Adol Tinrete - Risc

P1: GFU/LCCR/GAY P2: GCR/GFU
Journal of Youth and Adolescence [jya] PP210-342604 July 17, 2001 11:40 Style file version Nov. 19th, 1999
Journal of Youth and Adolescence, Vol. 30, No. 4, 2001
Course, Co-Occurrence, and Longitudinal

Associations of Emotional Disturbance and
Delinquency From Adolescence to Young
Adulthood: A Six-Year Three-Wave Study
Geertjan Overbeek,1 Wilma Vollebergh,2 Wim Meeus,3 Rutger Engels,4
and Eric Luijpers5
Received March 27, 2000; accepted March 28, 2001
Three questions were examined in this study: (a) What is the course of emotional
disturbance and delinquency during adolescence and young adulthood? (b) To
what extent do emotional disturbance and delinquency co-occur during adoles-
cence and young adulthood? and (c) What are the longitudinal associations be-
tween emotional disturbance and delinquency in different age and gender cate-
gories during adolescence and young adulthood? Data were used from a national
sample of 1,302 adolescents and young adults, who participated in a 6-year 3-wave
longitudinal study. Findings showed an increase of emotional disturbance and
delinquency from early to midadolescence, after which emotional disturbance sta-
bilized and delinquency declined into young adulthood. A significant but relatively
weak co-occurrence of emotional disturbance and delinquency was found. Multi-
group LISREL analyses demonstrated that a stability model with no cross-lagged
relations fit best for the total sample, and across age and gender categories. Thus,
co-occurrence of emotional disturbance and delinquency during adolescence and
young adulthood seems to result from associated but separate psychopathological
1 Professionally trained in Youth Studies, currently PhD student, Department of Child and Adolescent
Studies, Utrecht University, The Netherlands. Main interests include developmental processes between
internalizing and externalizing problems during adolescence and young adulthood. To whom corre-
spondence should be addressed at Department of Child and Adolescent Studies, Utrecht University,
Heidelberglaan 1, 3584 CS, Utrecht, The Netherlands; e-mail: [email protected].
2 Head, Research program, Psychiatric Epidemiology, Trimbos Institute, The Netherlands Institute of
Mental Health and Addiction and Professor in cross-cultural education, Katholic University, Nigmegen
(KUN).
3 Professor, Department of Child and Adolescent Studies, Utrecht University, The Netherlands.
4 Assistant Professor, Department of Child and Adolescent Studies, Utrecht University, The
Netherlands.
5 Senior Researcher, The Netherlands Institute of Health Promotion and Disease Prevention, Utrecht,
The Netherlands.
401
0047-2891/01/0800-0401$19.50/0 °
C 2001 Plenum Publishing Corporation
402 Overbeek et al.
processes. The strong stability of internalizing and externalizing behavior suggests

that prevention efforts should be aimed at children and young adolescents.
INTRODUCTION
During the past 2 decades, several community-based epidemiological studies

have focused on the course and comorbidity of internalizing and externalizing dis-
orders in adolescence and young adulthood (for reviews see Angold and Costello,
1993; Loeber and Keenan, 1994; McConaughy and Skiba, 1993; Zoccolillo, 1992).
However, there has been a growing awareness that the course and co-occurrence of
internalizing and externalizing psychosocial problems deserve attention as well.
It is important to study psychosocial problems in adolescence and young adult-
hood because they limit the daily functioning of youth, as do psychiatric disorders.
Furthermore, certain psychosocial problems during adolescence can be precursors
of later psychiatric disorders during young adulthood (Ferdinand et al., 1999).
Accordingly, the first aim of this study is to examine the course of emotional dis-
turbance and delinquency in a longitudinal survey among adolescents and young
adults. The second aim is to study the co-occurrence of emotional disturbance and
delinquency. Finally, the third aim is to investigate the longitudinal associations
underlying this co-occurrence of emotional disturbance and delinquency.
The Course of Internalizing and Externalizing Problems
Previous studies on the course of internalizing problems in adolescence have

documented an increase with age in the mean level of depressive mood (e.g., Elliott
et al., 1989; Kandel and Davies, 1982). Fewer studies have reported on the course
of internalizing problems in young adulthood, although there are some indica-
tions that internalizing problems stabilize during this life period (Ferdinand and
Verhulst, 1995b, 1996). Among others, Kandel and Davies (1982), Elliott et al.
(1989), Campbell et al. (1992), Hankin et al. (1997, 1998), and Leadbeater et al.
(1999) have also revealed that adolescent girls report more intense internalizing
problems than boys do. Moreover, girls generally start developing internalizing
problems earlier in adolescence than boys do (e.g., Angold and Rutter, 1992;
Hankin et al., 1998; Nolen-Hoeksema, 1990). Numerous studies have shown a
peak in delinquency at midadolescence as well as a steady decline from midadoles-
cence to young adulthood (e.g., Hirschi and Gottfredson, 1983; Loeber et al., 1998;
Moffit, 1993); likewise, many studies have shown that boys report a higher delin-
quent activity than girls do (e.g., Rantakillio et al., 1995). It should be emphasized
that differences in the course of internalizing and externalizing problems during
adolescence and young adulthood do not necessarily imply corresponding differ-
ences in the co-occurrence and longitudinal associations of such problems during
these life periods (Rowe et al., 1994). It does, however, point to the importance of
Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 403
examining possible differences in the co-occurrence and longitudinal associations

between emotional disturbance and delinquency across age and gender categories.
Co-Occurrence of Internalizing and Externalizing Problems
Previous research has pointed to the existence of a valid co-occurrence be-

tween internalizing and externalizing psychosocial problems during adolescence
and young adulthood. However, mixed results have been reported concerning the
strength of the association. Community-based epidemiological studies conducted
by Elliott et al. (1989) and Garnefski and Diekstra (1997) demonstrated that the
co-occurrence between conduct problems and depressive symptoms ranged from
31 to 57%. Furthermore, Ferdinand and Verhulst (1996) found correlations be-
tween anxiety-depression and delinquent behavior subscales in young adults (0.24
for males and 0.44 for females). On the other hand, studies on the co-occurrence of
internalizing problems and substance use have only found small or insignificant re-
lationships (Pertraitis et al., 1995). An explanation for these contrasting results can
be found in studies by Silbereisen and Noack (1988), Shedler and Block (1990),
Moffitt (1993), Maggs et al. (1997), and Loeber et al. (1998), who have shown
that a certain degree of norm-violating or delinquent behavior can be seen as a part
of normal development and, as such, is no correlate of poor emotional health. In
their opinion, it is plausible that only the more serious externalizing behaviors are
linked to internalizing problems. Thus, although it may be expected that emotional
disturbance and delinquency co-occur during adolescence and young adulthood,
no clear expectations can be put forward regarding the strength of this association.
Four Alternative Perspectives on the Co-Occurrence of Internalizing

and Externalizing Problems
Some authors have stressed the need for studies on explanatory mecha-
nisms that underlie the co-occurrence of internalizing and externalizing problems
(Angold and Costello, 1993; Caron and Rutter, 1991; Wittchen, 1996). In this
paper, we focus on the longitudinal associations between emotional disturbance
and delinquency. Knowledge about these longitudinal associations is important
because it provides more insight into the natural course and developmental history
of internalizing and externalizing problems during adolescence and young adult-
hood. According to Caron and Rutter (1991) and Angold and Costello (1993), a
number of different interpretations of co-occurrence may be distinguished.
According to the stability perspective, a co-occurrence of internalizing and
externalizing problems is caused by nonspecific (i.e., shared or overlapping) risk
factors. Hence, certain risk factors can lead to internalizing as well as external-
izing problems, and no cross-lagged relationships can be specified between these
404 Overbeek et al.
problems. Krueger et al. (1998) and Krueger (1999) state that an association be-
tween internalizing and externalizing problems is caused by separate but associated
core psychopathological processes that refer to internal or external modulations
of basic feelings of anxiety. Different people are supposed to internalize and ex-
ternalize their anxiety feelings to different levels, and maintain their “personal”
internalizing and externalizing habits over their life course.
A second interpretation, the mutual influence perspective, also is based on
the assumption that a co-occurrence of internalizing and externalizing problems
is caused by shared or overlapping risk factors. However, according to this in-
terpretation, internalizing and externalizing problems are assumed to be mutually
reinforcing throughout a certain time interval. Since the same risk factors form
the basis of both problems, the development of one problem is expected to lead
to an increased vulnerability for the other and vice versa. Therefore, cross-lagged
relationships would be apparent between internalizing problems and externalizing
problems in the same time interval.
A third interpretation is generally known as the acting out perspective. Psy-
choanalytic theorists have stated that conduct problems are often part of an inter-
nalizing problem that is “acted out.” The depressive feelings of children and young
adolescents are supposed to be masked by disruptive behavior and other symptoms
that fit into the broader category of externalizing problems (Carlson and Cantwell,
1980). In more general terms, it is often assumed that internalizing problems not
only predate, but also predict externalizing problems (e.g., Gold et al., 1989). Thus,
in this perspective merely one cross-lagged relationship exists from internalizing
problems to externalizing problems.
A fourth interpretation, the failure perspective, also contains the thesis that
one problem is constituting a risk factor for the other. However, unlike the acting out
perspective, this perspective states that externalizing problems predate and predict
internalizing problems. According to Capaldi (1992) and Patterson and Capaldi
(1990), for example, noxious behavior and a lack of skills may result in rejection
and a lack of support by important others (parents, teachers, and peers), which
can in turn lead to pervasive failure experiences in social situations and school.
These failure experiences are viewed as leading to an increased vulnerability for
depressive moods. Therefore, it is stated that only one cross-lagged relationship
exists from externalizing problems to internalizing problems.
Previous Research on Explanatory Mechanisms: Temporal Patterns

