Child Development, March/April 2003, Volume 74, Number 2, Pages 578–594

Parents Do Matter: Trajectories of Change in Externalizing and Internalizing

Problems in Early Adolescence
Nancy L. Galambos, Erin T. Barker, and David M. Almeida

This study examined the relative influence of three parenting behaviors (support, behavioral control, and
psychological control) and deviant peers on trajectories of externalizing and internalizing problems in early
adolescence. A white, working-to-middle-class sample of adolescents and their mothers and fathers in two-
earner families participated in a 312-year longitudinal study (N=109 families). The study began when the
adolescents were in sixth grade (M age=11.5 years). Analyses showed that parents’ firm behavioral control
seemed to halt the upward trajectory in externalizing problems among adolescents with deviant peers. Initial
levels of internalizing problems were higher among adolescents with parents who reported lower levels of
behavioral control and among adolescents with deviant peers. This study suggests that parenting exerts an
important influence in adolescents’ lives and may do so even in the face of potentially negative peer influence.

Several decades of research on parent–child relations adolescents’ association with deviant peers and
has led to the identification of three global, relatively externalizing problems such as drug use and delin-
independent dimensions of parental behavior: sup- quency (Brendgen, Vitaro, & Bukowski, 2000). The
port (responsiveness and connectedness to the child), focus of most contemporary developmental psychol-
behavioral control (regulation of the child’s behavior ogists, however, is on how parents and peers jointly
through firm and consistent discipline), and psycho- influence adolescents (Collins, Maccoby, Steinberg,
logical control (control of the child’s behavior Hetherington, & Bornstein, 2000). Therefore, in this
through psychological means such as love with- study, we examined how parenting (parental support,
drawal and guilt induction); (Barber, Olsen, & Shagle, behavioral control, and psychological control) and the
1994; Maccoby & Martin, 1983; Schwarz, Barton- deviant peer context are related to concurrent and
Henry, & Pruzinsky, 1985). Although each of these changing levels of individuals’ externalizing (e.g.,
parenting behaviors has been linked with indicators smoking, shoplifting) and internalizing (e.g., depres-
of adolescents’ behavior and psychosocial adjustment sive symptoms) problems over several years. These
(Barber et al., 1994; Forehand & Nousiainen, 1993), relations were observed from early to middle adoles-
there is a very limited body of research examining all cence because it is during this period of multiple,
three parenting behaviors in the same study as interrelated physical, social, and cognitive changes
unique predictors of adolescent development. Thus, and increasing levels of behavioral autonomy that
an important goal of this study was to understand young people become vulnerable to the potential
the relative and unique influence of support, beha- negative influences of peers, typically show an
vioral control, and psychological control on specific increase in some externalizing behaviors, and may
aspects of adolescent adjustment (Barber, 1997). first experience internalizing problems (Maggs, Al-
A second major issue concerns the combined meida, & Galambos, 1995; Moffitt, 1993; Petersen et al.,
influence of parents and peers on adolescent devel- 1993). Moreover, early adolescence is a sensitive
opment. Recent (and controversial) arguments pose period for parents because they must learn to facilitate
the question: Do parents matter? (Harris, 1995). After appropriate levels of autonomy in their children, relax
all, there is a strong and well-established link between some control, and remain supportive during a
demanding transition (Galambos & Ehrenberg, 1997).
Nancy L. Galambos and Erin T. Barker, Department of
The extent to which parents are successful in meeting
Psychology, University of Alberta; David M. Almeida, Division these challenges may influence how well adolescents
of Family Studies, University of Arizona. adjust to the many changes they experience.
This research was supported by a Social Sciences and
Humanities Research Council of Canada grant to the first author.
Parenting Dimensions and Adolescent Adjustment
The authors acknowledge the participation of the families who
participated in this project. The examination of parental support, behavioral
Correspondence concerning this article should be addressed to
Nancy L. Galambos, Department of Psychology, P217 Biological control, and psychological control as independent
Sciences Bldg., University of Alberta, Edmonton, AB T6G 2E9, r 2003 by the Society for Research in Child Development, Inc.
Canada. Electronic mail may be sent to galambos@ualberta.ca. All rights reserved. 0009-3920/2003/7402-0017
 Parents Do Matter 579

predictors of adolescent adjustment is important for Second, scholars have argued for the usefulness of
several reasons. First, studies that have examined typologies of parenting styles that consist of the
two or three of the parenting behaviors reveal aggregation of parenting behaviors (Baumrind, 1991;
interesting differential associations of parenting with Maccoby & Martin, 1983). For example, different
adolescent adjustment (Barber, 1997). For example, combinations of support and behavioral control
when compared with psychological control or have been used to operationalize four parenting
support, higher behavioral control has been linked styles: authoritative (high support/high control),
consistently with lower levels of externalizing authoritarian (low support/high control), indulgent
problems such as substance use, antisocial behavior, (high support/low control), and neglectful (low
and delinquency (Barber & Olsen, 1997; Eccles, support/low control); (Lamborn, Mounts, Steinberg,
Early, Frasier, Belansky, & McCarthy, 1997; Garber, & Dornbusch, 1991; Slicker, 1998; Steinberg, Lam-
Robinson, & Valentiner, 1997; Gray & Steinberg, born, Darling, Mounts, & Dornbusch, 1994). Such
1999; Herman, Dornbusch, Herron, & Herting, 1997; typologies assume that there are interactions among
Pettit, Laird, Dodge, Bates, & Criss, 2001). On the the parenting behaviors that constitute the types
other hand, higher psychological control has been (e.g., Support
Behavioral Control), yet not all
associated with internalizing problems such as studies have tested this assumption (e.g., Lamborn
depressed mood (Barber et al., 1994; Garber et al., et al., 1991; Steinberg et al., 1994). In those that have,
1997; Gray & Steinberg, 1999; Pettit et al., 2001). significant interactions largely have failed to materi-
Moreover, there is growing evidence that psycholo- alize (Barber et al., 1994; Garber et al., 1997; Herman
gical control is also related positively to externaliz- et al., 1997; Kurdek & Fine, 1994; Kurdek, Fine, &
ing problems (Barber, 1996; Barber & Olsen, 1997; Sinclair, 1995; Stice, Barrera, & Chassin, 1993; see
Eccles et al., 1997). Thus, psychological control Forehand & Nousiainen, 1993, for an exception).
seems to be more broadly related to adolescent These results cast doubt on the validity of the
problems than is behavioral control. parenting typologies and suggest that it is crucial to
With respect to the role of parental support vis-à- disaggregate and investigate the separate dimen-
vis behavioral and psychological control, two stu- sions that constitute overall parenting style (Barber,
dies have found that higher support, higher behav- 1997; Herman et al., 1997). Moreover, because so few
ioral control, and lower psychological control were studies have examined whether support, behavioral
related significantly to higher academic competence control, and psychological control interact in the
(Eccles et al., 1997; Gray & Steinberg, 1999). In a prediction of adolescent adjustment, it is important
similar study, Herman et al. (1997) reported that to conduct systematic testing of such interactions
support was related to higher grades, but lost to (Barber, 1997). In so doing, a more accurate and
behavioral and psychological control in regressions comprehensive picture will be obtained of the ways
used to predict other adolescent outcomes such as in which parenting behaviors are linked to adoles-
substance use. In their inclusive model, Garber et al. cent adjustment.
(1997) reported that both lower support and higher A third reason for considering the three parenting
psychological control were linked with adolescent behaviors simultaneously is to allow for the exam-
depression. Another study found that when parental ination of curvilinear relations between behavioral
support, behavioral control, and psychological con- control and adolescent adjustment. Authors have
trol were examined simultaneously, support was proposed that moderate levels of parental behavioral
linked negatively and strongly to externalizing and control are best for adolescents’ psychosocial adjust-
internalizing problems in adolescents, and behavior- ment, with high and low levels least effective
al control was not associated with either problem (Rollins & Thomas, 1979). Although several studies
(Fauber, Forehand, Thomas, & Wierson, 1990). have found curvilinear effects, results have been
Others (Costigan, 1996; Forehand & Nousiainen, inconsistent. There has been some support for the
1993) have also found that parental support was a ‘‘moderate control is best’’ hypothesis (Kurdek et al.,
primary predictor of adolescents’ adjustment, rela- 1995; Mason, Cauce, Gonzales, & Hiraga, 1996;
tive to psychological or behavioral control. Some Miller, McCoy, Olson, & Wallace, 1986; Stice et al.,
authors have suggested that parental support facil- 1993), but others have reported steep increases in
itates adolescents’ feelings of psychological well- adolescents’ psychosocial competence at the upper
being, gives them confidence, and leads generally levels of control (Costigan, 1996; Kurdek & Fine,
to social and academic competence (Barber, 1994). Clearly, further research is needed to test for
Maugahn, Olsen, & Thomas, 2002; Gray & Steinberg, and describe the specific nature of curvilinear
1999). relationships.
580 Galambos, Barker, and Almeida

