
48

63
37

93
7

94
6

Procedures
Currently, the YAP study consists of three times of assessment one year apart.
Time 1 has been completed and is comprised of individuals aged 9-13 years. At each
time of assessment the child is brought into the research offices by the caretaker. After
consent is obtained, the child is administered a host of questionnaires and activities
during a four-hour protocol including measurement of hormones via saliva, blood
pressure, height, weight, pubertal development, anxiety, aggression, depression,
delinquency, attachment, social networks, and community violence. The measures
used for the present study were gathered as a part of this protocol.

29
Measures
Pubertal Stage
Tanner Stages. The Tanner stages are represented by sets of serial line
drawings that depict the development of two different secondary sexual characteristics
(Morris & Udry, 1980). The pictures for female subjects are of breast development
and pubic hair growth, whereas the pictures for male subjects are of penis and testes
development and pubic hair growth. Each set comprises five pictures which range
from prepubertal to postpubertal development. After explanations of each picture in a
set are given, the individual indicates which picture most closely resembles their
current development. This rating produces a score from 1 to 5 with higher numbers
indicating more mature stages of development. The Tanner stages have shown high
validity and reliability (Marshall & Tanner, 1969, 1970). Self-report measures of the
Tanner stages have been found to be highly correlated with physician assessment and
are considered sufficient when rough estimation of pubertal stage is adequate (Dorn et
al., 1990). Scores on each drawing (breast/genital and pubic hair) were used as
separate indicators of pubertal development. Means, standard deviations, and ranges
for the present sample on the Tanner breast/genital and pubic hair stages can be found
in Tables 3 and 4.
Pubertal Development Scale. The PDS is a 4-item measure of physical
changes associated with pubertal development (height spurt, body hair, skin changes,
breast growth/deepening of voice). It was developed as an alternative to physician
rating measures and has shown to have adequate reliability and validity (Petersen, et

30
al., 1988). On a 4-point scale ranging from 1 (has not yet started) to 4 (has
completed) each subject is asked to indicate the level of development on each of the
physical changes. For the purpose of the present study the four items were averaged to
retain the original metric of the scale.

Table 3.
Descriptive Statistics for Tanner Stages, Pubertal Development Scale, and BMI
Time 1 Time 2
Males Females Males Females
Tanner breast/genital
N 235 211 205 186
Mean 2.27 1.85 2.85 2.35
Std Deviation 0.92 0.76 1.06 0.98
Range 1 to 5 1 to 5 1 to 5 1 to 5

Tanner pubic hair
N 236 211 205 186
Mean 2.26 1.99 2.88 2.69
Std Deviation 0.98 0.96 1.09 1.16
Range 1 to 5 1 to 5 1 to 5 1 to 5

Pubertal Development Scale
N 239 212 205 186
Mean 1.74 2.07 1.97 2.30
Std Deviation 0.51 0.72 0.55 0.67
Range 1 to 3.4 1 to 4 1 to 4 1 to 4
Cronbach's alpha 0.58 0.75 0.66 0.75

BMI
N 240 212 204 186
Mean 21.44 22.09 22.74 23.92
Std Deviation 5.05 5.12 6.38 5.77
Range
14.12 -
45.14
13.52 -
38.59
14.17 -
49.75
13.76 -
39.27

Body Mass Index. Body Mass Index (BMI) is an indicator of body composition that is
minimally invasive, comparable for both genders, and has been found to be correlated
with other measures of pubertal development (Frisch & Revelle, 1971). Early

31
maturing children of both sexes are both taller and heavier than their average- and
late-maturing peers (Malina & Bouchard, 1991). However it has been confounded
Table 4.
Means and Standard Deviations by Age and Gender for Tanner Stages, Pubertal Development
Scale, and BMI

Time 1 Time 2

Males
Mean (SD)
Females
Mean (SD)

Males
Mean (SD)
Females
Mean (SD)
Age 11.00 (1.16) 10.85 (1.16) 12.17 (1.22) 12.05 (1.22)
Tanner breast/genital
9 yrs 1.72 (.73) 1.26 (.45) -- --
10 yrs 1.85 (.77) 1.60 (.69) 2.05 (.72) 1.74 (.73)
11yrs 2.32 (.78) 1.87 (.71) 2.13 (.78) 1.79 (.85)
12 yrs 2.56 (.90) 2.15 (.58) 2.96 (.81) 2.36 (.80)
13 yrs 2.97 (.98) 2.80 (.70) 3.19 (.81) 2.84 (.95)
14 yrs -- -- 3.73 (1.43) 3.13 (.87)
15 yrs -- -- 3.50 (1.29) 3.20 (.84)

Tanner pubic hair
9 yrs 1.84 (.94) 1.19 (.48) -- --
10 yrs 1.80 (.79) 1.60 (.64) 2.05 (.84) 1.63 (.60)
11yrs 2.30 (.82) 1.97 (.89) 2.09 (.72) 2.13 (.96)
12 yrs 2.58 (.97) 2.58 (.87) 3.00 (.78) 2.76 (1.18)
13 yrs 2.97 (1.05) 3.20 (.89) 3.30 (.89) 3.19 (.86)
14 yrs -- -- 3.73 (1.40) 3.61 (.89)
15 yrs -- -- 3.75 (1.26) 4.00 (1.00)

Pubertal Development Scale
9 yrs 1.74 (.71) 1.53 (.47) -- --
10 yrs 1.70 (.55) 1.78 (.53) 1.66 (.40) 1.61 (.33)
11yrs 1.79 (.50) 2.04 (.63) 1.83 (.52) 2.02 (.64)
12 yrs 1.98 (.57) 2.45 (.65) 2.06 (.57) 2.38 (.62)
13 yrs 2.12 (.61) 2.99 (.70) 2.29 (.56) 2.75 (.50)
14 yrs -- -- 2.38 (.54) 2.82 (.47)
15 yrs -- -- 2.69 (.51) 2.90 (.29)

BMI
9 yrs 19.54 (3.46) 19.37 (4.84) -- --
10 yrs 20.08 (4.04) 20.29 (4.33) 19.65 (3.82) 21.68 (5.12)
11yrs 21.38 (5.27) 23.60 (5.15) 21.54 (4.61) 21.80 (5.13)
12 yrs 23.11 (5.81) 22.94 (4.09 22.25 (6.24) 25.17 (6.15)
13 yrs 22.81 (4.89) 24.89 (6.33) 23.47 (5.75) 24.10 (4.18)
14 yrs -- -- 24.82 (4.41) 26.18 (6.93)
15 yrs -- -- 22.75 (5.99) 26.47 (4.87)

32
with obesity. Body Mass Index norms for children and adolescents established by
the Centers for Disease Control indicate that a BMI less than the 5
th
percentile for age
and sex is considered underweight, between the 5
th
and 85
th
percentile is considered a
healthy weight, between the 85
th
and 95
th
percentile is at risk for being overweight,
and the 95
th
percentile or greater is considered overweight (CDC, 2006). For the
present study BMI was computed from the height and weight measurements obtained
during the health assessment portion of the interview. Two measurements of height
and three measurements of weight were averaged and used to compute BMI. The
formula for BMI is: BMI = ( Weight in Pounds / ( Height in inches ) x ( Height in
inches ) ) x 703.
Pubertal Timing
When the degree of physical development is standardized within same-age
peers, the resulting score can be used as an index of pubertal timing (Ge et al., 2001).
For the present study the scores on each of the Tanner Stage ratings (breast/genital and
pubic hair), the average of the PDS scores, and BMI were standardized within each
age (9, 10, 11, 12, 13) and gender (males and females). The resulting variable of
pubertal timing for each measure had a mean of 0 and a standard deviation of 1 with
higher scores indicating earlier maturation relative to peers. This continuous pubertal
timing variable was computed separately for Time 1 and Time 2 pubertal
development.
For substantive analyses this continuous method for measuring pubertal timing
was used rather than a categorical measure. Based on the multiple regression

33
techniques that were employed to test the specific aims, it has been found to be best
to retain the original continuous nature of the variables so that a loss of information
and power does not occur (Frazier et al., 2004). Simulation studies have shown that
hierarchical multiple regression procedures that retain the true nature of the continuous
variable result in fewer Type I and Type II errors for detecting moderator effects than
methods that use cut-points to create categorical groups (Bissonnette, Ickes, Bernstein,
& Knowles, 1990; Mason, Tu, & Cauce, 1996).
Delinquency
Adolescent Delinquency Questionnaire. Delinquency was measured via a self-
report computerized questionnaire: the Adolescent Delinquency Questionnaire (ADQ;
adapted from Huizinga & Morse, 1986). Self-report delinquency measures such as the
ADQ have been shown to have adequate test-retest reliability and validity compared to
official records. The ADQ is a 60-item measure that assesses the frequency of
delinquent acts, drug use, and substance use for the past 12 months and lifetime. For
the present study the 23 items assessing delinquency in the past 12 months were used.
Research has found that self-report delinquency measures are more accurate when the
reporting period is curtailed to past 12 months (Moffitt, 1996). The ADQ assesses
multiple types of delinquency and scales have been formulated in previous studies to
measure person offenses, property offenses, status offenses, nonviolent delinquency,
violent delinquency, general delinquency, and substance abuse. Psychometric analyses
were conducted to determine the scales to be used in the current study. The results of
these analyses can be found in the preliminary results section.

34
Chapter 5: Results
Preliminary Analyses
Missing Data Imputation
In order to obtain the maximum amount of data for the ADQ, missing data
were imputed. Data was imputed at the first order which is used for item level
missingness, instances in which a particular item on a measure was not answered. The
NORM (Schafer, 1999) software program was used to impute missing data using an
Estimation-Maximization (EM) algorithm (Dempster, Lair, & Rubin, 1977; Orchard &
Woodbury, 1972). Hence, missing data was replaced with a “best guess” using
multiple imputations that converge over iterations to a minimum-error estimation of
the missing data points. Missingness percentages show no more than 5% missingness
for any one case on the ADQ. Those cases missing more than 5% were omitted from
analyses. This resulted in 443 complete cases for Time 1 ADQ and 391 for Time 2.
For the substantive analyses Full Information Maximum Likelihood was
employed to handle variable level missing data. This method fits the model to the raw
data rather than the covariance matrix thus allowing the use of all available data
(Loehlin, 2004).
Tests for Normality
Prior to entering the variables into the substantive analyses, variable
distributions were tested for univariate normality. The delinquency items were found
to be skewed and were subsequently transformed using a square-root transformation
plus a constant (Tabachnick & Fidell, 2001).