and Longitudinal Associations
In part, the evidence for each of the 4 alternative perspectives is based on

very different types of empirical research: psychiatric–clinical studies, psychiatric–
epidemiological studies, and psychosocial–epidemiological studies. Psychiatric–
clinical studies are handicapped by Berkson’s bias: clinical samples contain a
disproportionately large number of patients having co-occurring problems (Caron

and Rutter, 1991). Moreover, they have generated contradictory results. Although
Puig-Antich (1982) found that major depressive disorder predated conduct dis-
order in preadolescent boys, Kovacs et al. (1988) showed that conduct disorder
generally postdated depression in pre- and early adolescents, while Harrington
et al. (1990) concluded that adolescents with both depressive and conduct dis-
orders had either a nosologically distinct syndrome or coped with depressions
of a secondary nature. Most psychiatric–epidemiological research has suggested
that externalizing disorders predate internalizing disorders (Anderson et al., 1987;
Block and Gjerde, 1990; Rohde et al., 1996). Although these studies provide in-
sight into the comorbidity of psychiatric disorders in adolescence, their results
cannot be directly related to the co-occurrence of psychosocial problems. Further-
more, psychiatric–epidemiological research has focused on the temporal patterns
between internalizing and externalizing problems, rather than on their longitudinal
associations.
Psychosocial–epidemiological research on explanatory mechanisms under-
lying the co-occurrence of internalizing and externalizing problems is relatively
scarce. Elliott et al. (1989) found evidence for a pattern in which minor offending
usually preceded the onset of mental health problems, and mental health prob-
lems generally preceded the onset of index offending in a general population
of adolescents. Only 1 study has focused on possible longitudinal associations
between internalizing and externalizing psychosocial problems. Capaldi (1992),
studying lower and working class, early adolescent boys from neighborhoods with
relatively high rates of delinquency, found that conduct problems at Grade 6 pre-
dicted increases in depressed mood by Grade 8. The results were concluded to
be consistent with a failure model, in which a lack of skills combined with nox-
ious behavior leads to pervasive failures and vulnerability to depressive moods.
However, the extent to which the results of this study can be generalized is ques-
tionable due to certain sample characteristics. Thus, no definite conclusions can
be drawn regarding each of the 4 alternative perspectives (i.e., stability, mutual
influence, acting out, failure). Moreover, possible age and gender differences in
the longitudinal associations between internalizing and externalizing problems
have remained unexamined. Therefore, this study concentrates on the longitudi-
nal associations between indicators of internalizing and externalizing problems
in a general population sample of both male and female adolescents and young
adults.
The Present Study
With the Utrecht Study of Adolescent Development (USAD, 1991–97), a

3-wave 6-year longitudinal study, we investigated the course, co-occurrence, and
longitudinal associations of emotional disturbance and delinquency in a large,
406 Overbeek et al.
community-based population of adolescents and young adults. Three main ques-

tions were addressed: (a) What is the course of emotional disturbance and delin-
quency during adolescence and young adulthood? (b) To what extent do emotional
disturbance and delinquency co-occur during adolescence and young adulthood?
and (c) What are the longitudinal associations between emotional disturbance and
delinquency in different age and gender categories during adolescence and young
adulthood? Regarding the first question, we expected that mean levels of emotional
disturbance and delinquency would increase from early until midadolescence,
after which emotional disturbance would stabilize and delinquency would decline
from late adolescence to young adulthood. Also, we expected that whereas girls
would have higher levels of emotional disturbance, boys would have higher levels
of delinquency. Regarding the second question, we expected a significant level of
co-occurrence between emotional disturbance and delinquency without any expec-
tations concerning the strength of this co-occurrence. We also had no expectations
with regard to the third question because previous research has not provided us
with conclusive evidence in favor or opposition of the stability, mutual influence,
acting out, or failure perspectives, and on possible age and gender differences in
the longitudinal associations between emotional disturbance and delinquency in
general population samples.
METHOD
Procedure and Sample Characteristics
Data for this study were collected as part of a broader longitudinal survey, the
Utrecht Study of Adolescent Development (USAD, 1991–97), in which the life
course trajectories of adolescents and young adults were examined (Helsen et al.,
2000; Meeus, 1996). To ensure the external validity of the research a nonclinical
sample was used, representative of the Dutch population for gender, age, religious
affiliation, residential status (e.g., living with parents, living alone, or living in a
student dormitory), and educational level (Meeus and ‘t Hart, 1993).
The first wave of the USAD was conducted in 1991. A national sample of
Dutch adolescents and young adults aged 12–24 was drawn from an existing
panel of 10,000 households. Subjects were interviewed in their home environment
by trained interviewers and were given a questionnaire to fill out on their own
and to send back to the research organization. Items about emotional disturbance
and delinquency were included in this self-report questionnaire because it was
expected that some respondents might have trouble answering these questions in
face-to-face interviews. Of all 3,394 adolescents and young adults who returned
the questionnaire, 3,186 respondents (89%) completed all questions regarding
emotional disturbance and delinquency, thereby constituting the baseline sample.
Four age groups were represented: early adolescence (12–14), midadolescence
(15–17), late adolescence (18–20) and young adulthood (21–24). Two follow-up
measurements took place in 1994 and 1997. About 61% of all baseline respondents
participated in the second wave of 1994, and approximately 66% of these also
cooperated in the third wave of 1997. The total response rate from Wave 1 to
Wave 3 was 41%.
An attrition analysis was carried out to test whether there were differences be-
tween the remaining longitudinal sample and those who dropped out from Wave 1
to Wave 3 according to gender, age, educational level, residential status, and
measures of emotional disturbance and delinquency. Logistic regression analy-
ses showed that attrition was associated with gender (B = −0.26, p < 0.01) and
age (B = −0.07, p < 0.01). Males and young adults dropped out more often
than females and adolescents. However, no significant differences were found be-
tween the participants and drop outs on measures of educational level, residential
status, emotional disturbance, and delinquency. Although the results of this study
must be interpreted with caution, the findings concerning emotional disturbance
and delinquency reported in this paper can be generalized to broader populations
of adolescents and young adults.
Finally, a total of 1,302 respondents participated in all 3 waves. This longitu-
dinal sample consisted of 550 boys (42%) and 752 girls (58%). The respondents
were evenly distributed over the 4 age categories (based on age at first wave): 321
early adolescents (25%), 341 midadolescents (26%), 261 late adolescents (20%),
and 379 young adults (29%). Further, the educational level of the respondents
could be differentiated as 18% low, 42% average, and 40% high. About 48% of all
respondents had a religious affiliation, while 52% had none. Approximately 78%
of all respondents lived at home with their parents, whereas 22% lived on their
own or in a student dormitory. In total, 99% of the sample consisted of adolescents
who were of Dutch origin, whereas 1% had a different ethnic background.
Measures
General Well-Being
General feelings of well-being were assessed using the Cantril Ladder

(Cantril, 1965). Respondents were asked to indicate on a 10-point scale how they
generally feel (1 – very bad, to 10 – very well).
Psychological Stress and Depressive Mood
Psychological stress and depressive mood were assessed, using a short version
of the General Health Questionnaire (GHQ; Goldberg, 1978; Kienhorst et al., 1990;
Meeus, 1993). The GHQ consists of 2 subscales “psychological stress” (6 items)
and “depressive mood” (4 items), and measures the degree to which psychological
408 Overbeek et al.
stress and depressive mood have recently been experienced. Respondents were
asked to indicate on a 4-point scale (1 – not at all, to 4 – much more than usual)
the extent to which symptoms of psychic stress (e.g., feeling tense and nervous)
or depression (e.g., feeling unhappy and dejected) had been experienced during
the past 4 weeks. Both scales had high internal consistencies: Cronbach’s alpha’s
for psychological stress and depressive mood were 0.89 and 0.84, respectively, at
Wave 1, 0.92 and 0.94, respectively, at Wave 2, and 0.91 and 0.92, respectively, at
Wave 3.
Suicidal Thoughts
The tendency to think about suicide was assessed with 1 item: “In the last
12 months, have you thought about committing suicide and putting an end to your
life?” (Diekstra et al., 1991). Respondents answered this question on a 4-point
scale (1 – never, to 4 – very often).
Emotional Disturbance
The measures mentioned earlier were highly interrelated, as was indicated