Shortcomings in Research on Parenting and Adolescent adolescents and with a dampening of the typical
Adjustment upward trajectory of alcohol misuse that occurs in
adolescence. Similarly, Simons, Chao, Conger, and
A shortcoming of previous work on parenting Elder (2001) used four waves of data to show that
behaviors and adolescents’ adjustment is heavy inept parenting (low monitoring, harsh discipline,
reliance on adolescents’ self-reports as indicators of hostility) increased adolescents’ affiliation with
parents’ behaviors and adolescents’ own problems deviant peers and their delinquent behavior across
(e.g., see Barnes, Reifman, Farrell, & Dintcheff, 2000; a 4-year period. In the current study, longitudinal
Gray & Steinberg, 1999; Stice et al., 1993). This data collected at five waves across the period of
practice creates a common method variance problem early adolescence were used to examine how
due to the use of a single reporter and, hence, may parenting in Grade 6 influenced trajectories of
exaggerate the true relations between parenting and adolescent problems.
adolescent adjustment. Some researchers have
avoided the common method variance problem by
Parenting Influences in the Context of Peers
using combined mother–adolescent data (Barber et
al., 1994) or by obtaining measures of adolescent Recent longitudinal studies show that adolescents
adjustment from independent reporters (Kurdek et with deviant peers engaged in higher levels of
al., 1995; Mason et al., 1996). In the current study, externalizing behavior (Mason et al., 1996; Scara-
combined reports from mothers and fathers were mella, Conger, Spoth, & Simons, 2002; Simons et al.,
used to measure parenting behavior, whereas ado- 2001). Moreover, there was an association between
lescents’ reports were used to assess their externaliz- having deviant friends and the presence of inter-
ing and internalizing problems. nalizing problems (i.e., depression; Brendgen et al.,
Another gap in the research to date is the dearth 2000). These studies also show that the relation
of longitudinal data examining the links between between deviant peers and adolescent adjustment
these three dimensions of parenting and adolescent depended on the quality of parenting. For example,
problems over time. Studies finding contempora- low parental support and low behavioral control had
neous links between dimensions of parenting and direct influences on adolescents’ higher levels of
adolescent adjustment are limited because it is externalizing problems (e.g., aggression, disruptive-
impossible to sort out whether a significant correla- ness; Kim, Hetherington, & Reiss, 1999; Pettit et al.,
tion is due to the parent’s influence on the 2001; Scaramella et al., 2002). There were also
adolescent, the adolescent’s influence on the parent, indirect effects of parenting behaviors that operated
or a third variable. Only through longitudinal through their impact on adolescents’ associations
studies that control for initial levels of adolescent with deviant peers. Specifically, poorer parenting
adjustment can conclusions be drawn about the (e.g., low involvement) increased the probability that
potential longer term effects of parenting on adoles- adolescents would affiliate with deviant peers, and
cents (Collins et al., 2000). Longitudinal studies of more association with deviant peers, in turn, was
parenting and adolescent development typically related to externalizing problems such as delin-
have examined changes in adolescents’ behaviors quency (Kim et al., 1999; Mason et al., 1996;
at only two time points (Barber, 1996; Herman et al., Scaramella et al., 2002; Simons et al., 2001). Given
1997; Mason et al., 1996; Steinberg et al., 1994; Stice & the importance of peers as contributors to adolescent
Barrera, 1995). Studies limited to two time points, behavior, in the present study, the relative influences
however, are unable to provide precise and reliable of parental support, behavioral control, and psycho-
estimates of intraindividual change. Multiwave, logical control on adolescent adjustment were
multiyear studies, on the other hand, enable the considered along with the influence of deviant peers.
precise modeling of individual growth trajectories,
determination of the rates of change in behavior, and
The Present Study
investigation of predictors of interindividual differ-
ences in intraindividual change (Bryk & Rauden- In the current investigation, all three parenting
busch, 1992; Willett, 1989). dimensions were examined as predictors of adoles-
In an exemplary multiwave study that followed cents’ externalizing and internalizing problems
13- to 16-year-old adolescents over a 7-year period, using data from a study that followed adolescents
Barnes et al. (2000) demonstrated that parents’ in two-parent families five times across a 3.5-year
higher behavioral control at Wave 1 was linked both period (Grade 6 to the summer following Grade 9).
with low initial (Wave 1) levels of alcohol misuse in Hierarchical linear modeling (HLM; Bryk &
 Parents Do Matter 581

Raudenbusch, 1992) was used for this purpose. The with 112 adolescents (62 girls, 50 boys) who were in
impact of initial (Grade 6) parenting behaviors on Grade 6 (M age 5 11.5 years, SD 5 .42) at the first
the initial level of adolescents’ externalizing and time of measurement. Data were collected on five
internalizing problems as well as on trajectories of occasions: winter 1988 (Time 1), summer 1988 (Time
change in these problems was examined. Following 2), winter 1989 (Time 3), summer 1990 (Time 4), and
previous investigations (e.g., Barber et al., 1994; summer 1991 (Time 5).
Barnes et al., 2000), we expected that parents’ higher With respect to the occupations held by the
behavioral control would be linked with lower initial parents, 25% of fathers were in professional/techni-
levels of and a dampening of the typical increase in cal occupations; 35% were in managerial, sales, or
externalizing problems across adolescence whereas clerical occupations; and 39% were in service,
parents’ higher psychological control would be unskilled, or skilled labor occupations. Correspond-
linked with higher initial levels of and an upward ing figures for mothers were 18%, 73%, and 10%,
increase in both externalizing and internalizing respectively. Mean family socioeconomic status
problems. Given the literature on the role of parental (SES) was 50.01 (SD 5 15.18), as assessed by the
support, we expected that support would be linked father’s score on the Blishen and McRoberts (1976)
inversely with externalizing and internalizing pro- occupational index for Canadian samples. Examples
blems. The current study also tested all two-way of occupations and their SES scores are motor
interactions among the three parenting behaviors in vehicle repair (32.8), real estate sales (50.1), and
the prediction of adolescents’ problems, as well as accounting (67.4). Thus, the sample is characterized
the curvilinear effect of behavioral control on these largely as working to middle class.
problems. The mothers of the participants worked for an
To examine parenting behavior in the context of average of 30.9 hours per week (SD 5 11.2), and
peers, adolescents’ reports of deviant peer associa- fathers worked an average of 42 hours per week
tions were used to disentangle (a) the concurrent (SD 5 10.3). The mothers had been employed for an
effects of Grade 6 parenting and deviant peers on average of 6.5 years (SD 5 5.4); fathers had been
initial levels and trajectories of change in adoles- employed an average of 17.2 years (SD 5 7.4). The
cents’ problems, and (b) the effects of Grade 6 mean number of children per family was 2.4
parenting on initial levels and trajectories of change (SD 5 0.9), and the mean number of years married
in adolescents’ problems, controlling for deviant was 14.7 (SD 5 4.5). Mothers and fathers had
peers at all occasions. We hypothesized that despite achieved similar levels of education (mothers:
an expected positive relation between deviant peers M 5 13.1 years, SD 5 2.2; fathers: M 5 13.8 years,
and adolescents’ problems, some parenting behav- SD 5 3.1). The mean age of the mothers was 37.4
iors would emerge as significant in the prediction of years (SD 5 3.9) and the mean age of fathers was 40
initial levels and trajectories of change in adoles- (SD 5 5.5).
cents’ problems (e.g., see Kim et al., 1999; Scaramella
et al., 2002). Finally, interactions between deviant
Procedure
peers and the parenting behaviors in the prediction
of initial levels and trajectories of change in At each time of measurement, questionnaires
adolescents’ externalizing and internalizing pro- were mailed individually to each family member to
blems were investigated. Based on the findings of complete and return by mail. Participants were
previous research (e.g., Mason et al., 1996), it was asked not to discuss the questionnaires with one
expected that among adolescents with deviant peers, another and were given separate return envelopes.
parents’ high behavioral control would halt the As an additional reassurance of confidentiality,
typical upward trajectory in externalizing problems. questions pertaining to adolescents’ externalizing
behaviors were printed and collated separately, and
an extra envelope was provided in which to seal
Method these questions on completion. Each participant
Participants received a token payment after participating (in-
creasing from $5 at Time 1 to $15 at Time 5).
The participants were adolescents, mothers, and Of the 112 families in the study, three were
fathers who participated in a 3.5-year longitudinal excluded because of missing data, leaving a sample
study of White, two-parent families in which both size of 109. The percentages of original participants
parents were employed (the Two-Earner Family providing data at each time of measurement were:
Study; Galambos & Maggs, 1991). The study began Time 2 (94%, n 5 103), Time 3 (83%, n 5 91), Time 4
582 Galambos, Barker, and Almeida

