Developmental Psychology 2010 American Psychological Association

2010, Vol. 46, No. 5, 1159 1175 0012-1649/10/$12.00 DOI: 10.1037/a0020659

Trajectories of Pure and Co-Occurring Internalizing and Externalizing

Problems From Age 2 to Age 12: Findings From the National Institute of
Child Health and Human Development Study of Early Child Care

Kostas A. Fanti Christopher C. Henrich

University of Cyprus Georgia State University

How and why do internalizing and externalizing problems, psychopathological problems from different
diagnostic classes representing separate forms of psychopathology, co-occur in children? We investigated
the development of pure and co-occurring internalizing and externalizing problems from ages 2 to 12
with the use of latent class growth analysis. Furthermore, we examined how early childhood factors
(temperament, cognitive functioning, maternal depression, and home environment) and early adolescent
social and behavioral adjustment variables were related to differential trajectories of pure and co-
occurring internalizing and externalizing problems. The sample (National Institute of Child Health and
Human Development Study of Early Child Care) consisted of 1,232 children (52% male). Mother reports
on the Child Behavior Checklist (Achenbach, 1991, 1992) were used to construct the trajectories of
externalizing and internalizing problems. Analyses identified groups of children exhibiting pure and
co-occurring internalizing and externalizing problems. Children exhibiting continuous externalizing or
continuous co-occurring internalizing and externalizing problems across the 10-year period under
investigation were more likely to (a) engage in risky behaviors, (b) be associated with deviant peers, (c)
be rejected by peers, and (d) be asocial with peers at early adolescence. However, children exhibiting
pure internalizing problems over time were only at higher risk for being asocial with peers as early
adolescents. Moreover, the additive effects of individual and environmental early childhood risk factors
influenced the development of chronic externalizing problems, although pure internalizing problems
were uniquely influenced by maternal depression. Results also provided evidence for the concepts of
equifinality and multifinality.

Keywords: internalizing, externalizing, co-occurrence, latent class growth analysis, developmental psy-
chopathology

Internalizing and externalizing problems may develop as early
as early childhood and can place children on a developmental
pathway to peer problems, negative interactions with parents,
delinquency, and other negative social and behavioral outcomes
(Coie & Dodge, 1998; Keiley, Lofthouse, Bates, Dodge, & Pettit,
2003; Kovacs & Devline, 1998). Internalizing problems appear in
the form of withdrawal, anxiety, fearfulness, and depression,
whereas externalizing problems take the form of hyperactivity,
aggression, defiance, and destructive behavior (Achenbach, 1991,
1992; Campbell, 1995). Early emotional and behavioral problems
have been found to precede child, adolescent, and adult antisocial
and depressive psychopathological problems. Therefore, it is of
great importance to identify children at risk for high and contin-
uous internalizing and externalizing problems early in develop-
ment (Gilliom & Shaw, 2004; Moffitt, 1993).
A question that remains unanswered in the literature is how and
why internalizing and externalizing problems, two psychopatho-
Kostas A. Fanti, Department of Psychology, University of Cyprus,
Nicosia, Cyprus; Christopher C. Henrich, Department of Psychology,
Georgia State University.
Correspondence concerning this article should be addressed to Kostas
A. Fanti, Department of Psychology, University of Cyprus, P.O. Box
20537, CY 1678, Nicosia, Cyprus. E-mail: kfanti@ucy.ac.cy
logical problems from different diagnostic classes representing
separate forms of psychopathology, co-occur in children (Oland &
Shaw, 2005). We examined the co-occurrence of internalizing and
externalizing problems from a developmental perspective, such
that co-occurrence was investigated in terms of varied patterns of
growth trajectories over the course of childhood and not just
similar levels at a given time. We also followed Kaplan, Crawford,
Cantell, Kooistra, and Deweys (2006) suggestion that the tempo-
ral term co-occurrence should be used when investigating devel-
opmental problems instead of the term comorbidity, which is used
to indicate the presence of at least two psychopathological prob-
lems that are independent of each other. According to previous
research, internalizing and externalizing problems can and often do
co-occur at every point in development at which they have been
measured (e.g., Gilliom & Shaw, 2004; Keiley et al., 2003; Young-
strom, Findling, & Calabrese, 2003), suggesting that an apprecia-
tion of the concept of co-occurrence is essential for explaining the
development and understanding the etiology of externalizing and
internalizing problems.
We investigated the co-occurrence between internalizing and
externalizing problems within a developmental psychopathology
conceptual framework. Developmental psychopathology is con-
cerned with individual differences in the origins, course, and
outcomes of normative and psychopathological developmental
processes (Cicchetti, 1984; Rutter & Sroufe, 2000). Furthermore,