by correlation coefficients ranging from 0.31 to 0.71 ( p < 0.01). To determine
whether it was possible to obtain a single score for the construct of emotional
disturbance, an exploratory factor analysis was conducted, using the 4-scale scores
as variables. Analysis resulted in a single-factor solution (see also Helsen et al.,
2000). At Wave 1 and Wave 2, this solution explained about 59% of the total
variance (with an average factor loading of 0.76 at Wave 1 and 0.71 at Wave 2),
while at Wave 3 the solution explained 56% of the total variance (with an average
factor loading of 0.64). Therefore, each respondent was assigned a factor score,
using a short regression method, for the construct of emotional disturbance at each
of the 3 waves.
Delinquency
Delinquency was assessed as the number of delinquent acts the respondents

reported over the past 12 months. The delinquency measure consists of 21 items
pertaining to 3 types of delinquent behavior: violent crime (e.g., “Have you ever
wounded somebody with a knife or other weapon?”), vandalism (e.g., “Have you
ever covered walls, buses, or entryways with graffiti?”), and crime against property
(e.g., “Have you ever bought something which you knew was stolen?”). Subjects
answered if they had behaved in one of these ways during the past 12 months
on a 2-point scale (0 – no, to 1 – yes). The scores on the 21 items were then
summed. Thus, a higher score on this scale indicates a higher delinquent activity.
The internal consistency of the scale was Cronbach’s α = 0.62 at Wave 1, 0.60 at
Wave 2, and 0.58 at Wave 3 (for more details on the reliability and validity of this
delinquency measure, see Luijpers, 1999; ‘t Hart, 1994).
Strategy of Analysis
First, descriptive analyses (standardized means and standard deviations) of

the variables were calculated for boys and girls, as well as for the different age
groups (12–14 year olds, 15–17 year olds, 18–20 year olds, and 21–24 year olds).
In addition, a MANOVA with repeated measures was conducted to analyze the
course of emotional disturbance and delinquency across the 3 waves, controlling
for the effects of age and gender. Second, correlation coefficients were computed
to examine the co-occurrence of emotional disturbance and delinquency. These
data were also used as input for the construction of covariance matrices for the
structural equation modeling analyses. Third, longitudinal equation modeling was
carried out using the LISREL 8.30-program (Jöreskog and Sörböm, 1993), to
analyze the longitudinal associations of emotional disturbance and delinquency
during adolescence and young adulthood. Respondents’ scores on the continuous
variables of emotional disturbance and delinquency were standardized and then
used to calculate their mutual covariance structure. The covariance matrices were
subsequently estimated in PRELIS 2.30, using the weighted least squares method,
because of the highly nonnormal distribution of the scores (Jöreskog and Sörböm,
1989).
Four different structural equation models were specified based on the per-
spectives presented earlier: a stability model, an acting out model, a failure model,
and a mutual influence model. In each of these models, emotional disturbance
and delinquency were specified as observed variables. Furthermore, each of these
4 models contained (a) stability paths from Wave 1 to Wave 2 and from Wave 2
to Wave 3 for emotional disturbance and delinquency, (b) co-occurrence paths
between emotional disturbance and delinquency at each of the 3 waves, and (c) es-
timates of residual information for emotional disturbance and delinquency at each
of the 3 waves (see Fig. 1). However, the stability model deviated from the other
3 models in 2 important respects. First, extra stability paths were specified from
Wave 1 to Wave 3 for both emotional disturbance and delinquency, while these
were not specified in the other 3 models. Second, no cross-lagged relations be-
tween emotional disturbance and delinquency were specified in the stability model,
as was done in the other 3 models. The acting out model contained cross-lagged
relations between emotional disturbance at Wave 1 and Wave 2 to delinquency at
Wave 2 and Wave 3, respectively. For the failure model, the cross-lagged relations
were specified in the opposite direction: from delinquency at Wave 1 and Wave 2
to emotional disturbance at Wave 2 and Wave 3, respectively. Finally, the mutual
influence model contained all of these cross-lagged relationships. The 4 competing
410 Overbeek et al.
Fig. 1. Four alternative structural models for the longitudinal associations of emotional disturbance
and delinquency (R indicates residual information).
models were fitted on the data of the total longitudinal sample. After this first test,
several multigroup analyses were carried out to test if the alternative models fit
across different age and gender groups.
Model fit was assessed by the following global fit measures: χ 2 , χ 2 /df ratio,
RMSEA, GFI, CFI, and AIC. The χ 2 provides a significance test of the null
hypothesis that the model is correct. However, this statistic is extremely vulnerable
to the effects of sample size. Large sample sizes lead to a tendency to reject a
model, even when most of the covariance in the data is accounted for (Williams
and Holahan, 1994). The χ 2 /df ratio, which indicates the fit of a model per df used,
was proposed by Jöreskog (1969) as a fit index that could account for sample size
effects associated with χ 2 . A threshold value smaller or equal to 2 is proposed
to indicate an acceptable fit (Wheaton et al., 1977). The Root Mean Square Error
of Approximation (RMSEA, see Steiger, 1990) is another measure suitable for
assessing the fit of a model per df used. A value of 0.05 or less indicates a close fit
of the model to the data (Browne and Cudeck, 1993). The Goodness of Fit Index
(GFI) is an estimate of the extent to which the sample variances and covariances
are reproduced by the hypothesized model. Bentler’s Comparative Fit Index (CFI,
see Bentler, 1989) is an incremental fit index derived from the comparison of
the hypothesized model with a null model in which no relationships between the
variables in a model are specified. For both the GFI and the CFI, a value of 0.90 and
higher indicates an acceptable fit. Finally, Akaike’s Information Criterion (AIC,
see Akaike, 1987) is used as a parsimony-based fit index. The model that yields
the smallest AIC-value can be considered the most parsimonious.
RESULTS
The Course of Emotional Disturbance and Delinquency
The standardized means and standard deviations for emotional disturbance

showed a strong continuity over the 3 waves for the total sample (Table I). Dif-
ference contrasts in a repeated measures MANOVA demonstrated that there were
no differences between the mean levels of emotional disturbance from Wave 1 to
Wave 2 (F(1, 1298) = 0.01, p > 0.05) and from Wave 2 to Wave 3 (F(1, 1298) =
0.01, p > 0.05). However, there was a clear difference between boys and girls in
their mean level of emotional disturbance. Overall, girls were more emotionally
disturbed (F(1, 1298) = 34.36, p < 0.001). Furthermore, there was a strong in-
teraction between the respondent’s age and mean level of emotional disturbance.
An increase in the mean level of emotional disturbance was apparent from early
adolescence to midadolescence (F(3, 1298) = 7.65, p < 0.001), while from late
adolescence to young adulthood the mean level of emotional disturbance sta-
bilized, F(3, 1298) = 3.59, p < 0.05. When the effects of both age and gen-
der on the longitudinal change of emotional disturbance were combined, it be-
came clear that for girls the increase of emotional disturbance occurred from
early to midadolescence, whereas for boys the mean level of emotional distur-
bance increased from midadolescence to late adolescence (F(4, 1299) = 3.87,
p < 0.01).
A different pattern emerged for the longitudinal change of delinquency dur-
ing adolescence and young adulthood (Table I). As with emotional disturbance,
the difference contrasts indicated that there was a strong continuity over the 3
waves for the total sample. No differences were apparent in the mean level of
delinquency from Wave 1 to Wave 2 (F(1, 1298) = 0.00, p > 0.05) and from
Wave 2 to Wave 3 (F(1, 1298) = 0.09, p > 0.05). However, gender had a strong
main effect (F(1, 1298) = 141.24, p < 0.001). Boys consistently reported higher
levels of delinquency than girls did. Another main effect was found for age
(F(3, 1298) = 23.71, p < 0.001). From adolescence to young adulthood, there
was a constant decline in the level of delinquency. There was also a strong in-
teraction between delinquency and age. While there was a sharp increase in the
delinquency of youth from early adolescence to midadolescence (F(4, 1298) =
10.29, p < 0.001), there was a steady decline in delinquency from late ado-
lescence to young adulthood (F(4, 1298) = 6.52, p < 0.001). When the effects
of both age and gender were examined, the increase in delinquency from
early adolescence to midadolescence appeared to be less intense for girls
(F(4, 1298) = 3.79, p = 0.01), and the decline in delinquency from late ado-
lescence to adulthood was stronger for girls than for boys (F(4, 1298) = 3.59,
p < 0.05).
Overall, the descriptive analysis showed that the mean level of emotional
disturbance increased from early through midadolescence, and stabilized during
P1: GFU/LCCR/GAY
412
Journal of Youth and Adolescence [jya]
P2: GCR/GFU
Table I. Standardized Means and Standard Deviations for Emotional Disturbance and Delinquency (N = 1302)
Gender Agea
Total Boys Girls 12–14 15–17 18–20 21–24
PP210-342604
Emot. dist.b
T1 −0.01 (1.00) −0.17 (0.88) 0.12 (1.06) −0.24 (0.92) −0.01 (1.01) 0.07 (1.00) 0.15 (1.01)
T2 −0.02 (1.00) −0.16 (0.91) 0.11 (1.05) 0.02 (1.09) −0.01 (0.98) −0.01 (1.00) 0.00 (0.93)
T3 −0.04 (1.00) −0.12 (0.88) 0.08 (1.07) 0.04 (0.98) −0.10 (0.88) −0.08 (0.85) 0.05 (1.13)
Delinquencyc
T1 0.01 (1.00) 0.28 (1.19) −0.20 (0.77) 0.01 (0.94) 0.26 (1.23) 0.00 (98) −0.23 (0.74)
July 17, 2001
T2 0.02 (1.00) 0.28 (1.20) −0.20 (0.77) 0.22 (1.22) 0.08 (1.09) −0.10 (0.88) −0.19 (0.69)
T3 0.02 (1.00) 0.29 (1.25) −0.21 (0.70) 0.28 (1.25) 0.05 (1.07) 0.03 (0.86) −0.26 (0.67)
Note. Emot. dist. = Emotional disturbance.