(70%; n 5 76), and Time 5 (69%; n 5 75). Time was logical control, respectively). Moreover, the mean
coded as the number of years that had passed since difference between mothers’ and fathers’ reports
the first occasion of measurement, thus reflecting the was  .23,  .04, and .15 for support, behavioral
uneven spacing of the intervals between occasions. control, and psychological control, respectively.
Specifically, time was coded as: Time 1 (0), Time 2 Close to 90% of the mother–father pairs differed by
(.5), Time 3 (1), Time 4 (2.5), and Time 5 (3.5). no more than 1 point on the 5-point scale. Thus, the
Participants who dropped out of the study did not practice of combining the scores in this sample did
differ from those who remained with respect to not average out extreme differences. There were six
demographic variables (e.g., education, age, SES). cases in which mother-only scores were used
Mothers and fathers who dropped out did not differ because fathers’ data were missing. There was one
from parents who remained on initial levels of all case in which father-only scores were used because
three parenting behaviors. As well, adolescents who mothers’ data were missing.
dropped out did not differ from those who stayed in Deviant peers. At each of Times 1 through 5, the
the study on initial levels of externalizing or mean of four items (Galambos & Maggs, 1991) was
internalizing problems and deviant peers. used to assess the extent to which the adolescent’s
friends engaged in misconduct or problem behavior,
such as shoplifting or damaging property. These
Measures
items (e.g., ‘My friends often get into trouble with
Parenting. Time 1 scores on the parent version of adults’’; ‘‘Some of my friends break other people’s
the 56-item Child’s Report of Parental Behavior things for fun’’) were rated on a 4-point scale
Inventory (CRPBI; Burger & Armentrout, 1971; ranging from 1 (disagree strongly) to 4 (agree strongly).
Schaefer, 1965) were used to assess mothers’ and Cronbach’s alpha ranged from .69 to .78 across the
fathers’ perceptions of their own support, behavioral five occasions of measurement. Higher scores on this
control, and psychological control in relation to their measure have been linked with adolescents’ higher
adolescent. The support measure consisted of the levels of problem behavior and lower levels of
mean of 24 items (e.g., ‘‘I almost always speak to our parental support (Galambos & Maggs, 1991; Galam-
child in a warm and friendly voice’’) rated on a 5- bos, Sears, Almeida, & Kolaric, 1995).
point scale ranging from 1 (very much unlike me) to 5 Externalizing problems. At Times 1 through 5,
(very much like me). We expanded the original 3-point externalizing problems were measured with the
response scale to increase variability. Cronbach’s mean of eight items from the Brown, Clasen, and
alphas were .92 for fathers and mothers. The Eicher (1986) misconduct scale and an additional 16
behavioral control measure consisted of the mean items from the Kaplan (1978) deviant response scale.
of 16 items (reverse scored) (e.g., ‘‘I let our child get Adolescents were asked ‘‘how many times in the
away with a lot of things’’; ‘‘I let our child stay up past month’’ they had engaged in 24 behaviors, rated
late if he/she keeps asking’’), also rated on the 5- on a 5-point scale ranging from 1 (never) to 5 (almost
point scale. Cronbach’s alphas were .80 for fathers every day). The items covered a range of behaviors
and .84 for mothers. The psychological control including disobedience to parents (e.g., ‘‘done some-
measure consisted of the mean of 15 items (e.g., ‘‘I thing that your parents told you not to do’’), school
say that someday our child will be sorry that he/she misconduct (e.g., ‘‘been suspended or expelled from
wasn’t better as a child’’; ‘‘I think our child is not school’’), substance use (e.g., ‘‘smoked marijuana’’),
grateful when he/she doesn’t obey’’) rated on a 5- and antisocial behavior (e.g., ‘‘damaged or destroyed
point scale (typically 16 items long; one item was public or private property on purpose’’). Cronbach’s
omitted at the request of school officials). Cronbach’s alphas ranged from .77 to .90 across the five times of
alphas were .87 for fathers and .85 for mothers. The measurement. Higher scores on this measure (in-
reliability and validity of these subscales are well dicating more problem behavior) were correlated
established (Schwarz et al., 1985). significantly with lower levels of impulse control
To combine mothers’ and fathers’ reports, the and mastery, and higher levels of peer involvement
mean of mothers’ and fathers’ scores on each scale and association with deviant peers (Galambos &
was calculated. Mothers’ and fathers’ reports were Maggs, 1991; Galambos et al., 1995).
combined because of evidence that aggregate scores Internalizing problems. At Times 1 through 4,
on these measures decrease systematic error var- adolescents completed the emotional tone subscale
iance and increase validity (Schwarz et al., 1985). The from the Self-Image Questionnaire for Young Ado-
alphas for the combined scales were high (.93, .81, lescents (Petersen, Schulenberg, Abramowitz, Offer,
and .87 for support, behavioral control, and psycho- & Jarcho, 1984). This subscale measures depressive
 Parents Do Matter 583

and anxious affect, with 11 items (e.g., ‘‘I frequently supportive used less psychological control. Higher
feel sad’’; ‘‘I am so very nervous’’), rated on a 6-point behavioral control was associated consistently with
scale ranging from 1 (does not describe me at all) to 6 lower levels of internalizing problems. Adolescents
(describes me very well). Mean scale scores were who reported more externalizing and internalizing
calculated so that higher scores indicated higher problems tended to associate with peers who were
levels of negative affect. Cronbach’s alpha across more deviant. Stabilities for externalizing problems
the four times of measurement ranged from .79 were generally moderately high across the five
to .86. Scores on this subscale were related signifi- occasions. Internalizing problems also showed mod-
cantly to other indices of internalizing problems erate stability across time.
such as depression, anxiety, and fears (Graber,
Brooks-Gunn, Paikoff, & Warren, 1994; Petersen Effects of Parenting on Trajectories of Externalizing and
et al., 1984). Internalizing Problems
The main data analysis used HLM (Bryk &
Results Raudenbusch, 1992; Mason, Wong, & Entwistle,
Descriptive Statistics and Correlations 1984), a method that allows simultaneous estimation
of both (a) a separate within-person model of
Table 1 presents means, standard deviations, and regression intercepts and slopes (i.e., growth trajec-
intercorrelations for the measures of externalizing tories) for each respondent, and (b) a between-
and internalizing problems, parenting behaviors, person model in which the within-person slopes and
and association with deviant peers. On average, intercepts are treated as dependent variables re-
adolescents reported relatively low levels of exter- gressed on person-level predictor variables. It is
nalizing problems and internalizing problems, and important to point out that this estimation procedure
their peers were not highly involved in deviant takes into consideration the amount of data available
behavior. Parents, on average, reported moderate to from each person, so that more weight is given to
high levels of support and behavioral control and persons with complete data than those with some
low to moderate levels of psychological control. missing data (Dempster, Laird, & Rubin, 1977). Thus,
Correlations show that parents who were more instead of deleting all of the respondent’s data

Table 1
Descriptive Statistics (M, SD, n) for and Pearson Correlations Among Parenting Behaviors, Externalizing and Internalizing Problems, and Deviant
Peers

Measure M SD n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

1. SPa 3.74 .38 109

2. BCa 3.92 .31 109 07
3. PCa 2.17 .42 109  29n  13
4. EPa (T1) 1.18 .19 107  11  14 13
5. EP (T2) 1.16 .19 101  12 03 31n 55n
6. EP (T3) 1.20 .23 91 01 02 18 62n 59n
7. EP (T4) 1.29 .33 77  05  35n 08 56n 36n 27n
8. EP (T5) 1.39 .41 75  01  17 15 37n 48n 22 69n
9. IPb (T1) 2.45 .84 108  07  29n 17 24n 02 06 07  15
10. IP (T2) 2.31 .75 103  16  19 08 31n 12 10 22  06 57n
11. IP (T3) 2.42 .89 90  18  26n 24n 11  06  01 30n 18 53n 44n
12. IP (T4) 2.42 .86 74  21  33n 18 14 06  12 32n 14 47n 43n 67n
13. DPc (T1) 1.74 .65 109  00  16 14 43n 22n 28n 26n 14 29n 22n 29n 18
14. DP (T2) 1.66 .64 103  06  28n 15 41n 34n 33n 29n 23n 39n 33n 33n 21 66n
15. DP (T3) 1.71 .65 91  09  20 08 40n 33n 51n 35n 20 36n 15 38n 08 46n 48n
16. DP (T4) 1.86 .68 76  13  27n 01 44n 42n 19 66n 47n 16 29n 20 36n 34n 35n 44n
17. DP (T5) 2.22 .71 74 08  10  18 32n 22 11 50n 54n 03  01 06 07 08 15 18 39n