1159
1160 FANTI AND HENRICH
this framework conceives of development as an active dynamic
process and is concerned with the time course of psychopathology
for different groups of individuals. In addition, central to the
developmental psychopathology approach is the belief that the
study of psychopathology can inform researchers understanding
of normal development and, conversely, that the study of norma-
tive development may shed light on the etiology and course of
psychopathology.
Following the developmental psychopathology perspective, the
current study investigated how children deviate from normal de-
velopment and follow trajectories of pure and co-occurring inter-
nalizing and externalizing problems from age 2 to age 12. We also
examined how early childhood risk factors measured during the
first 2 years of life, including child, family, and environmental risk
factors, forecast the development of pure or co-occurring trajec-
tories of behavioral and emotional problems. Finally, we investi-
gated how the differential trajectories of internalizing and exter-
nalizing problems are related to early adolescent behavioral and
social adjustment. Inclusion of early childhood risk factors and
early adolescent adjustment variables in the trajectory models
enabled us to investigate research questions key to developmental
psychopathology: These questions pertain to equifinality, in which
varied developmental pathways may lead to the same developmen-
tal outcome (i.e., to what extent trajectories of pure versus co-
occurring risks are differentially associated with early adolescent
functioning), and multifinality, in which a single vulnerability
factor or any one component or a single starting point may result
in diverse developmental outcomes (i.e., to what extent the early
childhood risk factors forecast differential trajectories of pure vs.
co-occurring risks; Cicchetti & Rogosch, 1996).
Development of Externalizing Problems
Externalizing behaviors, such as aggression and destructive
behaviors, are often used by toddlers to solve conflicts with peers
or playmates, and with the development of cognitive abilities and
the skills to regulate emotions, aggressive and externalizing prob-
lems decrease and diminish over the preschool and school age
period (Coie & Dodge, 1998). However, a small number of anti-
social children, representing 5% to 7% of the population, do not
outgrow the temper tantrums and the defiant and irritable behavior
that characterizes toddlers and follow a life-course persistent tra-
jectory of externalizing problems (Moffitt, 1993). Previous work
investigating heterogeneity in the course of externalizing prob-
lems, including studies that used the same data as we used in the
current investigation (Campbell, Spieker, Burchinal, Poe, & the
National Institute of Child Health and Human Development
[NICHD] Early Child Care Research Network, 2006; Fanti &
Henrich, 2007; NICHD Early Child Care Research Network,
2004) and studies that used different data sets (Broidy et al., 2003;
Nagin & Tremblay, 1999; Shaw, Gilliom, Ingoldsby, & Nagin,
2003), have identified children following chronic, moderate-
desisting, high-desisting, and low-externalizing problem trajecto-
ries, and those differences have been found as early as the second
year of life for both boys and girls. The majority of children exhibit
low levels of externalizing behavior over time, and most of the
children with moderate or high levels of externalizing behavior
problems early in development exhibit decreases in behavioral
problems after the preschool years. Only a small percentage of
children follow a high and consistent trajectory of aggression,
conduct problems, and delinquency. Children with chronic exter-
nalizing problems are also at higher risk of experiencing peer
rejection and of being involved with deviant peers, and these
experiences appear to maintain and exacerbate their behavioral
problems (Laird, Jordan, Dodge, Pettit, & Bates, 2001).
The life-course persistent trajectory of externalizing behavior
has been associated with prenatal and perinatal medical risks, and
these problems have been found to be related to infant neuropsy-
chological risk (Brennan, Hall, Bor, Najman, & Williams, 2003;
Moffitt, 1993). Neuropsychological deficits can then impair the
childs cognitive abilities and can also result in a difficult temper-
ament. Early deficiencies in cognitive functioning and difficult
temperament may set an individual on a pathway to exhibiting
chronic externalizing problems (Lynam, Moffitt, & Stouthamer-
Loeber, 1993; Miner & Clarke-Stewart, 2008; Moffitt, 1990,
1993). In addition, family adversity, maternal depression, and
sociodemographic risk factors, such as low socioeconomic status
and single parent status, are among the strongest predictors of later
externalizing problems (Ackerman, DEramo, Umylny, Schultz, &
Izard, 2001; Brennan et al., 2003; Gross, Shaw, & Moilanen, 2008;
Shaw, Owens, Vondra, Keenan, & Winslow, 1996).
Development of Internalizing Problems
In contrast to externalizing problems, internalizing problems on
average tend to increase gradually from infancy to early childhood
(Gilliom & Shaw, 2004), with girls showing a higher increase in
internalizing problems across time (Bongers, Koot, van der Ende,
& Verhulst, 2003). Furthermore, Achenbach, Howell, Quay, and
Conners (1991) compared children of different ages, from age 4 to
age 16, and provided evidence that among clinically referred
children, internalizing problems (somatic complaints, withdrawal,
anxiety, and depression) tend to increase with age. Increases in
internalizing problems might be the result of cognitive maturation
(Kovacs & Devline, 1998), because improvements in cognitive
abilities enable children to self-reflect and to remember and antic-
ipate negative or depressive events. As with externalizing prob-
lems, children at high risk for exhibiting continuous internalizing
problems from toddlerhood to adolescence have been identified
(Brendgen, Wanner, Morin, & Vitaro, 2005; Duggal, Carlson,
Sroufe, & Egeland, 2001; Sterba, Prinstein, & Cox, 2007). Sterba
et al. (2007) used the same data as we used in the current study to
investigate the development of internalizing problems from age 2
to age 11. In addition to identifying children at high risk for
developing internalizing problems, Sterba et al. provided evidence
for heterogeneity in the course of internalizing problems by iden-
tifying children exhibiting low, deceasing, and increasing internal-
izing problems across development. Children exhibiting internal-
izing problems are often unable to form good peer relationships,
and they are more likely to engage in isolating behaviors and social
withdrawal (Coie & Dodge, 1998; Hogue & Steinberg, 1995;
Oland & Shaw, 2005). Because of these isolating behaviors, chil-
dren who exhibit internalizing problems do not affiliate with
delinquent peers, and they are at lower risk for delinquent behav-
iors (Oland & Shaw, 2005).
Environmental risk factors, such as low familial sociodemo-
graphic status (SDS), exposure to a negative familial context, and
maternal depression, have been associated with high levels of
 INTERNALIZING AND EXTERNALIZING PROBLEMS
internalizing problems across development (Duggal et al., 2001;
McLeod & Shanahan, 1996). Further, as with the development of
externalizing problems, difficult temperament and low cogni-
tive abilities have been associated with the development of
internalizing problems as well (Booth-LaForce & Oxford, 2008;
Bub, McCartney, & Willett, 2007; Keenan, Shaw, Delliquadri,
Giovannelli, & Walsh, 1998).
Co-Occurrence Between Externalizing and
Internalizing Problems
Different behavioral syndromes tend to correlate with one an-
other, and children who score high in one syndrome tend to score
at average or high levels in other symptoms as well (Achenbach,
Conners, Quay, Verhulst, & Howell, 1989). Moreover, it has been
proposed that pure and co-occurring psychopathologies might rep-
resent distinct, meaningful syndromes (Angold & Costello, 1992;
Lilienfeld, 2003; OConnor, McGuire, Reiss, Hetherington, &
Plomin, 1998). In support of this idea, studies using cross-sectional
and longitudinal statistical methods have provided evidence of the
existence of pure internalizing, pure externalizing, and co-
occurring internalizing and externalizing problems (Epkins, 2000;
Gilliom & Shaw, 2004; Keiley, Bates, Dodge, & Pettit, 2000;
Keiley et al., 2003; Reitz, Dekovic, & Meijer, 2005; Youngstrom
et al., 2003). These studies also suggest that the co-occurrence
between internalizing and externalizing problems has an early age
of onset.
Even though co-occurrence can be thought of as a distinct
syndrome, common environmental and individual risk factors have
been proposed to account for the presence of pure and co-
occurring psychopathological problems (Angold & Costello, 1993;
Klein & Riso, 1993; OConnor et al., 1998). As shown in the
previous sections, family SDS, difficult temperament, negative
home environment, maternal depression, and cognitive deficien-
cies are associated with both internalizing and externalizing prob-
lems, providing evidence for the process of multifinality, because
the same starting point might result in diverse behavioral prob-
lems. These common risk factors could also account for the co-
occurrence between the two symptoms, with the co-occurrence
between emotional and behavioral problems being represented by
these overlapping risk factors and causal processes (Klein & Riso,
1993; Rutter, 1997). Research that tested this claim has found that
pure and co-occurring internalizing and externalizing problems are
associated with similar negative developmental processes, al-
though children exhibiting co-occurring behavioral and emotional
problems are faced with higher levels of these risk factors (Epkins,
2000; Ge, Best, Conger, & Simons, 1996; Keiley et al., 2003;
Lilienfeld, 2003; Milan, Pinderhughes, & the Conduct Problems
Prevention Research Group, 2006; Shaw et al., 2003).
Additionally, previous research has suggested that co-
occurrence may be related to the severity of psychopathology and
that individuals with co-occurring problems exceed individuals
exhibiting pure problems in terms of chronic history of mental
illness, more physical health problems, greater functional interfer-
ence in daily life, and more impaired adaptation across domains
such as education and social-support networks (Newman, Moffitt,
Caspi, & Silva, 1998). Furthermore, children with co-occurring
internalizing and externalizing problems are in greater danger of
experiencing negative relationships with peers, of being involved
1161
in delinquent acts, and of being affiliated with deviant peers when
compared with children with pure problem behaviors (Miller-
Johnson, Lochman, Coie, Terry, & Hyman, 1998; Oland & Shaw,
2005; Wright, Zakriski, & Drinkwater, 1999). For these reasons, a
greater understanding of co-occurrence is essential.
The current study investigated the co-occurrence between inter-
nalizing and externalizing problems (a) by identifying trajectories
over time that include both externalizing and internalizing behav-
iors, (b) by examining how early childhood risk factors are related
to the identified trajectories, and (c) by investigating how the
identified trajectories are related to early adolescent behavioral and
social adjustment. We expected to identify children at lower risk of
developing internalizing or externalizing problems (representing
the majority of the sample) and children at risk for chronically high
levels of pure externalizing problems, pure internalizing problems,
and co-occurring internalizing and externalizing problems from
age 2 to age 12. Early childhood risk factors, including negative
home environment, cognitive dysfunction, maternal depression,
and difficult temperament, are hypothesized to place children at
higher risk for developing co-occurring internalizing and external-
izing problems. On the other hand, children with pure internalizing
and externalizing trajectories are expected to experience milder
forms of early childhood risk factors in comparison with children
who have co-occurring problems over time. In addition, we antic-
ipated that low individual and contextual adversity would be
associated with reduced likelihood of developing any internalizing
or externalizing problems. Further, children who have co-
occurring internalizing and externalizing behavior problems over
time are expected to face more relationship problems with peers
and to be at higher risk of exhibiting delinquent behaviors and of
socializing with deviant peers during early adolescence in com-
parison with children following pure trajectories of behavioral and
emotional problems. Children exhibiting pure externalizing prob-
lems were also expected to be at higher risk for exhibiting risky
behaviors compared with children exhibiting pure internalizing
problems and children at lower risk in terms of both problem
behaviors. Finally, we investigated the processes of equifinality
and multifinality.
Method
The present study used data from the NICHD Study of Early
Child-Care, a study conducted by the NICHD Early Child Care
Research Network supported by NICHD through a cooperative
agreement that called for scientific collaboration between the
grantees and the NICHD staff. Participants were recruited from
different hospitals across10 locations in the United States (Little
Rock, Arkansas; Irvine, California; Lawrence, Kansas; Boston,
Massachusetts; Philadelphia, Pennsylvania; Pittsburgh, Pennsylva-
nia; Charlottesville, Virginia; Morganton, North Carolina; Seattle,
Washington; Madison, Wisconsin). A total of 8,986 women gave
birth during the sampling period (January 1991 and November
1991) across the different locations, and 60% (5,416) of those
women agreed to be contacted for a telephone interview. Of the
women who agreed to participate, 56% (3,015) were selected on
the basis of conditional random sampling. The conditional random
sampling procedure was used to ensure that the sample was rep-
resentative of single mothers, poverty status, ethnic minority, and
low maternal education. Participants from this sample were ex-
1162 FANTI AND HENRICH
cluded if (a) mothers were younger than 18 at the time of the
childs birth, (b) families were planning to move before the com-
pletion of the study, (c) children were born with disabilities or
remained in the hospital more than 7 days postpartum, and (d)
mothers could not communicate sufficiently in English (see http://
secc.rti.org). A total of 1,525 families were selected as eligible, but
only 1,364 completed the home interview when the infant was 1
month old, and these families made up the final sample of the
study. The studys final sample was diverse in terms of gender
(53% male), minorities (24% were minorities), maternal education
(11% of the mothers had not completed high school), and marital
status (14% were single). The average family income was 3.6
times the poverty threshold.
The trajectory analyses for the current study were based on
1,232 children (52% male) whose mothers completed the Child
Behavior Checklist (CBCL; Achenbach, 1991, 1992) during at
least two of the nine assessment periods, as was done in previous
research (NICHD Early Child Care Research Network,, 2004). The
sample used for the current study was diverse in terms of ethnicity
(77.5% were White, 6% were of Hispanic descent, 11.7% were
African American, 1.3% were Asian, and 3.5% represented other
minority groups). In addition, 70% of the samples population
reported that their income was above the poverty threshold, with a
mean annual income of $67,310; 20.4% scored below the poverty
threshold at some point during data collection, but not continu-
ously, with a mean annual income of $25,362; 9.6% scored below
the poverty threshold continuously during data collection, with a
mean annual income of $12,641. On the basis of interviews with
mothers during the 1st and 6th month after birth, 10% of mothers
had not completed high school, and 21% were single.
Externalizing and Internalizing Problems
Mothers rated their childrens externalizing and internalizing
problems at 2, 3, 4.5, 6, 7, 9, 10, 11, and 12 years of age with the
CBCL (Achenbach, 1991, 1992). There are two versions of the
CBCL: the preschool version for children ages 23 and the school-
age version for children ages 4 18. Both versions measure inter-
nalizing and externalizing problems, although some items vary to
capture developmental changes. For the present study, following
previous research (Gilliom & Shaw, 2004), items that appear on
both versions of the CBCL were used to maximize comparability
over time. Nine items were used to measure externalizing prob-
lems (doesnt seem to feel guilty after misbehaving, cant sit
still, restless, or hyperactive, cruel to animals, destroys his
own things, destroys things belonging to his family or others,
disobedient, gets in many fights, physically attacks people,
and temper tantrums or hot tempered). Six items were used to
measure internalizing problems (too fearful or anxious, self-
conscious or easily embarrassed, shy or timid, withdrawn,
doesnt get involved with others, unhappy, sad, or depressed,
and worries). Mothers rated how descriptive each item was of
the childs usual behavior now or within the past 6 months on a
3-point scale (0 not true, 1 somewhat or sometimes true, 2
very true or often true). The items were summed to represent a
count of the number of externalizing (Cronbachs ranged from
.75 to .80) or internalizing (Cronbachs ranged from .65 to .73)
behaviors weighted by frequency of occurrence.
Risk Indexes
Items indicating medical risk and early SDS were dichotomized
into 0 (no risk) or 1 (risk). The items were dichotomized to allow
the creation of an overall medical risk score and an overall SDS
risk score. These risk indexes were created for the current study.
Interviews with the mother during the 1st and 6th month after birth
were used to establish any health problems the mother had during
pregnancy and any perinatal infant health problems during the 1st
and 6th month of age. Mothers also reported whether they smoked
during and after pregnancy and whether they or the infants were
exposed to smoking by others (passive smoking). A score of 1 was
given for (a) smoking by the mother, (b) passive smoking, (c)
childs ear infection, (d) childs respiratory problems, (e) childs
injuries that required a visit to the doctor, and (f) maternal health
problems during pregnancy. The scores from these six items were
added together to create an overall medical risk index, which
ranged from 0 to 6.
The SDS risk index reflected mothers reports of family fi-
nances, maternal marital status, and maternal education, following
previous research (Henrich, Schwab-Stone, Fanti, Jones, & Ruch-
kin, 2004). Mothers reported the familys income from 1 month to
24 months, and an income-to-needs ratio was calculated from the
U.S. Census Bureau tables as the ratio of family income to the
appropriate poverty threshold for each household size and number
of children. The scores were averaged from 1 to 24 months, and
families below the poverty threshold received a score of 1. In
addition, mothers were asked whether they were married and
whether the father was living with them in the same household.
During the same interview, mothers also reported their educational
level. A score of 1 was given if the mother reported a low
educational level (lower than high school) or if the mother was a
single parent. The SDS risk index ranged from 0 to 3.
Early Childhood Risk Factors
Early difficult temperament. Temperament was assessed
twice with the Infant Temperament Questionnaire (Medoff-
Cooper, Carey & McDevitt, 1993). At the 1- and 6-month home
visits, mothers were asked to respond to 38 items that were
developmentally appropriate for young infants (rated on a 1 6
scale from almost never to almost always). The items provide
scores on five subscalesActivity, Adaptability, Approach,
Mood, and Intensityand all the items can be combined into a
single Difficult Temperament scale. Because the stability from 1 to
6 months was high, r .77, p .001, a Difficult Temperament
scale based on the average score of all the items from 1 and 6
months was created ( .70). Scores on the Infant Temperament
Questionnaire at 6 months have been shown to be related to the
development of childhood behavioral and emotional problems
(Booth-LaForce & Oxford, 2008; Miner & Clarke-Stewart, 2008).
Early cognitive developmental status. The Bayley Scales of
Infant Development-II (Bayley, 1993) were administered at 15 and
24 months. The Bayley Mental Development Index is the most
widely used measure of cognitive developmental status for chil-
dren in the first 2 years of life. It assesses memory, learning, and
problem solving; sensory perceptual acuity and discriminations;
early verbal communication; and the ability to form generaliza-
tions and classifications. The overall standardized score was based
 INTERNALIZING AND EXTERNALIZING PROBLEMS
on the average of the 15-month assessment and the 24-month
assessment. This variable was used as an indicator for the childs
early cognitive functioning. The stability from 15 to 24 months
was high, r .80, p .001.
Early home environment. Quality and quantity of the social
and physical resources and of support and stimulation in the
childrens home environment were assessed at ages 6 and 15
months with the Home Observation of the Measurement of the
Environment, Infant/Toddler version (Caldwell & Bradley, 1984).
This measure is a semistructured interview and observational
procedure designed to describe the stimulation and responsiveness
of mothers, their involvement with and acceptance of their chil-
dren, the availability of play and learning materials, and the
organization and variety of the physical environment. An example
of an interview item is, the family has a fairly regular and
predictable daily schedule for the child, and an example of an
observational item is, the house has at least one full shelf of
books. Each item is scored in a binary fashion (yesno), and the
information used to score those items is obtained during the course
of a home visit. The stability from 6 to 15 months was high, r
.72, p .001; therefore, the items from the 6- and 15-month
assessments were added together to reflect a positive and enriched
home environment ( .77).
Postpartum depression. When the study children were 1
month old, mothers filled out the Center for Epidemiological
Studies Depression Scale (Radloff, 1997). This scale is a widely
used self-report scale that measures symptoms of depression in
nonclinical populations. Respondents rate the frequency during the
past week of 20 symptoms. Response categories are rarely or none
of the time (less than 1 day), some or a little of the time (12 days),
occasionally or a moderate amount of time (3 4 days), and most
or all of the time (57 days). The items used to create the depres-
sion score had high internal consistency ( .91).
Early Adolescent Behavioral and Social
Adjustment Variables
Risky behavior at age 12. The Risky Behavior Protocol
(Conger & Elder, 1994) has three sections completed by the
mother, the child, and a friend of the studys child. Mothers
completed a 30-item questionnaire. The studys children and their
friends each completed 38 identical items that measure things
your friends do (19 items) and things you do (19 items). The
questions ask whether the child or friend experiments with weap-
ons, cigarettes, alcohol, or drugs. One total risk-taking score for
the children was created on the basis of the mother reports (
.71). The child self-reports and the friend reports resulted in a
variable for delinquency committed by friends ( .82) and
another variable for delinquency committed by the child ( .73).
Child behavior with peers at age 12. The Child Behavior
Scale (Ladd & Profilet, 1996) consists of 24 filler items and 35
items that are conceptually grouped into six subscales: Aggressive
With Peers (seven items), HyperactiveDistractible (four items),
Asocial With Peers (six items), AnxiousFearful (four items),
Prosocial With Peers (seven items), and Excluded by Peers (seven
items). For the NICHD Early Child Care Research Networks
(2004) Study of Early Child Care, mothers and teachers rated the
childrens behaviors with peers on a 3-point scale (0 not true,
1 sometimes true, 2 often true). The two subscales used in this
1163
study, which were based on the combined scores between the
mother and teacher reports, are Asocial With Peers (likes to be
alone; .87) and Excluded by Peers (peers avoid this child;
.91). Ladd, Herald-Brown, and Andrews (2009) provided
evidence for cross-informant correlations and adequate reliability
and validity of the peer relational constructs measured with the
Child Behavior Scale experienced by youths ages 5 through 13.
The Asocial With Peers and Excluded by Peers subscales con-
verged with peer reports on the same constructs and were both
related to relational competence constructs and a host of other
negative developmental processes.
Analyses
Latent class growth analysis (LCGA) in Mplus 5.1 (Muthen &
Muthen, 2007) was used to identify distinct groups of individual
trajectories for externalizing and internalizing problems and the
joint probabilities between externalizing and internalizing prob-
lems. Multinomial logistic regressions in SPSS 17.0 were per-
formed to identify early childhood risk factors that distinguished
membership in the identified groups, and analysis of covariance
(ANCOVA) was used to compare the different trajectory groups
on measures of early adolescent social and behavioral adjustment.
Patterns of Pure and Co-Occurring Externalizing and
Internalizing Problems Over Time
LCGA identifies heterogeneous classes by modeling a mixture
of distinct multivariate normal distributions. LCGA uses a poly-
nomial function to model the relationship between an attribute, in
this case externalizing or internalizing problems, and age (Muthen
& Muthen, 2007; Nagin, 1999; Nagin & Tremblay, 1999). The
function takes the following form:
ytj jo j1 Age t ,
where ytj is a latent variable that characterizes the level of exter-
nalizing or internalizing problems for participant at time t given
membership in class j. Aget is participant s age at time t, and
is a disturbance assumed to be normally distributed with zero mean
and constant variance. The models coefficients, jo and j1, deter-
mine the shape of the trajectory. The coefficients are superscripted
by j to denote that they are not constrained to be the same across
j classes and are free to vary, which allows for cross-class differ-
ences in the shape of developmental trajectories. We also investi-
gated whether change over time is linear or whether it follows a
more complex, curvilinear pattern over time. To retain children
with incomplete assessments in the analysis, we used full infor-
mation maximum likelihood fitting in the Mplus software.
Furthermore, because all the variables were highly skewed with
a large number of zeros at each time, a zero inflated Poisson model
was used (Nagin & Land, 1993). This model is a two-class mixture
model. The zero inflated Poisson model estimates a zero class,
including the children scoring zero in terms of externalizing or
internalizing problems, and a nonzero class, including the children
with random zeros or who score higher than zero at different times.
If a child is in the nonzero class, the probability of a zero count is
expressed as a Poisson distribution. However, if a child is in the
zero class, a zero count has a probability of one.
1164 FANTI AND HENRICH