11:40
a Longitudinal data are presented vertically for each of the age groups; cross-sectional data are presented horizontally.
b Standardized scores for emotional disturbance in the total sample range from −1.17 to 5.33 (T1), −1.26 to 5.77 (T2), and −1.28 to 5.34 (T3).
c Standardized scores for Delinquency in the total sample range from −0.59 to 5.96 (T1), −0.59 to 7.84 (T2), and −0.53 to 6.15 (T3).
Overbeek et al.
Style file version Nov. 19th, 1999
late adolescence and young adulthood. Girls had higher levels of emotional distur-
bance and also experienced a rise in emotional disturbance earlier in adolescence
than did boys. The mean level of delinquency also increased from early through
midadolescence, but declined again in late adolescence and young adulthood. Boys
reported higher levels of delinquency as well as a stronger increase in delinquency
than did girls, during adolescence.
Co-Occurrence of Emotional Disturbance and Delinquency
As can be seen in Table II, the co-occurrence of emotional disturbance and

delinquency was relatively weak in the total sample, as is indicated by correla-
tion coefficients ranging from 0.08 ( p < 0.001) to 0.10 ( p < 0.001). Neverthe-
less, all correlations for within-measurement association were significant, which
means that the co-occurrence of emotional disturbance and delinquency reached
a higher-than-chance level. Between measurement or cross-lagged associations
of emotional disturbance and delinquency were somewhat lower and not always
significant, ranging from 0.04 (ns) to 0.08 ( p < 0.001). In contrast to the relatively
weak co-occurrence and cross-lagged coefficients, the stability coefficients of emo-
tional disturbance and delinquency reached moderate values in the total sample.
These “normative” correlations, which inform us of the extent to which individuals
hold the same relative position to one another in a group over time (Verhulst and
Van der Ende, 1992), ranged from 0.31 ( p < 0.001) to 0.41 ( p < 0.001).
Descriptive as well as correlational analyses demonstrated the importance
of examining longitudinal associations between emotional disturbance and delin-
quency across different gender and age categories. First, the descriptive analysis
showed strong gender and age differences in the longitudinal changes of emo-
tional disturbance and delinquency. Second, the correlation coefficients between
emotional disturbance and delinquency also fluctuated across gender and age cat-
egories. Girls seemed to have a stronger co-occurrence of emotional disturbance
and delinquency than boys do and appeared to report higher cross-lagged as-
sociations as well. In addition, younger adolescents seemed to have a stronger
co-occurrence of emotional disturbance and delinquency as compared to older
adolescents and young adults, and they seemed to report higher cross-lagged asso-
ciations. Thus, both analyses indicated that the explanatory mechanisms underly-
ing the co-occurrence of emotional disturbance and delinquency might be different
across gender and age categories.
Longitudinal Associations of Emotional Disturbance and Delinquency
LISREL analysis of the total sample on the 4 alternative models showed that
the stability model offered the closest fit (Table III). Its χ 2 /df ratio almost reached
the threshold value of 2, while all other fit indices indicated a close fit of the
P1: GFU/LCCR/GAY
414
Table II. Correlation Coefficients Between Emotional Disturbance and Delinquency (N = 1,302)
Gender Agea
P2: GCR/GFU
Total Boys Girls 12–14 15–17 18–20 21–24
Co-occurrence and cross-lagged correlations

Emot. dist. T1 Delinquency T1 0.08∗∗∗ 0.06 0.17∗∗∗ 0.26∗∗∗ 0.11∗∗ 0.01 −0.04
Delinquency T2 0.04 0.01 0.14∗∗∗ 0.13∗∗∗ 0.08 0.01 −0.02
Delinquency T3 0.00 0.03 0.05 0.05 −0.02 0.08 0.01
PP210-342604
Emot. dist. T2 Delinquency T1 0.07∗∗ 0.08∗∗ 0.14∗∗∗ 0.11∗∗ 0.10∗∗ 0.06 0.00
Delinquency T2 0.10∗∗∗ 0.12∗∗∗ 0.16∗∗∗ 0.16∗∗∗ 0.13∗∗∗ 0.03 0.01
Delinquency T3 0.08∗∗∗ 0.12∗∗∗ 0.13∗∗∗ 0.13∗∗∗ 0.09∗∗ 0.07 −0.02
Emot. dist. T3 Delinquency T1 0.04 0.06 0.06∗∗ 0.07 0.08 −0.01 0.02
Delinquency T2 0.04 0.02 0.10∗∗∗ 0.05 0.05 0.01 0.03
Delinquency T3 0.10∗∗∗ 0.09∗ 0.19∗∗∗ 0.16∗∗∗ 0.11∗∗ 0.12∗∗ 0.05
Auto correlations: Emotional disturbance
July 17, 2001
Emot. dist. T1 Emot. Dist. T2 0.38∗∗∗ 0.41∗∗∗ 0.34∗∗∗ 0.44∗∗∗ 0.39∗∗∗ 0.32∗∗∗ 0.38∗∗∗
Emot. Dist. T3 0.31∗∗∗ 0.33∗∗∗ 0.28∗∗∗ 0.30∗∗∗ 0.25∗∗∗ 0.37∗∗∗ 0.36∗∗∗
Emot. dist. T2 Emot. Dist. T3 0.40∗∗∗ 0.41∗∗∗ 0.38∗∗∗ 0.30∗∗∗ 0.25∗∗∗ 0.37∗∗∗ 0.36∗∗∗
Auto correlations: Delinquency
11:40
Delinquency T1 Delinquency T2 0.41∗∗∗ 0.39∗∗∗ 0.35∗∗∗ 0.40∗∗∗ 0.44∗∗∗ 0.42∗∗∗ 0.35∗∗∗

Delinquency T3 0.35∗∗∗ 0.36∗∗∗ 0.22∗∗∗ 0.31∗∗∗ 0.45∗∗∗ 0.15∗∗∗ 0.41∗∗∗
Delinquency T2 Delinquency T3 0.39∗∗∗ 0.39∗∗∗ 0.27∗∗∗ 0.36∗∗∗ 0.44∗∗∗ 0.27∗∗∗ 0.34∗∗∗
Note. Emot. dist. = Emotional disturbance.

a Longitudinal data are presented vertically for each of the age groups; cross-sectional data are presented horizontally.
∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.
Overbeek et al.
P1: GFU/LCCR/GAY
Table III. Weighted Least Square Estimations of the Stability Model, Acting Out Model, Failure Model, and Mutual
Influence Model (N = 1,302)
P2: GCR/GFU
Fit index
χ 2 (df ) p-value χ 2 /df ratio GFI CFI RMSEA AIC
Whole sample
Stability 13.01 (6) 0.04 2.17 0.98 0.97 0.03 43.01
Acting out 67.63 (6) 0.00 11.27 0.92 0.78 0.09 97.63
PP210-342604
Failure 190.94 (6) 0.00 31.82 0.99 0.97 0.11 132.69

Mut. infl. 63.15 (4) 0.00 15.79 0.93 0.79 0.11 97.15
Multigroup: Gender differences
Stability 55.63 (21) 0.00 2.65 0.97 0.83 0.05 97.63
Acting out 107.13 (21) 0.00 5.10 0.95 0.58 0.08 149.13
Failure 98.83 (21) 0.00 4.70 0.96 0.62 0.08 140.83
July 17, 2001
Mut. infl. 94.42 (19) 0.00 4.97 0.96 0.63 0.08 140.42
Multigroup: Age differences
Stability 107.21 (51) 0.00 2.10 0.96 0.84 0.06 173.21
Acting out 167.29 (51) 0.00 3.28 0.92 0.67 0.08 233.29
11:40
Failure 161.81 (51) 0.00 3.17 0.92 0.68 0.08 227.81

Mut. infl. 160.33 (49) 0.00 3.27 0.92 0.68 0.08 230.33
Note. Mut. inf. = Mutual influence.