35
Psychometric Analyses
Factor Structure of the Adolescent Delinquency Questionnaire. Based on the
constructs hypothesized by Huizinga and Morse (1986) a three factor structure was
expected to provide the best measurement of the delinquency constructs for both
Timepoints. A restricted factor analyses was conducted to determine if the three
hypothesized scales provided the best measurable and reliable scales that can be
obtained from the items of the ADQ. The factor structure was tested using Structural
Equation Modeling (SEM) with Amos 5.0 (Arbuckle, 2003). Fit indices such as the χ
2
(chi-square) goodness-of-fit statistic and the root mean square error of approximation
(RMSEA) were used to evaluate the fit of the model to the data. Overall, a small χ
2
is
optimal and the RMSEA should be .08 or smaller when the model fit is good (Shapiro
& Levendosky, 1999; Browne & Cudeck, 1993).
Each item was specified to load on one and only one factor. The first item on
each latent factor was set to 1.0 to establish the metric. Due to the non-normal
distribution of the ADQ items, the Generalized Least Squares (GLS) estimation
method was used to obtain a solution. In studies with non-normal data GLS has been
found to provide more robust estimates of fit and parameter estimates (Byrne, 2001).
The three factor solution was found to have an adequate fit to the data for both Time 1
( χ
2
= 863.83 (227); RMSEA = .08) and Time 2 data ( χ
2
= 674.27 (227); RMSEA =
.07). The internal consistency reliability of each scale was assessed using coefficient
alpha (Guttman, 1945; Cronbach, 1951). The items, scales, factor loadings,
intercorrelations, and descriptive statistics for the ADQ data are in Tables 5, 6, and 7.

36
Table 6.
ADQ Factor Intercorrelations and Internal Consistency Reliabilities
Time 1 Time 2
Status Offences
Person
Offences
Property
Offences
Status
Offences
Person
Offences
Property
Offences
Status Offences .72** .60
Person Offences .76** .82** .58** .70
Property Offences .75** .81** .92 .63** .74** .83
Note. Cronbach’s alpha reliabilities are in the diagonals
Table 5.
Items and Scales of the Adolescent Delinquency Questionnaire
Item Factor Loading
Time 1 Time 2
Status Offences
Run away from home .47 .64
Skipped classes without an excuse .64 .56
Been suspended or sent home from school .48 .38
Lied about your age to get in someplace or buy something .70 .65
Stayed out on purpose past curfew .78 .62
Cheated on schoolwork .63 .36
Person Offences
Carried a hidden weapon? .63 .64
Made dirty crank phone calls .77 .52
Attacked someone with a weapon .79 .73
Hit someone with the idea of hurting them .72 .44
Used a weapon or force to get money or things from people .85 .83
Thrown things, like rocks or bottles, at people .67 .61
Gotten in physical fights .61 .57
Property Offences
Damaged or destroyed someone else’s property on purpose .80 .59
Set fire, or tried to set fire to a house, building, car, or other property .80 .51
Did things that you were supposed to pay for without paying .73 .70
Gone into or tried to go into a building to steal something .84 .77
Stolen or tried to steal money or things worth $20 or less .77 .78
Stolen or tried to steal money or things worth between $20 and $300 .84 .77
Snatched or grabbed someone’s purse or wallet or picked someone’s
pocket .83 .77
Taken something from a car that did not belong to you .89 .59
Knowingly bought, sold or held stolen goods or tried to do these things .79 .75
Tried to cheat someone by selling them something that was worthless .71 .62
Model fit: Time 1 χ2 =863.83 (227); RMSEA=.08; Time 2 χ
2
=674.27 (227), RMSEA=.07
Note. p<.01 for all loadings

37

A composite score was formed from sum of the items on each scale, these
scales were then tested for invariance between gender and between maltreatment and
comparison groups. That is, a latent factor of delinquency was tested with the three
delinquency scales (status offences, person offences, property offences) as manifest
indicators of the latent variable to determine if these three scales measured the same
latent construct in both males and females and both maltreatment and comparison
groups.
Evidence for invariance between genders was obtained using multiple group
structural equation modeling (MGSEM). First an unrestricted model was tested in
Table 7.
Descriptive Statistics for Delinquency Scales
Time 1 Time 2
Raw Transformed Raw Transformed
Status Offences
N 443 443 391 391
Mean 1.84 4.24 2.46 4.5
Std Deviation 3.47 2.06 3.79 2.04
Range 0-25 3-16.42 0-30 3-16.42
Cronbach's alpha .72 -- .60 --

Person Offences
N 443 443 391 391
Mean 1.44 4.39 1.23 4.27
Std Deviation 4.02 2.10 3.11 1.73
Range 0-35 3.5-19.15 0-25 3.5-16.53
Cronbach's alpha .82 -- .70 --

Property Offences
N 443 443 391 391
Mean 1.60 5.96 1.31 5.83
Std Deviation 5.59 2.90 4.04 2.20
Range 0-50 5-27.36 0-36 5-21.65
Cronbach's alpha .92 -- .83 --
Note. Delinquency scales transformed via square root + constant

38
which all measurement weights (factor loadings) were allowed to be estimated
separately for each gender. The unconstrained model was fully saturated, therefore
showing a perfect fit to the data. Next, measurement weights were restricted to be
equal across groups. A nonsignificant χ
2
difference between these two models
indicates invariance of measurement. For Time 1 data the χ
2
difference was not
significant ( Δχ
2
=.14 (2), p>.05), however for Time 2 data the difference was
significant ( Δχ
2
=10.43 (2), p<.01). These results support measurement invariance for
Time 1 delinquency scales but not Time 2. Because full measurement invariance was
not obtained, partial measurement invariance at Time 2 was tested. This was executed
by constraining one measurement weight at a time to be equal across groups to
examine the potential invariance of each specific parameter (Byrne, Shavelson, &
Muthén, 1989). The results of this analysis indicated that for Time 2 the property
measurement weight was invariant between genders as shown by a nonsignificant χ
2
difference ( Δχ
2
= 1.41 (1), p>.05). The results of the invariance analyses support
measurement constraints in subsequent structural models for all measurement weights
on delinquency for Time 1 and for the property measurement weight at Time 2.
The same procedure was used to test for measurement invariance between
maltreatment and comparison groups. The χ
2
difference between the unrestricted
model and the measurement weights restricted model was significant for the Time 1
data ( Δχ
2
=15.22 (2), p<.01) but not significant for the Time 2 data ( Δχ
2
=2.05 (2),
p>.05). Because full measurement invariance was not obtained for the Time 1 data
partial measurement invariance was then tested. The results of these analyses showed

39
that the person offences measurement weight was invariant between groups as
evidenced by a non significant χ
2
difference between the unrestricted model and
model in which the person offences measurement weight was constrained to be equal
across groups ( Δχ
2
=.34 (1), p>.05). These results support the use of equality
constraints for the person offence measurement weight for Time 1 data and for all
factors for Time 2 data when testing multiple group models between maltreatment and
comparison groups.
Construction of a Latent variable of pubertal timing. The measurement of
puberty was also examined to determine the most appropriate and reliable measure for
subsequent analyses. A restricted factor analysis was conducted using Maximum
Likelihood Estimation to determine if a common factor model for pubertal timing
could be constructed. A latent variable of pubertal timing was formed using four
manifest indicators: the standardized scores (within age and gender) of the Tanner
breast/genital stage, the Tanner pubic hair stage, the PDS, and BMI. Additionally this
latent variable model was tested without the inclusion of BMI. These two models were
first run using the total sample. The four variable model for pubertal timing showed an
adequate fit to the data, however BMI had a very low factor loading (Table 8). The
three variable model was fully saturated and thus a perfect fit to the data (Table 9).
Both models were then tested for invariance between males and females by the
same procedures as described for the delinquency analyses. For the Time 1 data the
four variable model showed a significant χ
2
difference between the unrestricted and
the measurement weights restricted models ( Δχ
2
=16.51(3), p<.001), the three variable

40
unrestricted and restricted models were also significantly different from each other,
but the difference was much smaller ( Δχ
2
=6.37 (2), p<.05).
Table 8. Table 9.
Factor Loadings for Four Factor Model of
Pubertal Timing
Factor Loadings for Three Factor Model of
Pubertal Timing
Factor Loading Factor Loading
Time 1 Time 2 Time 1 Time 2
Tanner breast/genital .79 .82 Tanner breast/genital .79 .82
Tanner pubic hair .80 .84 Tanner pubic hair .80 .84
PDS .43 .55 PDS .43 .55
BMI .17 .14 Model fit: Time 1 χ
2
= 0, RMSEA=.00;
Model fit: Time 1 χ
2
= 2.75(2), RMSEA=.03; Time 2 χ2 = 0, RMSEA=.00
Time 2 χ2 = 1.06(2), RMSEA=.00 Note. For all loadings p<.05
Note. For all loadings p<.05

For Time 2 data both models showed full measurement invariance. These invariance
analyses indicate that with the inclusion of BMI the model is further away from
measuring the construct equivalently in both groups. This result, along with the
superior fit of the three variable model and the nonsignificant correlations with the
ADQ scales provided support for the exclusion of BMI. The resulting three variable
model was tested for partial measurement invariance for Time 1 pubertal timing.
These analyses showed the Tanner breast/genital measurement weight to be invariant
between genders. Therefore, for the analyses regarding the structural paths, all
measurement weights were constrained for Time 2 puberty data and the Tanner
breast/genital measurement weight was constrained for Time 1 data.
Additionally, measurement invariance was tested for the three variable model
between maltreatment and comparison groups by the same procedure. The χ
2
difference between the unrestricted model and the measurement weights restricted

41
model was significant for both Time 1 ( Δχ
2
=10.61 (2), p<.01) and Time 2 data
( Δχ
2
=4.88 (2), p>.05). Therefore partial measurement invariance was examined for
both time points. The results of these analyses showed that the PDS measurement
weight was invariant for both Time 1 ( Δχ
2
=.00 (1), p>.05) and Time 2 (Δχ
2
=2.56 (2),
p>.05) across maltreatment status. The results support the use of equality constraints
for the PDS measurement weight for both Time 1 and Time 2 data.
Bivariate Relationships
Intercorrelations were computed between the variables of interest (Table 10).
Significant correlations were found among all the pubertal timing variables for Time 1
data (r = .10 to .13, p<.05) but only between Tanner breast/genital, Tanner pubic hair,
and PDS for Time 2 data. The three delinquency scales were found to have significant
intercorrelations for both Time 1 (r=.75 to .81, p<.01) and Time 2 data (r=.58 to .74,
p<.01). There were no significant correlations between BMI and any of the
delinquency scales for either Time 1 or Time 2.
Substantive Analyses
Specific Aim 1
To examine the relationship between pubertal timing and delinquency both cross-
sectionally and longitudinally across two time points and to determine whether these
relationships are different for males and females.
All analyses were conducted using structural equation modeling with
Maximum Likelihood Estimation in Amos 5.0. Foremost, good fitting models for the
total sample were obtained in order to specify baseline models. To examine the cross-

42
Table 10.
Correlations between Time 1 and Time 2 Variables

T1
TBG
T1
TPH
T1
PDS
T1
BMI
T1
Status
Offences
T1
Person
Offences
T1
Property
Offences
T2
TBG
T2
TPH
T2
PDS
T2
BMI
T2
Status
Offences
T2
Person
Offences
T1 TBG 1.00*
T1TPH .63** 1.00*
T1 PDS .35** .34** 1.00*
T1 BMI .10** .14** .13** 1.00*
T1 Status
Offences .17** .15** .15** .03** 1.00**
T1 Person
Offences .07** .09** .12** .01** .76** 1.00**
T1 Property
Offences .06** .05** .10** .01** .75** .81** 1.00**
T2 TBG .44** .38** .30** .14** .12** .07** .05** 1.00*
T2 TPH .39** .43** 30** .13** .10** .06** .01** .68** 1.00*
T2 PDS .29** .32** 40** .08** .12** .02** -.03** .45** .46** 1.00*
T2 BMI .14** .15** .15** .91** .13** .09** .07** .13** .10** .06** 1.00*
T2 Status
Offences .12** .16** .16** .03** .41** .32** .23** .24** .19** .24** .05 1.00*
T2 Person
Offences .11** .11** .11** .07** .43** .46** .30** .16** .13** .18** .05 .58** 1.00*
T2 Property
Offences .04** .03** .03** .04** .37** .38** .27** .13** .07** .12** .03 .63** .74**
*p<.05, **p<.01
Note. TBG= Tanner breast/genital stages; TPH=Tanner pubic hair stages; PDS=Pubertal Development Scale; BMI= Basal Metabolic Index;
all have been standardized within age and gender to obtain indices of pubertal timing

43
sectional relationships, two separate models were fit; one in which the measures of
pubertal timing and delinquency were both from Time 1 and a second in which the
measures of pubertal timing and delinquency were both from Time 2 (example of
these models in Figure 1). To test for the longitudinal effect, the measure of pubertal
timing was taken from Time 1, while the measure of delinquency was taken from
Time 2. In addition, the cross-lagged relationships were examined using the model in
Figure 2. Maltreatment status was included to account for the variance before
examining the relationship between pubertal timing and delinquency variable in each
model. The fit of each model was evaluated by the goodness fit statistics mentioned
previously. The goodness of fit statistics for these models are in Table 11 and the
parameter estimates are in Tables 12 and 13.