The LCGA model estimation in Mplus results in two outputs: peers, and peer exclusion. The ANCOVA analyses tested whether
(a) the shape and location of the different estimated class there were statistically reliable mean differences among the tra-
trajectories and (b) the posterior probability of class member- jectory groups after controlling for gender and adjusting the de-
ship. On the basis of Nagin and Tremblay (2001), when inspec- pendent variables for differences on covariates measured during
tion of the graphs suggests that a model with more classes the first 2 years of life: medical and SDS risk. As with simple
indicates the existence of similar classes of small theoretical im- regression, the covariates in the ANCOVA are used to account for
portance, the model with fewer and distinct classes is preferred. the variation they exert in the dependent variable, which increases
The average posterior probabilities and the entropy value are also the precision of comparisons among the groups.
taken under consideration to check for the precision of classifica-
tion and the degree to which the classes are distinguishable. Results
Average probabilities equal to or greater than .70 imply satisfac-
tory fit (Nagin, 2005), and an entropy value greater than .70 is Descriptive Statistics
preferred because it indicates clear classification and greater power
to predict class membership (Muthen, 2000). The means and standard deviations for externalizing and inter-
The model fit statistics used are the Bayesian information cri- nalizing problems from age 2 to age 12 are reported in Table 1.
terion (BIC) and the Lo, Mendel, Rubin (LMR) statistic. The BIC According to paired-sample t tests, there were significant mean-
is based on a maximization of a log likelihood function, and the level decreases in externalizing problems from ages 3 to 4.5,
model with a lower BIC is preferred (Schwartz, 1978). Moreover, t(1043) 13.80, p .001, from age 4.5 to 6, t(1022) 7.08, p
because the BIC criterion tends to favor models with fewer classes .001, from ages 6 to 7, t(992) 5.65, p .001, from ages 7 to 9,
by penalizing for the number of parameters (Bauer & Curran, t(948) 5.96, p .001, from ages 9 to 10, t(971) 4.19, p
2003), a likelihood statistic based on the sum of chi-square distri- .001, and from ages 10 to 11, t(981) 2.79, p .001, suggesting
butions was also used. Lo, Mendell, and Rubin (2001) adjusted the that there is a decrease in the average number of externalizing
likelihood ratio test so that it could be used in mixture modeling problems across time. Internalizing problems increased from ages
and to enable the comparison of nonnested models with different 2 to 3, t(1139) 6.25, p .001, from ages 3 to 4.5, t(1043)
numbers of classes. The LMR statistic tests k 1 classes against k 4.54, p .001, from ages 6 to 7, t(992) 3.59, p .001, and from
classes. A significant chi-square value ( p .05) indicates that the ages 10 to 11, t(981) 3.54, p .001. Furthermore, internalizing
k 1 class model is rejected in favor of the k class model. A problems decreased from ages 4.5 to 6, t(1022) 3.84, p .001,
nonsignificant chi-square value ( p .05) suggests that a model from ages 9 to 10, t(971) 1.99, p .001, and from ages 11 to
with one fewer class is preferred. 12, t(978) 2.90, p .001.
Finally, the intercept and growth parameters from the individual
class memberships for externalizing and internalizing problems Externalizing Problems Trajectories
were used to predict the probability of multiple class membership
in a joint statistical model. This statistical model, which is an To identify the optimal number of trajectories for externalizing
extension of the LCGA model, relates the entire longitudinal problems, models with one to six classes were estimated with the
course of behavioral and emotional problems and indicates the
probability that the identified subclasses co-occur between the two
types of behaviors (Muthen, 2000; Nagin & Tremblay, 2001). For Table 1
example, the chronic co-occurring group of children is composed Means, Standard Deviations, and Ranges for Externalizing and
of children who share common growth parameters for both high Internalizing Problems
externalizing and high internalizing problems. Thus, this model
Problem type
assigns class membership in trajectory classes across behaviors,
and age N M SD Min Max
taking into account longitudinal change over time. This type of
procedure is preferred over simple cross-tab analysis because it is Externalizing
based on latent classes and because it provides posterior probabil- 24 months 1,171 4.17 2.72 0 18
ities (Muthen, 2000; Nagin & Tremblay, 2001). 36 months 1,167 4.08 2.68 0 18
54 months 1,070 3.10 2.61 0 16
kindergarten 1,058 2.62 2.49 0 15
Early Childhood Risk Factors and Early Adolescent Grade 1 1,028 2.28 2.45 0 15
Grade 3 1,022 1.97 2.12 0 13
Adjustment Variables Grade 4 1,022 1.75 2.04 0 12
Grade 5 1,017 1.61 2.08 0 13
Multinomial logistic regression (MLR) analyses in SPSS were Grade 6 1,022 1.51 1.96 0 13
used to identify early childhood risk factors that discriminate Internalizing
among individuals with divergent pure or co-occurring develop- 24 months 1,171 1.16 1.26 0 8
mental trajectories. The analysis included four measures as inde- 36 months 1,167 1.41 1.42 0 11
pendent variablestemperament, cognitive functioning, maternal 54 months 1,070 1.60 1.51 0 8
kindergarten 1,058 1.44 1.52 0 8
depression, and familial environment controlling for gender, Grade 1 1,028 1.61 1.54 0 10
medical risk, and SDS risk. In addition, we used ANCOVA in Grade 3 1,022 1.64 1.70 0 12
SPSS to compare the identified trajectory groups in terms of early Grade 4 1,022 1.54 1.62 0 11
adolescent adjustment variables. The dependent variables were the Grade 5 1,017 1.68 1.71 0 10
Grade 6 1,022 1.55 1.75 0 11
childrens and the friends risky behaviors, asocial behavior with
 INTERNALIZING AND EXTERNALIZING PROBLEMS 1165

