LMS

$00?

This is to certify that the
dissertation entitled

COMPETENCE AND ITS LINK TO EXTERNALIZING AND
INTERNALIZING PROBLEMS: A LONGITUDINAL STUDY

presented by

VASILIKI MOUSOULI

has been accepted towards fulfillment
of the requirements for the

PhD. degree in School Psycholgqx—

 

0 ,
-’ f-‘Jw DcXJQJV 1/?MM&\

A

0 Major Profess’or’s Signature
S/s/oc

Date

MSU is an Afl‘innative Action/Equal Opportunity Institution

 

 

T LIBRARY a
MiChigau State
Unwemmy

 

 

 

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p;/CIRC/DateDue.indd-p.1

COMPETENCE AND ITS LINK TO EXTERNALIZING AND INTERNALIZING
PROBLEMS: A LONGITUDINAL STUDY

By

Vasiliki Mousouli

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Counseling, Educational Psychology, and Special Education

2006

ABSTRACT

COMPETENCE AND ITS LINK TO EXTERNALIZING AND INTERNALIZING
PROBLEMS: A LONGITUDINAL STUDY

By

Vasiliki Mousouli

The purpose of this study was to explore the empirical links between social and
academic competence and externalizing and internalizing problems over time. Four
different models of competence were examined, one with self-reported indicators of
competence, one with teacher-reported indicators of competence, an academic
competence model, and a model, which included teacher-student relationships as an
indicator of competence.

Results from path analyses showed a strong continuity from Fall of Year 1 to
Spring of Year 2 for both competence indicators (i.e., social skills, study skills, peer
support, and scholastic competence) and extemalizing-intemalizing problems. Important
findings emerged from the teacher-student relationships model, which showed that
positive teacher-student relationships significantly predicted high social and study skills
and low externalizing and internalizing problems almost two years later through the
mediating effects of prior levels of social and study skills and externalizing and
internalizing problems respectively. Moreover, externalizing problems directly predicted
lower teacher-student relationships while study skills predicted positive teacher-student
relationships 18 months later.

Results also demonstrated that externalizing problems consistently predicted low

scholastic self-competence over time while study skills predicted low externalizing

problems. An interesting finding was that internalizing problems predicted positive
scholastic self-competence. On the other hand, study skills predicted low internalizing
problems.

Competencies have been identified as protective factors against the development
of childhood psychopathology in previous studies. This study demonstrated that teacher-
student relationships and students’ study skills hold promise for preventing externalizing
and internalizing problems and enhancing students’ well-being. Given the accumulation
of empirical evidence regarding the importance of competence indicators, future research

should pay more attention to the empirical study of protective factors.

Capyright by
VASILIKI MOUSOULI
2006

To my beloved fiancé, Christos and my parents!

ACKNOWLEDGMENTS

The completion of this dissertation would not have been possible without the
constant support of many people. First, I would like to acknowledge my advisor, Dr. Jean
Baker and my committee members, Dr. Hi Fitzgerald, Dr. John Carlson, and Dr. Mark
Reckase. To my advisor, Dr. Jean Baker, there are no words to thank you for facilitating
my way towards my degree completion. Your guidance, continued support, and
flexibility made me feel secure and confident that I can succeed! To Dr. Hi Fitzgerald,
you challenged my thinking and helped me make this dissertation a quality work! But I
also need to thank you for your continued support throughout my studies at MSU. I’m
looking forward to continuing to work with you in the future. To Dr. John Carlson, your
thoughtful contributions and your continued support were very important to me. To Dr.
Mark Reckase, thank you very much for the time you devoted to this dissertation. Your
excellent teaching of statistical analysis has made this study possible for me. To Jessica
Barnes, your help and guidance with the data analysis have been very important to me. I
also want to thank Dr. Evelyn Oka. You’ve challenged my thinking and shaped my
identity as a School Psychologist.

I would like to express my deep love and gratitude to my fiance, Christos whose
constant support and caring helped me throughout my studies and made my stay in the
US. possible. Your continuous encouragement when things seemed difficult gave me
strength when I needed it most and helped me overcome every obstacle in my way.

A deep thank you goes to my parents who were beside me at every step I have

made even when their physical presence in my life in Michigan was not possible. Thank

vi

you not only for supporting my decisions but also for helping me take risks and push
myself beyond my comfort level! I would not have been here without you! Thank you to
my sweet sister, Natassa who sacrificed her summer vacations to be with me before my
defense! My last days in Michigan would not have been easy without you! And of course,
I cannot close this paragraph without thanking my brother whose love and support have
always been there for me!

Special thanks go to my roommates, Rozeta, Lela, and Irini! You stood to me like
family. There are no words to express my gratitude for the endless hours you spent
listening to me and supporting me when I needed you. Thank you for helping me not to
feel lonely when I could not reach my family. You’ll always be in my heart!

To my cohort and dearest friends, My, Sonia, Nora, and Jana! Your sweetness,
love, and support made my stay in Michigan so pleasant. How would the endless hours of
studying in the library and coffee shops have been possible without you? To Cheryl,
thank you for being a teacher and a friend to me! Your support and fi'iendship mean a lot
to me! Finally, to my dearest friends, Tania and Fofi whose love, reassurance, and belief

in me helped me tremendously!

vii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... ix
LIST OF FIGURES .................................................................................. x
CHAPTER I

INTRODUCTION .................................................................................... 1

CHAPTER 2

LITERATURE REVIEW ........................................................................... 5

CHAPTER 3

METHODS ............................................................................................ 33
CHAPTER 4

RESULTS ............................................................................................ 45
CHAPTER 5

DISCUSSION ....................................................................................... 71
APPENDIX A

Clinical/At-Risk Ranges ............................................................................ 89
APPENDIX B

Histograms for Estimated Sample .................................................................. 90
APPENDIX C

Study Measures .................................................................................... 104
REFERENCES .................................................................................... 108

viii

LIST OF TABLES

Table 1. Descriptive Statistics for Study Variables Pre—Imputation .......................... 40
Table 2. Descriptive Statistics for Study Variables Post-Imputation (Nfi42) .............. 41
Table 3. Correlations of the Variables Included in the Models (N=442) .................... 47
Table 4. Goodness of Fit Summary for Model 1 (N=442) .................................... 48
Table 5. Goodness of Fit Summary for Multiple Group Analysis for Model 1 based on
Gender (Boys N=208, Girls N=234) ............................................................. 51
Table 6. Goodness of Fit Summary for Multiple Group Analysis for Model 1 based on
Race (African-American N=252, White N=136) ............................................... 52
Table 7. Goodness of Fit Summary for Model lby Race (African-American N=252,
White N=136) ........................................................................................ 52
Table 8. Goodness of Fit Summary for Multiple Group Analysis for Model 1 based on
Grade Level (1-2 Grade at Time 1 N=21 1, 3-4 Grade at Time 1 N=231) .................. 55
Table 9. Goodness of Fit Summary for Model 2(N=442) ..................................... 56
Table 10. Goodness of Fit Summary for Multiple Group Analysis for Model 2 based on
Gender (Boys N=208, Girls N=234) ............................................................. 59
Table 11. Goodness of Fit Summary for Model 2by Gender (Boys N=208, Girls N=234)
Indirect Effects of Teacher-Student Relationships ............................................. 59
Table 12. Goodness of Fit Summary for Model 3(N=442) .................................... 62
Table 13. Goodness of Fit Summary for Multiple Group Analysis for Model 3 based on
Gender (Boys N=208, Girls N=234) .............................................................. 65
Table 14. Goodness of Fit Summary for Model 3 by Gender (Boys N=208, Girls
N=234)... ...................................................................................................................... 65
Table 15. Goodness of Fit Summary for Model 4-Teacher-Student Relationships
(N=442) .............................................................................................. 69
Table 16. Indirect Effects of Teacher-Student Relationships ................................. 69

ix

LIST OF FIGURES

Figure 1. Conceptual Model ....................................................................... 42
Figure 2. Conceptual Model for Self-Perceived Competence ................................. 43
Figure 3. Conceptual Model for Teacher-Reported Competence ............................ 44

Figure 4. Standardized Path Coefficients and Unexplained Variance for Model 1: Self
Reported Competence with Error CorrelatronsSO

Figure 5. Standardized Path Coefficients and Unexplained Variance for African-
American Students Model 1: Self-Reported Competence with Error
Covariances .......................................................................................... 53

Figure 6. Standardized Path Coefficients and Unexplained Variance for White Students
Model 1: Self-Reported Competence with Error Correlations. . 54

Figure 7. Standardized Path Coefficients and Unexplained Variance for Model 2:
Teacher-Reported Competence ................................................................... 58

Figure 8. Standardized Path Coefficients and Unexplained Variance for Boys Model 2:
Teacher-Reported Competence ................................................................... 60

Figure 9. Standardized Path Coefficients and Unexplained Variance for Girls Model 2:
Teacher-Reported Competence ................................................................... 61

Figure 10. Standardized Path Coefficients and Unexplained Variance for Model 3:
Academic Competence. . . . . . .............................................................................................. 63

Figure 11. Standardized Path Coefficients and Unexplained Variance for Boys Model 3:
Academic Competence ..................................................................................................... 66

Figure 12. Standardized Path Coefficients and Unexplained Variance for Girls Model 3:
Academic Competence ............................................................................ 67

Figure 13. Standardized Path Coefficients and Unexplained Variance for Teacher-Student
Relationship Model 4 .............................................................................. 7O

CHAPTER 1
INTRODUCTION

The growing number of children who experience mental health problems explains
the unremitting interest that clinicians and researchers show for the identification,
treatment, and prevention of childhood disorders. Epidemiological studies estimate that 1
in 5 children in the United States manifest some type of difficulty and l in 10 have a
clinical disorder that is associated with impaired functioning (see Mash & Dozois, 2003).
Both externalizing and internalizing problems have high prevalence in the childhood
population (Achenbach, Howell, McConaughy, & Stanger, 1995) and are associated with
a host of negative outcomes in children’s life, such as peer rejection, school failure, and
low self-esteem.

Although there is an extensive body of research regarding childhood
psychopathology, little attention has been given to children’s competencies and even less
attention to the relation of competence to psychopathology. However, increasing
evidence shows that competence and psychopathology are conceptually and empirically
related and need to be studied together (Garmezy & Masten, 1991; Garmezy, Masten, &
Tellegen, 1984; Masten &. Coatsworth, 1995; Masten & Curtis, 2000). Masten and Curtis
(2000) state that “though complex, the constructs of competence and psychopathology
each concern psychosocial adaptation and judgments about the individual differences in
the successfirlness of adaptive functioning” (p.529). Competence refers to the
effectiveness of individual adaptation in the environment and psychopathology, besides
various symptoms of the disorder, is ofien related to “impairment” in adaptive

firnctioning.

The two constructs of competence and psychopathology have historical,
conceptual, and empirical links (Masten & Coatsworth, 1995; Masten & Curtis, 2000).
The links between them started being closely studied 30 years ago with the emergence of
resilience research (Garmezy & Masten, 1991; Masten & Curtis, 2000). Resilience
research generated fi'om studies of psychopathology, in which researchers observed that
mentally ill patients with better premorbid competence had better prognosis than the rest
of the patients (Glick & Zigler, 1986; Glick, Zigler, Baldwin, Baldwin, & Cole, 1990;
Zigler & Phillips, 1961). Investigators also noted that children at risk for
psychopathology followed different pathways of development.

Competence and psychopathology have also strong empirical links (Cole, 1990;
Masten & Coatsworth, 1995; Masten & Curtis, 2000). Research shows that competent
children have the lowest scores on problem checklists while incompetent children have
the highest scores (Blechman, Tinsley, Carella, & McEnroe, 1985). After many decades
of research, Kohlberg, LaCrosse, and Ricks (1972) concluded that adult psychopathology
can be predicted only by broad indicators of childhood problems: school failure, poor
peer relationships, and antisocial behavior. Competence rather than the absence of
symptoms were more powerful for the prediction of adult problems (Kohlberg, LaCrosse,
& Ricks, 1972). This finding indicates that failure to achieve age-salient developmental
tasks like academic and social competence has more important implications for later
development than the absence of symptomatology.

Extensive research on risk and resilience has shown that there are many different
risk and protective factors associated with disorder. However, there is no direct link

between a specific risk factor and a single disorder, rather it seems that a variety of

adaptive functions are influenced at the same time leading to diverse symptomatology
(Coie et al., 1993). Poor peer relationships, for instance, predict different types of
problems, such as early academic problems, adolescent conduct problems, and
internalizing disorders (Coie et al., 1993 ). It is therefore, important to study multiple
factors simultaneously in order to uncover the relations between competence and
psychopathology.

Tracing the links between competence and psychopathology has significant
implications: it has the potential to inform theories about the etiology of childhood
psychological problems and provide valuable information about early intervention and
prevention (Coie et al., 1993; Masten & Coatsworth, 1995; Masten & Curtis, 2000;
Sroufe & Rutter, 1984). For instance, if competence can act as a protective factor against
the development of psychopathology, then intervening to enhance children’s social
competence in schools might be effective for the prevention of externalizing and
internalizing problems.

Despite these conceptual and empirical links and the implications of these
associations, there is little systematic research regarding the interrelatedness of
competence and psychopathology. The purpose of this study is to consider these two
constructs together in order to explore the ways by which competence and
psychopathology are linked. Specifically, in terms of psychopathology, the focus of this
study will be on externalizing and internalizing problems. In terms of competence, the
focus will be on social and academic competence. In the following chapter, I will define

these constructs and I will examine the conceptual and empirical links between

competence and psychopathology. Conceptual models regarding these linkages as well as

empirical studies in support of or contrary to these models will be presented.

CHAPTER 2

LITERATURE REVIEW

Theoretical Framework for Understanding the Relations between Competence and

Behavior Problems

This study is situated within a developmental psychopathology framework and a
risk and adaptation conceptual framework. Developmental psychopathology is based on
the premise that the study of normal and pathological should be jointly considered
(Sroufe, 1990). It is necessary to understand adaptational processes in order to begin
elucidating psychopathological processes and vice versa. Competence has been
considered a construct central to the understanding of adaptive and disordered behavior
(Garmezy & Masten, 1991). An increasing interest in positive psychology has shified
researchers’ focus ficm deficit perspectives of maladjustment to positive indicators of
adjustment (Compas, Connor-Smith, Saltzman, Thomsen, & Wadsworth, 2001; Cowen,
2000; Seligman & Csikszentmihalyi, 2000), which has led to the study of children’s
competencies, strengths, and the phenomena of resilience.

Child by Environment theories of risk and adaptation emphasize the importance
of risk and protective factors for children’s adjustment (Coie et al., 1993; Ladd, 1996;
Ladd & Burgess, 2001). They consider the relationships that children form with
significant adults and peers in different contexts as crucial for their psychological and
school adjustment (Ladd & Burgess, 2001). Risk factors are defined as relational and
environmental factors that increase the probability of adjustment difficulties, whereas
protective factors decrease the probability of adjustment problems and promote resilience

(Coie et al., 1993; Masten, 1994). For instance, teacher-student relationships and peer

support can act as a protective factor for children with belmvior problems and thus,
facilitate their adjustment.

Construct Definitions
Definition of Competence

Competence is closely related to adaptation and refers to the “capability for and
manifestation of favorable adaptation” (Masten & Coatsworth, 1995). Based on Masten
and Coatsworth’s review (1995), adaptation has been defined differently in different
theories. Some psychological theories have focused on internal adaptation, which
includes processes or outcomes related to self-regulation, self-perceptions, and
motivation while others place greater emphasis on external adaptation, such as academic
achievement, social relationships and interactions with the environment (Masten &
Coatsworth, 1995). Failure to achieve internal adaptation is often related to internalizing
problems, such as depression and anxiety, whereas failure to achieve external adaptation
is related to externalizing problems, such as aggression, peer rejection, and delinquency
(Masten & Coatsworth, 1995).

Ford (1985) recognized the complexity and ambiguity of the construct of
competence and emphasized that the concept needs to be operationalized more precisely
before being used in research. He indicated five interrelated yet distinct meanings of
competence generated from different theoretical perspectives at the time of his review: a)
competence as a motivational construct referring to the ability of the individual to pursue
goals and be effortful (e.g., Hatter, 1985a; Ryan & Deci, 2000) ; b) competence as a
subjective evaluation of one’s capability to affect the environment: self-efficacy (e. g.,

Bandura, Barbaranelli, Caprara, & Pastorelli, 1996; Bandura, Barbaranelli, Caprara, &

Pastorelli, 2001); c) competence as a personal “repertoire” of skills and abilities, such as
social skills (e.g., McGinnis & Goldstein, 1990). These skills represent the individual’s
potential and do not necessarily translate into actions or successful outcomes; d)
competence as effective behavior in relevant contexts, such as academic achievement,
accomplishment of social goals consistent with the individual’s developmental level, or
successful peer relationships (e.g., Bandura et al., 1996; Bandura et al., 2001; Waters &
Sroufe, 1983); and last, e) competence as a personality trait (e.g., Becker & King, 1964).

Masten and Coatsworth (1995) identified several common themes and differences
in definitions of the construct of competence. First, competence definitions recognize that
competence is a multidimensional construct that refers to mental processes or behavioral
outcomes regarding the “effectiveness of adaptation in the environment” (Masten &
Coatsworth, 1995). Processes underlying the construct of competence are complex and
require the integration of cognitive, affective, and motivational mental functions in order
to achieve adaptive behavior. Outcomes refer to observable manifestations of competence
through behaviors. Moreover, in most definitions, competence is considered a
developmental construct since indicators of competence change across development.
Differences in the definitions of competence include the emphasis on internal versus
external adaptation and processes versus manifestations.

According to Masten and Coatsworth (1995), competence is considered “a pattern
of effective performance in the environment, evaluated from the perspective of
development in ecological and cultural context” (p. 724). In their definition, the focus is
on both processes within the individual and transactions between the individual and the

environment The criteria for judging “effective performance in the environment” are

based on successful accomplishment of “salient developmental tasks”. These tasks might
be different based on the historical, cultural, and ecological context.