Note. SP 5 support; BC 5 behavioral control; PC 5 psychological control; EP 5 externalizing problems; IP 5 internalizing problems;
DP 5 deviant peers; T 5 time. Decimal points are omitted from correlations.
a
Possible range: 1 to 5. bPossible range: 1 to 6. cPossible range: 1 to 4.
n
po.05.
584 Galambos, Barker, and Almeida

because of a missed questionnaire, this approach has Level 2: b0i ¼p00 þ p01 ðSUPPORTÞ
the advantage of using all available data from a
þ p02 ðBEHAVIORAL CONTROLÞ
given respondent.
For illustrative purposes, the simple form of an þ p03 ðPSYCHOLOGICAL CONTROLÞ
HLM can be conceived of as two separate models,
one a within-person model (Level 1) and the other a þ U0i ð5Þ
between-person model (Level 2). To begin, we fit a
within-person model that assesses the individual b1i ¼p10 þ p11 ðSUPPORTÞ
trajectories of adolescent outcomes across the five
þ p21 ðBEHAVIORAL CONTROLÞ
occasions of measurement. This model can be
expressed as: þ p31 ðPSYCHOLOGICAL CONTROLÞ þ U0i :
Level 1: ADOLESCENT PROBLEMSti ð6Þ
¼ b0i þ b1i ðTimeti Þ þ R ti ; ð1Þ The models in Equations 5 and 6 tested whether the
where ADOLESCENT PROBLEMSti is the reported initial level (B0i) and rate of change in adolescent
problem (i.e., externalizing or internalizing) of behavior (B1i) varied according to parents’ support,
Adolescenti on Timet, Timeti indicates the number behavioral control and psychological control.
of years from the beginning of the study, B0i is the The initial HLM analyses explored a set of three
intercept indicating Adolescenti’s problem score at models for externalizing problems and internalizing
the initial time of measurement, B1i is the slope problems, respectively. All variables were grand-
indicating the linear rate of change in problems of mean centered to reduce cross-level confounding
Adolescenti, and Rti is the random component or (Jaccard, Turrisi, & Wan, 1990). Model 1 examined
error associated with each adolescent’s trajectory. To the average within-person rate of change in externa-
estimate average effects for the entire sample, the lizing problems and internalizing problems across
intercepts and slopes of the Level 1 within-person time (five waves for externalizing problems and four
model become the outcomes for the Level 2 between- waves for internalizing problems). Model 2 tested
person equations, as follows: the between-person differences in mean levels of
externalizing and internalizing problems in Grade 6
Level 2: b0i ¼ p00 þ U0i ð2Þ and the average rate of change in these problems as
b1i ¼ p10 þ U0i : ð3Þ a function of parenting behaviors. In Model 3 the
three 2-way interactions among the three parenting
Equation 2 shows that Adolescenti’s average pro- behaviors were added as predictors of Grade 6 levels
blem score in Grade 6 (B0i) is a function of the and rates of change in externalizing and internaliz-
intercept for the entire sampleFthe grand mean of ing problems.
the sampleFand a random component or error The results for externalizing problems are pre-
(U0i). Likewise, Equation 3 shows that Adolescenti’s sented in Table 2. In Model 1 there was a significant
time slope (B1i) is a function of the grand mean of the time effect. Adolescents increased their externalizing
entire sample and a random component or error problems across the five waves of measurement. In
(U0i). examining between-person predictors in Model 2,
The next step in the analyses assessed the extent none of the parenting variables were related to
to which initial levels and slopes are a function of externalizing problems in Grade 6. Across time,
parenting behaviors at the initial time of measure- however, behavioral control was related to externa-
ment. A distinctive feature of HLM is that the lizing problems. Adolescents whose parents used
intercepts and slopes are allowed to vary across more behavioral control increased their engagement
persons (Lee & Bryk, 1989). Therefore, between- in externalizing problems at a slower rate, relative to
person models of within-person trajectories can be adolescents whose parents were less behaviorally
formulated. controlling. In Model 3, there was a significant main
To examine whether parenting behaviors are effect of psychological control on initial levels of
associated with adolescent trajectories of externaliz- externalizing problems. Adolescents whose parents
ing problems and internalizing problems, we fit the used more psychological control reported more
following model externalizing problems. However, this finding
Level 1: ADOLESCENT PROBLEMSti was qualified by the significant interaction
between psychological control and behavioral
¼ b0i þ b1i ðTimeti Þ þ R ti ð4Þ control on initial levels of externalizing problems.
 Parents Do Matter 585

Table 2
Multilevel Results of the Effects of Parenting Behaviors on Initial Level and Change in Externalizing Problems

Model 1 Model 2 Model 3

Variable Coefficient (SE) Coefficient (SE) Coefficient (SE)

EP in Grade 6 1.16 (.02)n 1.15 (.02)n 1.16 (.02)n

Support (SP)  .02 (.05)  .01 (.05)
Behavioral control (BC) .02 (.07)  .00 (.06)
Psychological control (PC) .08 (.06) .10 (.05)n
SP
BC .19 (.14)
SP
PC  .06 (.12)
BC
PC .35 (.16)n
Rate of change in EP .06 (.01)n .06 (.01)n .06 (.01)n
Support (SP) .02 (.02) .02 (.03)
Behavioral control (BC)  .10 (.04)n  .09 (.04)n
Psychological control (PC) .03 (.04) .02 (.04)
SP
BC  .02 (.08)
SP
PC  .02 (.07)
BC
PC  .11 (.08)
Note. EP 5 externalizing problems. Level 1 N 5 449. Level 2 N 5 109.
n
po.05.

Follow-up analyses showed that the effect of higher of the relative and unique influence of three
psychological control on higher initial levels of parenting behaviors on trajectories of externalizing
externalizing problems was limited to families in problems confirmed the importance of behavioral
which the parents were also high on behavioral control.
control, B 5 .10, SE 5 .05, po.05. For families in A different picture emerged for trajectories of
which parents were low on behavioral control, there internalizing problems (see Table 3). The results of
was no relationship between psychological control Model 1 revealed that, on average, adolescents’
and externalizing problems, B 5 .01, SE 5 .07, p4.05. internalizing problems did not change across time.
Given that a goal of this study was to examine the Model 2 showed that parents’ behavioral control in
relative and unique influence of support, behavioral Grade 6 was significantly and negatively associated
control, and psychological control on adolescent with Grade 6 levels of internalizing problems.
adjustment, we calculated the proportion of be- Adolescents whose parents used more behavioral
tween-person variance in the time slopes accounted control had lower levels of internalizing problems.
for by each of the three parenting predictors, None of the main effects of parenting behaviors were
controlling for the other two parenting behaviors. related to internalizing problems across time. Never-
To do so, we compared tau (the between-person theless, we examined the relative and unique
variance in the slopes) for Model 2 with tau in three influence of the three parenting behaviors on
models. In each of the three models, one parenting between-person variance in the time slopes for
behavior was removed (reduced model). The pro- internalizing problems (comparing tau for Model 2
portion of between-person variance in the time in Table 3 with tau in reduced models in which the
slopes accounted for by a specific parenting behavior parenting behavior in question was removed). The
was calculated as the difference between tau for results were that support explained 7% of the
Model 2 and tau for the reduced model, divided by between-person variance in trajectories in internaliz-
tau for the reduced model (Nezlek, 2001). The results ing problems, whereas behavioral control and
were that behavioral control explained 7.5% of the psychological control explained 2.7% and .6% of
between-person variance in the time slopes, control- the variance, respectively.
ling for support and psychological control. Support We estimated a final model testing the curvilinear
and psychological control, however, each explained effects of behavioral control on initial levels and the
less than 1% of the variance in trajectories for trajectories of externalizing and internalizing pro-
externalizing problems, controlling for the other blems. We squared the behavioral control variable
parenting behaviors. Thus, this conservative test and added it to Model 2. This quadratic term was
586 Galambos, Barker, and Almeida

Table 3
Multilevel Results of the Effects of Parenting Behaviors on Initial Level and Change in Internalizing Problems

Model 1 Model 2 Model 3

Variable Coefficient (SE) Coefficient (SE) Coefficient SE)