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood
415
416 Overbeek et al.
Fig. 2. Parameter estimations for each of the 4 alternative structural models in the total sample. Only
significant path coefficients are presented. Superscript a indicates the stability model provided an
adequate fit for the total sample and the closest fit for different age and gender categories.
stability model to the data of the total sample. The acting out model and mutual
influence model both fitted poorly, as was indicated in higher values for the χ 2 /df
ratio, RMSEA, and AIC, and lower values for the GFI and CFI. Furthermore,
the specified cross-lagged relationship from emotional disturbance at Wave 1 to
delinquency at Wave 2 had a negative value instead of a positive one, which
was not in line with a priori specifications of the acting out model and mutual
influence model. For the failure model, high GFI and CFI values were obtained.
Nevertheless, its χ 2 /df ratio and AIC value were rather high, indicating that the
failure model fit less well than the stability model. Although all cross-lagged
relationships specified in the acting out model, failure model, and mutual influence
model attained significance (see Fig. 2), none of these improved the fit over the
already existent stability paths.
As is shown in Table III, a multigroup LISREL analysis of gender differences
made clear that the stability model best represented the longitudinal associations
between emotional disturbance and delinquency for both boys and girls. Although
the CFI indicated an insufficient fit, the other fit- indices (which also take the
parsimony of a model into account) presented another picture. The χ 2 /df ratio lay
close at a threshold value of 2, whereas the RMSEA indicated a close fit of the model
to the data, and the AIC also had a relatively low value. The acting out model, failure
model, and mutual influence model did not fit on the data of both gender groups,
as was clearly indicated by the various fit indices (Table III). As in the previous
analysis, the relationship from emotional disturbance at Wave 1 to delinquency at

Wave 2 had a negative value, that was not in line with a priori specifications in the
acting out model and mutual influence model. Although all specified cross-lagged
relationships between emotional disturbance and delinquency were significant over
a 3-year period for both boys and girls, none of these improved the fit over the
stability paths.
Finally, a multigroup LISREL analysis was conducted to examine whether
the longitudinal associations between emotional disturbance and delinquency were
the same across age groups. As with the multigroup analysis for gender, it became
clear that the stability model provided the best fit with the data of all age groups
(Table III). Apart from the CFI, the χ 2 /df ratio, GFI, RMSEA, and AIC all indicated
an adequate fit with the stability model. The other 3 models did not properly fit the
data for different age groups. As in the previous LISREL analyses, the relationship
from emotional disturbance at Wave 1 to delinquency at Wave 2 had a negative
value, that was not in agreement with the a priori specifications of the acting out
model and mutual influence model. Although all estimates of the cross-lagged
relations between emotional disturbance and delinquency reached significance
over a 3-year period, none of these improved the fit of the models over the stability
paths in the various models.
In conclusion, although descriptive and correlational analyses suggested that
there might be different longitudinal associations between emotional disturbance
and delinquency in different age and gender groups, multigroup LISREL analyses
clearly showed that only 1 model fit the data of different age and gender groups
adequately.6 Clearly, the stability model is the best representation of the longitu-
dinal associations between emotional disturbance and delinquency for both boys
and girls and for all age groups studied. Cross-lagged relationships between emo-
tional disturbance and delinquency are relatively weak (at least over these periods
of time), and have little predictive power as compared to the strong stability paths.
Because our LISREL analyses revealed no differences across gender and age cat-
egories, the stability model is presented for the total sample in Fig. 2.
DISCUSSION
Although previous studies have often relied on clinical samples (e.g.,

Harrington et al., 1990; Kovacs et al., 1988; Puig-Antich, 1982), samples con-
sisting only of males (e.g., Block and Gjerde, 1990; Capaldi, 1992), or samples
consisting of early adolescents (e.g., Anderson et al., 1987; Carlson and Cantwell,
1980), this study offered us the possibility to examine the course, co-occurrence,
6 Additional LISREL analyses were conducted for each gender and age group separately. The stability
model had an adequate fit in each of these different groups, whereas the acting out model, failure
model, and mutual influence model did not fit the data well. Additional data can be obtained from the
first author.
418 Overbeek et al.
and longitudinal associations of emotional disturbance and delinquency in a non-

clinical, general population sample of both male and female adolescents and young
adults.
Longitudinal Change and Stability of Emotional Disturbance

and Delinquency
This study demonstrated that from early through midadolescence the mean
level of emotional disturbance increases, which is in accordance with results of
previous research (Elliott et al., 1989; Kandel and Davies, 1982). Moreover, our
results showed that internalizing problems stabilize from late adolescence to young
adulthood, as was also indicated by studies of Ferdinand and Verhulst (1995,
1996). Further, our study corresponds to earlier findings that girls report higher
levels of emotional disturbance and start developing internalizing problems earlier
in adolescence than boys do (Campbell et al., 1992; Elliott et al., 1989; Hankin
et al., 1997, 1998; Kandel and Davies, 1982; Leadbeater et al., 1999). This study
also showed a peak in delinquency at midadolescence and a steady decline from
midadolescence until young adulthood, which is in agreement with research on
the age–crime curve (Hirschi and Gottfredson, 1983; Loeber et al., 1998; Moffitt,
1993). Our finding that boys report a higher delinquent activity than girls do has
been found in numerous other studies as well (e.g., Rantakillio et al., 1995). Thus,
our expectations concerning the course of emotional disturbance and delinquency
during adolescence and young adulthood were confirmed. Overall, the findings
indicate that adolescence is an important risk period for the development of more
severe internalizing and externalizing psychosocial problems, whereas emotional
disturbance stabilizes and delinquency decreases steadily during young adulthood.
According to Steinberg (1987), a substantial amount of adolescents’ risk
behavior might be caused by a difference in opinions between adolescents and
their parents over the pace of the process of becoming independent. The notion of
juvenile delinquency as a form of testing personal boundaries and exploring values
and beliefs fits in with this explanation, because adolescents will probably want to
“push their limits” and see how far they can go in certain life domains. Moreover,
it is closely linked to Arnett’s conception of adolescent risk behavior as a form
of thrill seeking (Arnett, 2000), which concerns the active seeking out of novel
and intense experiences. Probably, the decline of delinquent behavior in young
adulthood is due to the fact that other risk behaviors (e.g., binge drinking, risky
sexual behavior) become more prominent and replace earlier forms, since these
behaviors can be pursued more easily as soon as monitoring parents are absent
(Arnett, 1999, 2000).
Although the results demonstrated fluctuations in the mean level of emo-
tional disturbance and delinquency during adolescence and young adulthood, the
normative stability coefficients (i.e., auto correlations for emotional disturbance
and delinquency) were of a moderate strength, indicating that a large number of

adolescents and young adults are experiencing similar levels of emotional distur-
bance and delinquency relative to their peers over time. The 3-year and 6-year
stability coefficients found in this study were somewhat lower than those found in
other community-based studies among adolescents and young adults, but were of a
roughly comparable strength. Ferdinand et al. (1995), for example, found stability
estimates over a 4-year interval of anxious/depressed and delinquency syndromes
to be 0.43 and 0.41, respectively in 15–18-year-old adolescents. Achenbach et al.
(1995) found stability coefficients over a 3-year period of 0.53 and 0.44 for the
same anxious/depressed and delinquency syndromes, respectively, in 16–19-year-
old adolescents. Among researchers who have concentrated explicitly on the sta-
bility of depressive mood, Kandel and Davies (1986) found stability estimates over
a 9-year interval of 0.35 for boys and 0.44 for girls in a sample of adolescents aged
15–16.
Co-Occurrence of Emotional Disturbance and Delinquency
The co-occurrence correlations found in this study ranged from 0.08 to 0.10
over the 3 waves. Thus, when the focus is on the associations of internalizing
and externalizing psychosocial problems in a normal population of adolescents
and young adults rather than on psychiatric disorders, there is a significant but
low co-occurrence between emotional disturbance and delinquency. There are
some possible explanations for the different results of studies focusing on psychi-
atric disorders and psychosocial problems. For example, substance use in adoles-
cence is often considered to be a risk factor for later psychosocial problems (e.g.,
Aneshensel and Huba, 1983; Brook et al., 1998; Damphousse and Kaplan, 1998).
However, numerous studies have shown no or only small associations between
low self-esteem, depression, or psychological stress on the one hand and, for in-
stance, alcohol use on the other (Pertraitis et al., 1995). Thus, for the majority of
adolescents, alcohol use is quite normal and not a signal of threatened emotional
development. A similar argument can be made regarding delinquency. Previous
research has clearly shown that a certain amount of norm-violating or delinquent
behavior can be seen as a part of normal development and is not associated with
severe emotional problems in adolescence (Loeber et al., 1998; Maggs et al., 1997;
Moffitt, 1993; Shedler and Block, 1990; Silbereisen and Noack, 1988). Testing per-
sonal boundaries and exploring values and beliefs are normative behaviors during
adolescence and serve important developmental ends (Erikson, 1968; Havighurst,
1972). In addition, the relatively low co-occurrence found in this study seems to
indicate that only the more serious delinquent behaviors are linked to emotional
disturbance. It is possible that if we had concentrated on more serious externaliz-
ing behaviors, as for example Capaldi (1992) did, we would have found stronger
relationships with emotional disturbance.
420 Overbeek et al.
Longitudinal Associations Between Emotional