use of LCGA. As shown in Table 2, the BIC scores kept decreas- Internalizing Problem Trajectories
ing up to the five-class model. However, the six-class model had
a higher BIC, and the LMR statistic fell out of significance, Models with one to five classes were estimated with LCGA to
suggesting that the five-class model fit the data better. Addition- identify the optimal number of trajectories for internalizing prob-
ally, inspection of the graphs suggested that the six-class model lems. The BIC statistic changed dramatically from Class 2 to Class
split the lower class into two smaller groups, which were of limited 3, but the change was much smaller from Class 3 to Class 4, which
theoretical importance. Furthermore, the mean probability score suggests that the biggest improvement in fit occurred from the
for the five externalizing classes ranged from .77 to .94, and the two-class model to the three-class model (Table 2). In addition, the
LMR statistic fell out of significance for the four-class model,
entropy value was .76, suggesting that the classes were well
suggesting that the three-class model fit the data better. Moreover,
separated. Moreover, we compared the linear LCGA model to a
the four- and five-class models indicated the existence of two and
curvilinear one to investigate whether change is linear or curvilin-
three, respectively, very similar low classes of small theoretical
ear. The BIC (33,920.43) for the curvilinear model was lower in
importance. Accordingly, the more parsimonious three-class
comparison with the BIC (33,951.81) for the linear model, indi-
model was selected. Furthermore, the BIC (28,519.71) for the
cating that the curvilinear model better represented the data.
curvilinear model was lower compared with the BIC (28,559.65)
The trajectory groups identified with LCGA are shown in Figure
for the linear model, indicating that the curvilinear model better
1A, and the unstandardized intercepts and linear and quadratic represented the data. In addition, the mean probability score for the
terms are shown in Table 3. Children assigned to the low group three internalizing classes ranged from .87 to .90, and the entropy
(27%, n 332) exhibited some externalizing problems early on, was .76, suggesting that the classes were well differentiated in this
which declined to almost zero across time. Children in the model.
moderate-desisting group (31.6%, n 389) represented the largest Figure 1B shows the three-class model, and Table 3 shows the
class of children in the sample, suggesting that this group exhibited unstandardized intercept, linear, and quadratic terms for the dif-
normative levels of externalizing problems. Children in the mod- ferent groups. Children assigned to the low group (39.3%, n
erate group (19.4%, n 239) started at the same level as the 484) scored low on internalizing problems across time. Children in
moderate-desisting group but remained at moderate levels for the moderate group represented approximately half the sample
externalizing problems across time, suggesting that these children (42.6%, n 525), suggesting that moderate levels of internalizing
are at high risk for exhibiting continuous moderate externalizing problems are normative. Children in the high group (18.1%, n
problems. Children in the high-desisting group (13.6%, n 168) 223) started at higher levels of internalizing problems compared
started at high levels of externalizing problems but desisted to low with the other two groups and showed an increase over time.
levels across time. Children in the chronic group (8.4%, n 104) According to chi-square analyses, the different classes were not
started higher on externalizing problems compared with any of the differentiated in terms of ethnicity, although more girls were in the
other groups and remained high on externalizing problems, with a high-internalizing group, 2(1, N 1232) 4.11, p .01.
quadratic deceleration across the 10-year period. According to
chi-square analyses, the identified groups were not differentiated
Joint Occurrence
in terms of ethnicity, but some gender differences emerged. More
specifically, the low group was composed of more girls, 2(1, N Initially, a 15-class joint model representing all possible classes
1232) 8.17, p .001, and the chronic group was overrepre- between the three-class model for internalizing and the five-class
sented by boys, 2(1, N 1232) 11.05, p .001. model for externalizing problems (3 5) was estimated. Table 4
shows the posterior probabilities of children in each class on the
basis of the initial joint analysis. The entropy for the 15-class
model identified in Table 4 was .72, which is considered to be
Table 2
good. However, some identified classes had very low probabilities.
Model Fit Statistics
Specifically, Groups 4, 9, 12, and 13 had probabilities below .50
Problem type and (see Table 4), suggesting that around half of the children in each
model identified BIC Entropy LMR of these classes did not fit the category to which they were
assigned. The rest of the classes had good posterior probabilities
Externalizing (.75 to .90), indicating that these classes were distinct from other
1-class model 40,344.45 n/a n/a
2-class model 36,020.01 .87 p .01
classes. To ensure that the different classes were distinct from one
3-class model 34,525.04 .85 p .01 another, Groups 4, 9, 12, and 13 were excluded from the analyses.
4-class model 34,099.31 .80 p .01 The intercept and growth terms representing these classes were not
5-class model 33,951.81 .76 p .04 included in the joint analyses, and the children in these low-
6-class model 33,961.40 .72 p .56 probability classes were forced to identify with a different class.
Internalizing
1-class model 31,887.70 n/a n/a Table 4 also shows the posterior probabilities for the 11-class
2-class model 29,203.12 .82 p .01 model. Most of the children from the dropped groups identified
3-class model 28,559.65 .76 p .01 with children in Groups 7 and 10. The entropy for the 11-class
4-class model 28,406.75 .71 p .20 model was .77, and all the classes had good posterior probabilities
5-class model 28,290.08 .69 p .15
(.72 to .90), suggesting that the classes in the 11-class model were
Note. Values in bold indicate the selected model. BIC Bayesian more distinct than those in the 15-class model. Furthermore, the
information criterion; LMR Lo, Mendel, Rubin statistic. BIC (62,569.22) for the 11-group model was lower compared with
1166 FANTI AND HENRICH
Figure 1. Latent class growth analysis model for externalizing and internalizing problems.
the BIC (62,845.02) for the 15-class model, indicating that the
11-class model better represented the data.
Figure 2 demonstrates how the 11 identified classes changed
over time in terms of internalizing and externalizing problems.
Mplus allows for the identified longitudinal latent classes to be
used as a known latent categorical variable, enabling the investi-
gation of how membership in each group translates into longitu-
dinal change for each measured variable. This type of multiple-
group mixture modeling is used when there is one latent
categorical variable in the sample (i.e., individuals are placed in
different groups on the basis of their joint trajectories of internal-
izing and externalizing problems) that represents distinct groups of
individuals. When comparing Figures 1 and 2, we can conclude
that there are no considerable deviations in terms of the develop-
ment of externalizing and internalizing problems, after taking into
account the co-occurrence between these behavioral problems.
Furthermore, children in the chronic co-occurring group seem to
be at higher risk compared with the rest of the groups in terms of
both internalizing and externalizing problems.
Low-risk groups. On the basis of Figure 2, Groups 1, 2, 5,
and 6 might be considered to be at low or normative risk because
they exhibited low or moderate desisting forms of externalizing
problems and low or moderate internalizing problems. Children
assigned to Group 1 (17.1%, n 211) scored low in both problem
behaviors. This lower risk group was used as a normative com-
parison group for further analyses. Children in Group 2 (13.1%,
n 161) exhibited low internalizing and moderate desisting
externalizing problems. Children in Group 5 (8.1%, n 100)
exhibited low externalizing and moderate internalizing problems.
Children in Group 6 (15.5%, n 191) exhibited moderate inter-
nalizing and moderate desisting externalizing problems.
High-risk groups. Children assigned to Groups 3, 4, 7, 8, 9,
10, and 11 were considered to be at higher risk. These groups can
be divided in five different categories: (a) children in Group 4
(1.9%, n 24), the pure chronic externalizing group, and Group
8 (3.2%, n 39), the chronic-externalizing and moderate-
internalizing group, exhibited chronic externalizing problems and
normative levels (low and moderate) of internalizing problems, (b)
 INTERNALIZING AND EXTERNALIZING PROBLEMS 1167

Table 3 group, the pure chronic-externalizing group, and the pure

Unstandardized Growth Factor Parameter Estimates and moderate-externalizing group compared with the low-risk group.
Standard Errors (in Parentheses) Children characterized by difficult temperament were more likely
to be classified in the chronic co-occurring group, the high-
Problem type and internalizing and high-desisting externalizing group, the chronic-
group Intercept Linear slope Quadratic term
externalizing and moderate-internalizing group, and the pure
Externalizing chronic-externalizing group compared with the low-risk group,
Low 2.44 (0.085) .304 (.056) .007 (.008) with the chronic co-occurring and the pure chronic-externalizing
Moderate desisting 3.96 (0.052) .183 (.021) .002 (.002) groups having the higher ORs. Children with lower cognitive
Moderate 3.97 (0.180) .057 (.021) .005 (.003)
abilities, on the basis of the Bayley Mental Development Index,
High desisting 6.42 (0.074) .082 (.022) .007 (.003)
Chronic 7.54 (0.057) .010 (.019) .005 (.002) were more likely to be classified in the pure chronic-externalizing
Internalizing group compared with the low-risk group. Finally, children whose
Low 0.64 (0.087) .019 (.052) .002 (.007) mothers scored high on postpartum depression were less likely to
Moderate 1.28 (0.057) .079 (.020) .005 (.002) be classified in the low-risk group than all the higher risk groups,
High 2.31 (0.076) .126 (.018) .008 (.002)
with the chronic co-occurring group having the higher OR.
The analyses proceeded to compare the seven higher risk
p .05. p .01.
groups. The MLR comparing these groups was significant, 2(42,
N 569) 108.42, p .001. The findings suggested that
children in Group 3 (7.4%, n 91), the pure moderate- children born into a low-SDS environment were at higher risk for
externalizing group, and Group 7 (15.1%, n 186), the moderate- being classified in the chronic co-occurring group compared with
externalizing and moderate-internalizing group, exhibited moder- the pure moderate-externalizing group (OR 1.46, CI [1.08,
ate externalizing problems and normative levels (low and 2.00], p .05), the pure chronic-externalizing group (OR 1.54,
moderate) of internalizing problems, (c) children in Group 9 CI [1.01, 2.41], p .05), the pure internalizing group (OR 2.03,
(2.3%, n 28), the pure internalizing group, exhibited high levels CI [1.09, 3.86], p .05), the moderate-externalizing and
of internalizing problems and low levels of externalizing problems, moderate-internalizing group (OR 1.28, CI [1.00, 1.67], p
(d) children in Group 10 (12.6%, n 155), the high-internalizing .05), and the high-internalizing and high-desisting externalizing
and high-desisting externalizing group, exhibited high internaliz- group (OR 1.37, CI [1.04, 1.81], p .05). Furthermore, children
ing and high-desisting externalizing problems, and (e) children in faced with more early childhood medical risks had a higher like-
Group 11 (3.7%, n 46), the chronic co-occurring group, exhib- lihood of being in the pure moderate-externalizing group com-
ited chronic co-occurring internalizing and externalizing problems. pared with the chronic-externalizing and moderate-internalizing
Thus, there were children exhibiting pure externalizing problems group (OR 1.20, CI [1.07, 1.44], p .05) and the pure
(moderate and chronic), pure internalizing problems, high inter- internalizing group (OR 1.38, CI [1.06, 1.96], p .05).
nalizing and high-desisting externalizing problems, and co- Children who experienced a more positive and enriched early
occurring problems. home environment were more likely to be in the high-internalizing
According to chi-square analyses, the different groups were not and high-desisting externalizing group compared with the pure
differentiated in terms of ethnicity; therefore, ethnicity was not
included in further analyses. The groups exhibiting pure chronic Table 4
externalizing problems, 2(1, N 1232) 6.55, p .001, and Posterior Probabilities Based on the Joint Occurrence Analysis
moderate internalizing and chronic externalizing problems, 2(1, (N 1,232)
N 1232) 4.59, p .01, were overrepresented by boys.
Internalizing problem
MLR Analyses Externalizing
problem Low Moderate High
The analyses comparing the lower risk group with the rest of the
Initial group posterior probabilities
groups with the use of MLR was significant, 2(70, N 1232)
269.56, p .001. Table 5 incorporates odds ratios (ORs) to Low .8991 .8086 .76811
Moderate desisting .8052 .7617 .44512
compare each higher risk group with the low-risk group. In gen- Moderate .7613 .7538 .41113
eral, ORs reflect the odds likelihood of being in one group over the High desisting .4154 .4949 .74614
other, on the basis of the level of the independent variable. As Chronic .8595 .77210 .89015
shown in Table 5, children with higher SDS risk had a greater
Final group posterior probabilities
likelihood of being in the chronic co-occurring group compared
with the low-risk group. Moreover, children with more medical Low .7251 .8155 .8249
Moderate desisting .7822 .7876
risks had a higher likelihood of being in the high-internalizing and Moderate .7933 .8137
high-desisting externalizing group, the pure moderate- High desisting .81010
externalizing group, and the pure chronic-externalizing group Chronic .8374 .7868 .89911
compared with the low-risk group, with the pure moderate-
externalizing group having the higher OR. Children who experi- Note. The subscript numbers assign group membership. Values in bold
enced a less positive and enriched home environment were more indicate the groups that were excluded from the final joint occurrence
likely to be in the chronic-externalizing and moderate-internalizing analysis.
1168 FANTI AND HENRICH