Figure 1. Testing the linear relationship between the pubertal timing and delinquency
Tanner
breast &
genital
Status
Person
Property
Delinquency
Pubertal
timing
Tanner
pubic hair
PDS
Maltreatment
status

44

Figure 2. Testing cross-lag relationships between pubertal timing and delinquency
*Note: Manifest variables have been omitted for simplicity
Table 11.
Fit Statistics for Models Testing the Relationship between Pubertal Timing and Delinquency

χ
2
df RMSEA Δχ
2
Δdf p
T1puberty-T1delinquency
Total sample 23.90 12 .05
MG-unconstrained 29.70 24 .02
MG-measurement weights constrained 36.16 28 .03 6.46 4 ns
MG-structural weights constrained 45.35 30 .04 9.19 2 .01
MG-puberty to delinquency constrained 38.72 29 .03 2.56 1 ns
MG-maltreatment to delinquency constrained 42.34 29 .03 6.17 1 .05

T2puberty-T2delinquency
Total sample 31.12 12 .06
MG-unconstrained 49.36 24 .05
MG-measurement weights constrained 53.41 27 .05 4.05 3 ns
MG-structural weights constrained 55.20 29 .05 1.79 2 ns

T1puberty-T2delinquency
Total sample 16.69 12 .03
MG-unconstrained 28.02 24 .02
MG-measurement weights constrained 28.73 26 .02 .68 2 ns
MG-structural weights constrained 30.46 28 .01 1.73 2 ns

Cross lag model
Total sample 171.04 59 .07
Note. MG=multiple groups analysis
T2 Pubertal
timing
T2Delinquency
T1Pubertal
timing
T1Delinquency
Maltreatment
status

45

Table 12.
Measurement Model Estimates for Models Testing the Relationship between Pubertal Timing and
Delinquency

Unstandardized estimates (SE) Standardized estimates

Total
sample
Males Females
Total
sample
Males Females
T1pubertal
timing →
T1TannerBG 1.01 (.12) .99 (.14) 1.06 (.18) .80** .85** .75**
T1pubertal
timing →
T1TannerPH 1.00 1.00 1.00 .79** .86**
a
.71**
b

T1pubertal
timing →
T1PDS .55 (.07) .41 (.09) .76 (.13) .43** .35**
a
.54**
b

T1delinquency →
T1status offences .90 (.04) .92 (.05) .89 (.07) .84** .88** .75**
T1delinquency →
T1person
offences 1.00 1.00 1.00 .91** .90** .93**
T1delinquency →
T1property
offences 1.35 (.05) 1.35 (.07) 1.37 (.09) .89** .91** .85**
T2pubertal
timing →
T2TannerBG .99 (.09) .98 (.08) 1.01 (.17) .82** .90** .72**
T2pubertal
timing →
T2TannerPH 1.00 1.00 1.00 .83** .92** .71**
T2pubertal
timing →
T2PDS .67 (.07) .58 (.07) .84 (.14) .56** .54** .60**
T2delinquency →
T2status offences 1.02 (.07) .88 (.08) 1.34 (.14) .72** .69**
a
.75**
b

T2delinquency →
T2person
offences 1.00 1.00 1.00 .83** .87**
a
.75**
b

T2delinquency →
T2property
offences 1.37 (.08) 1.23 (.10) 1.66 (.17) .89** .88** .89**
*p<.05, **p<.01,
†
p<.10
Note. All parameter estimates are from unconstrained model
a, b
: parameter estimates with different subscripts are significantly different from each other based
on χ
2
diff

46
Table 13.
Structural Model Estimates for Models Testing the Relationship between Pubertal Timing and
Delinquency
Unstandardized estimates (SE) Standardized estimates

Total
sample
Males Females
Total
sample
Males Females
Time 1 Model
T1pubertal timing →
T1delinquency .32 (.14) .44 (.19) .05 (.18) .13* .17** .02
Maltreatment status →
T1delinquency .40 (.20) .87 (.31) -.10 (.23) .10* .19**
a
-.03
b

Time 2 Model
T2pubertal timing →
T2delinquency .36 (.11) .32 (.15) .34 (.14) .20** .17** .23*
Maltreatment status →
T2delinquency .26 (.16) .55 (.25) .07 (.18) .09 .16** .03
Time 1 and Time 2 Model
T1pubertal timing →
T2delinquency .21 (.11) .23 (.15) .17 (.18) .12
†
.13** .09
Maltreatment status →
T2delinquency .24 (.16) .46 (.23) .07 (.24) .08* .15** .02
Cross lag Model
T1pubertal timing →
T1delinquency .31 (.14) -- -- .13** -- --
T1pubertal timing →
T2delinquency -.25 (.16) -- -- -.13** -- --
T1pubertal timing →
T2pubertal timing .69 (.07) -- -- .67** -- --
T2pubertal timing →
T2delinquency .47 (.15) -- -- .25** -- --
T1delinquency →
T2delinquency .42 (.04) -- -- .53** -- --
Maltreatment status →
T1delinquency .34 (.21) -- -- .09** -- --
Maltreatment status →
T2delinquency .16 (.15) -- -- .05** -- --
*p<.05, **p<.01,
†
p<.10
Note. All parameter estimates are from unconstrained model
a, b
: parameter estimates with different subscripts are significantly different from each other
based on χ
2
diff

The Time 1 model showed a good fit to the data ( χ
2
= 23.90 (12),
RMSEA=.05). Parameter estimates indicated a significant relationship between Time

47
1 pubertal timing and Time 1 delinquency ( β=.13, p<.05) and between maltreatment
status and Time 1 delinquency ( β=.10, p<.05). The direction of the regression
coefficients indicates that earlier pubertal timing and maltreatment are related to
higher levels of delinquency. The main effect of pubertal timing on delinquency is
after accounting for maltreatment experience. The Time 2 model also showed a good
fit to the data ( χ
2
=31.12 (12), RMSEA=.06) and there was a significant relationship
between Time 2 pubertal timing and Time 2 delinquency ( β=.20, p<.01). The model
testing Time 1 pubertal timing and Time 2 delinquency showed a good fit as well
( χ
2
=16.69 (12), RMSEA=.03) with a marginally significant relationship between Time
1 pubertal timing and Time 2 delinquency ( β=.12, p<.10). The cross lag model fit the
data adequately ( χ
2
=171.04 (159), RMSEA=.07) and showed significant relationships
between Time 1 pubertal timing and Time 1 delinquency ( β=.13, p<.05), between
Time 2 pubertal timing and Time 2 delinquency ( β=.25, p<.01), between Time 1 and
Time 2 pubertal timing ( β=.67, p<.01), and between Time 1 and Time 2 delinquency
( β=.53, p<.01). The marginally significant relationship found in the previous model
between Time 1 pubertal timing and Time 2 delinquency disappeared after controlling
for Time 2 pubertal timing and Time 1 delinquency.
To examine gender differences in these relationships multiple group SEM was
used. Each of the models described above (except the cross lag due to limitations of
sample size) was run simultaneously for males and females. First, an unconstrained
model was run for both groups, this model provided a basis from which to compare
more restricted models. Next, the measurement weights were constrained to be equal

48
across groups as specified by the measurement model analyses. Lastly, structural
weights were constrained to be equal across groups. The change in the χ
2
provided
evidence as to whether parameters across groups were significantly different from
each other.
For the Time 1 data the unrestricted model showed adequate fit to both groups
( χ
2
=29.70 (24), RMSEA=.02). When the measurement weights were constrained to be
equal across groups there was not a significant decrement in the model fit ( Δχ
2
=6.46
(4), p>.05) indicating measurement invariance of both latent variables. The χ
2
difference between the measurement weights restricted model and the structural
weights restricted was significant (Δχ
2
=9.19 (2), p<.01) indicating that at least one of
the parameters was moderated by gender. Therefore, each structural parameter was
tested in turn to determine whether one or more were moderated by gender. A
significant decrement in the model constraining a specific parameter to be equal across
groups indicated an interaction effect. The χ
2
difference test showed that the parameter
between maltreatment status and Time 1 delinquency was significantly different for
each gender (Δχ
2
=6.17 (1), p<.05) (males β=.19, p<.01; females β=-.03, ns). The
direction of the regression coefficient indicates that maltreatment is more important in
predicting delinquency for males than females. The parameter between Time 1
pubertal timing and Time 1 delinquency was not significantly different between
genders ( Δχ
2
=2.56 (1), p>.05). Although not significantly different from each other,
the parameter estimate was statistically significant for males ( β=.17, p<.05) but not for
females ( β=.02, ns).

49
Results for the Time 2 data indicated a good fit to the data for both groups
when they were freely estimated ( χ
2
=49.36 (24), RMSEA=.05). Constraining the
measurement weights to be equal across groups did not significantly alter the fit from
the unrestricted model ( Δχ
2
=4.05 (3), ns), and restricting the structural weights did not
significantly alter the fit from the measurement model ( Δχ
2
=1.79 (1), ns). These
results indicate gender does not moderate either of the structural relationships.
Although the parameter estimates were not significantly different from each other, the
regression coefficients reached significance in one or both groups. As can be seen in
Table 10, Time 2 pubertal timing and Time 2 delinquency were significantly related
in both males ( β=.17, p<.05) and females ( β=.23, p<.05) indicating earlier pubertal
timing is associated with higher delinquency. However, maltreatment status and
delinquency were significantly related only in males ( β=.16, p<.05).
The unrestricted model for Time 1 pubertal timing and Time 2 delinquency
showed a good fit to the data ( χ
2
=28.02 (24), RMSEA=.02). When the measurement
weights were constrained to be equal across groups there was not a significant
decrement in the model ( Δχ
2
=.68 (2), p>.05), nor was there when the structural
weights were constrained ( Δχ
2
=1.73 (2), p>.05). Similar to the Time 2 model, these
results indicate that gender does not moderate any of the structural relationships in the
model. However, maltreatment status and Time 2 delinquency were significantly
related for males ( β=.15, p<.05) and not for females although the two groups were not
significantly different from each other. Time 1 pubertal timing and Time 2
delinquency were not significantly related in either gender.