Several studies have explored the different dimensions of competence (e.g.,
social, academic competence) and examined whether they are distinct dimensions or
reflect manifestations of global competence. Harter’s work on self-perceived competence
suggests that academic/scholastic and social competence are distinct dimensions of
competence (Harter, 1982, 1985a). Using factor analysis, Tanaka and Westerman (1988)
showed that competencies in the academic domain are separable fiem competencies in
the social domain measured with both self-perceptions (i.e., Harter’s measure) and
mother behavior ratings (i.e., Achenbach’s CBCL rating scale).

In this study, the construct of competence includes both internal processes fi'om
the child’s perspective (i.e., self-perceptions about academic competence and self-
perceptions about peer support) and judgments about competence by outside observers,
specifically teachers (e.g. teacher-rated social skills). Therefore, the focus is on both
observed skills and manifestations of behavior and self-reports of competence.
Competence, in addition to its biological definition, is also considered a contextual
concept because it is influenced and evaluated by the particular environment (i.e.,
classroom teachers). Based on these characteristics, competence in this study is defined in
terms of effective performance in age-salient developmental tasks (i.e., academic and
social) as judged both by teacher-rated manifestations of behavior in the school context
and student self-reports. This definition is congruent with Masten and Coatsworth’s
definition of competence presented above as well as other attempts to define the construct

(e.g., Ford, 1985; Waters & Sroufe, 1983).

Definition of Social Competence (Social Skills and Peer Support)

One social competence indicator considered in this study concerns social skills
rated by teachers. Social skills were first examined as a component of adaptive behavior.
Adaptive behavior, originally named “social maturity” originated from early work in
mental retardation (Karnphaus & F rick, 2002). In 1953, Doll noted that individuals with
mental retardation not only had intellectual deficits but they also experienced difficulties
with daily living skills (as cited in Kamphaus & Frick, 2002). He argued that assessment
of daily living skills was necessary. He created the Vineland Social Maturity Scale, later
called “Vineland Adaptive Behavior Scale”, which assesses communication skills, daily
living skills, socialization, coping skills, interpersonal-social skills and other dimensions
of competence. Thus, early on, social skills were considered an important contributor to
effective adaptation in the environment for children with mental retardation.

Several studies provided evidence for the importance of adaptive behavior for
children with mental retardation. For instance, one study using multitrait/multimethod
structural equation modeling showed evidence for the existence of four domains of
adaptive behavior: cognitive competence, social competence, social maladaptation, and
personal maladaptation in a sample of children with mental retardation (Widaman, Stacy,
& Borthwick—Duffy, 1993). Although there is no universally accepted definition of
adaptive behavior, it is often defined as the accomplishment of daily activities that are
necessary for social and personal independence (Kamphaus & F rick, 2002). Merrell and
Popinga (1994) examined the relationship between ratings of adaptive behavior and

social competence and they found moderate to strong relations between social

competence and adaptive behavior, especially the social and communication domains of
adaptive behavior suggesting that social competence is one aspect of adaptive behavior.

In the literature, there are several approaches to the operational definition of
social competence: social skills, sociometric status, friendships/relationships, and
functional outcomes (Rose-Krasnor, 1997). Based on the social skills approach, social
competence is conceptualized as a set of skills and is measured with the use of behavior
checklists. A difficulty with this approach concerns the identification of social skills that
lead to important social outcomes. Gresham (1995 ) reintroduced social validity, an
important component in the definition of social skills based on McFall’s work (1982). He
indicated that social skills are socially significant behaviors in that they lead to important
social outcomes (e.g., peer acceptance, friendships, positive self-esteem, and school
adjustment). The social validity definition is beneficial because it identifies social skills
in terms of the outcomes that they bring about.

Social competence and social skills have often been used interchangeably in the
literature. However, a distinction has been made between the two terms based on
McFall’s (1982) work. Social skills represent specific behaviors that others (e.g.,
teachers, peers) may judge as socially competent (Gresham, 1995). Therefore, social
skills are the behaviors, and social competence refers to the judgments of these behaviors
as competent. In this study, teacher-rated social skills are considered one indicator of the
construct of social competence.

The sociometric approach to social competence identifies students’ peer status
(e. g., popular, accepted, rejected). Relationship approaches to social competence assess

the quality of individual friendships. Functional approaches identify context-specific

10

social goals and focus both on outcomes of social behavior and on the mechanisms
behind these outcomes (e.g. information processing). The relations among these measures
of social competence (i.e., social skills, peer status, relationship, and firnctional
indicators) are moderate (Rose-Krasnor, 1997). This might be due to differences in the
methods and/or the raters. However, it is also possible that each indicator reflects a
different aspect of social competence (Rose-Krasnor, 1997).

Rose-Krasnor (1997) integrated the above approaches by viewing “social
competence” as an organizing construct that is defined as “effectiveness in interaction”
and has a variety of indicators: fi'iendship, peer status, social self-efficacy, social support,
and others. She conceptualized her theoretical model as a prism that has three different
hierarchical levels. The top level of the prism is called “Theoretical Level” and includes
“social competence” as a broad theoretical construct. The next level is called “Index
Level” and includes a variety of indicators of social competence as those mentioned
before (i.e., friendship, peer status, social support etc.). The third level called “Skill
Level” includes specific individual social skills, abilities, and values, such as problem-
solving and empathy. These skills are the “building blocks” or base of fi'iendships,
interactions, and peer status from the middle level. In the current study, aspects of Rose-
Krasnor’s prism are included. Specifically, social competence is considered a broad
theoretical construct whose indicators include self-perceptions of peer support and

teacher-rated social skills.

Definition of A cadernic Competence (Study Skills, Academic Self-Competence and

Grades)

11

Study skills have been defined as behaviors or strategies that facilitate the
processing of new material and promote learning (DiPema & Elliott, 2000). They have
also been defined as, “those competencies associated with acquiring, recording,
organizing, synthesizing, remembering, and using information and ideas” (Bos &
Vaughn, 2002). Study skills have been shown to be conducive to academic achievement.
Students with learning and behavior problems seem to have difficulty developing these
skills (Bos & Vaughn, 2002). DiPema and Elliott (2001) found significant moderate to
strong correlations between study skills and prior reading achievement (F38 for
Kindergarten to 2"" graders and F56 for 3rd to 6th graders). They suggested that study
skills become more important as students become older because the emphasis of the
instruction shifts from learning to read to reading to learn. Although the direct effect of
study skills to academic achievement was small (from .02 to .13) after controlling for
prior achievement, interpersonal skills, and motivation in their structural equation model,
the researchers believe that study skills are an important component of academic
competence (DiPema, Volpe, & Elliott, 2002). Another study found that study skills were
an important predictor of problem behavior (including externalizing and internalizing
problems) for female elementary-age students (Thorpe, Karnphaus, Rowe, &
Fleckenstein, 2000, August). In this study, study skills include strategies such as problem
analysis, organizational skills, homework completion, test confidence, and hard working.

Students’ self appraisals about their academic/scholastic competence constitute an
important dimension of academic competence, as well. Harter’s work on self-evaluations
of competence has focused on its multidimensional and hierarchical nature (Harter, 1982,

1985a, 1985b, 1999). Harter (1982) conceptualized five different dimensions of

12

competence (i.e., scholastic, social, conduct, athletic, and appearance) as subordinate
domains of a global construct of self-worth, which does not simply consist of the sum of
the specific competence dimensions. Each specific domain contributes to the overall self-
worth but the degree of that contribution is dependent upon the value that the specific
domain has in the child’s life. For instance, if a child believes that being a good student is
very important but he does not think that being a good athlete is so important, then
scholastic self-concept will weigh more on overall self-worth than athletic self-concept.
This study focuses on the specific domains of competence (i.e., academic and social).
Harter (1999) has considered the developmental aspects of self-perceived
competence and she found that children ages 4-7 were able to make evaluative judgments
of the five domains but these judgments were reduced to two broad dimensions;
cognitive-athletic competence formed one dimension and appearance-social-conduct
competence formed a second dimension. Children in middle to late childhood were able
to differentiate among the five competence domains. This developmental difference
might be attributable to children’s increased cognitive ability as well as their ability to
make social comparisons and to understand the way others judge them (Cole, Jacquez, &
Maschman, 2001 ). When students’ perceptions of their scholastic competence were
compared to teacher-rated measures of academic achievement, moderate relations were
discovered ranging from .30 for 3rd graders and increasing as the child advanced in the
grades reaching .60 in 6'“ grade (Harter, 1982, 1985a), thus indicating that children
become increasingly more accurate in judging their scholastic competence. In this study,
Harter’s scholastic self-perceived competence will be used in addition to teacher-rated

study skills and grades as an indicator of overall academic competence.

13

Definition of Psychopathology

Definitions of psychopathology reveal the conceptual links between competence
and psychopathology. Child psychopathology has been defined as an “adaptational
failure” in cognition, affect, physiology, and/ or behavior (Mash & Dozois, 2003). Mash
and Dozois (2003) explain the concept of “adaptational failure”:

“Adaptational failure may involve deviation from age-appropriate norms,
exaggeration or diminishment of normal developmental expressions,
interference in normal developmental progress, failure to master
developmental tasks, failure to develop a specific function or regulatory
mechanism, and/or the use of non-normative skills (e.g., rituals,
dissociation) as a way of adapting to regulatory problems or traumatic
experiences” (p. 6).

The role of adaptation and developmental benchmarks is evident in the above
definitions. Similarly, Masten and Coatsworth (1995) define psychopathology as
“patterns of maladaptive behavior or psychological distress that interfere with some
aspect of adaptation” (p.716). In the diagnostic manual of the American Psychiatric
Association (2000), criteria for disorder diagnosis or associated features of disorders
include competence characteristics. Significant impairment in social or academic
fimctioning is often one of the criteria for diagnosis. Internal symptoms or behaviors that
compromise external adaptation are also included in the descriptions of disorders.

In this study, I focus on externalizing (i.e., aggression, conduct problems,
hyperactivity) and internalizing problems (i.e., depression, anxiety, and somatization).

The terms “externalizing” and “internalizing” were first introduced by Achenbach and

14

Edelbrock in 1978 and refer to dimensions of child psychopathology that were
empirically derived. Children experiencing externalizing problems are also described as
undercontrolled, exhibiting disruptive behaviors while children experiencing internalizing

problems are described as overcontrolled, exhibiting internal adjustment difficulties.

Ecological-Transactional Model and Risk and Resilience Perspectives as a

Theoretical Framework for Understanding Competence and Behavior Problems

Child by Environment theories of risk and adaptation emphasize the importance
of risk and protective factors for children’s adjustment (Ladd, 1996; Coie et al., 1993).
They consider the relationships that children form with significant adults and peers in
different contexts as crucial for their psychological and school adjustment (Ladd &
Burgess, 2001). Risk factors are defined as relational and environmental factors that
increase the probability of adjustment difficulties, whereas protective factors decrease the
probability of adjustment problems and promote resilience (Masten, 1994; Coie et al.,
1993). For instance, teacher-student relationships and peer support can act as a protective

factor for children with behavior problems and thus, facilitate their adjustment.

Conceptual Models Linking Competence and Psychopathology
Conceptual links between competence and psychopathology are manifest in some
of the definitions presented above. In addition, there are a number of ways by which
competence and psychopathology are connected. Masten and Coatsworth (1995) have
identified four theoretical models regarding the relation between competence and

psychopathology. The first two models will be explored in depth in this study.

15

l. Psychopathology affects competence

  
  

    

Competence
Failure

 

Psychopathology

  
 

This conceptual link between competence and psychopathology is congruent with
a medical model of disorder (Masten & Coatsworth, 1995; Masten & Curtis, 2000). For
neurodevelopmental disorders, like schizophrenia and autism, this link is apparent.
However, for disorders like depression, the direction of this link is not as simple. There
are theories that suggest that competence failures contribute to depression and vice versa.
This leads us to the next theoretical model.

2. Competence failure leads to psychopathology

Psychopathology

 

Competence failures might lead to psychopathology. For instance, academic
failure and low peer support could lead to distress and depression. There are several
empirical models in support of the above conceptual link that will be presented in the
next section (see, Cole, Martin, & Powers, 1997; Patterson & Capaldi, 1990). The
purpose of this study is to clarify the direction of the relation between competence and
psychopathology (Models 1 & 2) using two waves of longitudinal data.

3. A third factor causes both competence and psychopathology

16

   

 
  

Psychopatholo
gy

    

Competence

The shared variance between competence and psychopathology might also be due
to common biological (e.g., temperamental factors, attention regulation, IQ) or
environmental (e. g. parenting, SES) factors that contribute to both lower competence and
symptoms of psychopathology (Masten & Coatsworth, 1995; Masten & Curtis, 2000).
Although this model has theoretical strength, it won’t be the focus of this study.

4. Competence and psychopathology arethe opposites of the same dimension of
behavior

Competence Psychopathology

 

Masten and Coatsworth (1995) refer to competence and psychopathology as
potentially being the two ends of the same dimension. For instance, academic
achievement could be considered as an indicator of competence, while academic failure
could be a symptom of psychopathology. This overlap between competence and
psychopathology criteria might also be explained by measurement problems (Masten &
Curtis, 2000). For instance, there is item overlap in measures used to assess rule-abiding
and rule-breaking behavior, which are considered indicators of social competence or
conduct problems respectively (Masten & Curtis, 2000). This conceptual model implies

that competence is the polar opposite of psychopathology and thus, it may be considered

17

that it is redundant to study competence in addition to psychopathology (Thorpe et al.,
2000, August). However, empirical evidence presented in the next section does not
support this conclusion.

In this study, the first two conceptual models will be explored and existent
empirical evidence in support or not of these conceptual models will be presented.
However, methodological limitations of the studies do not allow definite conclusions
about the direction of the effects based on the existing literature. For instance, there are
not many longitudinal studies that can demonstrate causal directions between competence
and psychopathology, and when they exist, they do not control for both competence and

symptoms at time 1 so that interpretations for the directions of the effects can be made.

Empirical Studies Linking Competence and Psychopathology

Psychopathology Leads to Competence Failures

A study using the 31-year-old New York longitudinal dataset, which consisted of
a primarily white, Jewish, middle to upper class sample, explored the relation between
two factors, Aggression and Affect (e. g. depression, anxiety) and three adjustment
outcomes (i.e., school, social, and family adjustment) at ages 1 to 14 (Lerner, Hertzog,
Hooker, Hassibi, & Thomas, 1988). A structural equation model showed that Aggression
in middle childhood (T irne 2) was a strong predictor of adolescent adjustment difficulties
(Time 3) while Affect did not predict adjustment independent of the aggression construct.
Both constructs (i.e., Aggression and Affect) showed high consistency from early
childhood (Time 1) to middle childhood (Time 2). This study provided some support that

psychopathology may lead to competence failures. However, the lack of control of the

18

competence indicators at the outset of the study prevents us from reaching definite

conclusions about the direction cf the effects.

Competence Failures Lead to Psychopathology

On the other hand, there is empirical support that competence failures contribute
to psychopathology. A longitudinal study of 5- to 13-year-old boys using the Dunedin
(New Zealand) dataset shed some light on the role that reading achievement can play in
the developmental trajectories of 4 groups of children: a) children with attention deficit
disorder (ADD) only, b) children with delinquency only, c) children with both ADD and
delinquency, and d) children with no disorder (Moffitt, 1990). It was the mixed group of
children with ADD and delinquency that had the worst reading achievement scores while
the other groups did not differ significantly at any age. This group experienced reading
problems right from the beginning of entering school while at the same time, their
antisocial problems increased. The author suggested that this temporal co—occurrence
might indicate that reading failure is a risk factor for the exacerbation of antisocial
behavior for boys with ADD. However, the design of the study does not allow for a
definite conclusion regarding the direction of this relation.

A tvvo-year longitudinal study examined the relation between academic and social
competence and externalizing problems (Lane, Gresham, MacMillan, & Bocian, 2001).
The researchers measured social competence with a teacher-rated social skills measure
and academic competence with a teacher-rated measure as well as reading and
mathematics scores on group achievement tests. They used three groups of fifth-grade

students: 1) students nominated “at-risk” for school failure by their third-grade teachers

19

and exceeding the 90m percentile on externalizing problems and hyperactivity based on
the ratings of their firth-grade teachers, 2) students nominated “at-risk” for school failure
by their third-grade teachers but not exceeding the 90'” percentile on externalizing
problems and hyperactivity based on the ratings of their fifth-grade teachers, 3) a
comparison or average group. They found that competence indicators as a whole
accounted for 57% of the variance in group differences and that both at-risk fifth-grade
groups had significantly lower scores on each measure of academic and social
competence at third grade relative to the comparison group. These findings provide
preliminary evidence for the importance of competence for differentiating between
students at-risk for school failure and average students.

A 2-year longitudinal study found that teacher-rated adaptive competencies at
Time I (specifically a composite score of social skills, study skills, adaptability, and
leadership) significantly predicted teacher-rated behavior problems (of both externalizing
and internalizing nature) at Time 2 for 5- to 8-year-old boys and girls after controlling for
externalizing and internalizing problems at Time 1 (Thorpe et al., 2000, August). In
addition, competencies negatively predicted disciplinary reports for girls only, again after
controlling for externalizing and internalizing problems. These findings suggest that the
predictive power of competence is over and beyond that accounted by externalizing and
internalizing behaviors.

The direction of the relation between competence and psychopathology has been
considerably explored in the depression literature, as well. Currently there are two
prominent models in the depression literature: 1) a competency-based model (Cole, 1990;

Cole et al., 1997; Cole, Martin, Powers, & Truglio, 1996; Masten & Curtis, 2000;

20

Patterson & Capaldi, 1990; Patterson & Stoolrniller, 1991; Wierzbicki & McCabe, 1988)
and 2) a dual failure model of depression (dual in terms of both academic failure and peer
rejection) (Hinshaw, 1992; Masten & Curtis, 2000; Patterson & Capaldi, 1990; Patterson
& Stoolmiller, 1991).