Figure 2. The development of externalizing and internalizing problems based on the final group probabilities.
The left part of each groups label represents internalizing problems, and the right part represents externalizing
problems. In the Results section, the groups are labeled as follows: (1) low risk, (2) low internalizing and
moderate-desisting externalizing, (3) pure moderate externalizing, (4) pure chronic externalizing, (5) low
externalizing and moderate internalizing, (6) moderate internalizing and moderate-desisting externalizing, (7)
moderate externalizing and moderate internalizing, (8) chronic externalizing and moderate internalizing, (9) pure
internalizing, (10) high internalizing and high-desisting externalizing, and (11) chronic co-occurring. extern.
externalizing; intern. internalizing; mod. moderate; des. desisting.

moderate-externalizing group (OR .88, CI [.81, .95], p .05),
the pure chronic-externalizing group (OR .88, CI [.77, .99], p
.05), the moderate-externalizing and moderate-internalizing group
(OR .93, CI [.87, .99], p .05), and the chronic-externalizing
and moderate-internalizing group (OR .87, CI [.80, .95], p
.05). Children whose mothers scored lower on depression in early
childhood were more likely to be in the pure moderate-
externalizing group compared with the chronic-externalizing and
moderate-internalizing group (OR .94, CI [.90, .98], p .05),
the pure high-internalizing group (OR .94, CI [.89, .99], p
.05), and the chronic co-occurring group (OR .92, CI [.89, .96],
p .001) and were more likely to be classified into the moderate-
externalizing and moderate-internalizing group compared with the
chronic co-occurring group (OR .95, CI [.92, .98], p .01).
Children characterized by difficult temperament were more
likely to be classified in the chronic co-occurring group com-
pared with the pure moderate-externalizing group (OR 1.54,
CI [1.07, 3.94], p .05), the pure high-internalizing group
(OR 1.87, CI [1.18, 6.05], p .05), and the moderate-
externalizing and moderate-internalizing group (OR 1.77, CI
[1.17, 3.73], p .05). Children with lower scores on the Bayley
Mental Development Index, were more likely to be classified in
the pure chronic-externalizing group compared with the pure
moderate-externalizing group (OR .95, CI [.92, .99], p
.05), the chronic-externalizing and moderate-internalizing
group (OR .96, CI [.92, .99], p .05), and the pure inter-
nalizing group (OR .94, CI [.89, .98], p .01). Also,
children with lower scores on the Bayley index were less likely
Table 5
Multinomial Logistic Regression Analyses Comparing the Higher Risk Groups With the Lowest Risk Group in Terms of Early Childhood Risk Factors (N 1,232)

Group comparison

Risk factor 11 vs. 1 10 vs. 1 9 vs. 1 8 vs. 1 7 vs. 1 6 vs. 1 5 vs. 1 4 vs. 1 3 vs. 1 2 vs. 1

Gender
OR 1.73 0.94 1.53 2.33 1.23 1.15 0.71 3.51 2.02 1.78
95% CI [0.91, 3.27] [0.60, 1.47] [0.68, 3.45] [1.26, 4.33] [0.82, 1.86] [0.75, 1.76] [0.43, 1.18] [1.22, 10.09] [1.20, 3.40] [1.16, 2.73]
SDS risk
OR 1.40 1.07 0.73 1.29 1.15 1.05 1.11 0.97 1.02 1.07
95% CI [1.13, 1.89] [0.83, 1.37] [0.38, 1.39] [0.97, 1.73] [0.91, 1.45] [0.82, 1.35] [0.82, 1.48] [0.63, 1.51] [0.77, 1.34] [0.83, 1.37]
Medical risk
OR 1.18 1.26 0.95 1.17 1.15 1.13 0.99 1.29 1.38 1.02
95% CI [0.93, 1.50] [1.06, 1.49] [0.68, 1.33] [0.94, 1.47] [0.98, 1.36] [0.95, 1.34] [0.81, 1.20] [1.03, 1.78] [1.14, 1.67] [0.85, 1.20]
Home environment
OR 1.01 1.05 1.07 0.92 0.99 1.05 1.06 0.93 0.93 1.02
95% CI [0.93, 1.12] [0.98, 1.13] [0.92, 1.24] [0.84, 0.99] [0.93, 1.06] [0.99, 1.14] [0.98, 1.15] [0.83, 0.99] [1.01, 1.15] [0.95, 1.10]
Difficult temperament
OR 2.71 2.37 1.38 2.27 1.62 1.65 1.06 2.75 1.78 1.74
95% CI [1.31, 5.58] [1.40, 3.99] [0.52, 3.70] [1.14, 4.52] [0.99, 2.63] [1.18, 3.22] [0.60, 1.90] [1.01, 7.56] [0.98, 3.24] [1.06, 2.88]
Maternal depression
OR 1.11 1.08 1.09 1.09 1.06 1.07 1.02 1.06 1.03 1.03
95% CI [1.07, 1.15] [1.05, 1.11] [1.04, 1.14] [1.05, 1.13] [1.03, 1.09] [1.03, 1.10] [0.98, 1.06] [1.01, 1.12] [0.99, 1.07] [0.99, 1.06]
Bayley
OR 0.98 0.98 1.02 0.99 0.98 0.99 0.99 0.95 0.99 0.98
95% CI [0.96, 1.01] [0.96, 1.00] [0.98, 1.06] [0.64, 1.02] [0.96, 1.00] [0.97, 1.01] [0.96, 1.01] [0.91, 0.98] [0.97, 1.02] [0.96, 1.00]
INTERNALIZING AND EXTERNALIZING PROBLEMS

Note. Comparisons are based on odds ratios (ORs). Group 1 is the low-risk group; Group 2 is the low-internalizing and moderate-desisting externalizing group; Group 3 is the pure moderate
externalizing group; Group 4 is the pure chronic-externalizing group; Group 5 is the low-externalizing and moderate-internalizing group; Group 6 is the moderate-internalizing and moderate-desisting
externalizing group; Group 7 is the moderate-externalizing and moderate-internalizing group; Group 8 is the chronic-externalizing and moderate-internalizing group; Group 9 is the pure internalizing
group; Group 10 is the high-internalizing and high-desisting externalizing group; Group 11 is the chronic co-occurring group. SDS sociodemographic status; Bayley Bayley Mental Development
Index.

p .05. p .01.
1169
1170 FANTI AND HENRICH

to be classified in the pure internalizing group than in the
moderate-externalizing and moderate-internalizing group
(OR 1.04, CI [1.01, 1.07], p .05).
ANCOVAs
ANCOVAs were used to investigate how the trajectory groups
were associated with early adolescent social and behavioral ad-
justment, controlling for gender and covariates measured during
early childhood (SDS risk and medical risk). Table 6 uses the
standardized mean difference statistic (Cohens d), as an effect size
to compare the higher risk groups with the low-risk group. The
effect size was computed as the difference between the means of
the two groups under comparison (after these means were adjusted
for the covariate effects) divided by the root-mean-square error for
the particular model. Cohens d is a scale-free measure of the
separation between two group means. After comparing the higher
risk groups with the normative group, the analyses proceeded to
identify the group at higher risk for early adolescent maladjust-
ment.
Risky behaviors. According to Table 6, the three chronic-
externalizing groups (Groups 11, 8, 4) and the pure moderate-
externalizing group engaged in more risky behaviors and had more
friends who committed risky behaviors compared with the low-
risk group. The strongest effect size for the child and friend risky
behaviors (child/friend report) was found for the pure chronic-
externalizing group, and the strongest effect size for the mother-
reported child risky behaviors was found for the chronic co-occurring
and the chronic-externalizing and moderate-internalizing groups.
The SDS risk index was a significant covariate for child- and
friend-reported risky behaviors committed by friends of the
studys children, F(1, 1232) 27.60, 2p .02, p .001, and by
the studys children, F(1, 1232) 38.75, 2p .03, p .001. The
medical risk index was a significant covariate for mother-reported
child risky behaviors, F(1, 1232) 12.84, 2p .01, p .001.
There were significant gender differences for child-reported risky
behaviors, F(1, 1232) 13.97, 2p .02, p .001, and mother-
reported risky behaviors, F(1, 1232) 7.37, 2p .01, p .001,
committed by the studys children, with boys scoring higher on
both the child-reported measure (for boys, M 2.60, SD 1.97;
for girls, M 1.98, SD 1.56) and the mother-reported measure
(for boys, M 7.89, SD 4.38; for girls, M 6.56, SD 4.13).
The analysis proceeded to compare the four groups of children
who scored higher than the low-risk group. No significant differ-
ences were found for friends risky behaviors, F(3, 200) 0.38,
2p .005, p .76, or the study childrens risky behaviors as
reported by the children, F(3, 200) 0.31, 2p .004, p .82.
Significant differences were found only for risky behaviors en-
gaged in by the studys children as reported by mothers, F(3,
200) 2.93, 2p .04, p .04, with the chronic co-occurring
group scoring higher than the pure moderate-externalizing group
(d .25, p .05).
Peer exclusion. As shown in Table 6, the only high-risk group
that was not differentiated from the low-risk group in terms of peer
exclusion was the pure internalizing group, and the strongest effect
size was found for the chronic co-occurring group. The medical
risk index, F(1, 1232) 4.41, 2p .01, p .01, and the SDS
risk index, F(1, 1232) 7.39, 2p .01, p .01, were
significant covariates for peer exclusion, although there were no
gender differences, F(1, 1232) 0.07, 2p 0, p .79. The
ANCOVA comparing the six groups of children who scored
higher than the low-risk group in terms of peer exclusion was
significant, F(5, 541) 4.20, 2p .04, p .001. The findings
suggested that the chronic co-occurring group scored higher on
peer exclusion compared with the high-internalizing and high-
desisting externalizing group (d .25, p .05), the moderate-
internalizing and chronic-externalizing group (d .23, p
.05), the moderate-externalizing and moderate-internalizing
group (d .26, p .05), the pure moderate-externalizing group
(d .22, p .05), and the pure chronic-externalizing group
(d .14, p .05).