IP in Grade 6 2.39 (.07)n 2.39 (.07)n 2.41 (.07)n

Support (SP)  .14 (.18)  .13 (.19)
Behavioral control (BC)  .60 (.19)n  .62 (.19)n
Psychological control (PC) .22 (.16) .24 (.16)
SP
BC  .34 (.44)
SP
PC .31 (.38)
SP
PC .33 (.42)
Rate of change in IP  .01 (.04)  .00 (.04)  .02 (.04)
Support (SP)  .04 (.11)  .07 (.11)
Behavioral control (BC)  .06 (.11)  .02 (.10)
Psychological control (PC) .05 (.12) .02 (.10)
SP
BC .38 (.24)
SP
PC  .19 (.25)
BC
PC  .41 (.26)
Note. IP 5 internalizing problems. Level 1 N 5 374. Level 2 N 5 109.
n
po.05.

not significant for either externalizing or internaliz- ioral control: (a) high peer deviance/low behavioral
ing problems. control, (b) high peer deviance/high behavioral
control, (c) low peer deviance/low behavioral con-
trol, and (d) low peer deviance/high behavioral
The Relative Effects of Parenting and Deviant Peers on
control. The analyses examined for each of these four
Trajectories of Externalizing Problems
groups the rate of change in externalizing problems
The next set of analyses explored the influence of across time. Figure 1, based on coefficients generated
peers relative to parents for externalizing problems. in these models, depicts the nature of the interaction.
All variables were grand-mean centered. Three Externalizing problems increased significantly in the
models were tested. Results for externalizing pro- high peer deviance/low behavioral control group
blems are presented in Table 4. Model 1 examined (B 5 .09, SE 5 .04, po.05), the low peer deviance/
between-person differences in initial levels and rates low behavioral control group (B 5 .07, SE 5 .03,
of change in externalizing problems as a function of po.05), and the low peer deviance/high behavioral
the three parenting behaviors and deviant peers at control group, (B 5 .12, SE 5 .03, po.05). However,
Grade 6. Deviant peers in Grade 6 were found to be there was no significant increase in the high peer
positively related to externalizing problems in Grade deviance/high behavioral control group (B 5 .02,
6, but not to rate of change in externalizing problems SE 5 .03, p4.05).
across time. Behavioral control was associated To explore this interaction further, contrasts were
negatively with rate of change in externalizing created that compared the high peer deviance/high
problems such that the increase in externalizing behavioral control group with each of the other three
was more rapid in adolescents with parents report- groups (a total of three contrasts). An HLM analysis
ing less behavioral control. was then conducted with the three contrasts entered
In Model 2, the three 2-way interactions between as predictors of both initial levels and rates of
deviant peers and each of the three parenting change in externalizing problems. With respect to
behaviors were added. New in this analysis was a initial levels, the results demonstrated that externa-
significant interaction between deviant peers and lizing problems in Grade 6 were significantly higher
behavioral control predicting rate of change in in the high peer deviance/high behavioral control
externalizing problems. The nature of this interac- group than in the two groups with low peer
tion was probed by conducting separate HLM deviance (when behavioral control was low,
analyses for each of four groups created by median B 5  .16, SE 5 .04, po.05, and when behavioral
splits on the variables of deviant peers and behav- control was high, B 5  .20, SE 5 .04, po.05).
 Parents Do Matter 587

Table 4
Multilevel Results of the Effects of Parenting and Deviant Peers in Grade 6, and Across-Time Deviant Peers
on Initial Level and Change in Externalizing Problems

Model 1 Model 2 Model 3

Variable Coefficient (SE) Coefficient (SE) Coefficient (SE)

Externalizing problems in Grade 6 1.16 (.02)n 1.16 (.02)n .91 (.03)n

Support (SP)  .02 (.04)  .02 (.05)  .04 (.03)
Behavioral control (BC) .04 (.06) .05 (.06) .12 (.04)n
Psychological control (PC) .06 (.05) .06 (.05) .02 (.04)
Deviant peers (DP) in Grade 6 .10 (.03)n .10 (.03)n
SP
DP  .02 (.06)
BC
DP  .01 (.10)
PC
DP  .02 (.09)
Rate of change in externalizing .06 (.01)n .06 (.01)n .03 (.01)n
Support (SP) .02 (.02) .03 (.02) .03 (.02)
Behavioral control (BC)  .10 (.04)n  .10 (.03)n  .08 (.03)n
Psychological control (PC) .03 (.04) .04 (.04) .03 (.03)
Deviant peers (DP) Grade 6  .01 (.02)  .02 (.02)
DP
SP .02 (.04)
DP
BC  .12 (.04)n
DP
PC .02 (.03)

DP as a time-varying covariate .15 (.02)n

Note. Level 1 N 5 449. Level 2 N 5 109.
n
po.05.

1.6 However, the third contrast was not significant,

indicating that Grade 6 externalizing problems were
similar in the two groups of adolescents with high
1.5 peer deviance (B 5  .04, SE 5 .05, p4.05).
With respect to rates of change in externalizing
Externalizing Problems

B=.09*

1.4 problems, the contrasts showed that the slope for

externalizing problems in the high peer deviance/
high behavioral control group was significantly
1.3 different (i.e., flatter) from the slopes for the other
B=.02
three groups: high peer deviance/low behavioral
1.2
control (B 5 .13, SE 5 .03, po.05), low peer de-
B=.07* viance/low behavioral control (B 5 .09, SE 5 .03,
po.05), and low peer deviance/high behavioral
1.1 control (B 5 .09, SE 5 .03, po.05). In short, although
B=.12* adolescents with deviant peers start out at a higher
level on externalizing problems, and there is a
1
general increase in externalizing problems across
1 5
Time
time, in only one group was this increase halted: the
group of adolescents whose parents exerted higher
High Peer Deviance/Low Behavioral Control
levels of behavioral control in the face of their
High Peer Deviance/High Behavioral Control
Low Peer Deviance/Low Behavioral Control
adolescents’ higher association with deviant peers.
Low Peer Deviance/High Behavioral Control In Model 3 of Table 4, deviant peers was included
at Level 1 as a time-varying covariate. This allowed
Figure 1. Trajectories of externalizing problems across five times of
measurement as a function of Grade 6 levels of parental us to test whether externalizing problems are higher
behavioral control and adolescents’ association with deviant when adolescents’ peers are more deviant. This also
peers. allowed us to control for the influence of deviant
588 Galambos, Barker, and Almeida

peers across all five waves of measurement. Deviant was controlled. Association with more deviant peers
peer association was related positively to more was positively related to internalizing problems
externalizing problems. More important though, across time. After controlling for deviant peers at
the association between behavioral control and rate all four waves of measurement, the associations
of change in externalizing problems remained between behavioral control and Grade 6 internaliz-
significant even when controlling for the effect of ing problems remained significant.
deviant peers over time.

The Relative Effects of Parenting and Deviant Peers on Follow-Up Analyses

Trajectories of Internalizing Problems
To examine whether controlling for gender would
Table 5 presents the results for internalizing affect the results, all models in Tables 2 through 5
problems, which followed the analytic plan for were recomputed, with gender entered as a covari-
examining the relative effects of parenting and ate at Level 2. The inclusion of gender did not affect
deviant peers on externalizing problems. Model 1 the pattern of results in any substantive manner.
examined the main effects of the three parenting There was a main effect of gender on initial levels
behaviors and deviant peers, showing a significant and rates of change in externalizing (but not
negative association between behavioral control and internalizing) problems. Boys demonstrated higher
Grade 6 levels of internalizing problems, as well as a levels of externalizing problems in Grade 6 than did
positive relation between deviant peers and initial girls, but girls accelerated faster over time compared
levels of internalizing problems. In Model 2 interac- with boys.
tions between deviant peers at Grade 6 and the three Given that the externalizing and internalizing
parenting behaviors were added. None of the problem scores were skewed, these scores were
interactions were significant. transformed using the natural logarithm. All anal-
In Model 3, deviant peers was added to Model 1 yses in Tables 2 through 5 were conducted again on
at Level 1 as a time-varying covariate. That is, the the transformed scores. The results were virtually
effect of deviant peers at each wave of measurement identical.