Disturbance and Delinquency
Despite the age and gender differences in the course and co-occurrence of
emotional disturbance and delinquency, the stability model provided an adequate
fit for the data of the total sample, and provided the closest fit for different age and
gender categories. The cross-lagged mechanisms that were examined in the co-
occurrence of emotional disturbance and delinquency did not fit the data well. This
finding stands in contrast with the results of psychiatric–clinical or psychiatric–
epidemiological studies, which have often pointed toward the temporal primacy
of depressive over conduct disorders or vice versa (e.g., Block and Gjerde, 1990;
Kovacs et al., 1988; Rohde et al., 1996). An explanation for this might be the
fact that whereas this study focused on delinquency, most other studies focused
on a broader category of conduct problems. It is likely that behaviors that are
seen as problematic but are not manifestations of delinquency (e.g., frequent argu-
ments with parents, lying, destroying one’s own property, teasing) have a stronger
relationship with emotional disturbances in adolescence and young adulthood.
The results of this study also contradict the findings of earlier psychosocial–
epidemiological research conducted by Elliott et al. (1989) and Capaldi (1992).
While these two studies pointed to the existence of specific temporal patterns
and cross-lagged longitudinal associations, our study demonstrated that a stability
model was the best representation for the longitudinal associations between inter-
nalizing and externalizing psychosocial problems. Two reasons might be put for-
ward to explain the different results. First, different samples were used. Capaldi’s
sample, for instance, consisted of early adolescent boys living in a neighborhood
with a relatively high delinquency rate, whereas the sample in this study consisted
of adolescent and young adult males and females from the general population.
Cross-lagged influence between various manifestations of internalizing and exter-
nalizing problems is probably only relevant for “at risk” populations. An indication
of this phenomenon might be the fact that co-occurrence appears to be related to a
more severe prognosis of psychic problems as well (Angold and Costello, 1993).
Accordingly, it might be stated that internalizing and externalizing problems must
develop beyond a certain “intensity-threshold” before they become risk factors for
the development of subsequent psychic problems. Second, different time intervals
were used between the measurements. In the studies of Elliott et al. and Capaldi
the time intervals between measurements were 1 year and 2 years, respectively,
while in this study the time interval spanned 3 years. Shorter time intervals may
lead to stronger cross-lagged relationships in longitudinal research, because over
longer periods of time certain intermittent variables (i.e., negative life events, status
transitions) have a higher chance of influencing these relationships. On the other
hand, it might be stated that even with shorter time intervals between waves, the
relatively low cross-sectional co-occurrence between emotional disturbance and
delinquency would still lead us to expect weak cross-lagged associations.
Stability of Psychosocial Problems During Adolescence

and Young Adulthood
The moderate stability and relatively low co-occurrence during adolescence

and young adulthood lead to a structural model in which emotional disturbance and
delinquency are predicted only by their earlier manifestations. This argues for the
importance of controlling for the influence of early manifestations of internalizing
and externalizing problems in the prediction of their subsequent development. The
relatively high stability of emotional disturbance and delinquency during adoles-
cence and young adulthood does not necessarily imply that once an internalizing
or externalizing problem has begun to develop, this process is unchangeable. It
merely implies that even if a person scores below a clinical–diagnostic range, this
person still might have the highest score relative to his or her peers (Krueger et al.,
1998).
The results of this study are in agreement with the assumptions of Krueger
et al. (1998) about “core psychopathological processes,” which refer to modula-
tions of basic feelings of anxiety that can be directed in one of two ways: internal
or external. This study showed that different adolescents internalize or externalize
their basic feelings of anxiety to different levels and, to a certain extent, maintain
their “personal” internalizing or externalizing habits at least until young adulthood.
Further, the results of our study indicate that a co-occurrence of internalizing and
externalizing problems might be caused by certain nonspecific risk factors, some
of which have been revealed in previous research. Parental depression, a family
history of mental illness or criminality, family discord, chronically adverse life
circumstances, alcoholism in a first degree relative, and acute adverse life events
all constitute risk factors for the development of both depressive and conduct
problems (Chiles et al., 1980; Costello, 1989; Rutter, 1989; Rutter and Quinton,
1984). Most of these risk factors seem to emphasize the importance of certain
environmental influences in the development of psychopathology. However, some
behavioral–genetic sibling studies have demonstrated that approximately half of
the covariance between depressive symptoms and antisocial behavior can be ex-
plained by a common genetic liability, while a relatively minor role was played by
shared or nonshared environmental influences (O’Connor et al., 1998a,b). Thus,
more research is needed before any firm conclusions on the origin of nonspecific
risk factors can be formulated.
Limitations
Some shortcomings of this study should be mentioned. First, our study cannot
elucidate whether the rather low estimates of the cross-lagged relationships (found
for the acting out model, failure model, and mutual influence model) have been
caused by the relatively long time interval between each of the waves. During
422 Overbeek et al.
the tumultuous years of adolescence, the precision of predictions can be nega-

tively influenced by a lack of insight into the individual’s development between
waves (Engels et al., 1999; Kraus, 1995). Thus, it remains possible that in a lon-
gitudinal study with 1 measurement each year, stronger cross-lagged paths would
appear. Although our study cannot show whether this is the case, the significant
cross-lagged paths over 3-year time intervals do give an indication of the exis-
tence of stronger cross-lagged relationships between emotional disturbance and
delinquency.
Second, although the measures that made up the construct of emotional distur-
bance can be considered valid and reliable instruments (Cantril, 1965; Goldberg,
1978; Kienhorst et al., 1990; Meeus, 1993), the delinquency-measure used in this
study seems to be only moderately reliable. This is probably due to the broad
range of different delinquent behaviors that were taken into account (i.e., violent
crime, vandalism, and crime against property) as well as the dichotomous answer-
ing options. Nevertheless, our results concerning the course of delinquency during
adolescence and young adulthood resembled those of numerous earlier studies,
while the relatively high stability of delinquency over the 3 waves also justifies the
use of this instrument.
Third, an attrition analysis showed that our results should be interpreted with
caution, because of the selective attrition of males and young adults in the sample.
On the other hand, no significant differences were found between the partici-
pants and drop outs on measures of emotional disturbance and delinquency, which
means that the findings concerning emotional disturbance and delinquency may
be generalized to broader populations of adolescents and young adults.
Implications for Future Research and Prevention Efforts
The temporal order of various internalizing and externalizing problems has

been the main focus of many studies, whereas research on longitudinal associations
underlying a co-occurrence of these problems is very scarce. Knowledge about the
latter subject is needed because it provides more insight into the natural course
and developmental history of internalizing and externalizing problems. It is import
that longitudinal research be designed to have 1 measurement each year. In that
way, a comparison can be made between short-term and long-term cross-lagged
relationships of the problems under study. It also seems important that future
research focuses systematically on the identification of nonspecific risk factors
for both the development of internalizing and externalizing problems. Finally, it
is important that research which seeks to explain manifestations of internalizing
and externalizing problems incorporates previous manifestations of these same
problems as independent predictors.
With regard to possible prevention efforts, the relatively high stability of emo-
tional disturbance and delinquency suggests that interventions should take place
as soon as possible. Children who have high levels of internalizing or external-