50
Specific Aim 2
To determine whether early, late, or off-time (early and late) pubertal timing is most
highly related to delinquency and whether this association differs between genders.
A quadratic term was added to the model for the next set of analyses.
Interaction and higher order terms can be included as latent variables, however there
are complexities in the computation and interpretation when using latent product
terms. Therefore, although the latent variable of pubertal timing was used in the
previous aim and found to have good measurement properties, to complete these
analyses a composite variable was formed by averaging the Tanner breast/genital,
Tanner pubic hair, and PDS scores. This composite of pubertal timing was squared to
form the quadratic term. First the analyses were conducted with the total sample.
Similarly to the previous set of analyses three main models were tested. One cross-
sectional model included only Time 1 data, the next included only Time 2 data, and
the longitudinal model used Time 1 pubertal timing and Time 2 delinquency. In all
models pubertal timing, pubertal timing
2
, and maltreatment status were modeled as a
manifest variables with direct effects on delinquency. Covariances were modeled
between pubertal timing and its squared product, between pubertal timing and
maltreatment status, and between pubertal timing
2
and maltreatment status. The
baseline model is shown in Figure 3. Fit statistics can be found in Table 14 and
parameter estimates in Table 15.

51

Figure 3. Testing a quadratic relationship between pubertal timing and delinquency

Table 14.
Fit Statistics for Models Testing Quadratic Relationship between Pubertal Timing and
Delinquency

χ
2
df RMSEA Δχ
2
Δdf p
T1puberty-T1delinquency
Total sample 17.92 6 .07
MG-unconstrained 22.61 12 .04
MG-measurement weights constrained 22.84 14 .04 .23 2 ns
MG-structural weights constrained 33.62 17 .05 10.77 5 .05
MG-puberty to delinquency
constrained 25.30 15 .04 2.46 1 ns
MG-maltreatment to delinquency
constrained 29.16 15 .05 6.31 1 .05
MG-puberty
2
to delinquency
constrained

23.15 15 .04 .31 1 ns

T2puberty-T2delinquency
Total sample 19.90 6 .08
MG-unconstrained 32.86 12 .07
MG-measurement weights constrained 33.65 13 .06 .79 1 ns
MG-structural weights constrained 35.95 16 .06 2.31 3 ns

T1puberty-T2delinquency
Total sample 11.39 6 .05
MG-unconstrained 15.97 12 .03
MG-measurement weights constrained 16.47 13 .02 .50 1 ns
MG-structural weights constrained 19.52 16 .02 3.05 3 ns
Note. MG=multiple groups analysis
Delinquency
Pubertal timing
Pubertal timing
2
Maltreatment
status

52
Table 15.
Parameter Estimates for Models Testing Quadratic Relationship between Pubertal Timing and Delinquency
Unstandardized estimates (SE) Standardized estimates
Total sample Males Females Total sample Males Females
Time 1 Model
T1pubertal timing
2
→T1delinquency -.08 (.07) -.02 (.02) -.10 (.08) -.05 -.01 -.10
T1pubertal timing →T1delinquency .28 (.10) .40 (.15) .09 (.12) .14** .18** .06
maltreatment status →T1delinquency .42 (.20) .88 (.31) -.09 (.23) .10* .19**
a
-.03
b

T1pubertal timing ↔T1pubertal timing
2
.49 (.07) .37 (.09) .63 (.11) .36** .28** .44**
T1pubertal timing ↔maltreatment status .04 (.03) .06 (.03) -.02 (.03) .06 .12 -.04

Time 2 Model
T2pubertal timing
2
→T2delinquency -.01 (.05) .04 (.09) -.10 (.08) .01 .03 -.10
T2pubertal timing →T2delinquency .28 (.08) .28 (.11) .29 (.12) .18* .18* .20*
maltreatment status →T2delinquency .26 (.15) .54 (.25) -.09 (.23) .10* .15* -.03
T2pubertal timing ↔T2pubertal timing
2
.37 (.08) .21 (.10) .54 (.12) .25** .15* .36**
T2pubertal timing ↔maltreatment status .00 (.04) -.02 (.03) -.02 (.03) .05 .03 .01

Time 1 and Time 2 Model
T1pubertal timing
2
→T2delinquency -.05 (.09) .01 (.09) -.11 (.08) -.04 .01 -.12
T1pubertal timing →T2delinquency .17 (.08) .18 (.12) .17 (.12) .12* .12 .12
maltreatment status →T2delinquency .24 (.16) .46 (.23) .07 (.24) .08 .15* .02
T1pubertal timing ↔T1pubertal timing
2
.49 (.07) .37 (.09) .63 (.11) .36** .28** .44**
T1pubertal timing ↔maltreatment status .02 (.02) .06 (.03) -.02 (.03) .06 .12 .00
*p<.05, **p<.01 Note. All parameter estimates are from unconstrained model;
a, b
: parameter estimates with different subscripts are significantly different from each other based on χ
2
diff

53
The first model which tested the cross-sectional relationship between Time 1
pubertal timing and Time 1 delinquency showed an adequate fit to the data ( χ
2
=17.92
(6), RMSEA=.07). As found previously in the models using the latent variable of
pubertal timing, there was a significant relationship between Time 1 pubertal timing
and Time 1 delinquency ( β=.14, p<.01) and between maltreatment status and Time 1
delinquency ( β=.10, p<.05). The Time 2 model also showed an adequate fit to the data
( χ
2
=19.90 (6), RMSEA=.08). Significant relationships emerged between Time 2
pubertal timing and Time 2 delinquency (β=.19, p<.01) and between maltreatment
status and Time 2 delinquency ( β=.10, p<.05). The longitudinal model fit the data well
( χ
2
=11.39 (6), RMSEA=.05), however the only significant relationship was between
Time 1 pubertal timing and Time 2 delinquency ( β=.12, p<.05). For all three models
the quadratic term was not significant indicating that a linear model is the best
explanation of the data.
To address whether the quadratic relationship may be moderated by gender,
the above described analyses were performed simultaneously for male and female
groups. That is, for each model, analyses were conducted in which first all the path
coefficients were freely estimated for each group, next the measurement weights were
constrained to be equal across groups, and finally the measurement and structural
weights were constrained across groups. If the model in which the structural paths
were constrained showed a decrement in fit over the model in which the paths were
unconstrained, it would provide evidence that gender moderated one or more of the
structural relationships in the model.

54
For the Time 1 data the unrestricted model showed adequate fit to both
groups ( χ
2
=22.61 (12), RMSEA=.04). When the measurement weights were
constrained to be equal across groups there was not a significant decrement in the
model fit (Δχ
2
=.23 (2), p>.05) indicating measurement invariance of the latent
delinquency variable. The χ
2
difference between the measurement weights restricted
and the structural weights restricted models was significant ( Δχ
2
=10.77 (3), p<.05)
indicating that at least one of the parameters was moderated by gender. Therefore,
each structural parameter was tested in turn to determine whether one or more were
moderated by gender. The χ
2
difference test showed that the only parameter moderated
by gender was that between maltreatment status and Time 1 delinquency ( Δχ
2
=6.31
(1), p<.05; males β=.19, p<.01, females β= -.03, ns). The parameter between Time 1
pubertal timing and Time 1 delinquency was not significantly different between
genders ( Δχ
2
=2.46 (1), p>.05) nor was the parameter between Time 2 pubertal timing
2

and Time 2 delinquency ( Δχ
2
=.31 (1), p>.05). These results do not find support for a
quadratic relationship between Time 1 pubertal timing and
Time 1 delinquency for either gender.
The results for the Time 2 data indicated a satisfactory fit to the data for both
groups when the parameters were freely estimated ( χ
2
=32.86 (12), RMSEA=.07).
Constraining the measurement weights to be equal across groups did not significantly
alter the fit from the unrestricted model ( Δχ
2
=.79 (1), ns), and restricting the structural
weights did not significantly alter the fit from the measurement model ( Δχ
2
=2.31 (3),
ns). These results indicate gender does not moderate any of the structural

55
relationships. There was no significant relationship found between pubertal timing
2

and delinquency for either gender.
The longitudinal unrestricted model showed a good fit to the data ( χ
2
=15.97
(12), RMSEA=.03). When the measurement weights were constrained to be equal
across groups there was not a significant decrement in the model ( Δχ
2
=.50 (1), p>.05),
nor was there when the structural weights were constrained (Δχ
2
=3.05 (3), p>.05).
Similar to the Time 2 model, these results indicate that gender does not moderate any
of the structural relationships in the model.
Overall, there was no evidence of a quadratic relationship between pubertal
timing and delinquency as shown by the lack of significant estimates between these
two variables in all models. The findings indicate that early pubertal timing is
significantly associated with higher delinquency for both genders, whereas late
pubertal timing is related to the lowest levels of delinquency.

Specific Aim 3
To determine whether maltreatment moderates the relationship between pubertal
timing and delinquency, and if the interaction effect differs between male and female
adolescents.
Two methods were used to test for moderating effects: MGSEM was used to
test for group differences between maltreatment and comparison groups, and the
introduction of an interaction term into the model along with MGSEM was used to test
for a three-way interaction.

56
First, to examine the moderator effect of maltreatment status multiple group
models were used to examine the direct effect of pubertal timing on delinquency
between maltreatment and comparison groups. Pubertal timing was modeled as a
latent variable with Tanner breast/genital, Tanner pubic hair, and PDS as indicators.
Delinquency was modeled as a latent variable with status, person, and property
offences as its indicators. A direct effect of pubertal timing on delinquency was
entered in the model. First the model was run for both groups simultaneously with no
parameter constraints. Then the measurement weights were constrained as dictated by
the measurement model analyses. Lastly the structural weight was constrained to be
equal across groups. These models were run for the cross-sectional Time 1 data, the
Time 2 data, and then longitudinally for Time 1 pubertal timing and Time 2
delinquency. Fit statistics and parameter estimates for all models tested can be found
in Tables 16 and 17.
The unconstrained model fit both groups adequately for the Time 1 data
( χ
2
=20.80 (16), RMSEA=.03). When the measurement weights were constrained to be
equal across groups there was not a significant decrement in the model fit ( Δχ
2
=.35
(2), p>.05) indicating measurement invariance of both latent variables. The χ
2
difference between the measurement restricted model and the structural weight
restricted was not significant ( Δχ
2
=2.47 (1), p>.05) indicating that the relationship
between pubertal timing and delinquency was not moderated by maltreatment status.
Although the parameter estimate was not significantly different between groups, the
association between Time 1 pubertal timing and Time 1 delinquency reached

57

Table 16.
Fit Statistics for Models Testing the Interaction between Pubertal Timing and Maltreatment
χ
2
df RMSEA Δχ
2
Δdf p
MG-maltreatment status
T1puberty-T1delinquency
unconstrained 20.80 16 .03
measurement weights constrained 21.15 18 .02 .35 2 ns
structural weights constrained 23.62 19 .02 2.47 1 ns