Table 5
Multilevel Results of the Effects of Parenting and Deviant Peers in Grade 6, and Across-Time Deviant Peers
on Initial Level and Change in Internalizing Problems

Model 1 Model 2 Model 3

Variable Coefficient (SE) Coefficient (SE) Coefficient (SE)

Internalizing problems in Grade 6 2.39 (.06)n 2.40 (.07)n 1.93 (.11)n

Support (SP)  .16 (.17)  .16 (.17)  .11 (.16)
Behavioral control (BC)  .51 (.17)n  .49 (.17)n  .49 (.17)n
Psychological control (PC) .17 (.16) .18 (.16) .18 (.15)
Deviant peers (DP) in Grade 6 .29 (.11)n .28 (.11)n
SP
DP  .16 (.26)
BC
DP  .18 (.23)
PC
DP  .30 (.24)
Rate of change in internalizing  .00 (.04)  .01 (.04)  .02 (.04)
Support (SP)  .04 (.11)  .06 (.13)  .04 (.11)
Behavioral control (BC)  .08 (.11)  .10 (.11)  .03 (.11)
Psychological control (PC) .06 (.12) .05 (.13) .06 (.12)
Deviant peers (DP) Grade 6  .06 (.06)  .08 (.06)
DP
SP  .18 (.24)
DP
BC  .06 (.13)
DP
PC .07 (.16)

DP as a time-varying covariate .27 (.06)n

Note. Level 1 N 5 374. Level 2 N 5 109.
n
po.05.
 Parents Do Matter 589

We also tested for a quadratic slope in the parenting and deviant peers on adolescents’ pro-
trajectory of externalizing problems. The quadratic blems; and (c) curvilinear effects of behavioral
effect was significant, because the first three times control in predicting adolescents’ initial problems
of measurement were 6 months apart in early and problem trajectories.
adolescence, after which the line accelerated. Which parenting behaviors predicted adolescents’
This effect was not included in the main analyses initial (Grade 6) externalizing problems? Parents’
because power at Levels 1 and 2 was too limited to higher levels of psychological control were related to
detect significance in the predictors of the quadratic adolescents’ higher levels of externalizing problems
trend. but only when parents also reported higher behav-
Finally, the interaction presented in Figure 1 was ioral control. This is an intriguing interaction,
puzzling. It was not surprising that externalizing especially considering that psychological and behav-
problems remained stable in adolescents with ioral control are typically inversely related (Barber,
deviant peers whose parents exerted high behavioral 1996). One possible explanation for this finding
control. It was surprising, however, that externaliz- concerns child effects on parenting behaviors. Long-
ing problems increased in adolescents with low peer itudinal studies show reductions in positive parent-
deviance and parents’ high behavioral control. How ing behaviors and increases in negative control over
can the difference between these two groups be time in response to children who demonstrate
explained? We speculated that high behavioral behavior problems (Pettit et al., 2001; Scaramella
control by parents of adolescents with deviant peers et al., 2002; Spieker, Larson, Lewis, Keller, &
might be more geared to monitoring their adoles- Gilchrist, 1999; Stice & Barrera, 1995). The coupling
cents’ friendships compared with parents who show of high psychological and high behavioral control
high behavioral control in the relative absence of may indicate that a coercive situation has developed
deviant peers. An examination of the items on the in which parents react to a misbehaving child by
behavioral control scale reveals that this scale may resorting to all available means of control.
not capture the specific behavior of monitoring Alternatively, based on findings that psychologi-
adolescents’ friendships. In Grade 6, however, cal control in adolescence may be anteceded by
adolescents completed an eight-item scale from the proactive parenting in childhood, Pettit et al. (2001)
CRPBI (Schaefer, 1965), which was meant to measure speculated that in an effort to prevent the occurrence
intrusiveness, but was included in our study as a of behavior problems, some parents engage in
monitoring measure (e.g., ‘‘My parents keep a ‘‘overmanagement.’’ The link between adolescents’
careful check on me to make sure that I have the externalizing problems and the combination of
right kind of friends’’; ‘‘My parents want to know parents’ high psychological and high behavioral
with whom I’ve been when I’ve been out’’). control might indicate that these parents are over-
Adolescents indicated the extent to which they managing their young adolescents in a way that is
agreed with each item on a scale from 1 (disagree developmentally inappropriate and intrusive. Par-
strongly) to 4 (agree strongly; a 5 .81). A nonsignifi- ents who exert control over their adolescents’
cant correlation between behavioral control and psychological and behavioral lives, whether these
monitoring confirmed that the two measures were parenting behaviors are antedated by child misbe-
orthogonal in this study, r 5.00, p4.05. Furthermore, havior or by good intentions, might be issuing a
a t test revealed significantly stronger perceptions of ‘‘double whammy’’ from which the adolescent
parent monitoring among adolescents in the high escapes by engaging in externalizing behaviors.
peer deviance/high behavioral control group Thus, behavioral control, which is generally under-
(M 5 2.54, SD 5 .60) compared with adolescents in stood to be a positive aspect of parenting, may not be
the low peer deviance/high behavioral control uniformly effective when it is combined with other
group, M 5 2.20, SD 5 .49, t(51) 5 2.30, po.05. less desirable parenting behaviors. Given that there
was no interaction between behavioral and psycho-
logical control in the prediction of trajectories of
Discussion
externalizing problems, it is difficult to know
This study examined (a) the relative influence of whether there are long-term implications of this
parental support, behavioral control, and psycholo- combination of parenting behaviors. Naturally, it
gical control and their interactions on adolescents’ will be important in future research to replicate the
initial levels of and trajectories of change from early interaction between psychological and behavioral
to middle adolescence in externalizing and inter- control and, if confirmed, to uncover in longitudinal
nalizing problems; (b) the relative influence of research its unfolding effects on adolescents.
590 Galambos, Barker, and Almeida

Behavioral control emerged as the sole significant viance. If adolescents’ close friends are norm
parenting behavior related to trajectories of change abiding, parents might not be aware that their
in externalizing problems. The typical trajectory in adolescents are at risk of engaging in problem
early adolescence toward higher levels of externaliz- behavior and might fail to exert the kind of
ing behaviors (Moffitt, 1993; Simons et al., 2001) behavioral control necessary to reduce that like-
seemed to be slowed down for adolescents whose lihood. Indeed, adolescents whose peers were less
parents reported firm behavioral control (and deviant in Grade 6 and whose parents reported high
accelerated for adolescents whose parents did not). behavioral control reported their parents as engag-
These results converge with earlier research showing ing in less monitoring compared with adolescents
that, on the whole, parents’ firm discipline and limit- whose peers were more deviant in Grade 6 and
setting behaviors may be key in preventing or whose parents reported high behavioral control.
reducing adolescents’ involvement in risky, aggres- Thus, the group of parents whose adolescents were
sive, or norm-violating activities (e.g., Barber et al., most at risk (those with more deviant peers) and
1994; Mason et al., 1996; Miller et al., 1986). who engaged in high behavioral control and high
How did parents’ behavioral control operate in monitoring were the parents who seem to have been
the context of deviant peers in explaining trajectories effective in preventing the escalation of externalizing
of change in adolescents’ externalizing problems? problems in their adolescents. The presence of
Although deviant peers and externalizing problems deviant peers may serve as an important cue to
were positively and significantly associated in Grade parents. Research shows that parents who are
6, parents’ behavioral control (and not deviant peers) concerned that their adolescents’ friends are a
predicted the rate of change in externalizing pro- negative influence make more active attempts to
blems across time. Moreover, even when the manage their adolescents’ friendships (e.g., by
significant relations between deviant peers and communicating disapproval) than do parents who
externalizing problems at each time of measurement are unconcerned about their adolescents’ friends
were controlled (as a time-varying covariate), par- (Tilton-Weaver & Galambos, 2003).
ents’ behavioral control in Grade 6 still seemed to How was parenting related to adolescents’ inter-
make a difference in the rate of change in externaliz- nalizing problems? Parents’ behavioral control was a
ing behavior. Finally, an interaction between behav- significant predictor of initial levels of internalizing
ioral control and deviant peers suggested that the problems, with adolescents whose parents reported
upward trajectory in externalizing problems among firm behavioral control showing lower levels of
adolescents with deviant peers appeared to be internalizing problems in Grade 6. Although psy-
halted if parents were higher on behavioral control. chological control is generally more strongly linked
These results confirm that having deviant peers is a to internalizing symptoms than is behavioral control,
risk factor for engaging in externalizing behaviors, associations between behavioral control and inter-
but parents may play a critical role in diminishing nalizing symptoms have been observed (e.g., Barber,
that risk. 1996; Kurdek & Fine, 1994). With respect to
It is interesting that adolescents with less deviant trajectories of change in internalizing problems, on
peers also showed increases in externalizing pro- average, adolescents’ internalizing problems neither
blems. It may be that these adolescents were increased nor decreased across time, and no parent-
engaged in adolescence-limited problem behavior ing behaviors attained significance in predicting
(behaviors such as drinking and shoplifting that fade individual trajectories. It is interesting that calcula-
by the end of adolescence; Moffitt, 1993). According tions for the amount of between-person variance in
to Moffitt’s (1993) maturity gap theory, the reason trajectories of internalizing explained by the three
some adolescents engage in adolescence-limited predictors showed that parental support accounted
problem behaviors is that they hope to attain the for the largest proportion (7%), followed by behav-
rewards associated with mature status (material ioral control (2.7%). Perhaps in a sample with more
goods, being seen to be mature by their peers) that change in internalizing problems, the effect of
they observe being obtained by antisocial teens. parental support would be significant.
Thus, adolescents with less deviant peers might The lack of significance of psychological control
mimic the activities of the more antisocial peers with for internalizing problems may be a measurement
whom they are not close friends. issue. Although the psychological control measure is
A puzzling issue is why parents’ high behavioral reliable and valid (Schwarz et al., 1985), Barber
control did not prevent an increase in externalizing (1996) developed a new scale to capture the proper-
problems among adolescents with low peer de- ties of psychological control better. It is also possible
 Parents Do Matter 591