A multitrait—multimethod structural equation model analyzed by Cole and
colleagues (1996) showed that social competence at Time 1 negatively predicted
depression 6 months later for sixth graders after controlling for depression at Time 1, thus
providing support for a social competence model of depression. However, the same
finding was not true for third graders, which the authors suggested being due to
developmental differences. For instance, family might be more salient than peers for
younger children or younger children might not have developed refined social
comparison skills yet. On the other hand, academic competence at Time 1 did not predict
depression at Time 2. The authors suggested that academic competence might affect
depression indirectly through its influence on social acceptance or other dimensions of
competence might be more relevant to depression (e.g., conduct-behavior competence).
On the other hand, depression at Time 1 did not predict social or academic problems 6
months later, thus not providing support for a dual failure model of depression. However,
as the authors indicated a 6-month interval might be short for such a change in social and
academic competence to occur.

On the other hand, a cross-sectional study by Patterson and Stoolrniller (1991)
using structural equation modeling found partial support for a dual failure model of
depression. Good peer relations were negatively related to depression for three different

samples of children (i.e., two cohorts of fourth grade boys at risk for antisocial behavior,

21

and 9- to 12-year old children from separated families). However, academic achievement
was negatively related to depression for one sample of children only (Patterson &
Stoolrniller, 1991 ). The researchers recognized that the cross-sectional nature of their
data didn’t allow them to reach definite conclusions about the direction of the relations.
That means that depression could lead to poor peer relations, which would render support
to a competency-based model of depression or the effects could be bidirectional.
Longitudinal studies that control for the dependent variable at the outset of the study are
necessary to clarify the direction of these relations.
Competence as 0 Protective F actor for Psychopathology

The role of competence as a protective factor has been explored, as well. A
longitudinal study using the Dunedin (New Zealand) dataset of l l- to 15-year-old
children examined the relation between social competence and DSM—III disorders
(McGee & Williams, 1991). The construct of social competence included indices of
parent and peer attachment, involvement in school activities, coping resources, and
children’s self-perceptions of their strengths. When examined concurrently, it was found
that at age 15, the probability of a DSM-III disorder increased six-fold with lower levels
of social competence. A stronger relation was found between externalizing disorders and
social competence for both boys and girls than between internalizing disorders and social
competence, which appeared only for girls. However, when examined longitudinally,
social competence did not directly predict disorder four years later. Extemalizing
disorders at age 15 were best predicted by the existence of an externalizing disorder at
age 11. Although competence did not predict disorder longitudinally, it did predict

persistence of the disorder over time. In other words, the presence of disorder together

22

with low social competence at age 11 did predict externalizing disorders at age 15
suggesting a protective role of social competence for the prognosis of disorder.

Another longitudinal study using data from the Ontario Child Health Study
focused on three types of disorder: conduct disorder, hyperactivity, and emotional
disorder for children ages 4 to 16 years old (Offord et al., 1992). “Problems getting
along” with family, teachers, or peers and school performance based on parental report
were used as prognostic and risk factors. Logistic regressions showed that “problems
getting along with others” at Time 1 for children with a disorder at that time had
prognostic value since it independently predicted the persistence of one or more disorders
four years later. Poor school performance did not predict disorder persistence. Neither
poor school performance nor problems with “getting along with others” predicted the risk
for disorder onset four years later. Moreover, children with conduct disorder had
significantly more problems “getting along with others” four years later compared to the
nondisordered group. Although social impairment is an associated feature of conduct
disorder, the authors suggested that it could be indicative of the severity of the disorder.
However, children with conduct disorder did not differ on school performance from the
nondisordered group. On the other hand, children with hyperactivity or emotional
disorder at Time 1 had more problems with school achievement at Time 2 compared to
the nondisordered group but they did not differ significantly from the nondisordered
group regarding “getting along with others”. Overall, in this study, social competence,
defined as “getting along” with peers, parents, and teachers, predicted the persistence of

disorder over time.

23

A cross-sectional study explored the relation between competence and behavior
problems as well as attitudinal indicators of competence using four competence groups:
1) Competent children (who scored above the median in both academic and social
competence), 2) Incompetent children (who scored below the median in both academic
and social competence), 3) Academically Skilled children (who scored above the median
only in the academic competence dimension), and 4) Socially Skilled children (who
scored above the median only in the social dimension) (Blechman, Tinsley, Carella, &
McEnroe, 1985). Children in the competence group differed significantly in social and
academic competence from children who had strengths only in one area. Teachers rated
them as having fewer behavior problems and being less inattentive and distractible than
children in the other three groups. Moreover, they were perceived by their peers as
happier than children in the incompetence or in the academically skilled group and they
self-reported significantly higher scholastic and social competence than the children in
the incompetence group. The academically skilled group scored higher in Harter’s
scholastic competence than the socially skilled group or the incompetent group and they
scored close to the mean in behavior problems. The socially skilled group on the other
hand was rated higher on conduct problems and nervousness compared to all the other .
groups. However, peers rated them happier than the incompetence or academically
skilled group “and they perceived themselves more socially competent than the other .
groups. It is apparent therefore that children who were competent in at least one domain
(i.e., academic or social) fared better than those who were competent in neither area.

Although the direction of the effects was not clear due to the cross-sectional nature of the

24

data, this study provided preliminary evidence that competent children had fewer
behavior problems and were perceived as happy and successful by self and others.

A 3-year longitudinal study used academic and social competence at Time 1 as a
predictor of “signs of disturbances” (i.e., academic problems, school behavior problems,
receipt of mental health services, child’s need for additional help, suicidal behavior, and
police contacts) at Time 2 (Stanger, Achenbach, & McConaughy, 1993 ). A large national
sample of 995 children having at least one of the disturbances and a matched control
sample was used. Using path analysis, it was found that low parent-rated school
competence added predictive power for academic and school behavior problems three
years later over and above that accounted by behavioral and emotional problems at Time
1. Moreover, low school competence at Time I predicted mental health services at Time
2 for boys. It also predicted parent-reported need for professional help and police contacts
at Time 2 for 6- to 11-year-old boys. In addition, social competence at Time 1 was a
direct significant predictor of suicidal behavior and mental health services at Time 2 for
12- to 16-year-old girls. It also predicted police contacts at Time 2 for girls and boys.
Overall, it was shown that social and school competence negatively predicted various
“signs of disturbances” three year later after controlling for behavioral/emotional
problems at Time 1, thus suggesting a protective function for children against the
development of a variety of problems.

Some researchers endorse a strength-based approach and they conceptualize
strengths as both external assets, resources (e.g., family and peer support) and internal
abilities (e.g., humor, ability to adapt) (e.g., Lyons, Uziel Miller, Reyes, & Sokol, 2000).

In a cross-sectional study, Lyons and colleagues (2000) found that strengths, such as

25

academic strengths, psychological strengths, and peer strengths, negatively predicted risk
behaviors (e.g. suicidality, dangerousness, runaway, or delinquency) in a group of
children and adolescents in residential placements after controlling for admission risk
levels. In addition, children’s higher levels of strengths were associated with positive
discharge placements (e. g., home of parent, adoptive home) compared to negative
placement (e.g., hospital, detention). The researchers concluded that “psychopathology is
not the opposite of strengths. Many children and adolescents with severe
psychopathology also have significant strengths. Others have few” (p. 180). Thus,
strengths assessment in addition to risk assessment might be important for the prediction
of future pathways.

In another cross-sectional study, Cohen and colleagues (1988) divided clinically-
referred children ages 6 to 12 years into three competence groups (i.e., high, mixed and
low) based on whether competence scores were above or below the median on four out of
five measures that included IQ, reading, self-esteem, locus of control, and coping style
(the remainder was designated as the mixed group). Children in the high competence
group were rated by teachers and parents as having fewer problems and being more
competent socially and academically. Children in the mixed competence group were
perceived similarly to the high competence group by teachers and parents although they
came from more stressful lower income families similarly to the low competence group.
The authors suggested that a child’s competencies protected them against stress and the
development of syrnptomatology.

In a cross-sectional study, Luthar (1991) explored competence as a protective

factor for adolescents experiencing stressful life events. She found that “social

26

expressiveness”, a type of social skills, moderated the effects of stress on “sociability”,
which was defined as the child’s popularity with peers. In other words, under conditions
of high stress, adolescents with high social expressivity were considered significantly
more sociable by peers than adolescents with low social expressivity. It seems therefore,
that social skills and specifically “social expressiveness” functioned in a protective way

for adolescents experiencing high levels of stress.

Teacher-Student Relationships as 0 Protective Factor

Recent findings support the importance of relational factors, such as teacher- V
student relationships for children’s adjustment. Specifically, Hamre and Pianta (2001)
found that negative teacher-student relationships at Kindergarten significantly predicted
academic (i.e., grades and standardized achievement scores) and behavior (i.e., work
habits and disciplinary infractions) outcomes through fourth grade after controlling for
gender, ethnicity, verbal IQ, and teacher-reported behavior problems. These predictions
held through eighth grade indirectly with teacher-student relationships mediated by
students’ previous levels of performance on the academic and social outcomes.

Another longitudinal study showed that teacher-student relationships predicted
school adjustment beyond children’s aggressive risk status (Ladd & Burgess, 2001).
Consistent with an additive model of development, teacher-child closeness predicted high
cooperative participation and school liking. Moreover, teacher-child conflict contributed
to higher attention problems and behavioral misconduct. It is therefore evident that
aspects of teacher-child relationships are important predictors of psychological and

school adjustment.

27

Competence for children with Extemalizing versus Internalizing Problems

A cross-sectional study of intemalizing-externalizing profile comparisons showed
differences in competence scores for 6- to 11-year-old clinically-referred boys
(McConaughy, Achenbach, & Gent, 1988). Boys with internalizing profiles derived from
the parent reports on the Child Behavior Checklist (CBCL) had higher scores on Total
Competence, as well as Activities, and Social scores than boys with externalizing
profiles. However, they had no differences on the School scale. Similarly, teacher ratings
(TRF) showed that boys with internalizing profiles had higher scores on total adaptive
flmctioning, as well as on working hard, behaving appropriately, learning, and happiness.
No differences were found on the school performance scale. Observational scores showed
that boys with internalizing profiles had higher on-task scores than boys with
externalizing profiles. Cognitive assessment indicated that boys with internalizing
profiles had higher Full Scale IQ on the WISC-R compared to boys with externalizing
profiles. Consistent with the parent and teacher ratings, there were no significant
differences on Total Achievement scores on the PIAT achievement test between the two
profile groups. However, boys with internalizing profiles reported more negative
statements about their real self and had a greater disparity between their ideal and real
self. Overall, children with internalizing profiles had better cognitive, social, and total
adaptive functioning than children with externalizing profiles.

Another cross-sectional study of both competence and extemalizing-intemalizing
problems of a 6— to 12-year—old clinically-referred sample found that parents who rated
their child high on internalizing problems, they also rated them low on social competence

while parents who rated their child high on externalizing problems, they rated them low

28

on both social and school competence as measured by the Achenbach Child Behavior
Checklist (Cohen et al., 1988). Teachers rated high parent-reported externalizing students
as low on academic performance, peer sociability, rule compliance, and frustration
tolerance, while they rated high parent-reported internalizing students as low on peer
sociability only. Moreover, using a measure of profile similarity, the authors compared
their three competence groups (i.e., high, mixed, low competence) to Achenbach and
Edelbrock’s clinical standardization sample and they found that within their high
competence group, children were classified as having internalizing (81%) rather than
externalizing problems and within the low competence group, they were classified as
having externalizing problems (67%). There was a similar percentage of extemalizers and
intemalizers in the mixed competence group. This finding is consistent with other studies
of clinically and nonclinically referred children with internalizing problems who have
higher rated and observed competence (i.e., social and academic) than children with

externalizing problems.

Research Questions

The goal of this study was to explore the empirical links between competence and
externalizing and internalizing problems using structural equation modeling (see Figure 1
for conceptual model). Specifically, one of the original research goals was to use
confirmatory factor analysis to estimate the convergent and discriminant factor validity of
the constructs involved in the study (i.e., academic and social competence). However, the
correlations between the indicators of the latent variables were small and nonsignificant,

which led to the decision to use path analysis to examine three different models, one with

29

the self-reported measures of competence, one with the teacher-reported indicators of
competence, and an academic competence model (see Figures 2-4).
Therefore, this study examined the following research questions and hypotheses:
1. Do Time 1 constructs predict Time 2 constructs after controlling for Time 1
constructs? Specifically:
a. Does teacher-rated and self-rated social and academic competence at Time
1 predict teacher-rated externalizing and internalizing problems at Time 2
afler controlling for Time 1 externalizing and internalizing problems?
Specific hypotheses were:
i. Social competence will be a stronger predictor of internalizing
problems than will academic competence.
ii. Both social and academic competence will predict externalizing
problems.
b. Do teacher-rated externalizing and internalizing problems at Time I
predict teacher-rated and self-rated social and academic competence at
Time 2 after controlling for Time 1 social and academic competence?
Specific hypotheses were:
i. Intemalizing problems will predict social competence more
strongly than academic competence.
ii. Extemalizing problems will predict both social and academic
competence.
iii. Extemalizing problems will predict social and academic

competence more strongly than internalizing problems.

30

2. Do constructs show stability from Time 1 to Time 2?
a. Teacher-rated Extemalizing Problems at Time lwill strongly predict
Extemalizing Problems at Time 2.
b. Similarly, teacher-rated Intemalizing Problems at Time 1 will moderately
predict Extemalizing Problems at Time 2.
c. Teacher-rated and self-rated Academic Competence will show consistency
from Time 1 to Time 2.
(1. Similarly, teacher-rated and self-rated Social Competence will show
consistency from Time 1 to Time 2.
3. The relation between teacher-rated Intemalizing and Extemalizing Problems was
explored.
a. Do Extemalizing Problems at Time I predict Intemalizing Problems at
Time 2?
b. Do Intemalizing Problems at Time 1 predict Extemalizing Problems at
Time 2?
4. Finally, do child characteristics, such as gender, race, and grade level moderate

the relations in the models?

Addition to the Original Hypotheses

5. The relations between Teacher-Student Relationships and the rest of the variables

in the model were explored.

31

a. It was hypothesized that Teacher-Student Relationships will predict
Extemalizing and Intemalizing Problems as well as Social and Study

Skills both concurrently and longitudinally.

32

CHAPTER 3
METHOD

Participants

Children. Participants in the current study were 442 first to fifth grade students
(ages 7-12 years, girls=234). At Year 1 Fall, the sample consisted of 24% first graders,
24% second graders, 25% third graders, and 27% fourth graders, and was proportionately
representative of the four participating schools. At Year 2 Spring, the sample consisted
of 24% second graders, 24% third graders, and 25% fourth graders, and 27% fifth
graders. The participating school district has a large population of at-risk students, with
72% of the student body participating in the free or reduced cost lunch program and a
large percentage of each school coming from public housing units. The district has a 47%
on time graduation rate from high school. Similar to the school population, 57% (n=252)
of the students in this sample were African-American, 31% (n=136) Caucasian, 7%
(n=29) Hispanic, 3% (n=12) Asian/Pacific Islander and 2% (n=8) multi-racial. Poverty
and race are confounded in this district with more African-American students eligible for

the free or reduced cost lunch program than students from other racial groups.

Teachers. A total of 45 teachers in the schools participated in the study for at
least one year. The sample was 96% female, 84% Caucasian, and 14% Afiican-
American. The teachers were experienced in terms of number of years teaching, with
over half of the sample having taught for 6 years or more and only 4% in their first year
of teaching. Sixty percent had graduate degrees. The teachers were similar between the
four participating elementary schools with respect to demographic composition,

educational level, and experience.

33

Procedure

All children in regular education, first through fifth grade classrooms were
recruited for participation. Permission forms in students’ native language were obtained
for almost 90% of the school’s population. Teacher participants provided signed
informed consent and received a small stipend for their participation. Teachers completed
behavior-rating scales for each of their participating students twice: in mid fall of the
school year and in late spring of the following school year for three years. In this study,
two waves of data were used: Year 1 Fall and Year 2 Spring. Students completed self-
report measures at the same periods as the teacher ratings. The measures were displayed
on an overhead projector and read aloud to control for reading differences among the
students. The order of the measures was counterbalanced between classrooms to control
for order effects. Data from student records were also collected at the end of the school

year.

Measures

Extemalizing Problems and Intemalizing Problems. Teachers completed a
standardized behavior rating scale, the Behavior Assessment System for Children
Teacher Rating Scales for Children (BASC TRS-C) (Reynolds & Kamphaus, 1998).
Teachers reported the frequency of children’s problem and adaptive behaviors by
responding to 148 items on a 4-point scale. In this study, the Extemalizing Problems
Composite, which includes the Hyperactivity scale, Aggression scale, and Conduct

problems scale and the Intemalizing Problems Composite, which includes the Depression

34

scale, the Anxiety scale, and the Somatization scale were used to measure students’
externalizing and internalizing problems. Both scales have appropriate reliability
(Extemalizing Problems Composite: Internal Consistency reliability=.94, Test-retest
reliability=.91, Interrater reliability=.79 and Intemalizing Problems Composite: Internal
Consistency reliability=.90, Test-retest reliability=.8 l , Interrater reliability=.69) and
validity.

Social Competence. The Social Skills scale of the BASC TRS-C was used to
assess teacher-rated social skills. The scale has an internal consistency of 0.92, a test-
retest reliability of 0.94, and an interraterreliability of 0.70. Sample items for this
subscale include: “offers to help other children”, “tries to bring out the best in other
people”, “politely asks for help”, “encourages others to do their best”. The BASC Social
Skills scale and the Social Skills Rating System (SSRS) (Gresham & Elliott, 1990)
moderately overlap (Reynolds & Kamphaus, 2002).