T2puberty-T2delinquency
unconstrained 32.56 16 .05
measurement weights constrained 38.12 19 .05 5.56 3 ns
structural weights constrained 38.58 20 .05 .45 1 ns

T1puberty-T2delinquency
unconstrained 27.25 16 .04
measurement weights constrained 29.46 19 .04 2.21 3 ns
structural weights constrained 31.10 20 .04 1.64 1 ns

Interaction term in model
T1puberty-T1delinquency
Total sample 16.86 6 .06
MG-gender: unconstrained 22.40 12 .04
measurement weights constrained 22.60 14 .04 .20 2 ns
structural weights constrained 34.80 17 .05 12.20 3 .01
puberty to delinquency constrained 25.99 15 .04 3.39 1 .10
maltreatment to delinquency
constrained 29.20 15 .05 6.60 1 .01
puberty x maltreatment constrained 23.40 15 .04 .80 1 ns

T2puberty-T2delinquency
Total sample 20.02 6 .07
MG-gender: unconstrained 30.62 12 .06
measurement weights constrained 31.46 13 .06 .84 1 ns
structural weights constrained 33.22 16 .05 1.77 3 ns

T1puberty-T2delinquency
Total sample 12.30 6 .05
MG-gender: unconstrained 19.94 12 .04
measurement weights constrained 20.49 13 .04 .55 1 ns
structural weights constrained 23.10 16 .03 2.61 3 ns
Note. MG=multiple groups analysis

58
Table 17.
Parameter Estimates for Multiple Group Models between Maltreatment and Comparison Groups

Unstandardized estimates (SE) Standardized estimates

Maltreated Comparison Maltreated Comparison
Time 1 Model
T1pubertal timing →
T1delinquency .58 (.26) .07 (.20) .15** .03**
T1delinquency →
T1person offences .75 (.03) .71 (.06) .92** .86**
T1delinquency →
T1status offences .71 (.04) .47 (.05) .87**
a
.68**
b

T1delinquency →
T1property offences 1.00 1.00 .88**
a
.94**
b

T1pubertal timing →
T1Tanner breast/genital 1.21 (.20) .68 (.11) .87**
a
.66**
b

T1pubertal timing →
T1Tanner pubic hair 1.00 1.00 .72**
a
.97**
b

T1pubertal timing →T1PDS .54 (.10) .55 (.10) .38** .59**
Time 2 Model
T2pubertal timing →
T2delinquency .44 (.21) .53 (.16) .16** .37**
T2delinquency →
T2person offences .70 (.05) .91 (.12) .84** .80**
T2delinquency →
T2status offences .73 (.06) .80 (.12) .73** .62**
T2delinquency →
T2property offences 1.00 1.00 .90** .83**
T2pubertal timing →
T2Tanner breast/genital 1.12 (.11) .76 (.13) .87**
a
.73**
b

T2pubertal timing →
T2Tanner pubic hair 1.00 1.00 .84**
a
.84**
b

T2pubertal timing →T2PDS .74 (.08) .53 (.11) .58** .50**
Time 1 and Time 2 Model
T1pubertal timing →
T2delinquency .43 (.23) .11 (.13) .14
†
.08
T2delinquency →
T2person offences .70 (.05) .87 (.12) .84** .78**
T2delinquency →
T2status offences .72 (.06) .76 (.12) .73** .61**
T2delinquency →
T2property offences 1.00 1.00 .90** .85**
T1pubertal timing →
T1Tanner breast/genital 1.28 (.22) .69 (.11) .90**
a
.66**
b

T1pubertal timing →
T1Tanner pubic hair 1.00 1.00 .70**
a
.97**
b

T1pubertal timing →T1PDS .55 (.10) .55 (.10) .37** .59**
*p<.05, **p<.01,
†
p<.10
Note. All parameter estimates are from unconstrained model; SE= Standard Error;
a, b
: parameter
estimates with different subscripts are significantly different from each other based on χ
2
diff

59
significance in the maltreated group ( β=.15, p<.05) but not in the comparison group
( β=.03, ns). This result indicates that earlier pubertal timing is significantly related to
higher delinquency in maltreated individuals.
The results for the Time 2 data indicated a satisfactory fit to the data for both
groups when the parameters were freely estimated ( χ
2
=32.56 (16), RMSEA=.05).
Constraining the measurement weights to be equal across groups did not significantly
alter the fit from the unrestricted model ( Δχ
2
=5.56 (3), ns), and restricting the
structural weights did not significantly alter the fit from the measurement model
( Δχ
2
=.45 (1), ns). These results indicate maltreatment status does not moderate the
structural relationship. The parameter estimate between Time 2 pubertal timing and
Time 2 delinquency was significant for both groups, indicating that earlier timing is
related to higher delinquency in both groups; comparison group ( β=.37, p<.01),
maltreatment group ( β=.16, p<.05).
The unrestricted model with Time 1 pubertal timing and Time 2 delinquency
showed a good fit to the data ( χ
2
=27.25 (16), RMSEA=.04). When the measurement
weights were constrained to be equal across groups there was not a significant
decrement in the model ( Δχ
2
=2.21 (3), p>.05), nor was there when the structural
weights were constrained ( Δχ
2
=1.64 (1), p>.05). Although the groups were not
significantly different from each other, there was a marginally significant relationship
between Time 1 pubertal timing and Time 2 delinquency for the maltreated group.
For all models, cross-sectional and longitudinal, maltreatment status did not
moderate the relationship between pubertal timing and delinquency. However for both

60
groups the direction of the coefficient indicates that earlier pubertal timing is
associated with higher delinquency.
The second method tested for a both a two-way interaction between pubertal
timing and maltreatment, and a three-way interaction between maltreatment status,
pubertal timing, and gender. The interaction term was computed by multiplying the
manifest composite variable of pubertal timing by maltreatment status (comparison=0,
maltreatment=1). As mentioned previously, although latent variable interaction terms
can be used there are computational and interpretational complexities. Therefore
manifest variables were used to model the main effects of pubertal timing,
maltreatment status, and pubertal timing x maltreatment status on the latent variable of
delinquency. Covariances were drawn between pubertal timing and the interaction
term, between pubertal timing and maltreatment status, and between maltreatment
status and the interaction term. First, this model was run with the total sample to test
the interaction effect of T1 pubertal timing x maltreatment on T1 delinquency, then
the effect of T2 pubertal timing x maltreatment on T2 delinquency, and lastly the
effect of T1 pubertal timing x maltreatment on T2 delinquency. The baseline model is
shown in Figure 4, fit statistics and parameter estimates can be found in Table 16 and
18 respectively.
The Time 1 model showed a satisfactory fit to the data ( χ
2
= 16.86 (6),
RMSEA=.06). Parameter estimates indicated a significant relationship between
maltreatment status and Time 1 delinquency ( β=.10, p<.05), but did not show a
significant relationship between the interaction term and delinquency. This result is

61
congruent with the findings from the previous method for testing interaction effects.
The Time 2 model also showed an adequate fit to the data ( χ
2
=20.02 (6),
RMSEA=.07). There was a significant relationship between Time 2 pubertal timing
and Time 2 delinquency ( β=.24, p<.05) but again not between the interaction term and
delinquency. The model testing Time 1 pubertal timing and Time 2 delinquency
showed a good fit ( χ
2
=12.30 (6), RMSEA=.05) but had no significant parameter
estimates. Therefore, for all models there was no effect of the interaction between
pubertal timing and maltreatment on delinquency.

Figure 4. Model testing the interaction between pubertal timing and maltreatment

A three-way interaction was then tested by conducting multiple group analyses
between genders for these same models. That is, for each model, first the path
coefficients were freely estimated for each group, next the measurement weights were
constrained to be equal across groups, and finally the measurement and structural
weights were constrained across groups. If the model in which the structural paths
were constrained showed a decrement in fit over the model in which the paths were
Pubertal timing
Maltreatment status
Maltreatment x
Pubertal timing

Delinquency

62
Table 18.
Parameter Estimates for Multiple Group Models between Gender with Interaction Term
Unstandardized estimates (SE) Standardized estimates

Total
sample Males Females
Total
sample Males Females
Time 1 Model
T1pubertal timing →T1delinquency .08 (.18) .37 (.27) -.25 (.20) .04 .16
a
-.17
b

maltreatment status →T1delinquency .42 (.20) .87 (.31) -.11 (23) .10* .19**
a
-.04
b

T1pubertal timing x maltreatment status →T1delinquency .23 (.21) .04 (.33) .40 (.23) .10 .01 .23
†

T1pubertal timing ↔maltreatment status .02 (.02) .06 (.03) -.02 (.03) .05 .12
†
-.04

Time 2 Model
T2pubertal timing →T2delinquency .35 (.14) .38 (.22) .26 (.15) .24* .22
†
.25
†

maltreatment status →T2delinquency .26 (.16) .57 (.26) .07 (.18) .09 .15* .03
T2pubertal timing x maltreatment status →T2delinquency -.10 (.17) -.09 (.27) -.09 (.18) -.06 -.04 -.07
T2pubertal timing ↔maltreatment status -.03 (.02) .01 (.03) -.02 (.03) -.01 .03 -.05

Time 1 and Time 2 Model
T1pubertal timing →T2delinquency .09 (.14) .20 (.23) .03 (.15) .06 .12 -.03
maltreatment status →T2delinquency .25 (.16) .53 (.26) .05 (.18) .08 .15* .02
T1pubertal timing x maltreatment status →T2delinquency .09 (.17) .02 (.28) .15 (.18) .05 .01 .12
T1pubertal timing ↔maltreatment status .02 (.02) .06 (.03) -.02 (.03) .05 .12
†
-.04
*p<.05, **p<.01,
†
p<.10 Note. All parameter estimates are from unconstrained model; SE= Standard Error
a, b
: parameter estimates with different subscripts are significantly different from each other based on χ
2
diff

63

unconstrained, it would provide evidence that gender moderated one or more of the
structural relationships in the model. Specifically, if placing an equality constraint on
the parameter between the interaction term and delinquency significantly degraded the
model, it would provide evidence for a three-way interaction.
For the Time 1 data, the unrestricted model showed adequate fit to both groups
( χ
2
=22.40 (12), RMSEA=.04). When the measurement weights were constrained to be
equal across groups there was not a significant decrement in the model fit ( Δχ
2
=.20
(2), p>.05) indicating measurement invariance the delinquency latent variable. The χ
2
difference between the measurement restricted model and the structural weights
restricted was significant ( Δχ
2
=12.20 (3), p<.01) indicating that at least one of the
parameters was moderated by gender. Therefore, each structural parameter was tested
in turn to determine whether one or more were moderated by gender. A significant
decrement in the model constraining a specific parameter to be equal across groups
indicated an interaction effect. The χ
2
difference test showed that two structural
parameters differed between genders. The parameter between maltreatment status and
Time 1 delinquency was significantly different for each gender ( Δχ
2
=6.60 (1), p<.01)
(males β=.19, p<.01; females β=-.04, ns) and between Time 1 pubertal timing and
Time 1 delinquency ( Δχ
2
=3.39 (1), p<.10) (males β=.16, ns; females β=-.17, ns). The
direction of the regression coefficient indicated that only for males maltreatment status
was related to higher delinquency. Although the estimates did not reach significance in
either group, the parameters between Time 1 pubertal timing and Time 1 delinquency
were in opposite directions for males and females. That is, for males earlier pubertal