that the scale used to measure internalizing pro- 1997)Findicators of positive adjustment that were
blems was too narrow in focus to detect the effects of not included in the present study. Moreover, the
psychological control. In light of these considera- finding that support paled in comparison with
tions and the body of research linking psychological behavioral control in the prediction of externalizing
control with internalizing symptoms (Barber et al., problems is consistent with some other research
1994; Gray & Steinberg, 1999), psychological control (Herman et al., 1997). Support may contribute to
should be studied further. enhanced psychological, social, and academic well-
It is noteworthy that deviant peer association was being, but when it comes to limiting troublesome
positively linked with internalizing problems in behavior, behavioral control may be most effective.
Grade 6, as well as at other times of measurement We expected parental support to be related to initial
as shown through concurrent correlations. Although levels and trajectories of internalizing problems. It
the relationship between deviant peers and externa- was not significantly related to either. It is interest-
lizing problems is well established and often ing, however, that support did explain the largest
studied, these results suggest that it might be portion of between-person variability in trajectories
worthwhile to engage in efforts to understand better of internalizing problems. Had there been more
the link between deviant peers and internalizing change over time in internalizing problems and had
problems. Indeed, Brendgen et al. (2000) found that the measurement of internalizing problems ex-
adolescents with deviant friends were more de- tended to age 15 rather than age 14 (as it did for
pressed than adolescents with nondeviant friends. externalizing problems), support might have
This result was attributed to several possibilities. emerged as significant.
First, adolescents who associate with deviant peers Although one goal of this study was to examine
may have similar backgrounds, attitudes, and the unique contributions of three parenting behav-
experiences that predispose them to depression iors to adolescent problems, another goal was to
(Elliott, Huizinga, & Ageton, 1985; Kandel, 1978). examine their interaction. Only 1 of 12 tests for two-
Second, the friendships of adolescents with deviant way interactions among the three parenting behav-
peers may be less supportive, more conflictual, and iors was significant (the Behavioral Control

more coercive than friendships among nondeviant Psychological Control interaction for initial levels
adolescents (Marcus, 1996). Third, friends who show of externalizing problems). This result is not surpris-
deviant behavior might mutually reinforce and ing given that there is little evidence for consistent
encourage negative thinking about their lives interactions in the literature. Likewise, no evidence
(Brendgen et al., 2000). Thus, adolescents with was found for a curvilinear effect of behavioral
deviant friends may bring some vulnerabilities to control on initial levels and trajectories of change in
their friendships that could lead to or amplify externalizing and internalizing problems. Whether
internalizing symptoms. this lack of curvilinearity is due to differences among
Overall, how did the three parenting dimensions samples (this is the only such study involving two-
compare as unique predictors of adolescents’ ex- earner families), different measures (other studies
ternalizing and internalizing problems? Behavioral have typically measured behavioral control with
control stood out as a significant predictor of change different scales), or the genuine lack of deviations
in externalizing problems from early to middle from linearity, we do not know. We believe, however,
adolescence, even overriding the known strong that it is important to examine curvilinearity in
relationship between deviant peers and externaliz- future research.
ing problems. Psychological control, on the other This study should be considered in light of its
hand, was expected to have a significant relationship limitations. The small and selective nature of the
with internalizing problems, but it did not, as sampleFpredominately White adolescents in two-
described previously. It is surprising that parental earner familiesFlimits the extent to which the
support was not a significant predictor of externaliz- results may be generalized. Indeed, these parents
ing or internalizing problems. and their adolescents may function better on average
The absence of parental support effects might be than those in a more representative sample. More-
attributable to the choice of dependent measures. over, whereas some parenting behaviors, particu-
That is, support has been linked most consistently larly behavioral control, emerged as significant,
with measures of adolescents’ academic achieve- there is much unexplained variability in initial levels
ment as well as psychosocial well-being and and trajectories of adolescents’ problems. Parenting
competence (e.g., Barber et al., 2002; Eccles et al., can contribute to important adolescent develop-
1997; Gray & Steinberg, 1999; Herman et al., ments, but it is important to consider other processes
592 Galambos, Barker, and Almeida

as well, including genetic relationships; the individ- tions, and self-reported behavior among adolescents.
ual characteristics and personalities of parents and Developmental Psychology, 22, 521–530.
adolescents; and attributes of the peer, family, Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear
neighborhood, and community contexts. models: Applications and data analysis. Newbury Park, CA:
In general, the results of this study call for the need Sage.
Burger, G. K., & Armentrout, J. A. (1971). Comparative
to consider a broad range of adolescent outcomes so
study of methods to estimate factor scores for reports of
that the specificity of parenting effects becomes clear. parental behaviors. Proceedings, 79th Annual Convention,
Future research could benefit from multiwave long- American Psychological Association, 6, 149–150.
itudinal investigations that follow parents and Collins, W. A., Maccoby, E. E., Steinberg, L., Hetherington,
children across childhood and adolescence to deter- E. M., & Bornstein, M. H. (2000). Contemporary research
mine not only how parenting contributes to changes on parenting: The case for nature and nurture. American
in adolescents’ adjustment, but how earlier child Psychologist, 55, 218–232.
behavior shapes parenting. Parents do matter, but it is Costigan, C. L. (1996). Parental values related to control in
important to explore how, when, and in what families with children with special needs. Unpublished
contexts their behaviors have the most impact. doctoral dissertation, Michigan State University.
Dempster, A. P., Laird, N. M., & Rubin, D. B. (1977).
Maximum likelihood from incomplete data via the EM
algorithm. Journal of the Royal Statistical Society, Series B,
39, 1–8.
References
Eccles, J. S., Early, D., Frasier, K., Belansky, E., & McCarthy,
Barber, B. K. (1996). Parental psychological control: K. (1997). The relation of connection, regulation, and
Revisiting a neglected construct. Child Development, 67, support for autonomy to adolescents’ functioning.
3296–3319. Journal of Adolescent Research, 12, 263–286.
Barber, B. K. (1997). Adolescent socialization in con- Elliott, D. S., Huizinga, D., & Ageton, S. S. (1985).
textFThe role of connection, regulation, and autonomy Explaining delinquent behavior and drug use. Beverly Hills,
in the family. Journal of Adolescent Research, 12, 5–11. CA: Sage.
Barber, B. K., Maughan, S. L., Olsen, J. A., & Thomas, D. L. Fauber, R., Forehand, R., Thomas, A. M., & Wierson, M.
(2002, April). Parental support, psychological control, and (1990). A mediational model of the impact of marital
behavioral control: Assessing the nature of effects. Paper conflict on adolescent adjustment in intact and divorced
presented at the Society for Research on Adolescence families: The role of disrupted parenting. Child Develop-
Meeting, New Orleans, LA. ment, 61, 1112–1123.
Barber, B. K., & Olsen, J. A. (1997). Socialization in context: Forehand, R., & Nousiainen, S. (1993). Maternal and
Connection, regulation, and autonomy in the family, paternal parenting: Critical dimensions in adolescent
school, and neighborhood, and with peers. Journal of functioning. Journal of Family Psychology, 7, 213–221.
Adolescent Research, 12, 298–315. Galambos, N. L., & Ehrenberg, M. F. (1997). The family as
Barber, B. K., Olsen, J. E., & Shagle, S. C. (1994). health risk and opportunity: A focus on divorce and
Associations between parental psychological and beha- working families. In J. Schulenberg, J. L. Maggs, & K.
vioral control and youth internalized and externalized Hurrelmann (Eds.), Health risks and developmental transi-
behaviors. Child Development, 65, 1120–1136. tions during adolescence (pp. 139–160). Cambridge,
Barnes, G. M., Reifman, A. S., Farrell, M. P., & England: Cambridge University Press.
Dintcheff, B. A. (2000). The effects of parenting on the Galambos, N. L., & Maggs, J. L. (1991). Out-of-school care
development of adolescent alcohol misuse: A six-wave of young adolescents and self-reported behavior. Devel-
latent growth model. Journal of Marriage and the Family, opmental Psychology, 27, 644–655.
62, 175–186. Galambos, N. L., Sears, H. A., Almeida, D. M., & Kolaric,
Baumrind, D. (1991). Effective parenting during the G. C. (1995). Parents’ work overload and problem
adolescent transition. In P. A. Cowan & M. Hetherington behavior in young adolescents. Journal of Research on
(Eds.), Family transitions (111–163). Hillsdale, NJ: Erl- Adolescence, 5, 201–223.
baum. Garber, J., Robinson, N. S., & Valentiner, D. (1997). The
Blishen, B. R., & McRoberts, H. A. (1976). A revised socio- relation between parenting and adolescent depression:
economic index for occupations in Canada. Canadian Self-worth as a mediator. Journal of Adolescent Research,
Review of Sociology and Anthropology, 13, 71–79. 12, 12–33.
Brendgen, M., Vitaro, F., & Bukowski, W. M. (2000). Graber, J. A., Brooks-Gunn, J., Paikoff, R. L., & Warren, M.
Deviant friends and early adolescents’ emotional and P. (1994). Prediction of eating problems: An 8-year study
behavioral adjustment. Journal of Research on Adolescence, of adolescent girls. Developmental Psychology, 30, 823–834.
10, 173–189. Gray, M. R., & Steinberg, L. (1999). Unpacking author-
Brown, B. B., Clasen, D. R., & Eicher, S. A. (1986). itative parenting: Reassessing a multidimensional con-
Perceptions of peer pressure, peer conformity disposi- struct. Journal of Marriage and the Family, 61, 574–587.
 Parents Do Matter 593