Table 6
Analyses of Covariance Comparing the Early Adolescent Adjustment of the Higher Risk Groups and the Lowest Risk Group
(N 1,232)

Group comparisons

Risk behavior F 2
p 11 vs. 1 10 vs. 1 9 vs. 1 8 vs. 1 7 vs. 1 6 vs. 1 5 vs. 1 4 vs. 1 3 vs. 1 2 vs. 1

Friend risky behaviors

(child/friend report) 4.51 .04 .22 .06 .09 .19 .03 .02 .04 .31 .25 .05
Child risky behaviors

(child/friend report) 3.99 .03 .21 .03 .11 .17 .03 .04 .09 .31 .24 .03
Child risky behaviors
(mother report) 21.45 .15 .31 .11 .02 .32 .11 .05 .01 .19 .21 .08
Asocial with peers

(mother/teacher report) 9.66 .07 .38 .22 .26 .11 .12 .11 .12 .27 .09 .05
Excluded by peers
(mother/teacher report) 9.05 .07 .37 .22 .09 .20 .20 .08 .05 .29 .21 .09

Note. The group comparisons are based on effect sizes (Cohens d). Group 1 is the low-risk group; Group 2 is the low-internalizing/moderate-desisting
externalizing group; Group 3 is the pure moderate externalizing group; Group 4 is the pure chronic-externalizing group; Group 5 is the low-externalizing/
moderate-internalizing group; Group 6 is the moderate-internalizing/moderate-desisting externalizing group; Group 7 is the moderate-externalizing/
moderate-internalizing group; Group 8 is the chronic-externalizing/moderate-internalizing group; Group 9 is the pure internalizing group; Group 10 is the
high-internalizing/high-desisting externalizing group; Group 11 is the chronic co-occurring group.