64

timing was related to higher delinquency, and for females later pubertal timing was
related to higher delinquency. Additionally, the interaction effect for Time 1 pubertal
timing x maltreatment on Time 1 delinquency was significant in females but not in
males. The significant interaction effect for females was summarized graphically by
computing predicted values for one standard deviation above and below the mean on
the predictor and moderator variables (Figure 5). This graph indicates that for
maltreated females earlier pubertal timing is related to higher delinquency whereas for
comparison females later pubertal timing is related to higher delinquency.
Females
1
1.5
2
2.5
3
3.5
4
4.5
5
Late Pubertal Timing Early Pubertal Timing
Delinquency
comparison
maltreated

Figure 5. Interaction effect of Time 1 pubertal timing x maltreatment status
on Time 1 delinquency

Results for the Time 2 data indicated a good fit to the data for both groups
when they were freely estimated ( χ
2
=30.62 (12), RMSEA=.06). Constraining the
measurement weights to be equal across groups did not significantly alter the fit from
the unrestricted model ( Δχ
2
=.84 (1), ns), and restricting the structural weights did not

65

significantly alter the fit from the measurement model ( Δχ
2
=1.77 (3), ns). These
results indicate gender does not moderate any of the structural relationships. Although
the parameter estimates were not significantly different from each other, the regression
coefficients reached significance in one or both groups. As can be seen in Table 18,
Time 2 pubertal timing and Time 2 delinquency were significantly related in both
males ( β=.22, p<.10) and females ( β=.25, p<.10) indicating earlier pubertal timing is
associated with higher delinquency. However, maltreatment status and delinquency
were significantly related only in males ( β=.15, p<.05). The interaction term did not
approach significance in either gender.
The unrestricted model for Time 1 pubertal timing and Time 2 delinquency
showed a good fit to the data ( χ
2
=19.94 (12), RMSEA=.04). When the measurement
weights were constrained to be equal across groups there was not a significant
decrement in the model fit ( Δχ
2
=.55 (1), p>.05), nor was there when the structural
weights were constrained ( Δχ
2
=2.61 (3), p>.05). Similar to the Time 2 model, these
results indicate that gender does not moderate any of the structural relationships in the
model. However, maltreatment status and Time 2 delinquency were significantly
related for males ( β=.15, p<.05) and not for females although the two groups were not
significantly different from each other. The relationship between maltreatment status
and pubertal timing was marginally significant for males ( β=.12, p<.10) but not
significant for females. This indicates that for males maltreatment is associated with
earlier pubertal timing. As with the Time 2 model, the interaction between pubertal
timing and maltreatment status was not significant in either gender.

66

Chapter 6: Discussion
This study examined the relationship between pubertal timing and delinquent
behavior in a sample of maltreated and comparison adolescents. Of primary interest
was the main effect of pubertal timing on delinquency as well as moderation of this
effect by gender and maltreatment status. The present study differs from the existing
research in several significant ways. The bulk of pubertal research has been conducted
with samples of middle-class white adolescents whereas the present study used a
sample primarily composed of African American and Hispanic adolescents from an
urban area. An additional strength of this sample is the equal composition of males
and females which afforded the examination of gender differences. This study is also
unique in the use of several indicators of pubertal development, and the construction
of a latent variable to measure pubertal timing. Furthermore, this is one of the first
studies to examine maltreatment status as a moderator of the relationship between
pubertal timing and delinquency.
The first aim of this study was to examine the main effect of pubertal timing on
delinquent behavior after accounting for maltreatment status. This was tested for the
total sample and then multiple group models were used to test gender as a moderator
of this relationship. The results showed that Time 1 pubertal timing was significantly
related to Time 1 delinquency but gender did not moderate this relationship. This
indicates that for both male and females, earlier pubertal timing at Time 1 is related to
higher delinquency at Time 1. This association also emerged for both genders between
Time 2 pubertal timing and Time 2 delinquency. Longitudinally, Time 1 pubertal

67

timing was predictive of Time 2 delinquency for the total sample. However, when
the effect of Time 1 delinquency on Time 2 delinquency was controlled for, the
relationship between Time 1 pubertal timing and Time 2 delinquency was no longer
significant. The stronger cross-sectional associations may indicate stage or tempo
effects that were not taken into account in this study. That is, early maturers at Time 1
may not all still be early in their pubertal maturation at Time 2, or being an early
maturer may be more salient at the younger ages. The only relationship found to be
moderated by gender was that between maltreatment status and Time 1 delinquency.
The relationship implies that maltreated males engage in more delinquent behavior
than comparison males. The cross-lag model also showed that Time 1 and Time 2
pubertal timing were significantly related as well as Time 1 and Time 2 delinquency.
Pubertal timing across the two time points was highly correlated indicating that most
but not all early maturers at Time 1 were also early maturers at Time 2. Overall,
consistent with previous research, earlier pubertal timing seems to be a risk factor for
delinquent behavior for both males and females (e.g. Ge et al., 2002; Kaltiala-Heino et
al., 2003; Graber et al., 1997; Caspi et al., 1993). The present findings add support to
this literature as well as extending the results to urban populations.
The second aim of this paper was to examine whether the relationship between
pubertal timing and delinquency was nonlinear. There is inconsistency in the literature
as to whether early or late timing is more related to delinquency. It may be that both
early and late pubertal maturation are risk factors for delinquency compared to on-
time development. However, a quadratic relationship between pubertal timing and

68

delinquency was not found, disconfirming this expectation. Therefore, based on the
findings a linear model is the best explanation of the data, with earlier pubertal timing
predicting higher delinquency in both genders within and across time.
The last aim of this study was to examine the interaction between pubertal
timing and maltreatment status. This was examined in two ways, first by conducting
multiple group analyses between maltreatment and comparison groups, and then by
introducing an interaction term into the model and estimating the main effect in males
and females in order to test a three-way interaction. Results from multiple group
models between maltreatment and comparison groups showed that maltreatment did
not moderate the relationship between pubertal timing and delinquency for the total
sample. However, Time 1 pubertal timing was significantly related to Time 1
delinquency in the maltreated group but not in the comparison group. This indicates
that for maltreated adolescents earlier pubertal timing is related to higher delinquency.
This is congruent with research findings supporting the contextual amplification of
pubertal timing effects (Obeidallah et al., 2004). However in this analysis the
maltreatment and comparison groups were not significantly different from each other
even though the parameter estimate was only significant in the maltreated group.
When the interaction term was added to the model, the relationship between
Time 1 pubertal timing and Time 1 delinquency was moderated by gender although
the parameter estimates did not reach significance in either group. A significant
interaction effect was found only for females between Time 1 pubertal timing x
maltreatment status and Time 1 delinquency. When graphed, this interaction indicates

69

that earlier pubertal timing is related to higher delinquency in maltreated females
whereas late pubertal timing is associated with higher delinquency in comparison
females. The experience of maltreatment for female adolescents appears to exacerbate
the risk associated with early pubertal timing, and is the reverse of the association
found for comparison females. For maltreated females the stress of dealing with
maltreatment experience may be difficult, but when it is compounded with early
maturation and aberrant physical development, the result seems to be higher levels of
delinquent behavior. For comparison females, late maturers may act out as a means of
coping with their relative physical immaturity or it may be an attempt to garner
attention that more physically mature girls are shown simply because of their physical
development. This interaction effect was only found for Time 1 data, it did not emerge
for Time 2 data, longitudinally, or for males. An explanation for the gender disparity is
that perhaps females are more vulnerable than males at an early age to contextual
stressors which amplify the pubertal transition. Or it may be that males and females
are at risk for different types of abuse, which are more or less associated with pubertal
development and delinquency. The emergence of an interaction effect only at Time 1
may be due in part to the closeness between the maltreatment experience and the onset
of puberty. The impact of maltreatment may be more critical at earlier ages when
adolescents have not yet fully developed adequate coping skills. These explanations
need to be investigated because at this point they are only conjecture. This study is the
first to find an interaction effect for pubertal timing x maltreatment status on

70

delinquency. In addition, this study is distinctive because of the comparison of
gender differences in the relationship between pubertal timing and delinquency.
This study contributes to the literature in several significant ways. Foremost,
this is the first study to use multiple indicators of pubertal development to construct a
latent variable and examine associations with multiple scales of delinquent behavior.
Consideration should be given to assessing multiple indicators of pubertal
development in order to capture the full array of associated changes. Second, these
relationships were examined both cross-sectionally and across two time points
whereas most studies have been restricted to cross-sectional designs, limiting what can
be inferred about the developmental processes taking place. Third, the composition of
the sample allowed for the clarification of gender differences in the associations
between pubertal timing and delinquency, as well as the extension of these
relationships to an ethnically diverse urban sample. An additional contribution of this
study is the examination of maltreatment experience as a moderator of the relationship
between pubertal timing and delinquency. The interaction effect found only for
females indicates that perhaps for males, the effects of pubertal timing are not
heightened by experience of maltreatment. One explanation for the absence of this
relationship in males could be related to the participants being in the early stages of
pubertal development at Time 1. At the very beginning of puberty the physical
changes may be more salient and obvious to females than to males as opposed to later
ages when almost all adolescents have some obvious sign of pubertal development.
Another explanation is that maltreated individuals may mature earlier than non-

71

maltreated individuals, which then puts them at higher risk for delinquent behavior.
There is some support for an association between sexual abuse and earlier maturation
for females (Brown et al., 2004; Zabin et al., 2005), however in the present study there
are no significant correlations between pubertal timing and maltreatment, therefore the
present findings cannot lend support to this supposition. Although the interaction
between pubertal timing and maltreatment was not found to heighten delinquent
behavior for males, the association between maltreatment status and delinquency
found only for males implies that maltreatment experience is more important in
predicting delinquency for males than females. Another interesting finding was that
for comparison females later pubertal timing was found to be related to higher
delinquency, which is in opposition to the bulk of the research (Ge et al., 2002;
Kaltiala-Heino et al., 2003; Graber et al., 1997; Obeidallah et al., 2004; Cota-Robles et
al., 2002). However, this is consistent with one study which found later pubertal
timing to be related to more behavior problems (Dorn, Susman et al., 1999).
There are several limitations of the current study that should be taken into
account. First is the dichotomization of maltreatment experience. Although many
studies find group differences simply when categorizing maltreatment versus non-
maltreatment groups (e.g. Pepin & Banyard, 2006; Hill, 2006), there are complexities
in specific characteristics of abuse that may affect the relationships examined in this
study. Although the relationship between child maltreatment and delinquency has been
demonstrated for all types of abuse and neglect, physical abuse in particular has been
linked to aggressive and violent behavior (Brown, 1984; Weatherburn & Lind, 1998).