Harris, J. R. (1995). Where is the child’s environment? A relation to adolescent sexual attitudes and behavior.
group socialization theory of development. Psychological Journal of Marriage and the Family, 48, 503–512.
Review, 102, 458–489. Moffitt, T. E. (1993). Adolescence-limited and life-course-
Herman, M. R., Dornbusch, S. M., Herron, M. C., & persistent antisocial behavior: A developmental taxon-
Herting, J. R. (1997). The influence of family regulation, omy. Psychological Review, 100, 674–701.
connection, and psychological autonomy on six mea- Nezlek, J. B. (2001). Multilevel random coefficient analyses
sures of adolescent functioning. Journal of Adolescent of event- and interval-contingent data in social and
Research, 12, 34–67. personality psychology research. Personality and Social
Jaccard, J., Turrisi, R., & Wan, C. K. (1990). Interaction effects Psychology Bulletin, 27, 771–785.
in multiple regression. Newbury Park, CA: Sage. Petersen, A. C., Compas, B. E., Brooks-Gunn, J., Stemmler,
Kandel, D. B. (1978). Similarity in real life adolescent M., Ey, S., & Grant, K. E. (1993). Depression in
pairs. Journal of Personality and Social Psychology, 36, adolescence. American Psychologist, 48, 155–168.
306–312. Petersen, A. C., Schulenberg, J. E., Abramowitz, R. H.,
Kaplan, H. B. (1978). Deviant behavior and self-enhance- Offer, D., & Jarcho, H. D. (1984). A Self-Image Ques-
ment in adolescence. Journal of Youth and Adolescence, 7, tionnaire for Young Adolescents (SIQYA): Reliability
253–277. and validity studies. Journal of Youth and Adolescence, 13,
Kim, J. E., Hetherington, E. M., & Reiss, D. (1999). 93–111.
Associations among family relationships, antisocial Pettit, G. S., Laird, R. D., Dodge, K. A., Bates, J. E., &
peers, and adolescents’ externalizing behaviors: Gender Criss, M. M. (2001). Antecedents and behavior-problem
and family type differences. Child Development, 70, 1209– outcomes of parental monitoring and psychological
1230. control in early adolescence. Child Development, 72,
Kurdek, L. A., & Fine, M. A. (1994). Family acceptance and 583–598.
family control as predictors of adjustment in young Rollins, B. C., & Thomas, D. L. (1979). Parental support,
adolescents: Linear, curvilinear, or interactive effects? power, and control techniques in the socialization of
Child Development, 65, 1137–1146. children. In W. R. Burr, R. Hill, F. I. Nye, & I. L. Reiss
Kurdek, L. A., Fine, M. A., & Sinclair, R. J. (1995). School (Eds.), Contemporary theories about the family (Vol. 1, pp.
adjustment in sixth graders: Parenting transitions, 317–364). New York: Free Press.
family climate, and peer norm effects. Child Develop- Scaramella, L. V., Conger, R. D., Spoth, R., & Simons, R. L.
ment, 66, 430–445. (2002). Evaluation of a social contextual model of
Lamborn, S. D., Mounts, N. S., Steinberg, L., & Dornbusch, delinquency: A cross-study replication. Child Develop-
S. M. (1991). Patterns of competence and adjustment ment, 73, 175–195.
among adolescents from authoritative, authoritarian, Schaefer, E. (1965). Children’s reports of parental behavior:
indulgent, and neglectful families. Child Development, 62, An inventory. Child Development, 36, 413–424.
1049–1065. Schwarz, J. C., Barton-Henry, M. L., & Pruzinsky, T. (1985).
Lee, V. E., & Bryk, A. S. (1989). A multilevel model of the Assessing child-rearing behaviors: A comparison of
social distribution of high school achievement. Sociology ratings made by mother, father, child, and sibling on the
of Education, 62, 172–192. CRPBI. Child Development, 56, 462–479.
Maccoby, E. E., & Martin, J. A. (1983). Socialization in the Simons, R. L., Chao, W., Conger, R. D., & Elder, G. H., Jr.
context of the family: Parent–child interaction. In M. (2001). Quality of parenting as mediator of the effect of
Hetherington (Ed.), Handbook of child psychology: Vol. 4. childhood defiance on adolescent friendship choices
Socialization, personality, and social development (pp. 1– and delinquency: A growth curve analysis. Journal of
101). New York: Wiley. Marriage and the Family, 63, 63–79.
Maggs, J. L., Almeida, D. M., & Galambos, N. L. (1995). Slicker, E. K. (1998). Relationship of parenting style to
Risky business: The paradoxical meaning of problem behavioral adjustment in graduating high school se-
behavior for young adolescents. Journal of Early Adoles- niors. Journal of Youth and Adolescence, 27, 345–372.
cence, 15, 344–362. Spieker, S. J., Larson, N. C., Lewis, S. M., Keller, T. E., &
Marcus, R. F. (1996). The friendships of delinquents. Gilchrist, L. (1999). Developmental trajectories of dis-
Adolescence, 21, 145–158. ruptive behavior problems in preschool children of
Mason, C. A., Cauce, A. M., Gonzales, N., & Hiraga, Y. adolescent mothers. Child Development, 70, 443–458.
(1996). Neither too sweet nor too sour: Problem peers, Steinberg, L., Lamborn, S. D., Darling, N., Mounts, N. S., &
maternal control, and problem behavior in African Dornbusch, S. M. (1994). Over-time changes in adjust-
American adolescents. Child Development, 67, 2115–2130. ment and competence among adolescents from author-
Mason, W. M., Wong, G. Y., & Entwisle, B. (1984). itative, authoritarian, indulgent, and neglectful families.
Contextual analysis through multilevel linear analysis. Child Development, 65, 754–770.
In S. Leinhardt (Ed.), Sociological methodology 1983–1984 Stice, E., & Barrera, M., Jr. (1995). A longitudinal examina-
(pp. 72–103). San Francisco: Jossey-Bass. tion of the reciprocal relations between perceived
Miller, B. C., McCoy, J. K., Olson, T. D., & Wallace, C. M. parenting and adolescents’ substance use and externa-
(1986). Parental discipline and control attempts in lizing behaviors. Developmental Psychology, 31, 322–334.
594 Galambos, Barker, and Almeida

Stice, E., Barrera, M., Jr., & Chassin, L. (1993). Relation of peer management behaviors. Manuscript submitted for
parental support and control of adolescents’ externaliz- publication.
ing symptomatology and substance use: A longitudinal Willett, J. B. (1989). Some results on reliability for the
examination of curvilinear effects. Journal of Abnormal longitudinal measurement of change: Implications for
Child Psychology, 21, 609–629. the design of studies of individual growth. Educational
Tilton-Weaver, L. C., & Galambos, N. L. (2003). Adolescents’ and Psychological Measurement, 49, 587–602.
characteristics and parents’ beliefs as predictors of parents’