izing problems early in life, run a risk of developing relatively higher levels of
these same problems in adolescence and young adulthood. To enable profession-
als to identify psychic problems at an early stage, the setup of a screening system
aimed at the psychosocial problems of youth may be helpful. Last, the results of
our study showed that in adolescence the mean level of emotional disturbance
increases, whereas young adulthood is a period in which emotional disturbance
stabilizes. Therefore, young adulthood seems to be a period in which the treatment
of internalizing problems becomes increasingly important, whereas in adolescence
(a life phase in which internalizing problems often have their onset) the main focus
should be on prevention efforts.
REFERENCES
Achenbach, T. M., Howell, C. T., McConaughy, S. H., and Stanger, C. (1995). Six-year predictors of
problems in a national sample of children and youth: II. Signs of disturbance. J. Am. Acad. Child
Adolesc. Psychiatry 34(4): 488–498.
Akaike, H. (1987). Factor analysis and AIC. Psychometrika 52: 317–332.
Anderson, J., Williams, S., McGee, R., and Silva, P. A. (1987). The prevalence of DSM-III disorders
in a large sample of preadolescent children from the general population. Arch. Gen. Psychiatry
44: 69–81.
Aneshensel, C. S., and Huba, G. J. (1983). Depression, alcohol use, and smoking over one year: A
four-wave longitudinal causal model. J. Abnorm. Psychol. 92(2): 134–150.
Angold, A., and Costello, E. J. (1993). Depressive comorbidity in children and adolescents: Empirical,
theoretical and methodological issues. Am. J. Psychiatry 150(12): 1779–1791.
Angold, A., and Rutter, M. (1992). Effects of age and pubertal status on depression in a large clinical
sample. Dev. Psychopathol. 4: 5–28.
Arnett, J. J. (1999). Adolescent storm and stress, reconsidered. Am. Psychol. 54: 317–326.
Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the
twenties. Am. Psychol. 55: 469–480.
Bentler, P. M. (1989). EQS: Structural Equations Program Manual. BMDP Statistical Software, Los
Angeles.
Block, J., and Gjerde, P. F. (1990). Depressive symptoms in late adolescence: A longitudinal perspec-
tive on personality antecedents. In Rolf, J., Masten, A. S., Cicchetti, D., Nuechterlein, K. H.,
and Weintraub, S. (eds.), Risk and Protective Factors in the Development of Psychopathology.
Cambridge University Press, New York, pp. 334–360.
Brook, J. S., Cohen, P., and Brook, D. W. (1998). Longitudinal study of co-occuring psychiatric
disorders and substance use. J. Am. Acad. Child Adolesc. Psychiatry 37(3): 322–330.
Browne, M. W, and Cudeck, R. (1993). Alternative ways of assessing model fit. In Bollen, K. A. and
Scott Long, J. (eds.), Testing Structural Equation Models. Sage, Newbury Park, CA, pp. 136–162.
Campbell, T. L., Byrne, B. M., and Baron, P. (1992). Gender differences in the expression of depressive
symptoms in early adolescents. J. Early Adolesc. 12: 326–338.
Cantril, H. (1965). The Pattern of Human Concerns. Rutgers University Press, NJ.
Capaldi, D. M. (1992). Co-occurrence of conduct problems and depressive symptoms in early adoles-
cent boys: II. A 2-year follow-up at grade 8. Dev. Psychopathol. 4: 125–144.
Carlson, G. A., and Cantwell, D. P. (1980). Unmasking masked depression in children and adolescents.
Am. J. Psychiatry 137(4): 445–449.
Caron, C., and Rutter, M. (1991). Comorbidity in child psychopathology: Concepts, issues and research
strategies. J. Child Psychol. Psychiatry 32(7): 1063–1080.
Chiles, J. A., Miller, M. L., and Cox, G. B. (1980). Depression in an adolescent delinquent population.
Arch. Gen. Psychiatry 37: 1179–1184.
424 Overbeek et al.
Costello, E. J. (1989). Developments in child psychiatric epidemiology. J. Am. Acad. Child Adolesc.
Psychiatry 28: 836–841.
Damphousse, K., and Kaplan, H. B. (1998). Intervening processes between adolescent drug use and
psychological distress: An examination of the self-medication hypothesis. Soc. behav. Pers. 26(2):
115–130.
Diekstra, R. W. F., Garnefski, N., De Heus, P., De Zwart, R., Van Praag, B. M. S., and Warnaar,
M. (1991). Scholierenonderzoek 1990. Gedrag en gezondheid van scholieren uit het voortgezet
onderwijs [Survey of pupils 1990; health and behavior]. NIBUD, Den Haag.
Elliott, D. S., Huizinga, D., and Menard, S. (1989). Multiple Problem Youth: Delinquency, Substance
Use, and Mental Health Problems. Springer-Verlag, New York.
Engels, R. C. M. E., Knibbe, R. A., and Drop, M. J. (1999). Predictability of smoking in adolescence:
Between optimism and pessimism. Addiction 94: 115–124.
Erikson, E. H. (1968). Identity: Youth and Crisis. Norton, New York.
Ferdinand, R. F., Stijnen, T., Verhulst, F. C., and Van der Reijden, M. (1999). Associations between
behavioural and emotional problems in adolescence and maladjustment in young adulthood.
J. Adolesc. 22: 123–136.
Ferdinand, R. F., and Verhulst, F. C. (1995). Psychopathology in Dutch young adults: Enduring or
changeable? Soc. Psychiatry Psychiatr. Epidemiol. 30: 60–64.
Ferdinand, R. F., and Verhulst, F. C. (1996). The prevalence of self-reported problems in young adults
from the general population. Soc. Psychiatry Psychiatr. Epidemiol. 31: 10–20.
Ferdinand, R. F., Verhulst, F. C., and Wiznitzer, M. (1995). Continuity and change of self-reported prob-
lem behaviours from adolescence into young adulthood. J. Am. Acad. Child Adolesc. Psychiatry
34(5): 680–690.
Garnefski, N., and Diekstra, R. F. W. (1997). “Comorbidity” of behavioral, emotional and cognitive
problems in adolescence. J. Youth Adolesc. 26(3): 321–338.
Gold, M., Mattlin, M., and Osgood, D. W. (1989). Background characteristics and responses to treatment
of two types of institutionalized delinquent boys. Crim. Just. Behav. 16(1): 5–33.
Goldberg, D. P. (1978). Manual of the General Health Questionnaire. General Practice Research Unit,
Horsham.
Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P. A., McGee, R., and Angell, K. E. (1998). De-
velopment of depression from preadolescence to young adulthood: Emerging gender differences
in a 10-year longitudinal study. J. Abnorm. Psychol. 107(1): 128–140.
Hankin, B. L., Roberts, J., and Gotlib, I. H. (1997). Elevated self-standards and emotional distress during
adolescence: Emotional specificity and gender differences. Cogn. Ther. Res. 21(6): 663–679.
Harrington, R., Fudge, H., Rutter, M., Pickles, A., and Hill, J. (1990). Adult outcomes of childhood
and adolescent depression. Arch. Gen. Psychiatry 47: 465–473.
Havighurst, R. J. (1972). Developmental Tasks and Education (3rd edn). McKay, New York.
Helsen, M., Vollebergh, W., and Meeus, W. (2000). Social support from parents and friends and
emotional problems in adolescence. J. Youth Adolesc. 29(3): 319–335.
Hirschi, T., and Gottfredson, M. (1983). Age and the explanation of crime. Am. J. Sociol. 89: 552–584.
Jöreskog, K. G. (1969). A general approach to confirmatory maximum likelihood factor analysis.
Psychometrika 34: 183–202.
Jöreskog, K. G., and Sörbom, D. (1989). LISREL 7: A Guide to the Program and Applications. SPPS,
Chicago.
Jöreskog, K. G., and Sörbom, D. (1993). LISREL 8: Structural Equation Modeling With the SIMPLIS
Command Language. SSI, Chicago.
Kandel, D. B., and Davies, M. (1982). Epidemiology of depressive mood in adolescents: An empirical
study. Arch. Gen. Psychiatry 39: 1205–1212.
Kandel, D. B., and Davies, M. (1986). Adult sequelae of adolescent depressive symptoms. Arch. Gen.
Kienhorst, C. W. M., Wilde, E. J., Van den Bout, J., and Diekstra, R. F. W. (1990). Psychometrische
eigenschappen van een aantal zelfrapportage-vragenlijsten over “(on) welbevinden” [Psycho-
metric characteristics of a number of self report measures of “(non) well being”]. Nederlands
Tijdschrift voor de Psychologie 45: 124–133.
Kovacs, M., Paulauskas, S., Gatsonis, C., and Richards, C. (1988). Depressive disorders in childhood.
III: A longitudinal study of comorbidity with and risk for conduct disorders. J. Affect. Disord. 15:
205–217.
Kraus, S. J. (1995). Attitudes and the prediction of behavior: A meta-analysis of the empirical literature.
Pers. Soc. Psychol. Bull. 21: 58–75.
Krueger, R. F. (1999). The structure of common mental disorders. Arch. Gen. Psychiatry 56: 921–
926.
Krueger, R. F., Caspi, A., Moffitt, T. E., and Silva, P. A. (1998). The structure and stability of common
mental disorders (DSM-III-R): A longitudinal–epidemiological study. J. Abnorm. Psychol. 107(2):
216–227.
Leadbeater, B. J., Kuperminc, G. P., Blatt, S. J., and Hertzog, C. (1999). A multivariate model of
gender differences in adolescents, internalizing and externalizing problems. Dev. Psychol. 35(5):
1268–1282.
Loeber, R., Farrington, D. P., Stouthamer-Loeber, M., Moffitt, T. E., and Caspi, A. (1998). The devel-
opment of male offending: Key findings from the first decade of the Pittsburgh Youth Study. Stud.
Crime Crime Prev. 7: 141–171.
Loeber, R., and Keenan, K. (1994). Interaction between conduct disorder and its comorbid conditions:
Effects of age and gender. Clin. Psychol. Rev. 14(6): 497–523.
Luijpers, E. T. H. (1999). Exploratie, sociale binding, en crimineel gedrag van adolescenten [Explo-
ration, social commitments, and criminal behavior of adolescents]. Unpublished manuscript.
Maggs, J. L., Frome, P. M., Eccles, J. S., and Barber, B. L. (1997). Psychosocial resources, adolescent
risk behaviour and young adult adjustment: Is risk taking more dangerous for some than others?
J. Adolesc. 20: 103–119.
McConaughy, S. H., and Skiba, R. J. (1993). Comorbidity of externalizing and internalizing problems.
Sch. Psychol. Rev. 22(3): 421–438.
Meeus, W. (1993). De psychosociale ontwikkeling van adolescenten [Psychosocial development in
adolescents]. In Meeus, W. and ‘t Hart, H. (eds.), Jongeren in Nederland: Een nationaal sur-
vey naar ontwikkeling in de adolescentie en naar intergenerationele overdracht [Young people
in the Netherlands: A national survey on development in adolescence and on intergenerational
transference]. Academische Uitgeverij Amersfoort, Amersfoort, pp. 31–55.
Meeus, W. (1996). Studies on identity development in adolescence: An overview of research and some
new data. J. Youth Adolesc. 25(5): 569–597.
Meeus, W., and ‘t Hart, H. (1993). Jongeren in Nederland: Een nationaal survey naar ontwikkeling in
de adolescentie en naar intergenerationele overdracht [Young people in the Netherlands: A na-
tional survey on development in adolescence and on intergenerational transference]. Academische
Uitgeverij Amersfoort, Amersfoort.
Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A develop-
mental taxonomy. Psychol. Rev. 100(4): 674–701.
Nolen-Hoeksema, S. (1990). Sex Differences in Depression. Stanford University Press, Stanford.
O’Connor, T. G., McGuire, S., Reiss, D., Hetherington, E. M., and Plomin, R. (1998a). Co-occurrence
of depressive symptoms and antisocial behavior in adolescence: A common genetic liability.
J. Abnorm. Psychol. 107(1): 27–37.
O’Connor, T. G., Neiderhiser, J. M., Reiss, D., Hetherington, E. M., and Plomin, R. (1998b). Genetic
contributions to continuity, change, and co-occurrence of antisocial and depressive symptoms in
adolescence. J. Child Psychol. Psychiatry 39(3): 323–336.
Patterson, G. R., and Capaldi, D. M. (1990). A mediational model for boys, depressed mood. In Rolf,
J., Masten, A.S., Cicchetti, D., Nuechterlein, K.H., and Weintraub, S. (eds.), Risk and Protec-
tive Factors in the Development of Psychopathology. Cambridge University Press, New York,
pp. 141–163.
Pertraitis, J., Flay, B. R., and Miller, T. Q. (1995). Reviewing theories of adolescent substance use:
Organizing pieces in the puzzle. Psychol. Bull. 117(1): 67–86.
Puig-Antich, J. (1982). Major depression and conduct disorder in prepuberty. J. Am. Acad. Child
Adolesc. Psychiatry 21(2): 118–128.
Rantakillio, P., Myhrman, A., and Koiranen, M. (1995). Juvenile offenders, with special reference to
sex differences. Soc. Psychiatry Psychiatr. Epidemiol. 30: 113–120.
Rohde, P., Lewinsohn, P. M., and Seeley, J. R. (1996). Psychiatric comorbidity with problem-
atic alcohol use in high school students. J. Am. Acad. Child Adolesc. Psychiatry 35(1): 101–
109.
Rowe, D. C., Vazsonyi, A. T., and Flannery, D. J. (1994). No more than skin deep: Ethnic and racial
similarity in developmental processes. Psychol. Rev. 101: 396–413.
426 Overbeek et al.
Rutter, M. (1989). Psychiatric disorder in parents as a risk factor for children. In Shaffer, D., Philips, J.,
and Enzer, N. B. (eds.), with Silverman, M. M., and Anthony, V. (Asst. eds.), Prevention of Mental
Disorders, Alcohol and Other Drug Use in Children and Adolescents. Office for Substance Abuse
Prevention, U.S. Department of Health and Human Sciences, OSAP Prevention Monograph 2.
Rockville, pp. 157–189.
Rutter, M., and Quinton, D. (1984). Parental psychiatric disorder: Effects on children. Psychol. Med.
14: 853–880.
Silbereisen, R. K., and Noack, P. (1988). On the constructive role of problem behavior in adolescence. In
Bolger, N. Caspi, A., Downey, G., and Moorehouse, M. (eds.), Persons in Context: Developmental
Processes. Cambridge University Press, New York, pp. 152–180.
Shedler, J., and Block, J. (1990). Adolescent drug use and psychological health: A longitudinal inquiry.
Am. Psychol. 45(5): 612–630.
Steiger, J. H. (1990). Structural model evaluation and modification: An interval estimation approach.
Multivar. Behav. Res. 25: 173–180.
Steinberg, L. (1987). Family processes in adolescence: A developmental perspective. Fam. Ther. 14:
77–86.
‘t Hart, H. (1994). Criminaliteit in latente klassen: Een voorlopige analyse van de structuur van
crimineel gedrag van jongeren [Criminal behavior in latent classes: A preliminary analysis of the
structure of adolescent’s criminal behavior]. Internal report, Utrecht University.
Verhulst, F. C., and Van der Ende, J. (1992). Six year developmental course of internalizing and
externalizing problem behaviours. J. Am. Acad. Child Adolesc. Psychiatry 5: 924–931.
Wheaton, D. E., Muthén, B., Alwin, D. F., and Sommers, G. F. (1977). Assessing reliability and stability
in panel models. In Heisse, D. R. (ed.), Sociological Methodology. Jossey Bass, San Fransisco,
pp. 84–136.
Williams, L. J., and Holahan, P. J. (1994). Parsimony-based fit indices for multiple-indicator models:
Do they work? Struct. Eqn. Model. 1(2): 161–189.
Wittchen, H. U. (1996). Critical issues in the evaluation of comorbidity of psychiatric disorders. Br. J.
Zoccolillo, M. (1992). Co-occurrence of conduct disorder and its adult outcomes with depressive and
anxiety disorders: A review. J. Am. Acad. Child Adolesc. Psychiatry 31(3): 547–556.