The Peer Support scale combining two subscales from the Classroom Life
Inventory (Johnson, Johnson, & Anderson, 1983) was used to assess students’
perceptions of support from their peers. The Student Academic subscale, which measures
academic support from peers and the Student Personal subscale, which measures caring
and interpersonal support from peers will be used to create the Peer Support scale. Each
of these subscales contains 4 items that use a 5 point Likert-type scale. The Cronbach
alpha reliabilities for student academic and personal support are .67 and .78 respectively.
Sample items include: “In this class other students like to help me learn”, “want me to do
my best schoolwork” and “In this class other students like me the way I am”, “care about

my feelings” for the peer academic and peer personal support subscales respectively.

35

Academic Competence. The modified Scholastic Competence subscale of the
Self-Perception measure (Harter, 1985b) was used as an indicator of self-reported
academic competence. The subscale includes 6 items of academic self-concept and asks
students to rate themselves in a 1-4 Likert-type format. The Internal Consistency
Reliability of the original measure is acceptable (Cronbach’s a=.82). Sample items
include: “I am very good at my schoolwork”, “I often forget what I learn”, and “I am
smart enough to do my schoolwork”.

The Study Skills scale of the BASC TRS-C was used as an indicator of teacher
rated academic competence. The scale has an internal consistency reliability of 0.93, a
test-retest reliability of 0.94, and an interrater reliability of 0.88. Sample items for this
subscale include: “reads assigned chapters”, “completes homewo ”, “asks to make up
missed assignments”, and “works hard even in courses he or she does not like”.

T eacher—Student Relationship Quality. Nine items from the Student-Teacher
Relationship Scale (STRS, Pianta & Nirnetz, 1991) were used as a measure of teacher-
child relationship quality. The scale includes both close, warm relationship items and
negative reactivity items, which were summed so that high scores represented positive
relationship quality. In this study, the internal consistency reliability of the measure was
.87. This brief measure was used to reduce the length of the teacher battery that was used
in the larger study. Baker (2005) suggests that this short measure does not represent the
three-factor structure of the original measure that was based on the attachment theory and
thus, it should not be compared directly with studies using the full STRS (e.g., Hamre &

Pianta, 2001). Instead, she suggests that it is conceptualized as a general measure of

36

relationship quality related to the concept connectedness as conceptualized by Ryan and

Deci (2000).

Data Analysis

Multiple data analyses were used to explore the above hypotheses. Initially,
descriptive analysis was conducted to examine the data regarding entry errors, patterns of
missing data, univariate and multivariate outliers, multicollinearity, and evidence for
univariate (i.e., skewness and kurtosis) and multivariate normality.

To explore the longitudinal paths between competence and psychopathology, a
structural equation modeling (SEM) using the AMOS 5 program (Arbuckle, 1997) was
used. SEM allows a researcher to specify apriori the relationships among the variables of
interest and examine direct and indirect effects among many difl’erent variables (both
exogenous and endogenous). Thus, it is a powerful technique that allows exploration of the
direction of the effects of a variety of variables together. Specifically, a path analysis was
used since the original hybrid model was abandoned for the reasons mentioned before.

Theoretical identification of the model was necessary before proceeding with the
analysis. In Models 1-4, the number of observations exceeded the number of free
parameters (e.g., in Model 1: v(v+1)/2=8"‘9/2=36, where v is the number of observed
variables, so number of observations (36)>number of parameters (31). Therefore, the
models were theoretically identified.

Several indices were used to assess the fit ofthe model based on Kline’s (1998)
recommendations. These indices include the generalized likelihood ratio (,6), the

generalized likelihood ratio adjusted (xz/df), Goodness of Fit Index (GFI), Comparative Fit

37

Index (CPI), Normed Fit Index (NF I), and Root Mean Squared Error of Approximation
(RMSEA). These six indices assess different aspects of model fit and have different criteria
for deciding when a model is considered to have good fit. In addition to the fit indices, I
inspected the correlation residuals (i.e., differences between observed and model-implied
correlations). Absolute values of correlation residuals less than .10 were desirable (Kline,
1998).

Finally, with a sample size of 442, there was sufficient power to stably estimate a

structural equation model (Chou & Bentler, 1995).

Missing Data Estimation

The current sample was drawn from a sample of 1,131 first through fifth grade
students from four elementary schools in a small city in the Southeast who participated in
a three-year cross-sequential longitudinal study. Each year, approximately 600 students
fiom fust to fifth grade were targeted. From these 600 students, approximately 450
students were followed fi'om Year 1 Fall to Year 2 Spring. However, only 192 had
complete data on all the variables in this study for both waves of data. Therefore, a
missing data estimation procedure using SYSTAT 10 Expectation Maximization (EM)
method was applied to estimate missing data The EM method is a full information
method of irnputing missing data by iterating through the existing data and fitting the best
values to the existing covariance structure (T aris, 2000). This method of handling missing
values produces less bias in the results than deleting cases or using the sample mean for

imputation (Rovine & Delaney, 1990; Taris, 2000).

38

In order to estimate missing values, one must be able to assert that data were
missing at random (MAR), which means that for any single variable, the pattern of
missing values did not rely on the subject’s true status on that variable (Kline, 1998). In
this study, I checked whether the data were missing at random or systematically by
creating dichotomous variables coded as missing-not missing for each variable in the
models. Then, these variables were correlated with all the variables in the study. Small
correlations indicated that the data were not missing systematically.

For the estimation, the data for the 6 main variables of this study (Extemalizing
Problems, Intemalizing Problems, Social Skills, Study Skills, Peer Personal Support,
Scholastic Competence) and the 3 demographic variables (child sex, grade level, and
ethnicity) were imputed together using 3 waves of data (i.e., Year 4, Year 5 Fall, and
Year 5 Spring). Only students who missed one wave of data (fi'om 300 to 440 students
depending on the variable) were used so that the estimation would be based on
information fiom at least two waves of data. The estimated sample consisted of 442
students from grades 1 to 5. The estimation of this data was based on Little MCAR test
statistic of 695.11 (df=683, [P 0.365), indicating that the data were missing completely at
random. No major changes existed in the means and standard deviations of the estimated

variables compared to the original variables (see Tables I & 2).

39

Table 1.

Descriptive Statistics for Study Variables Pre-Irnputation

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables N Mean St. Deviation
Extemalizing Problems] 573 52.27 11.82
Intemalizing Problems] 573 49.76 10.49
Social Skills] 573 50.22 11.02
Study Skills] 573 48.28 10.71
Peer Support] 533 4.79 2.1
Scholastic Competence] 537 6.43 1.7
Teacher-Student Re]. I 573 37.98 7.03
Extemalizing Problems2 44] 52.73 12.41
Intemalizing Problems2 44] 49.46 10.65
Social SkillsZ 44] 50.24 11.88
Study SkillsZ 44] 47.7 10.63
Peer Support2 479 4.49 2.12
Scholastic Competence2 476 6.44 1.55
Teacher-Student Rel. 2 430 37.1 7.46

 

40

 

Table 2.

Descriptive Statistics for Study Variables Post-Imputation 0V =4 42)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variables Mean SD Skew Kurtosis Median Range
Extemalizing Problemsl 51.96 11.19 1.2 1.06 49 40-95
Intemalizing Problemsl 49.36 10.02 2.0 5.4 47 39-10]
Social Skills] 51.45 10.51 -.02 -.46 51 26-77
Study Skills] 49.46 10.27 .02 -.70 49 26-73
Peer Support] 4.93 1.91 -.43 .04 5 0-8

Scholastic Competence] 6.61 1.32 -1.15 1.4 7 1-8

Teacher-Student Rel. 1 38.51 6.4 -l . 13 .74 41 14—45
Extemalizing Problems2 52.45 12.37 1.4 2.1 48 37-105
Intemalizing ProblemsZ 49.1 10.15 1.28 1.46 46 37-95
Social SkillsZ 50.69 11.7 .07 -.88 50 26-77
Study SkillsZ 48.36 10.79 .05 -l .07 48 26-73
Peer Support2 4.49 2.03 -.10 -.37 4 0-8

Scholastic CompetenceZ 6.44 1.49 -.53 -.95 7 2-8

Teacher-Student Rel. 2 37.61 7.23 -1.01 .54 40 10-45

 

41

 

 

 

 

 

 

 

 

 

 

 

Social
Skills
EP1
EP2
Scholastic
Competence

 

 

Figure 1. Conceptual Model

 

Peer
Support

 

 

 

Social
Competence1

Social
Competence

Academic
Competence1

Academic
Competence

\

Slit?!

 

 

 

 

42

 

 

1P1

 

1P2

 

 

 

 

EP1

 

 

 

EP2

 

 

 

Suppom

 

 

 

 

Peer
Suppor12

 

 

 

 

Scholastic

 

Coupeten

 

 

 

Scholastic
CoupetenZ

 

 

 

Figure 2. Conceptual Model for Self-Perceived Competence

43

 

 

1P1

 

 

 

 

 

Social
Skillst

 

 

 

1

 

Social
Slu'llsz

 

 

EP1

 

 

 

 

 

 

1

EP2,

 

 

1P1

 

 

’1 .l

 

 

Figure 3. Conceptual Model for Teacher-Reported Competence

Study
Slolsl

 

P2

 

 

 

 

Study
Skillsz

 

 

 

 

 

CHAPTER 4
RESULTS
Data Screening and Descriptive Analyses

Table 3 displays the correlations among all the variables in the models, as well as
their means and standard deviations. Correlations ranged fiom a low of .01 to a high of
.79. The self-report measures of peer support and scholastic competence had the lowest
correlations with the other variables in the models and they were nonsignificant most of
the times.

Visual inspection of the data and tests of univariate normality suggested no
significant violations of assumptions. Skewness ranged from -1.13 to +1.98 and kurtosis
ranged from -1.06 to +2.04 with the exception of Intemalizing Problems 1, which was
+5.33 (see Table 2 for skewness and kurtosis values). No problems with multicollinearity
were detected. Tests for multivariate outliers via calculation of Mahalanobis distances
revealed many observations that were falling far (p<.05) firm the centroid assuming a
multivariate normal distribution. Inspection of the raw data for these cases indicated no
errors in data entry. Given that the sample comes from a low income at-risk population,
these cases were judged to represent real variability in the sample and thus, they were
retained for the analyses.

In the following section, four models are presented. Model 1 examines the
relations between self-reported competence and extemalizing/intemalizing problems.
Model 2 examines the relations between teacher-reported competence and
extemalizing/intemalizing problems. Model 3 explores the relations between teacher-

reported study skills and self-reported scholastic competence and

45

extemalizing/intemalizing problems. Finally, in Model 4, a new variable, that of teacher-
student relationship is entered and the relation among this variable and the rest of the

variables in the model is explored.

Model 1: Self-Reported Competence

Model 1 tested the relations between teacher-reported externalizing and
internalizing problems and self-reported competence (i.e., peer support and scholastic
competence) (see Figure 4 for Model 1). The original model did not fit the data until
correlations between the errors were added (see Table 4 for Fit Indices). The correlations
between the errors indicate same rater bias. The same teacher rated both Extemalizing
and Intemalizing Problems at Time 2. Similarly, the same student reported both peer
support and scholastic competence. The error correlation between Intemalizing Problems
and Scholastic Competence indicates that there are common omitted causal variables
from the model. The model with the added error correlations fit the data well as indicated

by large GFI, NF 1, CF 1, small RMSEA, and small residuals (see Table 4).

46

tam—Ea u.

flcxxaEaeE. e\Sm Valuing Siamese S :8 Sonic. Entm»

 

55928

N

a

a q

5

Z

5

C

3

 

5.
Z.
5.
G.

3.

mfioBmmuem
w:::«:5_
Esau—ER
vac—nan.
moan—
maze:
maav.
maze:

woe.
mauve:
manage
98an
manage—55m
fizzzonau
rzuazzmmsm
wzzzoaau
moan.
ww=w~
mweaw
2&sz

moon
wagon»
more—ammo
02:an

agreemaaoa

”3&0:ng _

Honoratmaaaa

awn—eager”? n

:32. .38

—
.gii
amuse
Jeane
.oq
sou
.omno
.uuee
sumac
Joann
.o_
J_Aee
emcee

duo:

reea.
ra_..
._m..
.oc
.Na..
.aa..
r~_.
r~a..

.8:

due:

:3:

.3:
.3:

.3:

~

:8 he:
.3 duo:
.3: Aug...
.8 .S
2.32.. in.
.8 the:

.8 :3:

that.

tho...

r_a..
sno..

aac..

.3:
.ann
.32..
.3:

.80!

.oo
.uc..
.au..

.3:

.3:
.3

.3:

_
Lo.

.3:

.aa..

_

47

Table 4.
Goodness of F it Summary for Model 1 flV=442)

 

 

 

 

Model [(dt) {/df GFI AGFI NFI CFI RMSEA

SelfReport 1795" 22.4 .92 .64 .79 .79 .22
(8)

SelfReport 2.5 (5) .5 .99 .99 .99 1.00 .00

w/ Error

Correlations

*p<.001

Several findings regarding the direction and strength of the path coefficients are
noteworthy. The answer to the first research question: “Does social and academic
competence at Time I predict externalizing and internalizing problems at Time 2 after
controlling for Time 1 externalizing and internalizing problems?” is no. Self-reported
peer support and scholastic competence in the Fall of Year I did not predict externalizing
and internalizing problems in the Spring of Year 2 (see Figure 4 for path coefficients).

The answer to the second research question: “Do externalizing and internalizing
problems at Time I predict social and academic competence at Time 2 after controlling
for Time 1 social and academic competence?” is partly yes. Intemalizing problems
significantly predicted scholastic competence but not peer support. Similarly,
Extemalizing problems significantly predicted scholastic competence but not peer
support. Extemalizing problems did predict academic competence more strongly than
internalizing problems did (see Figure 4).

The third research goal was to examine the consistency of all the constructs fiom
Time 1 to Time 2. All hypotheses were supported. Extemalizing Problems at Time 1
sn-ongly predicted Extemalizing Problems at Time 2. Similarly Intemalizing Problems at

Time I predicted Intemalizing Problems at Time 2, although this relation was less strong

48

than that of the Extemalizing Problems. It is noteworthy that Academic and Social
Competence showed consistency fiom Time 1 to Time 2, as well (see Figure 4 for
specific paths).

The last research goal was to explore the relation between Intemalizing and
Extemalizing Problems. Extemalizing Problems at Time 1 significantly predicted high
Intemalizing Problems at Time 2. However, the opposite was not true. Intemalizing

Problems at Time 1 did not predict Extemalizing Problems at Time 2.

49

.07

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.18***
"a Peer
'44 Support]
.45“
._05 .21
.00 m.
-.05 pee, -19
Support2
EP1 '07
-.07 .79 “’1
71}- "°6 1 137*"
47 .15” 21
EP2 1P2
-.03
.01
Scholastic .09
Competen
.53 .79
.31‘” .11“
-.1
.22 3
Scholastic
'87 Competen2
-.14***

Figure 4. Standardized Path Coefficients and Unexplained Variance for Model 1: Self-
Reported Competence with Error Correlations.

(m p<.001, 1" p<.01, *p<.05)

50

The second phase of the modeling analyses explored the generalizability of the
model across different groups of students (i.e., gender, race, and grade level). First of all,
I explored whether gender moderated the relations specified in the model. One path
analysis was conducted using the two separate groups and having the paths unconstrained
(see Table 5 for Fit Indices for the unconstrained model). Then, a second analysis was
conducted after imposing cross-group equality constraints (see Table 5 for Fit Indices for
the invariant model). The change in the overall chi-square (3(sz (14)= 19.353) was not
statistically significant at p= .05, which means that the path coefficients as a set did not
differ significantly between boys and girls. Therefore, the original model (i.e., Model 1)
can be used for both boys and girls.

Table 5.

Goodness of F it Summary for Multiple Group Analysis for Model I based on Gender
(Boys N=208, Girls N =234)

 

 

 

 

Model 2? df {/df GFI AGFI NF] CF] RMSEA
3:32?“ 2&3)" 24 1.082 .986 .957 .971 .998 .014
33%“ ‘23 10 .662 .996 .973 .993 1.00 .000
11333:“? 19353 14

models (as)

 

ns=non significant at p= .05

Second, I explored whether race moderated the relations specified in the model.
Similar analyses were conducted. In Table 6, the Fit Indices for both the unconstrained
and constrained models are presented. The change in the overall chi-square (xzdjfl‘m
(14): 24.47) was statistically significant at p= .05, which means that the path coefficients

as a set differed significantly between African-American and White students. Therefore,

51

two separate models were further explored (see Figures 5 and 6). Fit coefficients for the
two different models are presented in Table 7.
Table 6.

Goodness of F it Summary for Multiple Group Analysis for Model I based on Race
(Afiican-American N =252, White N =136)

 

 

 

 

 

 

 

 

 

Model x2 df {/dt‘ GFI AGFI NF] CFI RMSEA
Invariant ,,

Model 39.36 24 1.64 .98 .93 .95 .98 .041
Unconstrained 14.89

Model (ns) 10 1.489 .99 .93 .98 .99 .036
Difference

btw the 2 24.47“ 14

models

ns=non significant

‘p= .05

Table 7.

Goodness of F it Summary for Model I by Race (Afiican-American N =25 2, White N =1 3 6)
Model 2:2 df f/df GFI AGFI NFI CFI RMSEA
Afiican- 2.33 5 .47 .998 .983 .995 1.00 .000
American (ns)

White 12.5“ 5 2.5 .98 .84 .95 .97 .1 1

ns= non-significant

*p<.05

52

.08

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.17“
m Peer
-43 Supportt
.45“
-10 .22
.00 m
Support2
.09
EP1
7 ‘09 .78 1P1
3+ "04 S [.37‘“
.42 .13, .20
EP2 IP2
-.O1
.02 ,
§cholastic .08
tmpeten
.58 .80
27*” .12
-.23
.30 3
plastic
35 ompeteni?
_.22***

Figure 5 Standardized Path Coefficients and Unexplained Variance for African-
American Students Model 1: Self-Reported Competence with Error Covariances.