72

Additionally, in a retrospective study, sexual abuse was related to earlier onset of
menstruation in females (Brown et al., 2004). Therefore, specific types of abuse may
be more or less associated with either puberty or delinquency.
A second limitation of this study is the young end of the age range of the
sample. Although the adolescents ranged from 9 to 13 years at the first assessment,
there is the possibility that some of those 9 year olds had not yet entered puberty
which then makes it difficult to measure. However, the strategy for constructing the
pubertal timing variable allowed for the use of those who had not yet begun puberty
by comparing them to others of the same age and gender. If all the 9 year olds had not
begun puberty then they would all be calculated as on-time in their development.
Because the pubertal timing variable was created as a measure relative to peers and
not relative to the national norms, it was a better reflection of how they may appear in
reference to their peer group. This is an important point because one proposed
mechanism by which early pubertal development may affect delinquency is through
the interaction of early maturers with older peers (Caspi et al., 1993; Haynie, 2003).
Those appearing more physically mature within their peer group may receive attention
from older peers who model and draw them into delinquent behavior.
A third limitation is that both the measures of pubertal development and
delinquency were obtained by self-report. Although self-reports of pubertal
development have been found comparable to physician-reports, pubertal ratings from
multiple informants would strengthen the validity of these measures. However, the use
of multiple informants, in particular physician examination, is not always a feasible

73

option. Multiple informants would add to the measurement of delinquency as well,
although using police reports or arrest records would discount a majority of
adolescents as many have not been involved in the juvenile justice system. Perhaps for
the assessment of more serious delinquency, official reports would be most useful.
Even within the present study the frequency of delinquent behavior was quite low.
This may be in part due to the young age range of the sample. Delinquency may be
more evident in older adolescents, therefore it would be informative to follow the
adolescents beyond the time point and age used in the present study.
An additional limitation to consider it that group differences in ethnicity were
not examined. There is some research that indicates African American girls enter
puberty the earliest of all ethnicities, which may out them at higher risk for transitional
difficulties associated with puberty (Chumlea et al., 2003). Although the present study
did not investigate ethnic differences per se, it is one of the first to establish the
relationship between early pubertal timing and delinquency for both genders in a
sample primarily composed of African American and Hispanic adolescents.
Lastly this study is limited in the examination of other contextual factors or
mediational processes. Although gender and maltreatment experience were examined
as moderators, there may be other factors related to parenting, peers, or neighborhoods
that affect these relationships. In addition, the present results do not address
mechanisms by which these associations may function.
Future research should continue to examine moderators of the relationship
between pubertal development and various maladaptive outcomes. In addition, effort

74

should be placed in identifying the mechanisms by which pubertal development
affects behavioral and mental health functioning. Maltreatment experience should be
examined more closely in regards to the characteristics of the abuse; frequency,
chronicity, duration, multiple types, multiple perpetrators, etc. Identifying those
characteristics most highly associated with problematic outcomes will allow future
studies to target those individuals at risk in an effort to reduce those difficulties. The
present findings support the use of multiple indicators of pubertal development and
should be considered in future research as this method may give a more complete and
accurate description of the pubertal process. Longitudinal studies will greatly enhance
our understanding of the developmental trajectories associated with pubertal
development. These studies should include children before they have begun puberty
and follow them through the completion. This will give an accurate measure of onset
of puberty, and then maltreatment experience occurring prior can be accounted for. In
addition, there may be differing associations between pubertal development and
various outcomes based on the age and developmental period. Some outcomes may be
more salient and critical to address early in adolescence, whereas other outcomes may
be more salient in later adolescence. By following children from before the onset of
puberty through adolescence and into early adulthood, the short and long-term impact
of pubertal development can be assessed. There may be certain outcomes which tend
to be more transient or moderators which may enhance the persistence of problem
behaviors. Knowledge of such risk factors can inform interventions aimed at reducing
the deleterious effects associated with off-time pubertal development.

75

References
Angold, A., Costello, E.J., & Worthman, C.M. (1998). Puberty and depression: the
roles of age, pubertal status and pubertal timing. Psychological Medicine, 28,
51-61.

Arbuckle, J.L. (2003). Amos 5.0 [Computer Software]. Chicago: Smallwaters.

Belsky, J., Steinberg, L., & Draper, P. (1991). Childhood experience, interpersonal
development and reproductive strategy: an evolutionary theory of socialization.
Child Development, 62, 647-670.

Bissonnette, V., Ickes, W., Bernstein, I., & Knowles, E. (1990). Personality
moderating variables: A warning about statistical artifact and comparison of
analytic techniques. Journal of Personality, 58, 567-587.

Bolton, F.G., Reich, J.W., & Gutierres, S.E. (1977). Delinquency patterns in
maltreated children. Victimology, 2, 349-357.

Brooks-Gunn, J., Warren, M.P., Rosso, J., & Gargiulo, J. (1987). Validity of self-
report measures of girls’ pubertal status. Child Development, 58, 829-841.

Brown, S. (1984). Social class, child maltreatment, and delinquent behavior.
Criminology, 22, 259-278.

Brown, J., Cohen, P., Chen, H., Smailes, E., & Johnson, J.G. (2004). Sexual
trajectories of abused and neglected youths. Journal of Developmental and
Behavioral Pediatrics, 25 (2), 77-82.

Browne , M.W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In
K.A. Bollen & J.S. Long (Eds.), Best methods for the analysis of change (pp.
136-162). Thousand Oaks, CA: Sage.

Byrne, B.M. (2001). Structural Equation Modeling with AMOS. Mahwah, New Jersey:
Lawrence Erlbaum Associates.

Byrne, B.M., Shavelson, R.J., & Muthén, B. (1989). Testing for the equivalence of
factor covariance and mean structures: The issues of partial measurement
invariance. Psychological Bulletin, 88, 456-466.

Caspi, A., Lynam, D., Moffitt, T.E., & Silva, P.A. (1993). Unraveling girls’
delinquency: biological, dispositional, and contextual contributions to
adolescent misbehavior. Developmental Psychology, 29(1), 19-30.

76

Centers for Disease Control (2006). Body Mass Index: About BMI for Children and
Teens. Retrieved February 1, 2007 from http://www.cdc.gov/nccdphp/dnpa/
bmi/childrens_BMI/about_childrens_BMI.htm

Chamberlain, P., & Moore, K.J. (2002). Chaos and trauma in the lives of adolescent
females with antisocial behavior and delinquency. Journal of Aggression,
Maltreatment, and Trauma, 6 (1), 79-108.

Chumlea, W., Schubert, C., Roche, A., Kulin, Lee, H., Himes, J., et al. (2003). Age at
menarche and racial comparisons in U.S. girls. Pediatrics, 111, 110-113.

Cicchetti, D. & Toth, S.L. (2000). Developmental processes in maltreated children. In
D.J. Hansen (Ed.), Nebraska Symposium on Motivation Vol. 46, 1998:
Motivation and child maltreatment (pp. 86-160). Lincoln, NE: University of
Nebraska Press.

Cohen, P., Cohen, J., Kasen, S., Velez, C.N., & Johnson, J. (1993). An
epidemiological study of disorders in late childhood and adolescence: I. Age
and gender specific prevalence. Journal of Child Psychology and Psychiatry,
34, 851-867.

Cota-Robles, S., Neiss, M., & Rowe, D.C. (2002). The role of puberty in violent and
nonviolent delinquency among Anglo American, Mexican American, and
African American boys. Journal of Adolescent Research, 17(4), 364-376.

Cronbach, L.J. (1951). Coefficient alpha and the internal structure of tests.
Psychometrika, 16, 297-334.

Davalos, D.B., Chavez, E.L., & Guardiola, R.J. (2005). Effects of parental school
support and family communication on delinquent behaviors in Latinos and
White non-Latinos. Cultural Diversity and Ethnic Minority Psychology, 11(1),
57-68.

Dempster, A.P., Laird, N.M., & Rubin, D.B. (1977). Maximum likelihood estimation
from incomplete data via the EM algorithm (with discussion). Journal of the
Royal Statistical Society, B39, 1-38.

Dick, D.M., Rose, R.J., Pulkkinen, L., & Kaprio, J. (2001). Measuring puberty and
understanding its impact: a longitudinal study of adolescent twins. Journal of
Youth and Adolescence, 30(4), 385-399.

77

Dorn, L.D., Susman, E.J., & Ponirakis, A. (1999). Pubertal timing and adolescent
adjustment and behavior: conclusions vary by rater. Journal of Youth and
Adolescence, 32(3), 157-167.

Dorn, L.D., Susman, E.J., Nottelmann, E.D., Inoff-Germain, G., & Chrousos, G.P
(1990). Perceptions of puberty: adolescent, parent, and health care personnel.
Developmental Psychology, 26, 322-329.

Dorn, L.D., Nottelmann, E.D., Susman, E.J., Inoff-Germain, G., Cutler, G.B. &
Chrousos, G.P. (1999). Variability in hormone concentrations and self-reported
menstrual histories in young adolescents: Menarche as an integral part of a
developmental process. Journal of Youth & Adolescence, 28(3), 283-304

Doerner, W.G. (1987). Child maltreatment seriousness and juvenile delinquency.
Youth and Society, 19, 197-244.

Duke, P.M., Litt, I.F., & Gross, R.T. (1980). Adolescents’ self-assessment of sexual
maturation. Pediatrics, 66, 918-920.

Elliott, D.S., & Huizinga, D. (1983). Social class and delinquency behavior in a
national youth panel: 1976-1980. Criminology, 21, 149-177.

Elliott, D.S., Huizinga, D., & Ageton, S.S. (1985). Explaining Delinquency and Drug
Use. Beverly Hills, CA: Sage Publications.

Fechner, P.Y. (2003). The biology of puberty: new developments in sex differences. In
C. Hayward (Ed.), Gender Differences at Puberty (pp. 17-28). Cambridge,
England: University Press.

Finklestein, J.W., D’Arcangelo, M.R., Susman, E.J., Chinchilli, V.M., Kunselman,
S.J., Schwab, J. et al. (1999). Self-assessment of physical sexual maturation in
boys and girls with delayed puberty. Journal of Adolescent Health, 25, 379-
381.

Flannery, D.J., Rowe, D.C., & Gulley, B.L. (1993). Impact of pubertal status, timing,
and age on adolescent sexual experience and delinquency. Journal of
Adolescent Research, 8(1), 21-40.

Frazier, P.A., Tix, A.P., & Barron, K.E. (2004). Testing moderator and mediator
effects in counseling psychology. Journal of Counseling Psychology, 51(1),
115-134.

78

Frisch, R.E. & Revelle, R. (1971). The height and weight of girls and boys at the
timeof initiation of adolescent growth spurt in height and weight and the
relationship to menarche. Human Biology, 43, 140-159.

Ge, X., Brody, G.H., Conger, R.D., Simons, R.L. & McBride-Murray, V. (2002).
Contextual amplification of pubertal transition effects on deviant peer
affiliation and externalizing behavior among African American children.
Developmental Psychology, 38(1), 42-54.

Ge, X., Conger, R.D., & Elder, G.H. (2001). Pubertal transition, stressful life events,
and the emergence of gender differences in adolescent depressive symptoms.
Developmental Psychology, 37(3), 404-417.

Ge, X., Kim, I.J., Brody, G.H., Conger, R.D., Simons, R.L., Gibbons, F.X. et al.
(2003). It’s about timing and change: pubertal transition effects on symptoms
of major depression among African American youths. Developmental
Psychology, 39(3), 430-439.