p .05. p .01.
 INTERNALIZING AND EXTERNALIZING PROBLEMS
Asocial behaviors with peers. As shown in Table 6, all
higher risk groups were more asocial with peers than the low-risk
group. The medical risk index, F(1, 1232) 3.75, 2p .01, p
.05, and the SDS risk index, F(1, 1232) 7.39, 2p .01, p .01,
were significant covariates for asocial behavior with peers, al-
though there were no gender differences, F(1, 1232) 0.27, 2p
0, p .60. The strongest effect size was found for the chronic
co-occurring group. The analysis proceeded to compare the nine
groups of children who scored higher than the normative group.
The ANCOVA comparing these groups in terms of being asocial
with peers was not significant, F(8, 860) 1.92, 2p .03,
p .06.
Discussion
The present study makes unique contributions by investigating
the development of co-occuring and distinct internalizing and
externalizing problems within a LCGA framework. In addition to
providing separate information for the development of internaliz-
ing and externalizing problems, the findings make three distinct
contributions. They (a) provide information on the developmental
pathways of pure and co-occurring internalizing and externalizing
problems, (b) identify early childhood risk factors related to co-
occurrence or their pure counterparts, and (c) identify ways in
which pure and co-occurring problem pathways are associated
with early adolescent social and behavioral adjustment.
When investigating the development of externalizing problems,
we identified children following trajectories of normative, moder-
ate, high-decreasing, and chronic externalizing problems, which is
in agreement with previous studies investigating the development
of externalizing problems (Broidy et al., 2003; Campbell et al.,
2006; Fanti & Henrich, 2007; Nagin & Tremblay, 1999; NICHD
Early Child Care Research Network, 2004; Shaw et al., 2003). In
terms of internalizing problems, the investigation is in agreement
with prior research in that the majority of the children participating
in the study exhibited low levels of internalizing problems from
toddlerhood to early adolescence and in that a minority of the
children were at higher risk of exhibiting continuous or increasing
internalizing problems (Sterba et al., 2007). However, the present
studys findings are unique because LCGA enabled the identifica-
tion of heterogeneous developmental patterns of pure or combined
internalizing and externalizing problems within a dynamic frame-
work by taking trajectories of change into account. Distinct devel-
opmental patterns of pure continuous externalizing (moderate and
chronic), pure internalizing, high-internalizing and high-desisting
externalizing, and co-occurring problems were identified.
The identification of distinct developmental patterns of high and
continuous internalizing and externalizing problems fits well with
previous suggestions that co-occurrence should be regarded as a
distinct syndrome (Angold & Costello, 1992; Lilienfeld, 2003;
OConnor et al., 1998). On the basis of prior literature (e.g.
Nottelman & Jensen, 1995; Oland & Shaw, 2005), we expected
that children experiencing higher levels of early childhood risk
factors would be more likely to exhibit combined internalizing and
externalizing problems and that children with chronic co-occurring
problems would have poorer social and behavioral adjustment in
early adolescence, compared with those with pure internalizing or
externalizing problems. Our findings provide partial support for
these expectations. In terms of early childhood risk processes,
1171
children with lower SDS were more likely to be classified as
having chronic co-occurring problems. Previous research has in-
dicated that children who come from low-SDS families are more
likely to be at high risk for both behavioral and emotional prob-
lems (Keenan, Shaw, Walsh, Delliquadri, & Giovannelli, 1997;
Keiley et al., 2000), and SDS risk has been considered as a general
stressor for psychopathology (Dodge, Pettit, & Bates, 1994; Dun-
can, Brooks-Gunn, & Klebanov, 1994). This finding provides
evidence for Angold and Costellos (1992) suggestion that one
possible explanation for the relation between behavioral and emo-
tional problems in children is that it represents undifferentiated
responding to stress. Our findings also indicate that the additive
effects of early childhood risk factors, including maternal depres-
sion and difficult temperament, may set the stage for the develop-
ment of co-occurring problems.
In terms of early adolescent adjustment, children with chronic,
co-occurring internalizing and externalizing problems were differ-
entiated from the rest of the sample in that they were more
excluded by peers during early adolescence, which might indicate
that the combination of internalizing and externalizing problems
acts as a general risk for being rejected by peers (Rudolph, Ham-
men, & Burge, 1994). For example, it may be that children who are
anxious/depressed as well as aggressive might be regarded as more
annoying by peers, compared with children who are just aggressive
or just depressed (Keiley et al., 2003). Furthermore, children with
chronic, co-occurring internalizing and externalizing problems
were also markedly asocial with peers. These findings indicate that
children with co-occurring problems do not have the necessary
social skills to associate with peers.
The investigation also identified a small number of children
exhibiting high pure externalizing problems. Children with lower
cognitive abilities in infancy were more likely to develop pure
externalizing problems. According to Moffitt (1993), mild cogni-
tive impairments early in development are associated to a host of
negative outcomes, including following a course of continuous
behavioral problems. Additionally, children living in an early
home environment low in support and stimulation were more
likely to develop continuous high or moderate externalizing prob-
lems but low or normative levels of internalizing problems. Dif-
ficult children may influence their parents negatively, resulting in
parents being less involved and providing a less positive or devel-
opmentally appropriate environment for these children (Shaw et
al., 2003). Eventually, a negative home environment may result in
the continuation of externalizing problems across development.
Thus, the current investigation provides evidence that low cogni-
tive abilities and a negative home environment might be more
strongly related to behavioral than emotional problems.
In contrast to our expectations, children following trajectories of
either pure externalizing or co-occurring internalizing and exter-
nalizing problems were not differentiated in terms of difficult
temperament and maternal depression. These findings indicate that
different individual and environmental factors similarly influence
the development of pure externalizing and co-occurring problems.
In addition, both children with pure externalizing problems and
children with co-occurring internalizing and externalizing prob-
lems had similarly high levels of deviant behaviors and friends
who committed risky behaviors during early adolescence, indicat-
ing continuity in their behavioral repertoire across time. Hence,
multiple developmental trajectories may lead to deviant behaviors,
1172 FANTI AND HENRICH
providing evidence for the concept of equifinality. The results also
indicate that it might be the continuation of externalizing prob-
lems, and not necessarily the levels of these problems, that lead to
deviant associations and behaviors later in life, because both
continuous moderate and chronic externalizing problems were
related to deviant behaviors and associations with deviant peers.
These findings are important because delinquent peer affiliations
may function to maintain and exacerbate childhood externalizing
problems (Laird et al., 2001).
The analyses also revealed a developmental pattern character-
ized by high internalizing problems and initially high but desisting
externalizing problems. In terms of multifinality, it is useful to
compare these children to those with co-occurring internalizing
and externalizing problems who had equivalent levels of internal-
izing and externalizing problems in early childhood and those with
pure externalizing problems who had equivalent levels of exter-
nalizing problems in early childhood. Children in these groups
were not differentiated in terms of medical risk, difficult temper-
ament, or maternal depression. As a result, these risk factors might
work together to set the stage for the initiation (but not the
continuation) of behavioral problems. However, children with a
more positive and enriched home environment were more likely to
have high internalizing problems and initially high but desisting
externalizing problems. On the basis of these findings, we might
conclude that the experiences of a positive home environment
might be sufficient for a sharp decline in externalizing problems.
Furthermore, children with greater SDS risk were more likely to
have co-occurring internalizing and externalizing problems, com-
pared with high internalizing but desisting externalizing problems,
again indicating that a positive and a low-stress familial environ-
ment might result in decreased externalizing problems but not
decreased internalizing problems. As a result, early environmental
experiences might be related to the resiliency of the high-
internalizing and high-desisting group, at least in the levels of
externalizing problems they exhibited. The findings also suggest
that the chronicity of externalizing problems for children who also
have high levels of internalizing problems cannot be predicted
from individual characteristics in early childhood.
In terms of equifinality, it is important to compare children with
high internalizing and high-desisting externalizing problems and
children exhibiting continuous high pure internalizing symptoms
because these two groups of children had similar levels of inter-
nalizing and externalizing problems at age 12. Our findings indi-
cated that difficult temperament and medical risk were associated
with initial high levels of externalizing problems but not with the
development of internalizing problems. Additionally, maternal de-
pression was associated with increased risk of being high in
internalizing problems as well as high in initial or chronic exter-
nalizing problems regardless of whether they were pure or co-
occurring. This finding suggests that maternal depression might be
a general risk marker for both internalizing and externalizing
problems and provides an example of multifinality in that a single
starting point may result in diverse outcomes. The impact of
maternal depression on childrens psychopathological problems is
well established, and there are a number of studies linking mater-
nal depression to childrens behavioral and emotional problems
(e.g., Gross et al., 2008; Shaw et al., 2003; Sterba et al., 2007).
Moreover, children with high internalizing and high-desisting
externalizing problems and those with pure internalizing problems
exhibited less deviant behaviors and associated less with deviant
peers, compared with those with pure externalizing problems.
According to prior research, anxiety and depression in the absence
of externalizing problems may serve as a protective factor against
later risky behaviors and being affiliated with delinquent peers
(Ialongo, Edelsohn, Werthamer-Larsson, & Crockett, 1996). How-
ever, children exhibiting chronic internalizing problems but not
continuous externalizing problems were similarly asocial with
peers, which is in accordance with Oland and Shaws (2005)
sociodevelopmental milestone model. The model suggests that
children exhibiting high levels of internalizing problems but low
levels of externalizing problems should also be more likely to
exhibit isolating behavior and to be withdrawn and avoidant be-
cause of the internalizing problems they exhibit. Furthermore,
because of these characteristics, children exhibiting high levels of
internalizing problems but low levels of externalizing problems are
less likely to be involved with delinquent peers and less likely to
exhibit delinquent behaviors. The studys findings align with this
theoretical description; asocial behavior with peers was the only
outcome that differentiated children with pure internalizing prob-
lems from those with neither internalizing nor externalizing prob-
lems. Thus, asocial behavior might be a common maladaptive
outcome for children following differential trajectories of exter-
nalizing and internalizing problems, providing an example of the
concept of equifinality. Children exhibiting externalizing symp-
toms tend to exhibit impulsive and undercontrolled behaviors
toward peers, which are considered by peers to be asocial behav-
iors (Calkins, Gill, & Williford, 1999), whereas children exhibiting
internalizing problems tend to be withdrawn and avoidant and
because of that they do not engage in friendly relations with peers,
which is also considered by peers to be asocial behavior (Oland &
Shaw, 2005; Rudolph et al., 1994).
The comparisons we discussed focus on the extremes of risk and
nonrisk. Children classified into these groupslow risk, chronic
co-occurring, high pure externalizing, pure internalizing, and high
internalizing and high-desisting externalizingtogether accounted
for less than half of the total sample. Thus, normative develop-
ment, at least in this sample, involves a degree of risk for devel-
oping various levels of internalizing and externalizing problems,
frequently in combination. Moreover, children who exhibit behav-
ior problems early in life do not necessarily continue to exhibit
these problems at later ages, and for some children, aggressive,
oppositional, and hyperactive behaviors early in life seem to be a
normal part of development. In addition, it is the children who
exhibited continuous behavioral problems who faced social and
delinquent problems. We also found that moderate internalizing
problems characterized the majority of the sample, suggesting that
some internalizing feelings might be a normal part of child devel-
opment. Comparisons between groups can also provide informa-
tion about the mechanisms involved in normal and psychopatho-
logical development, and our findings indicate that low individual
and environmental early risk processes protect the majority of
children from developing continuous behavioral problems.
Strengths, Limitations, Future Directions, and
Implications
Strengths of this investigation included a large sample of chil-
dren followed from birth to early adolescence. The nine data points
 INTERNALIZING AND EXTERNALIZING PROBLEMS
available for externalizing and internalizing problems enhanced
the reliability and flexibility of the longitudinal analyses (Singer &
Willett, 2003). Furthermore, the analyses incorporated data from
multiple informants and multiple methods. The investigation also
offered a number of methodological advances. Researchers have
been using statistical methods, such as correlations, clinical cutoff
scores, cluster analysis, and factor analysis, to identify syndromes
that tend to co-occur in the individual; however, these approaches
are not built to take longitudinal change into account, and at most,
these methods test the association of only two assessment periods
(Nagin & Tremblay, 2001). However, LCGA investigates co-
occurrence within a dynamic framework by taking trajectories of
change into account, by investigating nonlinear change, and by
using all the available longitudinal data in the analysis.
One limitation of the current study might be that the external-
izing and internalizing problem trajectory analyses relied exclu-
sively on mother reports. However, parents are considered to be a
critical source for reporting on their childs behavior (Shaw et al.,
2003), and mothers might be more observant of the childrens
internalizing symptoms compared with teachers or other infor-
mants (Keiley et al., 2000). Another limitation pertains to the
combinations of the two versions of the CBCL (the preschool
version for children ages 23 and the school-age version for
children ages 4 18), because behavioral and emotional problems
may manifest differently across the age range. For example, phys-
ical aggression is often used by toddlers to solve conflicts with
peers or playmates, although this type of aggression decreases over
the course of development and nonaggressive behaviors increase
(Tremblay, 2000). Additionally, the inclusion of other biological,
individual, and sociocultural risk factors and the consideration of
both risk and protective developmental processes might further
elucidate the differences between pure and co-occurring problems.
For instance, the presence of callous unemotional traits has been
related to higher externalizing problems, although these traits
suppress the development of internalizing problems (Frick & Ellis,
1999).
Investigating the trajectories of behavioral and emotional prob-
lems over time can also inform the timing and construction of
interventions (Loeber & Farrington, 1994). The current study
suggested that high pure and co-occurring internalizing and exter-
nalizing problems may start as early as the second year of life, and
according to Dodge and Pettit (2003), preventive interventions
should start early in life before antisocial outcomes or other types
of psychopathology become inevitable. In support of this idea,
prior work provided evidence that interventions during toddler-
hood are successful in reducing externalizing (Dishion et al.,
2008), internalizing (Shaw, Connell, Dishion, Wilson, & Gardner,
2009), and co-occurring behavioral and emotional problems (Con-
nell et al., 2008). The current findings may have the ability to
inform the construction of interventions by providing information
on unique or common factors influencing the development of pure
externalizing, pure internalizing, and co-occurring behavioral and
emotional problems.
In conclusion, the current investigation provides important in-
formation for the co-occurrence between different behavioral
problems, which is considered to be a major research challenge for