(1m p<.001, " p<.01, *p<.05)

53

.21

 

.02

Figure 6. Standardized Path Coefficients and Unexplained Variance for White Students
Model I: Self-Reported Competence with Error Correlations.

(Im p<.001, H p<.01, *p<.05)

54

Finally, I explored whether grade level moderated the relations specified in the
model. Specifically, the sample was divided into two groups: a) 1"‘-2"d grade students at
Time 1 were 2“"-3rd grade students at Time 2 and b) 3m-4d‘ grade students at Time 1 were
4m-5'h grade students at Time 2. Analyses similar to gender and race were conducted for
grade level. In Table 8, the Fit Indices for both the unconstrained and constrained models
are presented. The change in the overall chi-square (xzdim (14): 22.21) was not
statistically significant at p= .05, which means that the path coefficients as a set did not
differ significantly between early elementary and late elementary students. Therefore, the
same model can be used for both groups (i.e., Model 1).

Table 8.

Goodness of F it Summary for Multiple Group Analysis for Model I based on Grade Level
(1-2 Grade at Time 1 N=21], 3-4 Grade at Time I N=23I)

 

 

 

 

Model {(dt) df f/df GFI AGFI NF] CFI RMSEA
3:32? 2;: 24 1.14 .99 .95 .97 .996 .018
33W” :33 10 .53 .997 .98 .99 1.00 .000
533:“? 2221 14

models (ns)

 

ns=non significant at p= .05

Model 2: Teacher-Reported Competence

Model 2 tested the relations between teacher-reported externalizing and .
internalizing problems and teacher-reported competence (i.e., social skills and study
skills) (see Figure 4 for Model 2). Similarly with Model 1, correlated errors were allowed
between pairs of measures that used the same method/rater. The model with the added

error correlations fit the data adequately as indicated by the GP], NP], and CF]. However,

55

the other fit indices are not satisfactory: the x2 is significant, the x2 /df is above 3, the
RMSEA is above .05 and the AGFI is marginal (see Table 9 for Fit Indices). The

correlation residuals are less than .10, which means that the fit of different parts of the

 

 

 

model is relatively good.

Table 9.

Goodness of F it Summary for Model 2(N=442)

Model x1 (dt) )8 /df GFI AGFI NFI CFI RMSEA
Teacher 1 1.57“ 5.79 .99 .88 .995 .996 .104
Report w/ (2)

Error

Correlations

‘p<.01

Several findings regarding the direction and strength of the path coeflicients are
noteworthy. The answer to the first research question: “Does social and academic
competence at Time I predict externalizing and internalizing problems at Time 2 after
controlling for Time 1 externalizing and internalizing problems?” is partly yes. Teacher-
reported study skills in the Fall of Year I predicted externalizing and internalizing
problems in the Spring of Year 2 (see Figure 4 for path coefficients).

The answer to the second research question: “Do externalizing and internalizing
problems at Time 1 predict social and academic competence at Time 2 after controlling
for Time ] social and academic competence?” is partly yes. Intemalizing problems
significantly predicted social skills but not study skills. Extemalizing problems did not
predict any of the competence indicators (see Figure 4 for path coefficients).

The third research goal was to examine the consistency of all the constructs from

Time 1 to Time 2. All hypotheses were supported. Extemalizing Problems at Time 1

56

strongly predicted Extemalizing Problems at Time 2. Similarly Intemalizing Problems at
Time I predicted Intemalizing Problems at Time 2, although this relation was less strong
than that of the Extemalizing Problems. It is noteworthy that Social and Study Skills
showed strong consistency fi'om Time 1 to Time 2, as well (see Figure 4 for specific
paths). These findings are consistent with the Model 1 findings.

The last research goal was to explore the relation between Intemalizing and
Extemalizing Problems. Extemalizing Problems at Time ] significantly predicted
Intemalizing Problems at Time 2. However, the opposite was not true. Intemalizing
Problems at Time 1 did not predict Extemalizing Problems at Time 2. These findings are

consistent with the Model ] findings.

57

-.63

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-.32
Social
Skills1
.5
.31 -08
.79
Social
SklllSZ .1o*
EP1
.60 1 1P1
, - 09 .
.ofi 34*“
.48 07* 23
EP2 ‘33 1P2
-.2
-.1s** Study -.41
Skllls1
.52 -57 .77
.71***
.67 _ 04 -°6
1
Stud
.49 " -
_.39 Skllls
-.26

Figure 7. Standardized Path Coefficients and Unexplained Variance for Model 2:

Teacher-Reported Competence. (All correlation between the exogenous variables and the
errors were significant at p<.001)

(*" p<.001, " p<.01, *p<.05, Significant Paths are in bold)

58

Similarly with Model 1, the second phase of the modeling analyses explored the
generalizability of the model across different groups of students (i.e., gender, race, and
grade level). In Model 2, gender moderated the relations specified in the model, while
race and grade level did not (see Table 10 for Fit Indices). Therefore, separate models
were explored only for boys and girls (see Figures 5 & 6 for Models and Table 11 for Fit
Indices). Several findings were noteworthy for boys. First of all, similarly to the total
sample model, Study Skills at Time 1 continued to significantly predict Extemalizing and
Intemalizing Problems at Time 2. In addition, Social Skills at Time 1 significantly
predicted Extemalizing and Intemalizing Problems at Time 2. All variables showed
strong consistency fiem Time 1 to Time 2. The model for girls differs from the boys’
model and the total sample model in that only Extemalizing Problems at Time 1
significantly predicted Social Skills at Time 2.

Table 10.

Goodness of F it Summary for Multiple Group Analysis for Model 2 based on Gender
(Boys N =208, Girls N =234)

 

 

 

 

Model {(dt) df ledf GFI AGFI NFI CFI RMSEA
Invariant "

Model 37.13 18 2.063 .98 .92 .98 .99 .049
Unconstrained "

Model 13.33 4 3.33 .99 .87 .99 .996 .073
Difference

btw the2 2380* 14

models

 

("1* p<.001, ** p<.01, *p<.05)

 

 

 

Table 11.

Goodness of F it Summary for Model 2by Gender (Boys N =208, Girls N =23 4)

Model 12 (df) df 1: /df GFI AGFI NFI CFI RMSEA
Boys 679* 2 3 4 .99 .86 .99 .995 .11

Girls 65* 2 3.27 .99 .88 .99 .996 .1

 

59

ns= non-significant
*** p<.001, **p<.01, *p<.05

-.58

 

 

 

 

 

 

 

EP1
' { lab
EP2
-.19*
.55 -.64
.63
.50
-.36

 

.70

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-.37
Social
Skills1
.5
.30 .19
.75
Social
SkillsZ .03
1P1
-.04 -
.41“
.10 .29
“-35 1P2
-.2
Study -.43
Skills1
.71
.77“ .07
0
Studé
Skllls
-.19

Figure 8. Standardized Path Coefficients and Unexplained Variance for Boys Model 2:

Teacher-Reported Competence.

(*** p<.001, ** p<.0], *p<.05, Significant Paths are in bold.)

60

-.61

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Social
Skillet
.27
Social
SkillsZ
EP1
.73
.6
.44 .01
EP2 --42
-.08 Study
Skills1
.56 -.64
.64‘”
.69 _.1 1
1 7
Stud
.53 °
_.43 Skills

 

 

-.0

.10

-.08

-.17

.06

-.35

 

 

 

 

 

 

 

-.36
.79
1P1
.31***
.22
1P2
-.48
.79

Figure 9. Standardized Path Coefficients and Unexplained Variance for Girls Mode] 2:

Teacher-Reported Competence.

(m p<.001, ** p<.01, *p<.05, Significant Paths are in bold.)

61

Model 3: Academic Competence Model (T eacher-Reported Study Skills & Self-Reported
Scholastic Competence)

Model 3 tested the relations between teacher-reported externalizing and
internalizing problems, teacher-reported study skills and self-reported scholastic
competence (see Figure 7 for Model 3). Similarly with the previous models, correlated
errors were allowed between pairs of measures that used the same method/rater. The
model with the added error correlations fit the data adequately as indicated by the fit

indices in Table 12.

 

 

Table 12.

Goodness of F it Summary for Model 3(N =4 42)

Model {(dt) f/df GFI AGFI NF] CFI RMSEA
Academic 7.18(3) 2.39 .996 .95 .995 .997 .056
Competence (ns)

Model

 

ns= non-significant

62

-.67***

 

 

 

 

 

 

 

44*"
-.1
EP1
.GH.O7
481 EP2
+ .04
.52
-.39***
.24 *
.88

-.02

.48

Study
Skills1

 

34*“

.52

 

Stud
Skills

 

 

 

_. 26***

.05

Scholasti
Competen

 

-.1 .29“

 

Scholastic

 

 

 

 

-.‘l

.06 ‘13

 

 

 

 

-.09*

 

.00

 

 

.09
.77

.11*

Competen2.- 12

_.14***

Figure 10. Standardized Path Coefficients and Unexplained Variance for Model 3:

Academic Competence.

as” p<.001, u p<.Ol, *p<.05 (Significant Paths are in bold.)

63

Several findings regarding the direction and strength of the path coefficients are
important in this model. Study Skills at Time 1 significantly predicted both Extemalizing
and Intemalizing Problems at Time 2 while Extemalizing and Intemalizing Problems at
Time I predicted self-perceived Scholastic Competence at Time 2.

A finding that was consistent in all three models involved the consistency of the
constructs fi'om Time 1 to Time 2. Again, Extemalizing Problems at Time 1 strongly
predicted Extemalizing Problems at Time 2. Similarly Intemalizing Problems and
Scholastic Competence at Time I predicted Intemalizing Problems and Scholastic
Competence at Time 2 respectively, although these relations were less strong than that of
the Extemalizing Problems and Study Skills. It is noteworthy that Study Skills showed
strong consistency from Time 1 to Time 2 (see Figure 7 for the specific paths).

Finally, the relation between Intemalizing and Extemalizing Problems was
reversed in this model compared to the previous models. Intemalizing Problems at Time
] significantly predicted Extemalizing Problems at Time 2.

Again, the second phase of the modeling analyses explored the generalizability of
the model across different groups of students (i.e., gender, race, and grade level). In this
model, similarly to Model 2, gender moderated the specified relations, while race and
grade level did not (see Table 13 for Fit Indices). Therefore, separate models were
explored only for boys and girls (see Figures 8 & 9 for Models and Table 14 for Fit
Indices). Several pathways were important for boys. First of all, similarly to the total
sample model 3 and to the model 2, Study Skills at Time 1 significantly predicted
Extemalizing Problems at Time 2. Consistently with previous models, all variables

showed strong consistency from Time 1 to Time 2. In the model for girls, study skills

negatively predicted Intemalizing Problems at Time 2 while Time 1 Extemalizing

Problems negatively predicted Scholastic Competence at Time 2. An interesting finding

was that Intemalizing Problems at Time 1 positively predicted Scholastic Competence at

Time 2. Similarly to all previous models, all variables showed strong consistency from

Time 1 to Time 2.

Table 13.

Goodness of F it Summary for Multiple Group Analysis for Model 3 based on Gender

(Boys N =208, Girls N =234)

 

 

 

 

 

 

 

 

 

Model {(dtl df {/df GFI AGFI NFI CFI RMSEA

Invariant "

Model 39.47 20 1.97 .98 .92 .97 .99 .047

Unconstrained *

M ode] 14.72 6 2.45 .99 .90 .99 .99 .033

Difference

btw the 2 24.75* 14

models

(***p<.001, "p<.01, *p<.05)

Table 14.

Goodness of F it Summary for Model 3 by Gender (Boys N =208, Girls N =23 4)

Model {(dtl df {/df GFI AGFI NFI CFI RMSEA

Boys 9.4* 3 3.13 .99 .87 .99 .99 .102

Girls 5.32 3 1.77 .99 .93 .99 .997 .058
tnS)

ns= non-significant

*** p<.001, ** p<.01, *p<.05

65

.42 V. -43

 

Study
Sk1||s1

.80“
.51

 

Stud
Skills .03 25

 

 

 

t
“H'io. 1:.
.441 EP2 Rag Ali”. \0

- Scholastic
V Competen

6 ”3 6

-.09 .34*** .00

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    
    

 

Scholastic
Q Competen2-14

 

 

 

-.08

Figure I I . Standardized Path Coefficients and Unexplained Variance for Boys Model 3:
Academic Competence.

m p<.001, ** p<.01, *p<.05 (Significant Paths are in bold.)

66

 

EP1

NH

-44 EP2

 

 

 

 

 

.02

.56
-.41

.22

.55

-.12

.87

.52

 

Study
Skllls1

-.48

.65”

.48

-.23

 

Stud
Skills

 

-.10

.06

 

.06

 

-.34

.02

 

Scholastic
Dampeten‘

 

 

 

-.26 25*“

 

Scholastic
30mpeten1

 

 

 

 

 

 

 

1P1
-.10
.29***
.22

l 2
.08

 

 

.13
.78_

.23”

3.13

-.17

Figure 12. Standardized Path Coefficients and Unexplained Variance for Girls Model 3:

Academic Competence.

*** p<.001, ** p<.01, *p<.05 (Significant Paths are in bold.)

67

Model 4: Teacher-Student Relationship Model

In Model 4, teacher-student relationships were added as a positive indicator of
school adjustment and competence (see Figure 13 for Model 4). Similarly with the
previous models, correlated errors were allowed between pairs of measures that used the
same method/rater. The model with the added error correlations fit the data adequately as
indicated by the fit indices in Table 15. There are several important findings in this
model. Student-Teacher Relationships strongly predicted low Extemalizing and
Intemalizing Problems, as well as high Social and Study Skills concurrently at both Time
1 and 2. More importantly, positive teacher-student relationships at Time 1 significantly
predicted high social and study skills and low externalizing and internalizing problems at
Time 2 through the mediating effects of prior levels of social and study skills and low
externalizing and internalizing problems respectively (see Table 16 for Indirect Effects).

Moreover, Extemalizing Problems at Time 1 directly predicted low Teacher-
Student Relationships at Time 2 and Social and Study Skills at Time I predicted positive
Teacher-Student Relationships at Time 2. Similarly with previous models, Extemalizing
Problems at Time 1 indirectly predicted low study and social skills through the mediating
effects of prior levels of social and study skills. Intemalizing problems at Time ]
indirectly predicted low study skills through the mediating effects of prior levels of study
skills. However, Intemalizing problems at Time 1 directly predicted high social skills at
Time 2. Finally, consistently with all previous models, all variables showed consistency
from Time 1 to Time 2 with one exception. Teacher-Student Relationship at Time 1 did

not predict Teacher-Student Relationship at Time 2.

68

Table 15.

Goodness of F it Summary for Model 4-Teacher-Student Relationships (N =4 42)

 

 

 

 

 

 

 

 

 

 

Model {(dt) {/df GFI AGFI NFI CF] RMSEA
Teacher- 14. 14 1 .09 .99 .97 .995 1 .00 .014
Student Rel (13) (ns)
Model
ns=non-significant
Table 16.
Indirect Efl'ects of Teacher-Student Relationships
Outcomes

Predictor Extemalizing Intemalizing Social Study

Problems 2 Problems 2 Skills2 SkillsZ
Teacher— -.3]**"‘ -.16*** .15“ .15***
Student R 1
Extemalizing - - -.12*"”" -.l6***
Problemsl
Intemalizing - - .03“ -.04*“
Problems] ~
*" p<.001, **p<.01, *p<.05

69

Tea Rel1

.07

.69

 

Skills A

—.12

53/

Figure 13. Standardized Path Coefficients and Unexplained Variance for Teacher-
Student Relationship Model 4. (All correlations between exogenous and errors were
significant at least at p<.01).

"* p<.001, *" p<.01, *p<.05 (Significant Paths are in bold.)

70

CHAPTER 5
DISCUSSION

The purpose of this study was to evaluate the longitudinal relations between
externalizing-internalizing problems and competence indicators (i.e. academic and social
competence and teacher-student relationships) from Fall of Year 1 to Spring of Year 2 for
a low-income, predominantly African-American sample of first to fiflh graders. Path
Armlysis allowed for the exploration of the relations among these variables
simultaneously in the same model. Moreover, it allowed for the prediction of Time 2
constructs after controlling for Time 1 constructs. The models tested in this study provide
a number of insights regarding the relations among these constructs.

Five major sets of findings emerged. First, externalizing problems at Time I
consistently predicted low scholastic self-competence at Time 2 in almost all models. On
the other hand, study skills at Time 1 moderately predicted low externalizing problems at
Time 2. A surprising finding was that internalizing problems at Time I predicted positive
scholastic self-competence at Time 2. On the other hand, study skills at Time 1
moderately predicted low internalizing problems at Time 2. Third, externalizing problems
at Time 1 moderately predicted high internalizing problems at Time 2 in the majority of
the models. On the other hand, internalizing problems at Time I predicted low
externalizing problems at Time 2 in the academic competence model. Fourth,
competence indicators (i.e., social skills, study skills, peer support, and scholastic
competence) and extemalizing-intemalizing problems showed strong continuity from
Time 1 to Time 2. Finally, important findings emerged from the follow-up model of

teacher-student relationships. It was found that positive teacher-student relationships

71

significantly predicted high social and study skills and low externalizing and internalizing
problems almost two years later through the mediating effects of prior levels of social and
study skills and externalizing and internalizing problems respectively. Moreover,
externalizing problems directly predicted lower teacher-student relationships while study
skills predicted positive teacher-student relationships 18 months later. Each of these

findings is elaborated below.

Extemalizing Problems and Competence (Models [-3)

One of the goals of this study was to clarify the relation between externalizing
problems and competence indicators, such as academic and social competence. Several
significant findings are worth our attention. First, it was found that earlier externalizing
problems consistently predicted Iow scholastic self-competence across all models. It
seems that externalizing problems influence the perception that a child has about his or
her academic competence. Students with behavior problems often receive little positive
reinforcement, are more frequently punished and have conflictual relationships with their
teachers, which affect the way they view themselves. It is also possible that externalizing
problems interfere with academic achievement and thus, children are actually less
competent academically (Hinshaw, 1992). On the other hand, low scholastic competence
did not predict later externalizing problems. Perceiving oneself as low in scholastic
competence does not lead a child to exhibit behavior problems.

It is interesting that externalizing problems did not predict later study skills while
high study skills predicted low externalizing problems 18 months later. This finding

suggests that being a student with good academic skills protects against the development

72

of behavior problems at school. Previous studies have found that good academic skills
reduce the risk of disorder (Rae-Grant, Thomas, Offord, & Boyle, 1989). Stanger and
colleagues (1993) found that high academic competence predicted lower school behavior
problems and mental health services as reported by parents three years later. Similarly,
Thorpe and colleagues (2000, August) found that study skills significantly predicted
lower behavior problems for girls two years later.

Extemalizing problems predicted lower peer support for white students and lower
social skills for girls only. However, they were not related to social skills and peer
support for the sample as a whole. Previous studies have been unable to find significant
effects of externalizing problems on social competence. For instance, a longitudinal study
found that prosocial behavior as rated by self, peers, and teachers predicted peer social
preference (i.e., peers selection of whom they like to play or study with) and academic
achievement 5 years later while aggression did not (Caprara, Barbaranelli, Pastorelli,
Bandura, & Zimbardo, 2000). It might be that externalizing problems influence social
competence through the mediating effects of prior levels of externalizing problems.
Future studies should explore the indirect effects of externalizing problems on social
competence. On the other hand, peer support was not a direct predictor of externalizing
problems 18 months later. Previous studies have found that social skills and peer support
had little predictive significance for either externalizing or internalizing problems (e.g.,
Hetherington, Cox, & Cox, 1985; Hirsch & DuBois, 1992).

Differences in the results of social competence in this study compared to previous
studies rrright be due to the different indicators of social competence that were explored

in this study (i.e., social skills and self-perceived social support). There is growing

73

evidence that fiiendships and peer relationships have positive effects on students’ well
being (e.g., Ladd & Troop-Gordon, 2003; Parker, Rubin, Price, & DeRosier, 1995). It
would be recommended that future studies examine the relational characteristics of peer
fi'iendships and their interaction with students’ social skills. It might be that social skills
lmve an indirect positive effect through peer relationships on adjustment. Researchers
should also be careful of not considering a set of social skills as equivalent of the
multidimensional construct of social competence(Rose-Krasnor, 1997). It might be that
some indicators of social competence (e. g., peer close relationships) are more important

than others (e.g., specific social skills) for predicting children’s adjustment.

Intemalizing Problems and Competence (Models [—3)

Contrary to externalizing problems which predicted low scholastic self-
competence, internalizing problems consistently predicted high levels of scholastic self-
competence. Although this finding seems counterintuitive at first, it might mean that
students who tend to internalize their symptoms, actually do well academically, and thus,
have a positive perception of their scholastic competence. High objective academic
competence may therefore be a third factor in the relation between internalizing problems
and self-perception of scholastic competence. It might also be that students who have
internalizing tendencies are error prone in their self-appraisals and they wrongly perceive
themselves as doing well scholastically. Future research should explore these
possibilities. Similarly, internalizing problems predicted high social skills. Students who

have the tendency to internalize their problems may also be perceived by their teachers as

74

cooperative, polite, and helpful to their peers, especially because they are not disruptive
to the class.

Intemalizing problems did not predict later study skills. It seems that internalizing
problems do not affect students’ academic skills directly. Previous studies have found
that students with internalizing profiles have milder difliculties and better adaptive skills
compared to students with externalizing profiles (e.g., McConaughy etal., 1988). It may
be that the internalizing nature of their problems does not affect their academic
performance directly. Teachers’ difficulty in reporting students’ internalizing difficulties
should be however noted.

On the other hand, high study skills predicted low internalizing problems 18
months later. Being a good student academically as reported by one’s teacher seems to
protect children against internalizing problems. Similar findings have been reported by a
longitudinal study where dissatisfaction with grades and failure to do homework
predicted depression one year later while major life events, daily hassles, and other
psychosocial factors did not (Lewinsohn et al., 1994). It seems therefore that academic
competence can protect students fi‘orn internalizing problems.

On the other hand, scholastic self-competence did not have any predictive power
over internalizing problems. Not considering oneself competent in the academic domain
does not necessarily lead to internalizing problems. It might be that other aspects of
competence are more important, such as general self-worth. Harter (1999) suggests that
students construct a general perception about themselves that is over and above their
performance judgments in specific domains of competence, such as academic and social

competence. Having a low perception of one’s academic competence may not influence

75

one’s overall sense of self-worth. Therefore, it is recommended that future studies include
a measure of general self-worth, as well.

It is interesting that internalizing problems did not predict peer support and peer
support did not predict internalizing problems either. Lewinsohn and colleagues (1994)
reported similar findings where self-rated social competence and social support from
peers did not predict depression, although they were concurrently related to depression.
Similarly, Hirsch and DuBois (1992) found that peer support did not predict
symptomatology two years later after controlling for prior levels of symptoms unless
students were going through a transition fiom elementary to junior high school. However,
they found significant effects of symptomatology on peer support after controlling for
initial levels of support.

There are many potential reasons why this study did not find significant effects
between peer social support and internalizing problems. First, the measure of support
used in this study was about the subjective experience of feeling socially cared and liked
by one’s peers. It did not refer to specific fiiendships. It might be that having a close
friendship is more important for preventing psychopathology than feeling supported by
one’s peers in general. Second, it might be that the subjective experience of being
supported by one’s peers is more important for students during transition periods when
they have not yet built their own close network of peers (see Hirsch & DuBois, 1992) or
during periods of stress when peer support might be more needed. It might also be that
peer support functions indirectly through its influence on adjustment indicators, such as
grades and classroom behavior. Finally, the sample of this study was a normal, school-

based, non-clinical sample, which means that the levels of internalizing behaviors were

76

lower as a whole, and thus, less interfering with students’ adjustment. Future research
needs to explore these possibilities before concluding that the influence of peer support
on psychopathology might be limited.

Overall, internalizing problems had less predictive significance compared to
externalizing problems. This is consistent with previous findings (e.g., Hetherington et
al., 1985; Koot & Verhulst, 1992; McConaughy et al., 1988). As speculated above, there
are several reasons why frndings regarding internalizing problems are less stable than
those of externalizing problems. First, it might be that the non-disruptive and thus, less
observable nature of internalizing problems makes teachers not reliable reporters of this
type of problems. Hinshaw (1992) also reports that rating scales may be less sensitive to
capturing internalizing compared to externalizing behaviors. It might also be that
teachers feel more empathetic towards students with internalizing problems and thus,
tend to be positively biased towards them. Finally, it might be that internalizing problems
do indeed have milder effects on students’ adjustment compared to the more debilitating

effects of externalizing problems.

Relation between Intemalizing and Extemalizing Problems (Models 1-3)

Extemalizing problems at Time 1 moderately predicted high internalizing
problems at Time 2 in the majority of the models. This finding corroborates previous
cross-sectional studies, which indicate a moderate relation between externalizing and
internalizing problems (Hinshaw, Han, Erhardt, & Huber, 1992; Reynolds & Kamphaus,

1998). Moreover, longitudinal research has shown that higher levels of externalizing

77

problems are associated with higher levels of internalizing problems (Capaldi, 199];
Gilliom & Shaw, 2004; Hetherington et al., 1985; Keiley, Bates, Dodge, & Pettit, 2000).

On the other hand, internalizing problems predicted lower externalizing problems
18 months later. However, this finding was not consistent in all models and was not as
strong as the positive effects from externalizing to internalizing problems. Similarly,
Keiley and colleagues (2000) reported a marginally significant effect where children with
high mother-reported internalizing problems in kindergarten tended to have low
externalizing problems.

It is clear in the literature that externalizing and internalizing problems often co-
occur. However, the direction of their relation is not always clear. In this study, it is noted
that the relation fi'om externalizing problems to internalizing problems is moderate and
positive while the opposite direction is smaller and negative. These findings shed some
light on the relation between these constructs. It seems that children who have high prior
levels of externalizing problems tend to develop later internalizing problems while
children who are initially high in internalizing problems tend to remain low in
externalizing problems. Previous research has shown that children with externalizing
problems develop internalizing problems over time due to the rejection fi'om peers and
adults that they experience because of their externalizing problems (e.g., Kupersmidt &
Coie, 1990; Kupersmidt, Coie, & Dodge, 1990; Ladd & TroopdGordon, 2003; Volling,
MacKinnon Lewis, Rabiner, & Baradaran, 1993). However, students who initially
experience internalizing problems might have the tendency to react to problems by
internalizing their symptoms and thus, remain intemalizers and do not develop

externalizing problems over time. Future research is needed to corroborate these findings.

78

Construct Stability (Models 1 -3)

Extemalizing problems showed high stability from Fall of Year 1 to Spring of
Year 2. Standardized path coefficients ranged fiom .45 to .71. A considerable body of
longitudinal studies lms reported similar regression weights for intervals up to ten years
(e.g., Capaldi & Stoohniller, 1999; Lerner etal., 1988; Moffitt, Caspi, Rutter, & Silva,
2001; Stanger et al., 1993). Consistent with developmental models of psychopathology,
this study shows that teacher-reported externalizing problems are stable and predictive
over time.

Intemalizing problems showed lower stability compared to externalizing
problems but still high enough to be considered strong based on Cohen’s (1988) criteria‘.
Standardized regressions weights ranged from .34 to .40. Similarly, studies of depression
have found correlational estimates of the same magnitude (e.g., Cole et al., 1996). It is
evident therefore, that internalizing problems show continuity over time, especially when
one takes into account that teachers are not as good reporters of internalizing problems,
as other raters, such as mothers. Teachers do not often have the opportunity to observe
internalizing problems in school due to the subtle and private nature of internalizing
behaviors (Keiley et al., 2000).

The continuity of competence indicators, such as social skills, study skills, peer
support, and scholastic competence has not been explored as extensively as the stability
of behavior problems in previous studies. It is important therefore to examine their
continuity more closely. Study skills showed strong stability over a period of 18 months

(i.e., path coefficients ranged from .62 to .74). One of the few competence studies has

 

' Effect sizes of .15 are considered medium, whereas effect sizes of at least .35 are large (Cohen, 1988 p.
413).

79

reported a sfiong stability coefficient (.73) within a period of two years for teacher-rated
academic competence (Lane et al., 2001). In addition, Masten and colleagues (1995) have
reported moderate continuity in academic competence (.36) within a period of ten years.

Self-perceptions of Scholastic Competence showed moderate stability (i.e.,
regression weight= .29). There is scarce evidence regarding the stability of scholastic
competence over time. For instance, Cole and colleagues (1996) reported a very strong
stability of academic competence, consisting of self-reports, teacher, parent ratings, and
peer nominations (.96). However, this strong finding is probably due to the short time
span of their study (i.e., 6 months). Further research is necessary to corroborate the
stability of self-reported scholastic competence.

Teacher-rated social skills and self-perceptions of peer support showed moderate
to high stability, as well, with path coefficients ranging fiom .41 to .59 and with boys
showing stronger stability of social skills (.59) than girls (.43). It seems that students who
are perceived by their teachers as socially skilled and feel supported by their peers
continue to be perceived socially competent and feel supported 18 months later. Previous
studies show similar findings. For instance, Hetherington, Cox, and Cox (1985) report a
high continuity of social competence for girls (Pearson r=.42) and a stronger one for boys
(r=.65) in a period of six years. Similarly, Lane and colleagues (2001) report a strong
stability coefficient (.66) for teacher-reported social skills within a period of two years.
Finally, a longer longitudinal study indicates that teacher reports of social competence at
age 6 highly predict peer social competence in adolescence (Hussong, Zucker, Wong,

Fitzgerald, & Puttler, in press).

80

The stability of competence markers reported here is consistent with a
developmental fiamework of psychopathology. Dispositional, biological factors and
environmental conditions may influence the development of stable patterns of behavior

for both competence and psychopathology.

Teacher-Student Relationships (Model 4)

Positive teacher-student relationships significantly predicted high social and study
skills through the mediating effects of initial levels of social and study skills. It is
particularly interesting that warm teacher-student relationships as rated by teachers also
significantly predicted teacher-rated low externalizing and internalizing problems almost
two years later through the mediating effects of initial levels of externalizing and
internalizing problems. These findings suggest therefore that the quality of teacher-
student relationships contribute to enhancing students’ competence and decreasing both
externalizing and internalizing problems. Previous longitudinal evidence support these
results (Hamre & Pianta, 2001; Howes, 2000; Ladd & Burgess, 2001). This study adds to
the existing literature by underscoring the importance of teacher-student relationship as a
protective factor for enhancing students’ competence and reducing students’ behavior
problems.

In addition, positive teacher-student relationships were strongly associated with
lower externalizing and internalizing problems and higher social and study skills when
measured concurrently either at Time 1 or Time 2. These results are comparable to

previous concurrent studies that have found that teacher-student relationships are strongly

81

 

associated with students’ positive outcomes, such as classroom adjustment, social skills,
school liking, and school involvement (Baker, 2005; Birch & Ladd, 1997).

A novelty about the current study is that it looks at the opposite direction of the
above longitudinal relations, as well. It was found that high study skills directly predicted
positive teacher-student relationships 18 months later. It might be that teachers feel closer
to students who are motivated to learn and work hard. Students with good study skills
may also be more open to a supportive relationship with their teachers. Moreover, high
externalizing problems directly predicted lower teacher-student relationships. These
findings suggest that there are probably bidirectional effects between teacher-student
relationships and competence indicators and behavior problems. Given that children are
active agents in their environment, it is very probable that there are reciprocal relations
among these variables. Therefore, children’s behavior and competence levels directly
influence their relationship with their teachers and teacher-student relationships indirectly
affect children’s levels of competence and behavior problems.

It is interesting however that internalizing problems predicted positive teacher-
stlnent relationships while positive teacher-student relationships indirectly predicted
lower internalizing problems almost two years later. It is possible that teacher-student
relationships protect students fiom internalizing problems but at the same time, teachers
may be more prone to build warm relationships with children with internalizing problems
who does not disrupt others due to the internalizing nature of their problems.

In addition to teacher-student relationships, this model explored the relations
between competence and problem behaviors. Similar to the models presented above, this

model verified the strong continuity of competence indicators and problem behaviors

82

over a period of 18 months. In addition, externalizing problems negatively affected the
development of both study and social skills through the mediating effects of initial levels
of social and study skills. This finding suggests that behavior problems can indirectly
influence the development of students’ competence. Intemalizing problems, on the other
hand, negatively predicted the development of study skills but positively affected later
social skills. As mentioned above, it may be that teachers positively view and are more
empathetic with students whose problems do not disrupt others. This positive perception
of their non-disruptive behavior may lead teachers to believe that students are behaving
appropriately towards others. However, it should be noted that the magnitude of these
relations was small.

In sum, this model provided evidence that beyond students’ initial levels of
competence and problems of both externalizing and internalizing nature, positive and
warm teacher-student relationships are salient predictors of later social and behavioral
adjustment in school. However, the opposite direction of these relations was evident, as
well. Competence and behavior problems directly predicted teacher-student relationships
longitudinally suggesting the importance of mutual effects between these variables. It
seems that the relations between teacher-student relationships and behavior problems and
competence are more complex than previously thought and research designs that consider

the bidirectional effects are needed for uncovering the nature of these relations.

Limitations

Several limitations require caution in the interpretation of these results. First,

there is shared source variance between the measures of behavior problems and most of

83

the competence measures. Teachers rated both students’ behavior problems and
competence with the exception of self-report measures, which may contribute to some
inflated findings. Behavior ratings and self-reports were used to measure the constructs
under investigation in this study. Observations could be used as a more objective measure
of students’ behavior. Moreover, this study lacks measures of family variables. Parents’
reports would enrich this study by providing data regarding children’s problem behavior
and competence in non-school contexts. Differences in perspective need to be taken into
account because it is possible that they make unique contributions to the understanding of
competence. In addition, including family variables would indicate risk and protective
factors that could influence the relations studied here and help us uncover different
psychological mechanisms (e.g., Loukas, Zucker, Fitzgerald, & Krull, 2003). Therefore,
it is recommended that filture studies use multiple raters, methods, and contexts.

Another limitation of this study is the short time span of studying the longitudinal
relations among the constructs of interest (i.e., 18 months). The processes examined in
this study covered a particular developmental period (i.e., first to fifth graders), and thus,
inferences regarding the present findings may be specific to this period. Having multiple
data points covering longer periods of children’s development might have provided
insight on different developmental patterns. For instance, different dimensions of
competence acquire precedence during adolescence (e.g., occupational competence,
competence with romantic relationships) (Masten et al., 1995). Statistical methods that
allow for the analysis of multiple data points would be recommended for future research

(e.g., Hierarchical Linear Modeling).

84

Third, race and socio-economic status were confounded in this study, and
restraints imposed by the schools regarding the collection of individual SES data made it
impossible to differentiate the effects of race and SES. Therefore, the present findings
regarding race should be viewed with caution since they may represent difference in SE8.
Data on both race and SES should be pursued in firture research, especially because
recent studies point to the importance of moderating effects based on race (e.g., Meehan,
Hughes, & Cavell, 2003).

There are antecedent variables that this study was unable to control, such as IQ,
language skills, neurodevelopmental delays, and biological influences. This lack of
neurobiological variables is also worsened by the absence of early childhood data. Recent
research has been more definite regarding the influences of early development and
neurobiological factors on later adaptation (e.g., Fitzgerald & Zucker, 2005; Martel et al.,
2004; Shonkoff & Phillips, 2000). Therefore, it is strongly recommended that future
research account for early adjustment and behavior problems.

Finally, this study explored three indicators of competence (i.e., social
competence, academic competence, and teacher-student relationships). However, there
are other competence dimensions, such as behavioral-conduct competence (see Masten et
al., 1995) and athletic-physical competence (see Harter, 1985a), which would be worth
exploring in the same model as social and academic competence. Moreover, it would be
important to examine the relational-friendship dimension of social competence. Research
has shown that friendships have important influence on children’s adjustment (e.g.,
Parker & Asher, 1993) and may help investigators uncover developmental mechanisms of

the influence of social competence on future adaptation.

85

Implications for Practice

This study holds several implications for practice. Given the protective efi’ects of
the quality of teacher-student relationships documented in this study as well as previous
research (e.g., Baker, 2005; Birch & Ladd, 1997; Hamre & Pianta, 2001; Pianta &
Nimetz, 1991), it seems important that prevention and early intervention efforts focus on
strategies to enhance teacher-students relationships. In this study, the quality of teacher-
student relationships was a good indirect predictor of both low externalizing and
internalizing problems and high social and study skills. Therefore, interventions that have
the improvement of teacher-student relationships as a goal may be able to prevent the
exacerbation of externalizing-intemalizing problems as well as enhance students’ study
and social skills. Baker (2005) suggests that treatment programs which include teacher-
student relationship components be used in school-based practice (e.g., Cowen et al.,
1996; Hughes & Cavell, 1999).

In addition, this study consistently showed that study skills protected students
against externalizing and internalizing problems. Although it is premature to suggest
practical applications of this finding, it indicates that academic strengths may protect
students against externalizing and internalizing problems. Previous research has shown
that study skills (e.g., Thorpe et al., 2000, August)and an academic focus in classrooms
fimction as a protective factor for psychopathology (e.g., Kasen, Johnson, & Cohen,
1990). Therefore, interventions that focus on enhancing students’ academic strengths may
protect students against the development of externalizing and internalizing problems.

Finally, the strong continuity for both behavior problems and competence

indicators suggest the need for early interventions. Shaping these behaviors and

86

enhancing children’s competencies early before they become stable characteristics will
help prevent behavior problems and build children’s strengths, which in turn will protect

them from adjustment difliculties.

Conclusions

The contribution of this research lies in examining the longitudinal relations of
behavior problems and competence markers in the same model. Few studies include
measures of competence and even fewer examine both externalizing and internalizing
problems and competence together. Moreover, in the literature, models with either
externalizing or internalizing problems have been examined and thus, one does not know
whether the predictive power derives fiom one or the other dimension (Masten &
Coatsworth, 1995). In this study, externalizing and internalizing problems were explored
in the same model and thus, it was possible to examine their different effects and the
relation between them.

Competencies have been identified as protective factors against the development
of childhood psychopathology (Coie et al., 1993; Masten & Curtis, 2000). This study
concludes that teacher-student relationships and students’ study skills hold promise for
preventing externalizing and internalizing problems and enhancing students’ well-being.
Given the accumulation of empirical evidence regarding the importance of competence
indicators, researchers should pay more attention to the empirical study of protective

factors.

87

APPENDICES

88

CLINICAL/AT-RISK RANGES

APPENDIX A.

Students at Clinical or At-Risk Ranges for Extemalizing-Intemalizing Problems and

 

 

 

 

 

Competence
At—Rislc Clinical At-Risk Clinical
T=60- 70 T above 70 T=30-40 T below 30
Extemalizing 55 38 Social 46 15
Problems 1 Skills 1
Extemalizing 57 45 Social 76 14
Problems 2 Skills 2
Intemalizing 29 22 Study 89 10
Problems 1 Skills 1
Intemalizing 46 23 Study 1 14 10
Problems 2 Skills 2

 

 

 

 

 

89

 

APPENDD( B.

HISTOGRAMS FOR ESTIMATED SAMPLE

 

120-

100: F7321:

 

 

40—1

 

<O=O=DOHTI

 

20-i

$1 2f . Yfflhfiem
40.00 50.00 60.“) 0.00 80.“) 90.00

Extemalizing Problems Time 1

my 5.74:1

 

 

 

 

 

 

 

Figure 1. Distribution of Extemalizing Problems at Time 1

90

100.00

 

Mean=51.96
Std. Dav.=11.19
N=442

 

70--1 F."

r;-
60-1 1‘1"
1,51%
in. i
it“ 1'91
3. ‘ .nl
504 . .- ;
rt-f: 1" '

 

 

 

<0:O:QO"'TI

 

 

 

 

 

 

 

 

 

 

 

Intemalizing Problems at Time 1

Figure 2. Distribution of Intemalizing Problems at Time 1

91

 

 

Mean = 49.3602
Std. Dev. = 10.02
N = 442

 

100-

60—4

(030:80‘1‘1

20-1

 

0—1

J"

K :—

4 '-‘ . '
r.' .

.-
.'. I‘ll-Ix r‘r

 

. .

. .._ _‘
. , -_.,

_ L‘fll: 12'

.“fl ‘ '4 ' ”1:,"

._..
-..'.-

1"““7‘5.

a 4
"NE!“ 335:3.“ ”1.1;;
. 31f: . r_,_ "'r.‘
-.

Mean = 52.454
~':-.i:-. . Std. Dev. = 12.37
«was . . "=49

40.00 60.00 80.00 100.00
Extemalizing Problems Time 2

 

 

 

 

 

 

 

~J‘.

 

 

Figure 3. Distribution of Extemalizing Problems at Time 2

92

 

1004

60-4 {9

"/4; I

 

 

‘(OSOCQG‘TI

Mean=49.1035
Std. Dev. = 10.15
N=442

 

  

 

 

 

 

 

 

 

 

 

30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00

Intemalizing Problems Time 2

Figure 4. Distribution of Intemalizing Problems at Time 2

93

urn-Eu 55.3.

 

 

~<osocno~1
8
1
' \
l
I
ll

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1%
Mean = 51.4467
_ ' -; , 3 . SM.Dev.=10.5
3 l ‘ ’l ' ' ‘« '~ : ‘ ' ’ | ' I N = 442
20.00 30.00 40.00 50.00 60.00 70.00 80.00

Social Skills Time 1

Figure 5. Distribution of Social Skills at Time 1

94

 

40-

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F30“ r _
, _ ,/_
e / ‘ - .
«1 z3 I :-
"2., _/. . , _
e / _
n " , . '.‘
c /._
y .
10— ‘
. Msan=50.6851
» :2 .5 .. » su.oav.=11.69
G l l ‘ | N=442
20.00 30.00 40.00 50.00 60.00 70.00 80.00

Social Skills Time 2

Figure 6. Distribution of Social Skills at Time 2

95

Mean =- 49.4622
Sid. Dev. =- 10.27

N8442

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

40—4

J
m

m w

FFOQUOflCy

00.00

70.00

 

Study Skills Time 1

l

bution of Study Skills at Time

D”

Figure 7.

96

 

8
1

 

F
r
e
(I
u 200 A
n
c
y
10— .r ,,

 

 

 

iii

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

., Mean = 48.3813
- {2% Sid. Dev. = 10.70
0 is N = 442
40.00 50.00 60.00 70.00 00.00

Study Skills Time 2

Figure 8. Distribution of Study Skills at Time 2

97

.st.=1.

 

 

 

 

 

 

 

 

 

 

 

 

 

704

60-1

 

0.“)

Peer Support Time 1

Figure 9. Distribution of Peer Support at Time 1

98

 

1% £3"

7 '
‘

4.»
1...
1;; M

O .' h.)- ‘ a .
.. -_ ,u A... .
9 “a“... ‘ 1 . . \.

., ".
‘kal’m

‘».
.v-I- .1
, t

l
1"“!

1‘

 

‘ 'ss 15

F

l'

0

q

u 1 ,i
O

n

c

1 ‘°"

 

Mean = 4.4937
Std. Dev. . 2.03
N =- 442

 

   

ti. ‘5.
__ m... .-,-.'~1

 

     

6.00
Peer Support Time 2

Figure 10. Distribution of Peer Support at Time 2

 

140-1

120‘

i

<GSOBQO‘TI

 

Figure I 1. Distribution of Scholastic Competence at Time 1

 

 

 

 

 

 

 

 

 

1 .00 2.00 3.00 4.00 5.00

 

7.00

Scholastic Competence Time 1

100

 

 

Meen=6.6091
Std.Dev.=1.32
N=442

 

 

 

150-

\i'_ A. ..‘_,-;.
~‘\\' A . I e

. . ,...
um
“'4.

r.,
:1“ u, '
‘ .

«xi-1i .
3»: inf/:1 ' .
.. J -‘ ' ".

V'.‘

90%

é-.. .,
, ..
farm.

r‘tg, .‘

5..
.51-3.
.,.-

604

(OQOCQQW'I'I

 

30-1

 

 

 

,
‘ I
I4
n..
‘ g-
\. .
.__H . ...
_._ .

2.00 3.00 4.00 5.00

Scholastic Competence Time 2

 

 

0‘

Figure 12. Distribution of Scholastic Competence at Time 2

101

 

80d

'11
8
L

~<oao=ao~
8
L

 

8
l

 

 

..,'. .. . . . , .. : . . .,
;. :ij 1 .rA , .‘ -,.l 1.1.; .b‘" = w 512
5 {1' 7 s ,1. '5. i," f: ‘
3 . ‘ : ‘ I - '7: , ’
J .. . .=e.4

 

 

 

 

t!"
T
f

..

ll

W
his
I?
'61
7r
r I

 

 

10.!” 40.00 50.00

TeacherStudent Relationships Time 1

Figure 13. Distribution of Teacher-Student Relationships at Time 1

102

 

80-1

8
n

<oaecn0-I'n
8
1

 

3
1

 

 

 

 

 

 

 

 

 

 

 

40.00

TeacberStudent Relationships Time 2

Figure 14. Distribution of Teacher-Student Relationships at Time 2

103

 

Mean = 37.6075
Std. Dev. - 723
N = 432

PPN?‘

99’3”.“

”$950991”?

)0

”$999993?

APPENDD( C
STUDY MEASURES
Student Personal Support (Johnson et al., 1983)

In this class, other students think it is important to be my friend.
In this class, other students like me the way I am.

In this class, other students care about my feelings.

In this class, other students really care about me.

Schola_stic Competence (Harter, 1985b)

I feel that I am very good at my school work.
I am smart enough to do my school work.

I do very well at my school work.

I can figure out the answers to school work.

Teacher-student Relationships (Pianta & Nimetz, 1991)

I share an affectionate relationship with this child.

This child and I seem to frequently snuggle with each other.

This child feels that I treat him/her family.

I often feel myself becoming frustrated with this child.

This child accepts help from me when I offer it.

It is easy to joke and have fun with this child.

Dealing with this child drains my energy.

Despite my best efforts, I am uncomfortable with how this child and I have gotten
along.

This child openly shares his/her feelings and experiences with me.

Behavior Assessment System for Children Subscales
(BASC, Reynolds & Kamphaus, 1998)

BASC- Social Skills

Volunteers to help with things.

Says “please” and “thank you”.

Encourages others to do their best.

Has a sense of humor.

Compliments others.

Tries to bring out the best in other people.

Politely asks for help.

Congratulates others when good things happen to them.

104

9. Makes suggestions without offending others.
10. Admits mistakes.
1 1. Offers to help other children.

12. Shows interest in others’ ideas.

BASC-Study Skfl

Reads assigned chapters.

Studies with other students.

Analyzes the nature of a problem before starting to solve it.
Does extra credit.

Works hard even in courses he or she does not like.
Appears confident before tests.

Reeds.

Completes homework.

Asks to make up missed assignments.

10. Has good study habits.

11. Uses the school library.

12. Is well organized.

PFHQMPP’N?‘

The BASC Extemalizing Problem_s Compofl scales are:
l. Hyperactivity
2. Aggression
3. Conduct Problems

Aggr_ession

"Shows off. "

Dares other children to do things.
Hits other children.

Teases others.

Argues with parents.

Is a "sore loser.”

Is critical of others.

Argues when denied own way
Is cruel to animals

10. Complains about rules.

11. Breaks other children's things.
12. Calls other children names.

13. Argues when denied own way.

99.“???pr

105

Hyperactiviy

Cannot wait to take turn.

Leaves seat during meals.

Throws tantrums.

Needs too much supervision.

Is restless during movies.

Fiddles with things while at meals.
Interrupts others when they are speaking.
Is overly active.

Makes loud noises when playing.
Climbs on things.

“PPNP‘MPP‘NT

.9

Conduct Problems

Runs away from home.

Has fiiends who are in trouble.

Is in trouble with the police.

Uses foul language.

Shows a lack of concern for others' feelings.
Has to stay after school for punishment.
Lies to get out of trouble.

Gets into trouble in the neighborhood.
Gets in trouble.

Has been suspended from school.

Lies.

-Pwsewswwr

#O

The BASC Intemalizing Problerg Composite scales are:
1. Anxiety
2. Depression
3. Somatization

flflfl

Worries.

Is too serious.

Worries about what parents think.

Says, “I get nervous during tests” or “Tests make me nervous.”
Tries too hard to please others.

Is afraid of dying.

Worries about things that cannot be changed.
Worries about what teachers think.

Says, “I’m afraid I will make a mistake.”
Says, “I’m not very good at this.”

Worries about schoolwork.

““PWSP‘S‘PPNT‘

F‘O

106

PPHP‘MPP’N!‘

I—il—Il-d
Pi"?

“99°N9‘9PP’PT‘

Depression

Says, "I don't have any friends."
Says, "I want to kill myself."

Cries easily.

Says, "Nobody understands me."

Is easily frustrated.

Is easily upset.

Complains about not having friends.
Changes mood quickly.

Says, ”I want to die” or "I wish I were dead."
Says, "Nobody likes me.”

Is sad.

Says, "I'm so ugly."

Sorfitization

Complains of pain.

Vomits.

Has ear infections.

Makes frequent visits to the doctor.
Has headaches.

Has allergic reactions.

Gets sick.

Has difficulty breathing.
Complains of heart beating too fast.
Has stomach problems.

107

REFERENCES

Achenbach, T. M., Howell, C. T., McConaughy, S. H., & Stanger, C. (1995). Six-year
predictors of problems in a national sample of children and youth: 1. Cross-
inforrnant syndromes. Journal of the American Academy of Child & Adolescent
Psychiatry, 34, 336-347.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental
disorders : DSM-[V-TR: Washington, DC : American Psychiatric Association.

Baker, J. A. (2005). Contributions of Teacher-Child Relationships to Positive School
Adj ustrnent during Elementary School. Manuscript submitted for publication.

Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Multifaceted
impact of self-efficacy beliefs on academic functioning. Child Development,
67(3), 1206-1222.

Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (2001). Self-efficacy
beliefs as shapers of children's aspirations and career trajectories. Child
Development, 72(1), 187-206.

Becker, W. C., & Krug, R. S. (1964). A circumplex model for social behavior in children.
Child Development.

Birch, S. H., & Ladd, G. W. (1997). The teacher-child relationship and children's early
school adjustment. Journal of School Psychology, 35 (I), 61-79.

Blechman, E. A., Tinsley, B., Carella, E. T., & McEnroe, M. J. (1985). Childhood
competence and behavior problems. Journal of A bnormal Psychology, 94(1), 70-
77.

Bos, C. S., & Vaughn, S. (2002). Strategies for teaching students with learning and
behavior problems. Boston: Allyn and Bacon.

Capaldi, D. M. (1991). Co-occurrence of conduct problems and depressive symptoms in
early adolescent boys: I. Familial factors and general adjustment at Grade 6.
Development and Psychopathology, 3(3), 277-300.

108

Capaldi, D. M., & Stoolrniller, M. (1999). Co-occurrence of conduct problems and
depressive symptoms in early adolescent boys: III. Prediction to young-adult
adjustment. Development & Psychopathology, 1 I (1 ), 59-84.

Caprara, G. V., Barbaranelli, C., Pastorelli, C., Bandura, A., & Zimbardo, P. (2000).
Prosocial foundations of children's academic achievement. Psychological Science,
I I (4), 302-306.

Chou, C. P., & Bentler, P. M. (1995). Estimates and tests in structural equation modeling.
In R. H. Hoyle (Ed.), Structural equation modeling: Concepts, issues, and
applications. (pp. 37-55). Thousand Oaks, CA: Sage Publications.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, N.J.:
L. Erlbaum Associates.

Cohen, N. J ., Kershner, J., & Wehrspann, W. (1988). Correlates of competence in a child
psychiatric population. Journal of Consulting and Clinical Psychology, 56(1), 97-
103.

Coie, J. D., Watt, N. R, West, S. G., Hawkins, J., Asamow, J. R., Markrnan, H. J., et al.
(1993). The science of prevention: A conceptual framework and some directions
for a national research program. American Psychologist, 48(10), 1013-1022.

Cole, D. A. (1990). Relation of social and academic competence to depressive symptoms
in childhood. Journal of A bnormal Psychology, 99(4), 422-429.

Cole, D. A., Jacquez, F. M., & Maschman, T. L. (2001). Social origins of depressive
cognitions: A longitudinal study of self-perceived competence in children.
Cognitive Therapy and Research, 25(4), 377-395.

Cole, D. A., Martin, J. M., & Powers, B. (1997). A competency-based model of child
depression: A longitudinal study of peer, parent, teacher, and self-evaluations.
Journal of Child Psychology and Psychiatry and Allied Disciplines, 38(5), 505-
5 14.

Cole, D. A., Martin, J. M., Powers, B., & Truglio, R. (1996). Modeling causal relations
between academic and social competence and depression: A multitrait-
multimethod longitudinal study of children. Journal of A bnormal Psychology,
105(2), 258-270.

109

Compas, B. E., Connor-Smith, J. K., Saltzrnan, H., Thomsen, A. H., & Wadsworth, M. E.
(2001). Coping with stress during childhood and adolescence: Problems, progress,
and potential in theory and research. Psychological Bulletin, 127(1), 87-127.

Cowen, E. L. (2000). Now that we all know that primary prevention in mental health is
great, what is it? Journal of Community Psychology, 28(1), 5-16.

Cowen, E. L., Hightower, A. D., Pedro-Carroll, J. L., Work, W. C., Wyman, P. A., &
Haffey, W. G. (1996). School-based prevention for children at risk: The Primary
Mental Health Project. Washington, DC: American Psychological Association
URL: h_ttp://www.z_rpa.org@oks.

DiPema, J. C., & Elliott, S. N. (2000). Academic Competence Evaluation Scales. San
Antonio, TX: The Psychological Corporation.

DiPema, J. C., Volpe, R. J ., 8: Elliott, S. N. (2002). A model of academic enablers and
elementary reading/language arts achievement. School Psychology Review, 31 (3),
298-3 12.

Fitzgerald, H. E., & Zucker, R. A. (2005). Growing up in an alcoholic family: Pathways
of risk aggregation for alcohol use disorders. In K. Freeark & W. D. 111 (Eds),
The crisis in youth mental health: Families, children, communities (V 01. 3). West
Port, CT: Praeger Press.

Ford, M. E. (1985). The concept of competence. Themes and Variations. In H. A.
Marlowe, R. B. Weinberg & Florida Mental Health Institute. (Eds), Competence
development : theory and practice in special populations (pp. xiii, 243 p.).
Springfield, 111., U.S.A.: C.C. Thomas.

Garmezy, N., & Masten, A. S. (1991). The protective role of competence indicators in
children at risk. In E. M. E. G. Cummings, Anita L.; Ed; et a1 (Ed.), Life-span
developmental psychology: Perspectives on stress and coping (pp. 151-174):
Hillsdale, NJ, Englandiates or Hillsdale, NJ. : L. Erlbaum Associates.

Garmezy, N., Masten, A. S., & Tellegen, A. (1984). The study of stress and competence
in children: A building block for developmental psychopathology. Child
Development, 55(1), 97-11 1.

Gilliom, M., & Shaw, D. S. (2004). Codevelopment of externalizing and internalizing
problems in early childhood. Development and Psychopathology, 16(2), 313-333.

110

Glick, M., & Zigler, E. (1986). Premorbid social competence and psychiatric outcome in
male and female nonschizophrenic patients. Journal of Consulting and Clinical
Psychology, 54(3), 402-403.

Glick, M., Zigler, E. F ., Baldwin, A. L., Baldwin, C., & Cole, R. E. (1990). Premorbid
competence and the courses and outcomes of psychiatric disorders. & Stress-
resistant families and stress-resistant children. In J. E. E. M. Rolf, Ann S.; Ed; et
a1; (Ed.), Risk and protective factors in the development of psychopathology. (pp.
497-513 & 257-280): New York, NY, USsity Press or Cambridge ; New York :
Cambridge University Press.

Gresham, F. M. (1995). Best practices in social skills training. In A. Thomas & J. Grimes
(Eds), Best Practices in School Psychology 111 (pp. 1021 -103 1). Washington,
DC: National Association of School Psychologists.

Gresham, F. M., & Elliott, S. N. (1990). Social skills rating system Manual. Circle Pines,
MN: American Guidance Service.

Hamre, B. K., & Pianta, R. C. (2001). Early Teacher-Child Relationships and the
Trajectory of Children's School Outcomes through Eighth Grade. Child
Development, 72(2), 625-638.

Harter, S. (1982). The Perceived Competence Scale for Children. Child Development,
53(1), 87-97.

Harter, S. (1985a). Competence as a dimension of self-evaluation. Toward a
comprehensive model of self-worth. In R. L. Leahy (Ed.), The Development of the
self (pp. 55-121). New York: Academic Press.

Harter, S. (1985b). Manual for the self-perception profile for children. University of
Denver.

Harter, S. (1999). The construction of the self: a developmental perspective. New York:
Guilford Press.

Hetherington, E., Cox, M., & Cox, R. (1985). Long-term effects of divorce and
remarriage on the adjustment of children. Journal of the American Academy of
Child Psychiatry, 24(5), 518-530.

111

Hinshaw, S. P. (1992). Extemalizing behavior problems and academic underachievement
in childhood and adolescence: Causal relationships and underlying mechanisms.
Psychological Bulletin, 1 I I (1), 127-155.

Hinshaw, S. P., Han, S. S., Erhardt, D., & Huber, A. (1992). Intemalizing and
externalizing behavior problems in preschool children: Correspondence among
parent and teacher ratings and behavior observations. Journal of Clinical Child
Psychology, 21(2), 143-150.

Hirsch, B. J ., & DuBois, D. L. (1992). The relation of peer social support and
psychological symptomatology during the transition to junior high school: A two-
year longitudinal analysis. American Journal of Community Psychology, 20(3 ),
333-347.

Howes, C. (2000). Social-emotional classroom climate in child care, child-teacher
relationships and children's second grade peer relations. Social Development,
9(2), 191-204.

Hughes, J. N., & Cavell, T. A. (1999). School-based interventions for aggressive
children: PrimeTime as a case in point. In S. W. Russ & T. H. Ollendick (Eds),
Handbook of psychotherapies with children and families (pp. 419-446).
Dordrecht, Netlrerland.

Hussong, A. M., Zucker, R. A., Wong, M. W., Fitzgerald, H. E., & Puttler, L. I. (in
press). Social Competence in Children of Alcoholic Parents Over Time.
Developmental Psychology.

Johnson, D. W., Johnson, R. T., & Anderson, D. (1983). Social interdependence and
classroom climate. Journal of Psychology: Interdisciplinary & Applied 114(1),
135-142.

Karnphaus, R. W., & Frick, P. J. (2002). Clinical assessment of child and adolescent
personality and behavior. Needham Heights, MA: Allyn & Bacon.

Kasen, S., Johnson, J ., & Cohen, P. (1990). The impact of school emotional climate on
student psychopathology. Journal of A bnormal Child Psychology, 18(2), 165-177.

Keiley, M. K., Bates, J. E., Dodge, K. A., & Pettit, G. S. (2000). A cross-domain growth
analysis: Extemalizing and internalizing behaviors during 8 years of childhood.
Journal of Abnormal Child Psychology, 28, 161-179.

112

Kline, R. B. (1998). Principles and practice of structural equation modeling. New York:
Guilford Press.

Kohlberg, L., LaCrosse, J ., & Ricks, D. (1972). The predictability of adult mental health
from childhood behavior. In B. B. Wolrnan (Ed.), Manual of child
psychopathology. New York: McGraw-Hill.

Koot, H. M., & Verhulst, F. C. (1992). Prediction of children's referral to mental health
and special education services from earlier adjustment. Journal of Child
Psychology and Psychiatry and Allied Disciplines, 33(4), 717-729.

Kupersmidt, J. B., & Coie, J. D. (1990). Preadolescent peer status, aggression, and school
adjustment as predictors of externalizing problems in adolescence. Child
Development, 61(5), 1350-1362.

Kupersmidt, J. B., Coie, J. D., & Dodge, K. A. (1990). The role of poor peer relationships
in the development of disorder. In S. R. Asher & J. D. Coie (Eds), Peer rejection
in childhood (pp. 274-305). NY: University Press.

Ladd, G. W. (1996). Shifting ecologies during the 5-7 year period: Predicting children's
adjustment during the transition to grade school. In A. Samerofl‘ & M. Haith
(Eds), The Five to seven year shift (pp. 363-386). Chicago: University of Chicago
Press.

Ladd, G. W., & Burgess, K. B. (2001). Do relational risk and protective factors moderate
the linkages between childhood aggression and early psychological and school
adjustment? Child Development, 72(5), 1579-1601.

Ladd, G. W., & Troop-Gordon, W. (2003). The role of chronic peer difficulties in the
development of children's psychological adjustment problems. Child
Development, 74(5), 1344-1367.

Lane, K. L., Gresham, F. M., MacMillan, D. L., & Bocian, K. (2001). Early detection of
students with antisocial behavior and hyperactivity problems. 24(3), 294-308.

Lerner, J. V., Hertzog, C., Hooker, K. A., Hassibi, M., & Thomas, A. (1988). A
longitudinal study of negative emotional states and adjustment from early
childhood through adolescence. Child Development, 59(2), 356-366.

113

Lewinsohn, P. M., Roberts, R. E., Seeley, J. R., Rohde, P., Gotlib, I. H., & Hops, H.
(1994). Adolescent psychopathology: 11. Psychosocial risk factors for depression.
Journal of Abnormal Psychology, 103(2), 302-315.

Loukas, A., Zucker, R. A., Fitzgerald, H. E., & Krull, J. L. (2003). Developmental
trajectories of disruptive behavior problems among sons of alcoholics: Effects of
parent psychopathology, family conflict, and child undercontrol. Journal of
Abnormal Psychology, 112(1), 119-131.

Luthar, S. S. (1991). Vulnerability and resilience: A study of high-risk adolescents. Child
Development, 62(3), 600-616.

Lyons, J. S., Uziel Miller, N. D., Reyes, F., & Sokol, P. T. (2000). Strengths of children
and adolescents in residential settings: Prevalence and associations with
psychopathology and discharge placement. Journal of the American Academy of
Child and Adolescent Psychiatry, 39(2), 176-181.

Martel, M. M., Nigg, J. T., Wong, M. W., Fitzgerald, H. E., Jester, J. M., Puttler, L. 1., et
al. (2004). Childhood and adolescence resiliency, regulation, and executive
ftmctioning in relation to adolescent problems and competencein a sample at risk
of substance abuse. Manuscript submitted for publication.

Mash, E. J ., & Dozois, D. J. A. (2003). Child psychopathology: A developmental-
systems perspective. In R. A. Barkley & E. J. Mash (Eds), Child
psychopathology. New York: Guilford Press.

Masten, A. S. (1994). Resilience in individual development: successful adaptation despite
risk and adversity. In E. W. G. M.C. Wang (Ed.), Educational Resilience in inner
city America: Challenges and prospects (pp. 3-25). Hillsdale, NJ, England:
Lawrence Erlbaum Associates, Inc.

Masten, A. S., & Coatsworth, J. D. (1995). Competence, resilience, and psychopathology.
In D. J. Cohen & D. Cicchetti (Eds), Developmental psychopathology, Vol. 2:
Risk, disorder, and adaptation. Oxford, England: John Wiley & Sons.

Masten, A. S., Coatsworth, J. D., Neemann, J., Gest, S. D., Tellegen, A., & Garmezy, N.
(1995). The structure and coherence of competence from childhood through
adolescence. Child Development, 66(6), 1635-1659.

114

Masten, A. S., & Curtis, W. J. (2000). Integrating competence and psychopathology:
Pathways toward a comprehensive science of adaption in development.
Development & Psychopathology Special Issue: Reflecting on the past and
planning for the fixture of developmental psychopathology, 12(3 ), 529-550.

McConaughy, S. H., Achenbach, T. M., & Gent, C. L. (1988). Multiaxial empirically
based assessment: Parent, teacher, observational, cognitive, and personality
correlates of child behavior profile types for 6- to 11-year-old boys. Journal of
Abnormal Child Psychology, 16(5), 485-509.

McFall, R. M. (1982). A review and reformulation of the concept of social skills.
Behavioral Assessment, 4(1), 1-33.

McGee, R., & Williams, S. (1991). Social competence in adolescence: Preliminary
findings fi'om a longitudinal study of New Zealand 15-year-olds. Psychiatry,
54(3), 281-291.

McGinnis, E., & Goldstein, A. P. (1990). Skill-streaming in early chiltflwod: Teaching
prosocial skills to the preschool and kindergarten child: Charnpaign, IL, US:
Research Press.

Meehan, B. T., Hughes, J. N., & Cavell, T. A. (2003). Teacher-Student Relationships as
Compensatory Resources for Aggressive Children. Child Development, 74(4),
1 145-1 1 57.

Merrell, K. W., & Popinga, M. R. (1994). The alliance of adaptive behavior and social
competence: An examination of relationships between the Scales of Independent
Behavior and the Social Skills Rating System. Research in Developmental
Disabilities, 15(1), 39-47.

Moffitt, T. E. (1990). Juvenile delinquency and Attention Deficit Disorder: Boys'
developmental trajectories from age 3 to age 15. Child-Development, 61(3), 893-
910.

Moffitt, T. E., Caspi, A., Rutter, M., & Silva, P. A. (2001). Sex drferences in antisocial
behaviour: Conduct disorder, delinquency, and violence in the Dunedin
Longitudinal Study. New York, NY: Cambridge University Press.

Offord, D. R., Boyle, M. H., Racine, Y. A., Fleming, J. E., Cadrnan, D. T., Blum, H. M.,
et al. (1992). Outcome, prognosis, and risk in a longitudinal follow-up study.

115

Journal of the American Academy of Child and Adolescent Psychiatry, 31(5),
916-923.

Parker, J. G., & Asher, S. R. (1993). Friendship and fiiendship quality in middle
childhood: Links with peer group acceptance and feelings of loneliness and social
dissatisfaction. Developmental Psychologr, 29(4), 611-621.

Parker, J. G., Rubin, K. H., Price, J. M., & DeRosier, M. E. (1995). Peer relationships,
child development, and adjustment: A developmental psychopathology
perspective. In D. Cicchetti & D. J. Cohen (Eds), Developmental
psychopathology, V012: Risk, disorder, and adaptation (pp. 96-161). Oxford,
England: John Wiley & Sons.

PattersOn, G. R., & Capaldi, D. M. (1990). A mediational model for boys' depressed
mood. In A. S. Masten & J. E. Rolf (Eds), Risk and protective factors in the
development of psychopathology. New York: Cambridge University Press.

Patterson, G. R., & Stoolnriller, M. (1991). Replications of a dual failure model for boys'
depressed mood. Journal of Consultiog and Clinical Psychology, 59, 491-498.

Pianta, R. C., & Nimetz, S. L. (1991). Relationships between children and teachers:
Associations with classroom and home behavior. Journal of Applied
Developmental Psychology, 12(3), 379-393.

Rae-Grant, N., Thomas, B. H., Offord, D. R., & Boyle, M. H. (1989). Risk, protective
factors, and the prevalence of behavioral and emotional disorders in children and
adolescents. Journal of the American Academy of Child & Adolescent Psychiatry,
28(2), 262-268.

Reynolds, C. R., & Kamphaus, R. W. (1998). Behavior Assessment System for Children
Manual. Circle Pines, MN: American Guidance Service.

Reynolds, C. R., & Kamphaus, R W. (2002). The clinician 's guide to the Behavior
Assessment System for Children (BASC). New York, NY: Guilford Press.

Rose-Krasnor, L. (1997). The nature of social competence: A theoretical review. Social
Development, 6(1), 111-135.

116

Rovine, M., J ., & Delaney, M. (1990). Missing Data Estimation in Developmental
Research. In A. v. Eye (Ed.), Statistical methods in longitudinal research. (V 01. I,
pp. 35-79). Boston: Academic Press, Inc.

Ryan, R M., & Deci, E. L. (2000). Self-determination theory and the facilitation of
intrinsic motivation, social development, and well-being. American Psychologist,
55, 68-78.

Seligman, M. E. P., & Csikszentmihalyi, M. (2000). Positive Psychology: An
Introduction. American Psychologist, 55(1), 5-14.

.A. Aw;

Shonkoff, J. P., & Phillips, D. A. (2000). From Neurons to Neighborhoods. Washington
DC: National Academy Press.

Sroufe, L. A. (1990). Considering normal and abnormal together: The essence of
developmental psychopathology. Development-and-Psychopathology., 2(4), 335-
347.

 

w... I |‘ .

Sroufe, L. A., & Rutter, M. (1984). The domain of developmental psychopathology.
Child Development, 55(1), 17-29.

Stanger, C., Achenbach, T. M., & McConaughy, S. H. (1993). Three-year course of
behavioral/emotional problems in a national sample of 4- to l6-year-olds: 3.
Predictors of signs of disturbance. Journal of Consulting and Clinical Psychology,
61(5), 839-848.

Tanaka, J. S., & Westerman, M. A. (1988). Common dimensions in the assessment of
competence in school-aged girls. Journal of Educational Psychology, 80(4), 579-
584.

Taris, T. W. (2000). A primer in longitudinal data analysis. London: SAGE.

Thorpe, J ., Kamphaus, R, Rowe, E., & F leckenstein, L. (2000, August). Longitudinal
eflects of child adaptive competencies and, externalizing and internalizing
behavior problems on behavioral and academic outcomes. Poster session
presented at the annual meeting of the American Psychological Association,
Washington, DC.

117

Volling, B. L., MacKinnon Lewis, C., Rabiner, D., & Baradaran, L. P. (1993). Children's
social competence and sociometric status: Further exploration of aggression,
social withdrawal, and peer rejection. Development and Psychopathology, 5(3),
459-483.

Waters, E., & Sroufe, L. (1983). Social competence as a developmental construct.
Developmental Review, 3(1), 79-97.

Widaman, K. F ., Stacy, A. W., & Borthwick-Duffy, S. A. (1993). Construct validity of
dimensions of adaptive behavior: A multitrait-multimethod evaluation. American
Journal on Mental Retardation, 98(2), 219-234.

Wierzbicki, M., & McCabe, M. (1988). Social skills and subsequent depressive
symptomatology in children. Journal of Clinical Child Psychology, 1 7(3 ), 203-
208.

 

Zigler, E., & Phillips, L. (1961). Social competence and outcome in psychiatric disorder. ‘;
Journal of A bnormal and Social Psychology.

 

118

IIIIIIIIIIIIIIIIIIIIIIIIIIIII

"”11111111“111111111'1""