Adol Tinrete - Risc

Uploaded by

Copyright:

Available Formats

Adol Tinrete - Risc

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Adol Tinrete - Risc

Uploaded by

Copyright:

Available Formats

P1: GFU/LCCR/GAY P2: GCR/GFU

Journal of Youth and Adolescence, Vol. 30, No. 4, 2001

Course, Co-Occurrence, and Longitudinal

402 Overbeek et al.

processes. The strong stability of internalizing and externalizing behavior suggests

During the past 2 decades, several community-based epidemiological studies

The Course of Internalizing and Externalizing Problems

Previous studies on the course of internalizing problems in adolescence have

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 403

examining possible differences in the co-occurrence and longitudinal associations

Co-Occurrence of Internalizing and Externalizing Problems

Previous research has pointed to the existence of a valid co-occurrence be-

Four Alternative Perspectives on the Co-Occurrence of Internalizing

404 Overbeek et al.

Previous Research on Explanatory Mechanisms: Temporal Patterns

In part, the evidence for each of the 4 alternative perspectives is based on

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 405

disproportionately large number of patients having co-occurring problems (Caron

The Present Study

With the Utrecht Study of Adolescent Development (USAD, 1991–97), a

406 Overbeek et al.

community-based population of adolescents and young adults. Three main ques-

Procedure and Sample Characteristics

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 407

General feelings of well-being were assessed using the Cantril Ladder

Psychological Stress and Depressive Mood

408 Overbeek et al.

The measures mentioned earlier were highly interrelated, as was indicated

Delinquency was assessed as the number of delinquent acts the respondents

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 409

First, descriptive analyses (standardized means and standard deviations) of

410 Overbeek et al.

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 411

The Course of Emotional Disturbance and Delinquency

The standardized means and standard deviations for emotional disturbance

Note. Emot. dist. = Emotional disturbance.

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 413

Co-Occurrence of Emotional Disturbance and Delinquency

As can be seen in Table II, the co-occurrence of emotional disturbance and

Longitudinal Associations of Emotional Disturbance and Delinquency

Total Boys Girls 12–14 15–17 18–20 21–24

Co-occurrence and cross-lagged correlations

Delinquency T1 Delinquency T2 0.41∗∗∗ 0.39∗∗∗ 0.35∗∗∗ 0.40∗∗∗ 0.44∗∗∗ 0.42∗∗∗ 0.35∗∗∗

Note. Emot. dist. = Emotional disturbance.

Failure 190.94 (6) 0.00 31.82 0.99 0.97 0.11 132.69

Failure 161.81 (51) 0.00 3.17 0.92 0.68 0.08 227.81

Note. Mut. inf. = Mutual influence.

416 Overbeek et al.

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 417

analysis, the relationship from emotional disturbance at Wave 1 to delinquency at

Although previous studies have often relied on clinical samples (e.g.,

418 Overbeek et al.

and longitudinal associations of emotional disturbance and delinquency in a non-

Longitudinal Change and Stability of Emotional Disturbance

Emotional Disturbance and Delinquency From Adolescence to Young Adulthood 419

and delinquency) were of a moderate strength, indicating that a large number of

Co-Occurrence of Emotional Disturbance and Delinquency

420 Overbeek et al.