developmental psychopathology (Rutter & Sroufe, 2000). For ex-
ample, the studys findings on co-occurrence can inform previous
research concerned with children following a chronic trajectory of
1173
externalizing problems (Broidy et al., 2003; Nagin & Tremblay,
1999; NICHD Early Child Care Research Network, 2004; Shaw et
al., 2003). Much research has focused on finding possible ways to
identify these children, and the present study offers an additional
perspective by providing evidence that they can be differentiated
in terms of internalizing problems. In addition, our findings sug-
gest that following a path of continuous externalizing or co-
occurring internalizing and externalizing problems may lead to
engagement in deviant behaviors, association with deviant peers,
rejection by peers, and asocial behavior with peers. Moreover,
multiple child and environmental risks operate in conjunction with
each other to influence the development of chronic externalizing
and co-occurring externalizing and internalizing problems. How-
ever, experiencing a home environment high in support and stim-
ulation may lead to decreases in externalizing problems but not to
decreases in internalizing problems. Furthermore, even though
children exhibiting pure internalizing problems are asocial with
peers, they are at low risk for exhibiting deviant behaviors during
early adolescence. The continuous internalizing problems of these
children seemed to be uniquely predicted by maternal depression.
Finally, our work suggests that taking into account the concept of
co-occurrence is essential for explaining the development and
understanding the etiology of externalizing and internalizing prob-
lems.
References
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/418
and 1991 Profile. Burlington, VT: Department of Psychiatry, University
of Vermont.
Achenbach, T. M. (1992). Manual for the Child Behavior Checklist/23
and 1992 Profile. Burlington, VT: Department of Psychiatry, University
of Vermont.
Achenbach, T. M., Conners, C. K., Quay, H. C., Verhulst, F. C., & Howell,
C. T. (1989). Replication of empirically derived syndromes as a basis for
taxonomy of child/adolescent psychopathology. Journal of Abnormal
Child Psychology, 17, 299 323.
Achenbach, T. M., Howell, C. T., Quay, H. C., & Conners, C. K. (1991).
National survey of problems and competencies among 4- to 16-year-
olds: Parents reports for normative and clinical samples. Monographs of
the Society for Research in Child Development, 56, 1131.
Ackerman, B. P., DEramo, K. S., Umylny, L., Schultz, D., & Izard, C. E.
(2001). Family structure and externalizing behavior of children from
economically disadvantaged families. Journal of Family Psychology, 15,
288 300.
Angold, A., & Costello, E. (1992). Comorbidity in children and adoles-
cents with depression. Child and Adolescent Psychiatric Clinics of North
America, 1, 3151.
Angold, A., & Costello, E. (1993). Depressive comorbidity in children and
adolescents: Empirical, theoretical, and methodological issues. Ameri-
can Journal of Psychiatry, 150, 1779 1791.
Bauer, D., & Curran, P. (2003). Distributional assumptions of growth
mixture models: Implications for overextraction of latent trajectory
classes. Psychological Methods, 8, 338 363.
Bayley, N. (1993). Bayley Scales of Infant DevelopmentII. San Antonio,
TX: Psychological Corporation.
Bongers, I. L., Koot, H. M., van der Ende, J., & Verhulst, F. C. (2003). The
normative development of child and adolescent problem behavior. Jour-
nal of Abnormal Psychology, 112, 179 192.
Booth-LaForce, C., & Oxford, M. L. (2008). Trajectories of social with-
drawal from Grades 1 to 6: Prediction from early parenting, attachment,
and temperament. Developmental Psychology, 44, 1298 1313.
1174 FANTI AND HENRICH
Brendgen, M., Wanner, B., Morin, A. J. S., & Vitaro, F. (2005). Relations
with parents and with peers, temperament, and trajectories of depressed
mood during early adolescence. Journal of Abnormal Child Psychology,
33, 579 594.
Brennan, P. A., Hall, J., Bor, W., Najman, J. M., & Williams, G. (2003).
Integrating biological and social processes in relation to early-onset
persistent aggression in boys and girls. Developmental Psychology, 39,
309 323.
Broidy, L. M., Nagin, D. S., Tremblay, R. E., Bates, J. E., Brame, B.,
Dodge, K. A., . . . Vitaro, F. (2003). Developmental trajectories of child-
hood disruptive behaviors and adolescent delinquency: A six-site cross-
national study. Developmental Psychology, 39, 222245.
Bub, K. L., McCartney, K., & Willett, J. B. (2007). Behavior problem
trajectories and first-grade cognitive ability and achievement skills: A
latent growth curve analysis. Journal of Educational Psychology, 99,
653 670.
Caldwell, B. M., & Bradley, R. H. (1984). Home Observation of the
Measurement of the Environment. Little Rock, AK: University of Ar-
kansas at Little Rock.
Calkins, S. D., Gill, K., & Williford, A. (1999). Externalizing problems in
two-year-olds: Implications for patterns of social behavior and peers
responses to aggression. Early Education and Development, 10, 267
288.
Campbell, S. B. (1995). Behavior problems in preschool children: A
review of recent research. Journal of Child Psychology and Allied
Disciplines, 36, 113149.
Campbell, S. B., Spieker, S., Burchinal, M., Poe, M. D., & the National
Institute of Child Health and Human Development Early Child Care
Research Network. (2006). Trajectories of aggression from toddlerhood
to age 9 predict academic and social functioning through age 12. Journal
of Child Psychology and Psychiatry, 47, 791 800.
Cicchetti, D. (1984). The emergence of developmental psychopathology.
Child Development, 55, 17.
Cicchetti, D., & Rogosch, F. (1996). Equifinality and multifinality in
developmental psychopathology. Development and Psychology, 8, 597
600.
Coie, J. D., & Dodge, K. A. (1998). Aggression and antisocial behavior. In
W. Damon & N. Eisenberg (Eds.), Handbook of child psychology: Vol.
3. Social, emotional, and personality development (pp. 779 862). New
York, NY: Wiley.
Conger, R. D., & Elder, G. H. (1994). Families in troubled times: Adapting
to change in rural America. New York, NY: de Gruyter.
Connell, A., Bullock, B. M., Dishion, T. J., Shaw, D., Wilson, M., &
Gardner, F. (2008). Family intervention effects on co-occurring early
childhood behavioral and emotional problems: A latent transition anal-
ysis approach. Journal of Abnormal Child Psychology, 36, 12111225.
Dishion, T. J., Shaw, D. S., Connell, A., Gardner, F., Weaver, C., &
Wilson, M. (2008). The family check-up with high-risk indigent fami-
lies: Outcomes of positive parenting and problem behavior from ages 2
through 4. Child Development, 79, 13951414.
Dodge, K., & Pettit, G. S. (2003). A biopsychosocial model of the devel-
opment of chronic conduct problems in adolescence. Developmental
Psychology, 39, 349 371.
Dodge, K. A., Pettit, G. S., & Bates, J. E. (1994). Socialization mediators
of the relation between socioeconomic status and child conduct prob-
lems. Child Development, 65, 649 665.
Duggal, S., Carlson, E. A., Sroufe, A., & Egeland, B. (2001). Depressive
symptomatology in childhood and adolescence. Development and Psy-
chopathology, 13, 143164.
Duncan, G. J., Brooks-Gunn, J., & Klebanov, P. K. (1994). Economic
deprivation and early childhood development. Child Development, 65,
296 318.
Epkins, C. (2000). Cognitive specificity in internalizing and externalizing
problems in community and clinic-referred children. Journal of Clinical
Child Psychology, 29, 199 208.
Fanti, K. A., & Henrich, C. C. (2007). The relation of home and childcare/
school environment to differential trajectories of externalizing problems.
International Journal About Parents in Education, 1, 117123.
Frick, P. J., & Ellis, M. (1999). Callous unemotional traits and subtypes of
conduct disorder. Clinical Child and Family Psychology Review, 2,
149 168.
Ge, X., Best, K. M., Conger, R. D., & Simons, R. L. (1996). Parenting
behaviors and the occurrence and co-occurrence of adolescent depres-
sive symptoms and conduct problems. Developmental Psychology, 32,
717731.
Gilliom, M., & Shaw, D. S. (2004). Codevelopment of externalizing and
internalizing problems in early childhood. Development and Psychopa-
thology, 16, 313333.
Gross, H. E., Shaw, D. S., & Moilanen, K. L. (2008). Reciprocal associ-
ations between boys externalizing problems and mothers depressive
symptoms. Journal of Abnormal Child Psychology, 36, 693709.
Henrich, C. C., Schwab-Stone, M., Fanti, K., Jones, S. M., & Ruchkin, V.
(2004). The association of community violence exposure with middle
school achievement: A prospective study. Journal of Applied Develop-
mental Psychology, 25, 327348.
Hogue, A., & Steinberg, L. (1995). Homophily of internalizing distress in
adolescent peer groups. Developmental Psychology, 31, 897906.
Ialongo, N., Edelsohn, G., Werthamer-Larsson, L., & Crockett, L. (1996).
The course of aggression in first-grade children with and without co-
morbid anxious symptoms. Journal of Abnormal Child Psychology, 24,
445 456.
Kaplan, B., Crawford, S., Cantell, M., Kooistra, L., & Dewey, D. (2006).
Comorbidity, co-occurrence, continuum: Whats in a name? Child:
Care, Health and Development, 32, 613756.
Keenan, K., Shaw, D. S., Delliquadri, E., Giovannelli, J., & Walsh, B.
(1998). Evidence for the continuity of early problem behaviors: Appli-
cation of a developmental method. Journal of Abnormal Child Psychol-
ogy, 26, 441 454.
Keenan, K., Shaw, D. S., Walsh, B., Delliquadri, E., & Giovannelli, J.
(1997). DSMIIIR disorders in preschool children from low-income
families. Journal of the American Academy of Child and Adolescent
Psychiatry, 36, 620 627.
Keiley, M. K., Bates, J. E., Dodge, K. A., & Pettit, G. S. (2000). A
cross-domain growth analysis: Externalizing and internalizing behaviors
during 8 years of childhood. Journal of Abnormal Child Psychology, 28,
161179.
Keiley, M. K., Lofthouse, N., Bates, J. E., Dodge, K. A., & Pettit, G. S.
(2003). Differential risks of covarying and pure components in mother
and teacher reports of externalizing and internalizing behavior across
ages 5 to 14. Journal of Abnormal Child Psychology, 31, 267283.
Klein, D., & Riso, L. (1993). Psychiatric disorders: Problems of boundaries
and comorbidity. In C. Costello (Ed.), Basic issues in psychopathology
(Vol. 9, pp. 19 66). New York, NY: Guilford.
Kovacs, M., & Devline, B. (1998). Internalizing disorders in childhood.
Journal of Child Psychology and Psychiatry, 1, 47 63.
Ladd, G. W., Herald-Brown, S. L., & Andrews, R. K. (2009). The Child
Behavior Scale (CBS) Revisited: A longitudinal evaluation of CBS
subscales with children, preadolescents, and adolescents. Psychological
Assessment, 21, 325339.
Ladd, G. W., & Profilet, S. M. (1996). The Child Behavior Scale: A
teacher-report measure of young childrens aggressive, withdrawn, and
prosocial behaviors. Developmental Psychology, 32, 1008 1024.
Laird, R. D., Jordan, K. Y., Dodge, K. A., Pettit, G. S., & Bates, J. E.
(2001). Peer rejection in childhood, involvement with antisocial peers in
early adolescence, and the development of externalizing behavior prob-
lems. Development and Psychopathology, 13, 337354.
Lilienfeld, S. (2003). Comorbidity between and within childhood external-
 INTERNALIZING AND EXTERNALIZING PROBLEMS
izing and internalizing disorders: Reflections and directions. Journal of
Abnormal Child Psychology, 31, 285291.
Lo, Y., Mendell, N., & Rubin, D. B. (2001). Testing the number of
components in a normal mixture. Biometrika, 88, 767778.
Loeber, R., & Farrington, D. P. (1994). Problems and solutions in longi-
tudinal and experimental treatment of child psychopathology and delin-
quency. Journal of Consulting and Clinical Psychology, 62, 887900.
Lynam, D., Moffitt, T., & Stouthamer-Loeber, M. (1993). Explaining the
relation between IQ and delinquency: Class, race, test motivation, school
failure, or self-control? Journal of Abnormal Psychology, 102, 187196.
McLeod, J. D., & Shanahan, M. J. (1996). Trajectories of poverty and
childrens mental health. Journal of Health and Social Behavior, 37,
207220.
Medoff-Cooper, B., Carey, W. B., & McDevitt, S. C. (1993). Early infancy
temperament questionnaire. Journal of Developmental and Behavioral
Pediatrics, 14, 230 235.
Milan, S., Pinderhughes, E. E., & the Conduct Problems Prevention Re-
search Group. (2006). Family instability and child maladjustment tra-
jectories during elementary school. Journal of Abnormal Child Psychol-
ogy, 34, 4356.
Miller-Johnson, S., Lochman, J. E., Coie, J. D., Terry, R., & Hyman, C.
(1998). Comorbidity of conduct and depressive problems at sixth grade:
Substance use outcomes across adolescents. Journal of Abnormal Child
Psychology, 26, 221232.
Miner, J. L., & Clarke-Stewart, K. A. (2008). Trajectories of externalizing
behaviors from age 2 to age 9: Relations with gender, temperament,
ethnicity, parenting, and rater. Developmental Psychology, 44, 771786.
Moffitt, T. E. (1990). The neuropsychology of delinquency: A critical
review of theory and research. In N. Morris & M. Tonry (Eds.), Crime
and justice (Vol. 12, pp. 99 169). Chicago, IL: University of Chicago
Press.
Moffitt, T. E. (1993). Adolescence-limited and life-course persistent anti-
social behavior: A developmental taxonomy. Psychological Review,
100, 674 701.
Muthen, B. (2000). Latent variable mixture modeling. In G. A. Marcou-
lides & R. E. Schumacker (Eds.), Advanced structural equation model-
ing: New developments and techniques (pp. 133). Hillsdale, NJ: Erl-
baum.
Muthen, L. K., & Muthen, B. (2007). Mplus users guide. Los Angeles,
CA: Authors.
Nagin, D. S. (1999). Analyzing developmental trajectories: A semipara-
metric, group-based approach. Psychological Methods, 4, 139 157.
Nagin, D. S. (2005). Group based modeling of development. Cambridge,
MA: Harvard University Press.
Nagin, D. S., & Land, K. (1993). Age, criminal careers, and population
heterogeneity: Specification and estimation of a nonparametric mixed
Poisson model. Criminology, 31, 327362.
Nagin, D. S., & Tremblay, R. E. (1999). Trajectories of boys physical
aggression, opposition, and hyperactivity on the path to physically
violent and nonviolent juvenile delinquency. Child Development, 70,
11811196.
Nagin, D. S., & Tremblay, R. E. (2001). Analyzing developmental trajec-
tories of distinct but related behaviors: A group based method. Psycho-
logical Methods, 6, 18 34.
National Institute of Child Health and Human Development Early Child
Care Research Network. (2004). Trajectories of physical aggression
from toddlerhood to middle childhood. Monographs of the Society for
Research in Child Development, 69, 1128.
Newman, D., Moffitt, T., Caspi, A., & Silva, P. (1998). Comorbid mental
disorders: Implications for treatment and sample selection. Journal of
Abnormal Psychology, 107, 305311.
Nottelman, E. D., & Jensen, P. S. (1995). Comorbidity of disorders in
1175
children and adolescents: Developmental perspectives. In T. H. Ollen-
dick & R. J. Prinz (Eds.), Advances in clinical child psychology (Vol. 17,
pp. 109 155). New York, NY: Plenum.
OConnor, T. G., McGuire, S., Reiss, D., Hetherington, E. M., & Plomin,
R. (1998). Co-occurrence of depressive symptoms and antisocial behav-
ior in adolescence: A common genetic liability. Journal of Abnormal
Psychology, 107, 2737.
Oland, A. A., & Shaw, D. S. (2005). Pure versus co-occurring externalizing
and internalizing symptoms in children: The potential role of socio-
developmental milestones. Clinical Child and Family Psychology Re-
view, 8, 247270.
Radloff, L. S. (1997). The CES-D Scale: A self-report depression scale for
research in the general population. Applied Psychological Measurement,
1, 385 401.
Reitz, E., Dekovic, M., & Meijer, A. M. (2005). The structure and stability
of externalizing and internalizing problem behavior during early adoles-
cence. Journal of Youth and Adolescence, 34, 577588.
Rudolph, K. D., Hammen, C., & Burge, D. (1994). Interpersonal function-
ing and depressive symptoms in childhood: Addressing the issues of
specificity and comorbidity. Journal of Abnormal Child Psychology, 22,
355371.
Rutter, M. (1997). Comorbidity: Concepts, claims and choices. Criminal
Behavior and Mental Health, 7, 265285.
Rutter, M., & Sroufe, L. A. (2000). Developmental psychopathology:
Concepts and challenges. Development and Psychopathology, 12, 265
296.
Schwartz, G. (1978). Estimating the dimension of a model. Annals of
Statistics, 6, 461 464.
Shaw, D. S., Connell, A., Dishion, T. J., Wilson, M. N., & Gardner, F.
(2009). Improvements in maternal depression as a mediator of interven-
tion effects on early childhood problem behavior. Development and
Psychopathology, 21, 417 439.
Shaw, D. S., Gilliom, M., Ingoldsby, E. M., & Nagin, D. S. (2003).
Trajectories leading to school-age conduct problems. Developmental
Psychology, 39, 187200.
Shaw, D. S., Owens, E. B., Vondra, J. I., Keenan, K., & Winslow, E. B.
(1996). Early risk factors and pathways in the development of early
disruptive behavior problems. Development and Psychopathology, 8,
679 699.
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis:
Modeling change and event occurrence. Oxford, England: Oxford Uni-
versity Press.
Sterba, S. K., Prinstein, M. J., & Cox, M. J. (2007). Trajectories of
internalizing problems across childhood: Heterogeneity, external valid-
ity, and gender differences. Development and Psychopathology, 19,
345366.
Tremblay, R. E. (2000). The development of aggressive behavior during
childhood: What have we learned in the past century? International
Journal of Behavioral Development, 24, 129 141.
Wright, J., Zakriski, A., & Drinkwater, M. (1999). Developmental psycho-
pathology and the reciprocal patterning of behavior and environment:
Distinctive situational and behavioral signatures of internalizing, exter-
nalizing, and mixed-syndrome children. Journal of Consulting and Clin-
ical Psychology, 67, 95107.
Youngstrom, E., Findling, R., & Calabrese, J. (2003). Who are the comor-
bid adolescents? Agreement between psychiatric diagnosis, youth, par-
ent, and teacher report. Journal of Abnormal Child Psychology, 31,
231245.
Received February 4, 2009
Revision received June 15, 2010
Accepted June 21, 2010