Gomme, I.M., Morton, M.E., & West, W.G. (1984). Rates, types, and patterns of male
and female delinquency in an Ontario county. Canadian Journal of
Criminology, 26 (3), 313-323.

Graber, J.A., Lewinsohn, P.M., Seeley, J.R., & Brooks-Gunn, J. (1997). Is
psychopathology associated with the timing of pubertal development? Journal
of the American Academy of Child and Adolescent Psychiatry, 36(12), 1768-
1776.

Guttman, L. (1945). A basis for analyzing test-retest reliability. Psychometrika, 10,
255-282.

Haynie, D.L. (2003). Contexts of risk? Explaining the link between girls’ pubertal
development and their delinquent involvement. Social Forces, 82(1), 355-397.

Hayward, C. (2003). Methodological concerns in puberty-related research. In C.
Hayward (Ed.), Gender Differences at Puberty (pp. 1-14). Cambridge,
England: University Press.

Herrera, V.M., & McCloskey, L.A. (2001). Gender differences in the risk for
delinquency among youth exposed to family violence. Child Abuse and
Neglect, 25, 1037-1051.

Hill, J. (2006). Child maltreatment and depression in adults: Implications for
prevention. Clinical Nueropsychiatry: Journal or Treatmetn Evaluation, 3(1),
23-28.

79

Huizinga, D., & Elliott, D.S. (1986). Reassessing the reliability and validity of self-
report delinquency measures. Journal of Quantitative Criminology, 2(4), 293-
327.

Huizinga,, D., Elliott, D.S., & Menard, S. (1989). Multiple Problem Youth. New York:
Springer-Verlag.

Huizinga, D., & Morse, B. (1986) Adolescent Delinquency Questionnaire (ADQ).
Unpublished instrument.

Ireland, T.O., Smith, C.A., & Thornberry, T.P. (2002). Developmental issues in the
impact of child maltreatment on later delinquency and drug use. Criminology,
40, 359-399.

Kaltiala-Heino, R., Marttunen, M., Rantanen, P., & Rimpela, M. (2003). Early puberty
is associated with mental health problems in middle adolescence. Social
Science and Medicine, 57, 1055-1064.

Kaplan, S.J., Pelcovitz, D., & Labruna, V. (1999). Child and adolescent abuse and
neglect research: a review of the past 10 years. Part I: physical and emotional
abuse and neglect. Journal of the American Academy of Child and Adolescent
Psychiatry, 38 (10), 1214-1222.

Loehlin, J.C. (2004). Latent Variable Models (4
th
Ed.). New Jersey: Lawrence
Erlbaum Associates.

Loftus, E.F., & Marburger, W. (1983). Since the eruption of Mt. St. Helens, has
anyone beaten you up? Improving the accuracy of retrospective events with
landmark events. Memory and Cognition, 11(2), 114-120.

Malina, R.M. & Bouchard, C. (1991). Growth, Maturation, and Physical Activity.
Champaign, IL: Human Kinetics Books.

Marshall, W.A., & Tanner, J.M. (1969). Variations in Patterns of pubertal changes in
girls. Archives of Disease in Childhood, 44, 291-303.

Marshall, W.A., & Tanner, J.M. (1970). Variations in Patterns of pubertal changes in
boys. Archives of Disease in Childhood, 45, 13-23.

Mason, C.A., Tu., S., & Cauce, A.M. (1996). Assessing moderator variables: Two
computer simulation studies. Educational and Psychological Measurement, 56,
45-62.

80

McCord, J. (1983). A forty year perspective on child abuse and neglect. Child
Abuse and Neglect, 7, 265-270.

Moffitt, T.E. (1996). Measuring children’s antisocial behaviors. Journal of the
American Medical Association, 275(5), 403-404.

Morris, N.M., & Udry, R.J. (1980). Validation of a self-administered instrument to
assess stage of adolescent development. Journal of Youth and Adolescence,
9(3), 271-280.

NCCAN (National Center of Child Abuse and Neglect) (1996). Study Findings: Study
of National Incidence and Prevalence of Child Abuse and Neglect: 1993.
Washington, D.C.: Department of Health and Human Sciences.

Obeidallah, D., Brennan, R.T., Brooks-Gunn, J., & Earls, F. (2004). Links between
pubertal timing and neighborhood contexts: implications for girls’ violent
behavior. Journal of the American Academy of Child and Adolescent
Psychiatry, 43(12), 1460-1468.

Orchard, T., & Woodbury, M.A. (1972). A missing information principle: Theory and
applications. In Proceedings of the 6
th
Berkeley Symposium on Mathematical
Statistics and Probability, 1 (pp. 697-715). Berkeley, CA: University of
California Press.

Pepin, E.N. & Banyard, V.L. (2006). Social support: A mediator between child
maltreatment and developmental outcomes. Journal of Youth and Adolescence,
35(4), 617-630.

Petersen, A.C., Crockett, L.J., Richards, M., & Boxer, A. (1988). A self-report
measure of pubertal status: reliability, validity, and initial norms. Journal of
Youth and Adolescence, 17, 117-133.

Rivera, B., & Widom, C.S. (1990). Childhood victimization and violent offending.
Violence and Victims, 5, 19-35.

Robbins, R.L.N., & Price, R.K. (1991). Adult disorders predicted by childhood
conduct problems: Results from the NIMH Epidemiologic Catchment Area
Project. Psychiatry, 54, 116-132.

Schafer, J.L. (1999). NORM: Multiple imputation of incomplete multivariate data
under a normal model (Version 2.0), Computer statistical software available
from http://www.stat.psu.edu/~jls/misoftwa.html.

81

Shapiro, D.L., & Levendosky, A.A. (1999). Adolescent survivors of childhood
sexual abuse: the mediating role of attachment style and doping in
psychological and interpersonal functioning. Child Abuse and Neglect, 23(11),
1175-1191.

Smith, C., & Thornberry, T.P. (1995). The relationship between childhood
maltreatment and adolescent involvement in delinquency. Criminology, 33,
451-477.

Tabachnick, B.G., & Fidell, L.S. (2001). Using Multivariate Statistics (4
th
Ed.).
Needham Heights, MA: Allyn & Bacon.

Taylor, S.J., Whincup, P.H., Hindmarsh, P.C., Lampe, F., Odoki, K., & Cook, D.G.
(2001). Performance of a new pubertal self-assessment questionnaire: a
preliminary study. Pediatric and Perinatal Epidemiology, 15, 88-94.

Weatherburn, D., & Lind, B. (1998). Poverty, parenting, peers, and crime-prone
neighborhoods. Canberra, ACT, Australia: Australian Institute of Criminology.

Williams, J.M. & Dunlop, L.C. (1999). Pubertal timing and self-reported delinquency
among male adolescents. Journal of Adolescence, 22, 157-171.

Zabin, L.S., Emerson, M.R., & Rowland, D.L. (2005). Childhood sexual abuse and
early menarche: the direction of their relationship and its implications. Journal
of Adolescent Health, 36, 393-400.

Zingraff, M.T., Leiter, J., Myers, K.A., & Johnson, M.C. (1993). Child maltreatment
and youthful problem behavior. Criminology, 31, 173-202.

Abstract (if available)

Abstract There is significant disparity in the existing literature on pubertal development as to whether early, late, or both early and late timing are related to delinquent behavior, and whether this association varies based on gender. The purpose of this study was to examine the relationship between pubertal timing and delinquent behavior in a sample of maltreated and comparison adolescents. The sample for the present study was drawn from a larger research project examining the effects of maltreatment on adolescent development. The design was cross-sequential with the age of subjects ranging from 9-13 years at the initial assessment with subsequent assessments one year apart. Data for the present study was taken from the first and second assessments. The first aim of this study was to examine the relationship between pubertal timing and delinquency cross-sectionally and longitudinally for the total sample, and then to test for gender differences using multiple group models. The second aim examined the possibility of a nonlinear relationship between pubertal timing and delinquency. That is, whether early, late, or early and late timing contributed to higher levels of delinquency. The third aim was to examine the interaction of pubertal timing and maltreatment to determine whether the experience of maltreatment along with off-time pubertal development creates an increased risk for delinquent behavior, and whether this relationship differs between genders. The results showed that for both males and females earlier pubertal development relative to peers was related to higher levels of delinquency. Multiple group models showed that maltreatment group did not moderate the relationship between pubertal timing and delinquency when examining both genders together. However, there was evidence of an interaction effect for females.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Examining the longitudinal relationships between community violence exposure and aggressive behavior among a sample of maltreated and non-maltreated adolescents

PDF

Untangling the developmental relations between depression and externalizing behavior among maltreated adolescents

PDF

Trajectories of internalizing problems among maltreated girls and boys: differences by maltreatment type and developmental timing

PDF

The impact of child maltreatment on the mental health of adolescents: a longitudinal study of social anxiety and self-perception

PDF

The development of electrodermal fear conditioning from ages 3 to 8 years and relationships with antisocial behavior at age 8 years

PDF

Maltreated adolescents and their families: a longitudinal examination of family functioning, parenting attitudes, & youth mental health

PDF

Multiple informant-rated psychopathy in young adolescents: a multitrait-multimethod investigation of the Antisocial Process Screening Device

PDF

Parents' attunement: relation to adolescent problem behavior and the moderating role of marital conflict

PDF

Emotion regulation as a mechanism linking parents’ marital aggression to adolescent behavioral problems: a longitudinal analysis

PDF

Social status, perceived social reputations, and perceived dyadic relationships in early adolescence

PDF

The link between maternal depression and adolescent daughters' risk behavior: the mediating and moderating role of family

PDF

Observed and self-reported childrearing in mothers with a history of sexual abuse

PDF

The relationship between teacher assessment practices, student goal orientation, and student engagement in elementary mathematics

PDF

Individual, family, and peer influences on substance use, delinquency, and problem behaviors among at-risk Asian American and Pacific Islander adolescents

PDF

Social network influences on depressive symptoms among Chinese adolescents

PDF

Typologies of male and female adolescent perpetrators of dating violence: three subtypes for males and four subtypes for females, and differences among them

PDF

The role of depression symptoms on social information processing and tobacco use among adolescents

PDF

Psychosocial and behavioral ractors associated with emotional eating in adolescents

PDF

Neighborhood context and adolescent mental health: development and mechanisms

PDF

Childhood adversity across generations and its impact on externalizing behavior

Asset Metadata

Creator Negriff, Sonya (author)

Core Title The relationship between pubertal timing and delinquent behavior in maltreated male and female adolescents

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 03/27/2007

Defense Date 03/05/2007

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

Tag delinquency,maltreatment,OAI-PMH Harvest,pubertal timing

Language English

Advisor Trickett, Penelope K. (committee chair), Manis, Franklin R. (committee member), Mennen, Ferol E. (committee member), Schwartz, David (committee member)

Creator Email negriff@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m325

Unique identifier UC1502150

Identifier etd-Negriff-20070327 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-323301 (legacy record id),usctheses-m325 (legacy record id)

Legacy Identifier etd-Negriff-20070327.pdf

Dmrecord 323301

Document Type Dissertation

Rights Negriff, Sonya

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu