READING FIRST, OR IS IT? AN EXAMINATION OF THE INTERRELATIONSHIP
BETWEEN READING ACHIEVEMENT AND BEHAVIORAL PROBLEM TRAJECTORIES
ACROSS ELEMENTARY SCHOOL FOR CHILDREN FROM DISADVANTAGED
CIRCUMSTANCES
By
Teresa Parton Clark

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILSOPHY
School Psychology
2011

ABSTRACT
READING FIRST, OR IS IT? AN EXAMINATION OF THE INTERRELATIONSHIP
BETWEEN READING ACHIEVEMENT AND BEHAVIORAL PROBLEM TRAJECTORIES
ACROSS ELEMENTARY SCHOOL FOR CHILDREN FROM DISADVANTAGED
CIRCUMSTANCES
By
Teresa Parton Clark
Reading is a necessary skill in our modern society and is increasingly critical for
success in our highly technological society. Yet many children from low-SES backgrounds
struggle to develop reading proficiency (Lee, Grigg, & Donahue, 2007) and continuing concerns
about this pervasive relationship has led to an increased focus on creating more effective
interventions to address these differences in reading development. Despite these efforts this
disparity in reading proficiency continues (Lee, Grigg, & Donahue, 2007), but most of the
intervention efforts to date have been focused almost exclusively on improving a limited set of
early reading sub-skills and ignores the larger context of child development within a low-SES
family. This heavy focus on ameliorating early deficits in reading proficiency is founded upon a
belief in a Matthew Effect for reading. The Matthew Effect model hypothesizes that early deficits
are expected to lead to a cumulative disadvantage over time, with lower entry skills leading to a
slower rate of growth in future reading skills. However, there is contradictory findings
concerning the existence of a Matthew Effect for reading with several studies suggesting that
children follow a more compensatory trajectory in reading development (Aarnouste et al., 2001;
Parrila et al., 2005; Phillips et al., 2002) and others suggesting that that there are multiple
trajectories of reading development (Aunola et al., 2002; Lepannen et al., 2004; Parrila et al.,
2005). Further, very few studies have specifically looked at the reading development of
American children from low-SES families. If factors other than initial endowment are related to

the slower rate of growth for children from low-SES families, then early intervention alone may
not be sufficient to effectively address the reading deficits typically found in this population. One
of these other factors may be the higher rate of behavior problems typically displayed by these
children (NCES, 2000; Zill et al., 1995). A better understanding of the reading development of
children from low-SES families is essential to guide effective interventions to address their
persistent deficit in reading proficiency. In order to extend the research regarding the reading
development of children from low-SES families, the present study used the Early Childhood
Longitudinal Study-Kindergarten Cohort (ECLS-K) dataset to examine: 1.) What trajectories the
reading development of children follow and whether the Matthew Effect model is the most
representative model for the reading development of children from low-SES families and 2.)
Whether the presence of comorbid behavior problems is related to the slower growth in reading
skills commonly found for children from low-SES families. Results suggested that on average
children from low-SES families did enter school with lower initial reading skills and a higher
rate of teacher reported behavior problems. Further children followed multiple trajectories of
reading development, with the largest percentage following a parallel trajectory and two smaller
groups following either a cumulative or compensatory trajectory. Socioeconomic status was
associated with the type of trajectory followed; children from lower SES families were
significantly more likely to follow a cumulative trajectory consistent with the Matthew Effect.
However, analyses showed that children from low-SES families demonstrated slower growth in
reading even when initial reading scores were controlled for suggesting that other factors than
reading proficiency at school entry may be associated with this slower growth. There was limited
support for the hypothesis that a higher rate of behavior problems was associated with the slower
reading growth of children from low-SES families.

DEDICATION
This dissertation is dedicated in memory of my father, Willis D. Parton, who always emphasized
the importance of education. His constant support and encouragement was invaluable in helping
me achieve my goals.
I would also like to dedicate this dissertation work in memory of Dr. Jean Baker, who taught me
many valuable lessons about life, research and being a professional. Her patience and mentorship
were greatly appreciated.

iv

ACKNOWLEDGEMENTS

My most sincere appreciation and thanks to Dr. Sara Bolt and Dr. Kim Maier for
agreeing to serve as co-chairs of my committee. Their feedback and assistance was invaluable in
the completion of this project. I would also like to extend special gratitude to Dr. Kim Maier
whose support and encouragement has been invaluable over the last couple of years. I also wish
to thank my committee members who were more than generous with their expertise and precious
time. Thank you Dr. John Kosciulek, Dr. Summer Ferreri and Dr. Nell Duke for agreeing to
serve on my committee. Finally, I would like to acknowledge Dr. John Carlson and Dr. Evelyn
Oka for the guidance and assistance provided as I have matriculated through the program.

v

TABLE OF CONTENTS

LIST OF TABLES

viii

LIST OF FIGURES

ix

CHAPTER 1
INTRODUCTION

1

CHAPTER 2
LITERATURE REVIEW
Socioeconomic Status
SES and Reading Achievement
SES and Early Reading
Matthew Effect
National Policy Implications
Trajectories of Reading Development
SES and Behavior Problems
Reading Achievement and Behavior Problems
Summary and Hypotheses

6
7
11
12
14
20
22
29
30
41

CHAPTER 3
METHOD
Participants
Measures
Child Outcomes
Family Measures
Data Analysis

45
45
46
46
55
57

CHAPTER 4
RESULTS
Simplex Model for Reading
Growth Curve Analysis for Reading
Simplex Model for Reading and Behavior
Growth Curve Model for Reading and Behavior
Clustering by Cases Analysis
Cluster membership by categorical SES and Behavior

63
64
67
71
74
77
84

CHAPTER 5
DISCUSSION
Reading Developmental Trajectories
Reading Development and SES
Other Factors to be Considered
Clustering by Cases Analysis

86
86
89
91
97

vi

Cluster Membership Relationship with SES and Behavior
Reading Development and Behavior
Limitations of this Study
Implications of Findings for Practice and Future Research

99
102
105
108

APPENDICES
Appendix A: Tables
Appendix B: Figures

110
111
135

REFERENCES

145

vii

LIST OF TABLES
Table 1 Comparison of Reading IRT Scores by Grade Level for Included and Excluded
Samples
112
Table 2: Comparison of Teacher-Rated SRS Scores by Assessment Period for Included
and Excluded Samples.
113
Table 3: Sociodemographic Information Comparison Between Excluded and Included
Samples
115
Table 4: Sociodemographic Information for Final Sample Used in this Study

116

Table 5: Means and Standard Deviations for Reading Assessments by Grade Level

117

Table 6: Means and Standard Deviations for Proficiency Probabilities by Grade and
Skill Level
118
Table 7: Means and Standard Deviations for Teacher-rated SRS by Assessment
Period

120

Table 8: GSS Occupational Codes by Parent for Initial Kindergarten Sample

121

Table 9: Reading IRT Scores by SES Quintile & Measurement Point

122

Table 10: Behavior Ratings by SES Quintile & Measurement Point

123

Table 11: Univariate Simplex Model for Reading with SES

124

Table 12: Univariate Piecewise Growth Curve Model for Reading with SES

125

Table 13: Univariate Simplex Model for Behavior with SES

126

Table 14: Multivariate Simplex Model for Reading and Behavior with SES

127

Table 15: Univariate Piecewise Growth Curve Model for Behavior with SES

129

Table 16: Multivariate Growth Curve Model for Reading and Behavior with SES

130

Table 17: Comparison of IRT Reading Scores and Proficiency Scores by Cluster

132

viii

LIST OF FIGURES
Figure 1: Univariate Simplex Model for Reading Development

136

Figure 2: Univariate Latent Growth Curve Model for Reading Development

137

Figure 3: Univariate Piecewise Growth Curve Model for Reading Development

138

Figure 4: Univariate Simplex Model for Behavior Ratings

139

Figure 5: Multivariate Simplex Model for Reading and Behavior

140

Figure 6: Univariate Latent Growth Curve Model for Behavior Ratings

141

Figure 7: Univariate Piecewise Growth Curve Model for Behavior Ratings

142

Figure 8: Multivariate Piecewise Growth Curve Model for Reading & Behavior

143

Figure 9: Comparison of Reading Trajectories by Clusters

144

ix

CHAPTER 1
INTRODUCTION
Reading is a necessary skill in our society and is increasingly critical for success in our
highly technological society. Yet approximately 30 million American adults leave school without
having attained the necessary proficiency skills in reading (Kutner et al., 2007). While children
from all SES backgrounds can struggle in developing proficient reading skills, children from
low-SES families are historically more likely to become poor readers (Lee, Grigg, & Donahue,
2007; Lonigan, Burgess, Anthony, & Barker, 1998; Molfese, Modglin, & Molfese, 2003).
Continuing concerns about this pervasive relationship between low-SES and poor reading skills
has led to an increased focus on creating more effective interventions to address these differences
in reading development, but unfortunately this disparity in reading skills continues (Gamse,
Bloom, Kemple, Jacob, 2008; Lee, Grigg, & Donahue, 2007; Russell et al., 2007). Much of the
research which has examined the relationship between low-SES and poor reading skill has
narrowly focused on improving a limited set of reading sub-skills and ignores the larger context
of child development within a low-SES family (Bowey, 1995; Dickinson, McCabe,
Anastasopoulos, Peisner-Feinberg, & Poe, 2003; Foorman, Anthony, Seals, & Mouzaki, 2002;
Torgeson, 1998; Whiteley, Smith, & Connors, 2007; Yeh, 2003). It is necessary to develop a
deeper understanding of the proximal processes that influence the reading development of
children from low-SES families in order to create more effective interventions to address these
deficits in reading proficiency.
Most of the current interventions to address these differences in reading proficiency are
focused on ameliorating deficits in early reading sub-skills before or soon after school entry
(Snow, Burns & Griffin, 1998). Previous research has indicated that children from low-SES

1

families enter school with less knowledge of print concepts, lower phonemic awareness, a less
developed understanding of the alphabetic principle and smaller vocabulary (Adams, 1990;
Anthony & Lonigan, 2004; NCES, 2000). The Matthew Effect model hypothesizes that these
early deficits are expected to lead to a cumulative disadvantage for children from low-SES
families, with lower entry skills leading to a slower rate of growth in future reading skills. This
expectation of a Matthew Effect has led to an increased emphasis on targeting interventions to
reach “at-risk” populations early in their educational trajectories. If rates of gain are relative and
proportional to initial endowment, as predicted by the Matthew Effect model, then strengthening
the early reading skills of children from low-SES families should lessen the academic gap by
placing these children on an equal footing with their more advantaged peers. However, if factors
other than initial endowment are related to this slower rate of growth for this group of children,
then early intervention alone may not be sufficient to effectively address the reading deficits
typically found in this population. At present, federal policy is almost exclusively focused on
addressing this deficit in the reading skills of children from low-SES families through early
reading intervention. Early Reading First and Reading First, both billion-dollar-a-year federal
initiatives have been in place to address the deficits in the reading skills of children from lowSES families. However, early evaluations of these initiatives have found that children are not
making as many gains as hoped (Gamse et al., 2008; Russell et al., 2007), a finding consistent
with earlier research showing that children with weak oral language and impoverished home
environments are less likely to respond to early literacy instruction (Al Otaiba, Conner, Lane,
Kosanovich, Schatschneider, Dyrlund et al., 2008; Al Otaiba & Fuchs, 2002; Lonigan, 2003;
Torgeson, 2000; Torgeson, Wagner, Rashotte, Rose, Lindamood, Conway, & Garvan, 1999).

2

While these programs are fairly new and more time or some adjustments may be needed to see
the expected gains, it is still unclear if providing early intervention to place these children on an
equal footing with their more advantaged peers is sufficient to close the gap in reading
proficiency long-term. It is clear that more needs to be known about the reading development of
children from low-SES families. In particular, what trajectories of reading development do
children from low-SES families follow and is the Matthew Effect model the best representation
of their reading development?
While there has been some evidence to support the Matthew Effect model (Cunningham
& Stanovich, 1997; Durham et al., 2007; Juel, 1988), many of the studies have significant
limitations and other studies have provided evidence of a compensatory model of reading
development (Aarnouste et al., 2001; Parrila et al., 2005; Phillips et al., 2002). Many children
who enter school with lower reading skills follow a compensatory trajectory of reading
development and experience more growth than students with higher initial skills even in the
absence of any additional educational interventions (Phillips et al., 2002; Spira, Bracken, &
Fischel, 2005), while other students continue to struggle with reading in spite of receiving
evidence-based interventions targeted at improving reading proficiency (Al Otaiba et al., 2008;
Al Otaiba & Fuchs, 2002; Rabiner, Coie et al., 2000; Torgeson, 2002; Torgeson, 1999). There is
also evidence that children’s reading development follows many different trajectories (Aunola et
al., 2002; Lepannen et al., 2004; Parrila et al., 2005), but no study has identified what trajectories
the reading development of children from low-SES families may follow. Identifying the specific
developmental reading trajectories followed by children from low-SES backgrounds is essential
for creating the most effective educational interventions to improve the reading proficiency of
this population. Some students who enter school with poor reading skills manage to benefit from

3

classroom instruction, while others struggle and continue to fall behind. Delineating specific
trajectories and then identifying the factors related to these trajectories can support more
informed, effective, research-based intervention strategies to improve the reading proficiency of
these children.
One such factor that may be related to slower growth in reading skills for children from
low-SES backgrounds is comorbid behavior problems. The same demographic factors related to
lower early reading skills are also associated with increased problem behaviors (NCES, 2000;
Zill et al., 1995), which have been shown to hinder the effectiveness of classroom instruction
(Gunter & Coutinho, 1997; Hamre & Pianta, 2001; Kochenderfer & Ladd, 1996). There has been
a consistent relationship found between reading difficulties and behavior problems, with
anywhere from 50% to 85% of children with reading difficulties also being described as having
behavior problems (Greenbaum et al., 1996; Nelson et al., 2004). There is evidence that early
behavior problems that predate school entry or that occur before the child has experienced a
history of learning failure are associated with slower growth in reading skills (Deater-Deckard et
al., 2009; Rabiner, Coie et al., 2000; Smart et al., 2001). There is also evidence that children with
behavior problems have more difficulty effectively engaging in classroom instruction (DuPaul &
Stoner, 1994; Mantzicopoulos, 2005; Rothbart & Bates, 1998) and often do not benefit from
evidence-based reading interventions (Al Otaiba & Fuchs, 2002; Nelson, Benner, & Gonzalez,
2003; Torgeson et al., 1999). Both of these findings suggest that early behavior problems may be
associated with slower reading growth. The comorbid behavior problems of children from lowSES families may not be the cause of their reading difficulties, but it is certainly a relationship in
need of closer examination.

4

Developing a more explicit understanding of these issues should provide a foundation
for further inquiry that will guide the development of more effective interventions to address the
persistent reading deficits experienced by children from low-SES backgrounds. Given the
importance of reading in our society and the critical need to identify the more proximal processes
that influence the reading development of children from low-SES backgrounds to inform
effective intervention strategies, the present study addressed two questions:
1.) What trajectories does the reading development of children follow and is the Matthew
Effect model the most representative model for the reading development of children
from low-SES families?
2.) Is the presence of comorbid behavior problems related to the slower growth in
reading skills commonly found for children from low-SES families?

5

CHAPTER 2
LITERATURE REVIEW
Reading is a necessary skill for participation in modern American society. Every day
Americans are inundated with written information. Public signs, menus, written notices,
brochures, medication labels, job applications, and rental agreements are just a few examples of
the written material encountered daily. Individuals who are not proficient in reading are at a
significant disadvantage in every domain of American life. When compared to adults with
average reading skills, adults with less proficient reading skills are less likely to be employed full
time, receive significantly lower salaries when employed and are more likely to receive public
assistance (Finnie & Meng, 2005; Kutner, Greenberg, Jin, Boyle, Hsu, & Dunleavy, 2007). They
are also less likely to vote, to participate in community service activities and as parents are less
likely to support literacy in the home and to be involved in their children’s education (Kutner et
al., 2007). Unfortunately, approximately 30 million American adults lack the ability to read and
understand information found in the short, simple texts required by everyday literacy activities
and more than 50 percent of this population live in households with incomes of less than 20
thousand dollars a year (Kutner et al., 2007). Clearly, children who become adults without
proficient reading skills are at an increasing disadvantage in our society and this disadvantage is
likely to be exacerbated by the increasing literacy demands required by a highly technological
society. This awareness has led to increased emphasis on improving the reading skills of all
children, but especially those from low socioeconomic status (SES) backgrounds.

6

Socioeconomic Status
Current thinking recognizes that healthy child development occurs within a context of
adequate resources and appropriate parenting (Parker, Piotrkowski, Horn, & Green, 1995). SES,
as a global index of family resources, has been shown to be moderately associated with the
nature of the developmental context. As early as 1929, Van Alstyne recognized SES to be a
distal variable that indirectly affects child development through its influence on such factors as
the family’s lifestyle, availability of resources, and sociopolitical status. Numerous studies since
then have further supported the relationship between SES and various health, psychological,
academic, and social outcomes (Bradley & Corwyn, 2002; Duncan & Brooks-Gunn, 1997;
Evans, 2004).
Children raised in families with a higher SES are more likely to benefit from their
parents’ greater knowledge about typical child development and effective parenting strategies
(Benasich & Brooks-Gunn, 1996; Goodenow & Collins, 1990; Laosa, 1982; McGillicuddyDeLisi & Sigel, 1995; Reich, 2005), which helps parents be better equipped to create an
environment more appropriate to a child’s emerging abilities and to promote the child’s healthy
development (Benasich & Brooks-Gunn, 1996; Miller, 1988). This greater level of parental
knowledge has been associated with more cognitively stimulating home environments (Benasich
& Brooks-Gunn, 1996); higher cognitive scores for children (Dichtelmiller, Meisels, Plunkett,
Bozynski, Claflin, & Mangelsdorf, 1992); higher maternal sensitivity (Tarabulsy, Bernier,
Provost, Maranda, Larose, Moss, Larose, & Tessier, 2005; Van Doesum, Hosmanm, RiksenWalraven, & Hoefnagels, 2007); improvements in children’s behavior (Benasich & BrooksGunn, 1996; Neitzel & Stright, 2004); less child aggression (Benzies, Keown, Magill-Evans,
2009; Côté, Boivin, Nagin, Japel, Xu, Zoccolillo, Junger, & Tremblay, 2007); and reduced

7

parental stress, lowered child maltreatment, and overall better child outcomes (Culp, Culp,
Blankemeyer, & Passmark, 1998; Honig & Wittmer, 1991). Further, parents from higher SES are
more likely to use more effective problem-solving strategies during interactions with their
children (Blechman & McEnroe, 1985). Neitzel and Stright (2004) found that on average
mothers with more education were better at regulating the task difficulty for their children while
encouraging the child’s active role in the problem-solving task. More educated mothers were
more likely to provide more cognitive and emotional support for their children and were better
able to successfully modify their interactions to support children with more difficult
temperaments.
Also, parents of higher socioeconomic status are more likely to have higher status
occupations, which may also influence parenting practices. Earlier work by Kohn (1959, 1963,
1977, 1979) found a relationship between occupational status and parenting practices, which has
been supported in more recent work (Cooksey, Menaghan, & Jekielek, 1997; Luster, Rhoades, &
Haas, 1989; Mengahan & Parcel, 1991, 1995; Parcel & Menaghan, 1990; Piotrkowski & Katz,
1982; Rogers, Parcel, & Menaghan, 1991). According to Kohn, parents working in jobs with low
occupational status are more likely to value conformity to external authority and to emphasize
obedience and good manners, while parents possessing jobs with higher status are more likely to
value autonomy, self-control and personal responsibility. Luster and colleagues (1989)
demonstrated that these work-related value systems do affect parenting practices in a distinct
manner. A mother’s value of conformity was associated with lower levels of maternal
involvement, lower maternal warmth and less reading to the child. Additionally, mothers who
highly valued conformity were more likely to use harsh, restrictive discipline and were less
likely to encourage exploration.

8

Other factors associated with occupational status have also been shown to affect
parenting practices. Parents whose jobs involve more complex work (occupations usually
associated with higher SES), requiring them to interact more with people and ideas than with
objects, tend to create more cognitively enriching and emotionally supportive home
environments for their children (Cooksey, Menaghan, & Jekielek, 1997; Menaghan & Parcel,
1991, 1995; Parcel & Menaghan, 1990; Piotrkowski & Katz, 1982; Rogers, Parcel & Menaghan,
1991). Job stress and dissatisfaction has also been associated with more negative developmental
outcomes potentially through its association with the negative quality of marital and child
relationships (Kinnunen, Gerris, & Vermulst, 1996; Wilson, Ellwood, & Brooks-Gunn, 1995).
Further non-standard work schedules (e.g., late night shifts, weekends) have been related to
lower family functioning, more parent distress and more hostile and ineffective parenting (Han,
2008; Joshi & Bogen, 2005; Strazdins, 2006). While it is true that families from any SES
category can experience job stress, job dissatisfaction and non-standard work schedules,
typically low-SES families are overrepresented within these categories (Strazdins, 2006). Parents
with less education and fewer job skills tend to have a more limited range of jobs to choose from
and are often required to take the less desirable positions (Rogers, Parcel, & Menaghan, 1991;
Strazdins, 2006; Wilson, Ellwood, & Brooks-Gunn, 1995).
The lower level of economic resources associated with low-SES also can have a profound
influence on developmental outcomes, such as lower levels of physical, cognitive and
social/emotional development, as well as lower levels of academic achievement and more
negative self-esteem (Bradley & Corwyn, 2002; Duncan & Brooks-Gunn, 1997; Hanson,
McLanahan, & Thomson, 1997). From an investment perspective the parents of low-SES
families may not have sufficient economic resources to support the healthy development of their

9

children (Bradley & Corwyn, 2002, Conger & Donnellan, 2007; Duncan & Magnuson, 2003;
Haveman & Wolfe, 1994; Linver, Brooks-Gunn, & Kohen, 2002). This lack of resources can
often have adverse effects before birth because low-income families frequently do not have
sufficient resources to provide adequate maternal nutrition, to obtain the necessary prenatal care
or to afford appropriate housing (Larson, 2007; U.S. Department of Health & Human Services,
2005). Children from low-income families are more likely to be born prematurely, at low birth
weight or with some type of birth defect (Crooks, 1995, U.S. Department of Health & Human
Services, 2005; Vrijheid, Dolk, Stone, Alberman, & Scott, 2000). After birth, inadequate
nutrition and poor quality housing can place these children at greater risk for a number of other
health problems and physical injuries (Aber, Bennett, Conley, & Li, 1997; Brooks-Gunn &
Duncan, 1997; Chen, Matthews, & Boyce, 2002; Guo & Harris, 2000; Sampson, 2003; Scholer,
Hickson, & Ray, 1999) and these greater health risks are exacerbated by the lack of access to
adequate healthcare (Ma, Gee, & Kushel, 2008; Owens, Zodet, Berdahl, Dougherty, McCormick,
& Simpson, 2008; Porr, Drummond, & Richter, 2006). This lack of economic resources can also
affect the parents’ ability to create a cognitively stimulating home environment. Children from
low-income families often have access to less reading material in the home and are less likely to
go on trips, visit a library or museum, or to attend a theatrical performance (Bradley, Corwyn,
Burchinal, McAdoo, & Garcia-Coll, 2001; Brooks-Gunn, Klebanov, & Liaw, 1995; Entwisle,
Alexander, & Olson, 1994; Guo & Harris, 2000). Further, the stress often experienced by
families with few resources, such as difficulty obtaining adequate basic necessities (e.g., food,
clothing, housing), inability to pay bills, poor or non-existent healthcare, job instability and
higher rate of daily hassles has been associated with lower quality of parenting (Conger &
Conger, 2002; Conger et al., 2002; McLoyd, 1998). These financially related stressors predict

10

increased parental emotional distress, more marital conflict and less supportive parenting leading
to poorer emotional, behavioral, cognitive and physical outcomes for the children (Conger, 1995;
Conger et al., 2002).

SES and Reading Achievement
While children from all SES backgrounds can struggle developing proficient reading
skills, children from families with a lower socioeconomic status are at substantially higher risk to
become poor readers (Bowey, 1995; Lonigan, Burgess, Anthony, & Barker, 1998; Molfese,
Modglin, & Molfese, 2003; Whitehurst, 1997). Socioeconomic status, as a global index of all
family resources, social, cultural as well as economic, has consistently been associated with
lower reading proficiency for children. The most recent National Assessment of Educational
Progress (NAEP; Lee, Grigg, & Donahue, 2007) report showed only 17 percent of fourth graders
and 15 percent of eighth graders from low-income families had reached proficiency in reading in
comparison to the 44 percent of fourth graders and the 40 percent of eighth graders from middle
to upper income families who reached the same level (Lee, Grigg & Donahue, 2007). These
findings are not surprising; the NAEP has reported similar discrepancies in reading achievement
for children from low-income families in every report since 1971. Continuing concern about this
persistent relationship between low-SES and lower reading proficiency has led to an increased
focus on creating more effective interventions to address these differences. Yet much of this
growing body of research narrowly focuses on improving a limited set of reading sub-skills
(Bowey, 1995; Dickinson, McCabe, Anastasopoulos, Peisner-Feinberg, & Poe, 2003; Foorman,
Anthony, Seals, & Mouzaki, 2002; Torgeson, 1998; Whiteley, Smith, & Connors, 2007; Yeh,
2003) ignoring the larger context of child development within a low-SES family. According to

11

Bronfenbrenner’s socio-ecological model (1979, 1999) child development is conceptualized as
occurring within a dynamic environment consisting of nested interactive and interdependent
systems that directly and indirectly influence the developmental course. Many of the
environmental factors (i.e., inadequate resources, maternal education, and parental support) that
influence the development of early reading skills also influence socioemotional development,
making it critical to identify the association between these interdependent systems
(socioemotional development and reading development) for children from low-SES
backgrounds. This study will provide a more thorough understanding of the reading development
of children from low-SES backgrounds by exploring the nature of this dynamic interaction
between the growth in reading skills and socioemotional development.

SES and Early Reading
Many have argued the reading difficulties experienced by children from low-SES
backgrounds begin before school entry (Burns, Griffin, & Snow, 1999; Entwisle & Alexander,
1993; McCardle, Scarborough, & Catts, 2001; Shonkoff & Phillips, 2000; Storch & Whitehurst,
2002; Torgeson, 2002). Research, using very different measurements of SES, suggests that on
average many children from low-SES families experience less rich language environments and
enter school with fewer linguistic skills than their more advantaged peers (Evans, 2004;
Gershoff, 2003; Hart & Risley, 1995). On average, children from disadvantaged backgrounds
have less exposure to books at home and are less likely to be read to by parents (Evans, 2004;
Federal Interagency Forum on Child and Family Statistics, 2005; Korat, Klein, & Segal-Drori,
2007; Lee & Burkam, 2002; U.S. Dept. of Ed., NCES, 2000; Whitehurst & Lonigan, 1998).
Many of these children also receive less language exposure and what language exposure they do

12

have is often of lower linguistic quality in comparison to children from higher SES families. Hart
and Risley (1995) found that SES was correlated with the number of words spoken per hour by
the parent; in a typical hour-long observation period they found that the average family receiving
welfare addressed 616 words to the child compared to the 2153 spoken by the average parent
from the higher SES family. This more limited early literacy environment has repeatedly been
associated with lower early reading skills for children from low-SES families (Morrow,
O'Connor' & Smith, 1990; Snow, 1991).
By the time children enter kindergarten these differences in their early literacy
environments have had a significant influence on the early reading skills of children from
different backgrounds. Children from low-SES families tend to enter school with lower levels of
oral language, phonological sensitivity, phonemic awareness, letter knowledge, knowledge of
print concepts; all skills which have been associated with much lower reading proficiency
(Aram, 2005; Bowey, 1995; Christian, Morrison, & Bryant, 1998; Hood, Conlan, & Andrews,
2008; Korat, Klein, & Segal-Drori, 2006; Spira, Bracken, & Fischel, 2005; Walker, Greenwood,
Hart & Carta, 1994). A survey of 2,000 children about six months from entering kindergarten
drawn from the 1993 National Household Education Survey (NHES) found there was a
significant difference in the skills children bring with them upon school entry. Only 57% of the
children from low-SES families showed signs of what they termed literacy indicators (e.g.,
recognizes most or all letters, pretends to read or reads stories and writes own name) compared
to 76% of children from families with higher SES (Zill, Collins, West, & Hausken, 1995). This
finding seems to be related to the level of maternal education because only 49% of the children
with mothers with less than a high school diploma demonstrated these literacy indicators
compared to 74% of the children with mothers with a high school diploma or higher. Similar

13

findings were reported by the U.S. Department of Education, National Center for Education
Statistics (NCES) for a nationally representative sample of 22,000 children entering kindergarten
in the fall of 1998. Forty-nine percent of children from families receiving welfare scored in the
lowest quartile on a reading assessment (measuring knowledge of print conventions, letter
recognition, phonological sensitivity, word reading and comprehension of written text) compared
to 22% of children from families who have never received welfare. Fifty-two percent of children
with mothers with less than a high school diploma scored in the lowest quartile on the reading
assessment compared to 32% for children with mothers with a high school diploma and 8% for
children with mothers with a Bachelor’s degree or higher (NCES, 2000). In sum, these findings
support the contention that children from low-SES families are more likely to enter school with
lower levels of oral language, phonological sensitivity, phonemic awareness, letter knowledge,
and knowledge of print concepts than their same age peers from higher SES families.

Matthew Effect
The difference in entry skills is expected to create a cumulative advantage for children
from higher SES families, with superior entry skills leading to a faster rate of growth in future
skill development, often called the Matthew Effect. This “rich get richer” or Matthew Effect
model was first proposed by Merton (1968) to explain variation in scientific productivity. Merton
developed this principle drawing from the Gospel of Matthew (25:29; KJV- “For unto every one
that hath shall be given, and he shall have abundance: but from him that hath not shall be taken
away even that which he hath”). His observations of career advancement in scientific fields led
him to conclude that the initial advantage of working with prominent scientists in prestigious
universities leads to better job placement and more frequent citations in the scientific literature.

14

Over time this early advantage provides cumulative benefits such that these better known
individuals receive greater recognition for similar scientific observations or discoveries than their
less well-known peers. Walberg and Tsai (1983) applied the Matthew Effect model to an
educational setting and described the cumulative advantages of early educational experiences for
later science achievement. Their “fan-spread” hypothesis suggested that rates of gain are relative
and proportional to initial endowment and when plotted over time the increasing variance will
lead to a fan-spread of data points. Based on these earlier findings, Stanovich (1986) concluded,
“ [this] cumulative advantage phenomenon is almost inextricably embedded within the
developmental course of reading progress.” Students entering school with fewer reading skills
would be at a decided disadvantage when compared to their more skilled peers.
The Matthew Effect model of Stanovich (1986) can be described as a set of interrelated
mechanisms: reciprocal causation, developmentally limited relationships and organismenvironment correlation. Reciprocal causation occurs when individual differences in a particular
process cause differential gains in reading efficiency and in turn reading itself causes further
individual differences in the process in question. For example, phonological awareness is
necessary to gain sufficient early word recognition skills that lead to reading independence.
Reading itself facilitates increased phonological awareness and better word recognition skills.
Further, the additional practice in reading gained by the more proficient reader leads to a more
rapid development of reading fluency. Higher rates of fluency reduce the cognitive demands of
reading and allow these cognitive resources to be allocated to the higher-level processes of text
integration and comprehension. Reading then facilitates further growth in reading
comprehension ability by increasing general knowledge, improving syntactic awareness and
augmenting vocabulary. However, Stanovich (1986) suggests these relationships are

15

developmentally limited. That is, individual differences in processes that cause variance in
reading ability early in development at some point cease to be causal factors. Although
phonological awareness is critical early in the reading developmental trajectory, it becomes less
important as direct visual access predominates and is less likely to be related to differences in
reading ability in middle school and beyond. Finally, Stanovich (1986) describes the effects of
three types of organism-environment correlations: active, evocative and passive. Active
organism-environment correlations occur when individuals select and shape their environment
by the choices they make. Students experiencing early success in reading are more likely to find
reading enjoyable and be motivated to engage in reading activities. These children create
environments that are conducive to continued improvement in reading by associating with peers
who also enjoy reading, gaining access to reading materials, and finding opportunities to read.
This is not usually the case for children who have struggled to develop reading skills; these
children usually find reading aversive and lack motivation to engage in reading activities.
Evocative organism-environment correlations occur when the individual evokes certain
responses from the environment. Successful readers who enjoy reading are likely to evoke more
support from family members, teachers and peers than less successful readers. Finally, a passive
organism-environment correlation is a relationship between the individual and the environmental
quality that is completely beyond the control of the individual. For instance, innate ability,
neighborhood environment, or the quality of education they receive. Stanovich (1986) predicted
the effect of these mechanisms would inevitably lead to a “rich get richer” or cumulative
advantage phenomenon within reading development, such that children entering school with
higher early reading skills would develop reading proficiency at a faster rate.

16

This differential in skill growth should lead to divergent achievement over time, possibly
explaining the substantial discrepancy in reading achievement found between children from
different socioeconomic backgrounds. Juel (1988) provided evidence for this “cumulative
advantage” reporting the probability that a child would remain a poor reader at the end of fourth
grade if a child was a poor reader at the end of first grade was .88, and the probability that a child
would become a poor reader in fourth grade if the child had average reading skills in first grade
was .12. Cunningham and Stanovich (1997) provided support for a reciprocal relationship between
reading and print exposure. The first grade reading of 27 children from a predominately middleclass school predicted significant variance in the eleventh grade print exposure when controlling
for eleventh grade reading comprehension and cognitive ability. The speed of initial reading
acquisition was moderately correlated with reading comprehension (.44), vocabulary (.37), print
exposure (.48) and general knowledge (.33) in the eleventh grade. A more recent study (Durham,
Farkas, Hammer, Tomblin, & Catts, 2007) indicated kindergarten language skills have a
statistically significant association with the development of reading skills in second and fourth
grade. The oral language skills measured in kindergarten explained 75% of the variance in
second grade reading scores and an additional 26% of the variance in fourth grade even when
controlling for the effect of second grade scores. This finding is substantial considering the
second grade reading scores explained 68% of the variance in fourth grade scores.
Other studies have also shown a significant relationship between various measures of
early reading skills and later reading achievement, but explaining considerably less variance than
the study conducted by Durham and colleagues. An Australian study of 479 children from nine
schools indicated that letter naming and phonemic segmentation measured at the beginning of
kindergarten explained 51 percent of the variance in first grade reading achievement (Share,

17

Jorm, Maclean, & Matthews, 1984). Another study looking at the reading development of 286
Australian children from Kindergarten to the end of sixth grade showed that early reading skills
measured in kindergarten (e.g., letter naming, phonemic awareness, sentence memory and
vocabulary) were the single most important predictor of reading ability in sixth grade. Students
who were the poorest readers in the early years of elementary school remained the poorest
readers at the end of sixth grade. The kindergarten measures explained 34 percent of the variance
in first and second grade reading scores and 49 and 44 percent of the third and sixth grade
reading scores (Butler, Marsh, Sheppard, & Sheppard, 1985). A study of 95 students from two
schools in North Carolina showed that early reading skills (e.g., alphabet recognition, concept of
word in text, spelling with beginning and ending consonants, and word recognition) in
kindergarten predicted up to 31 percent of the variance in reading achievement in first and
second grade (Morris, Bloodgood, & Perney, 2003). A study of 38 Israeli students from a
predominately low-SES population showed early reading skills (e.g., word recognition, word
writing, orthographic awareness, phonological awareness) measured in kindergarten were highly
correlated with reading skill at the end of second grade (e.g., spelling, word writing, text reading
and reading comprehension), explaining 27 percent of the variance in second grade reading
scores (Aram, 2005).
According to this cumulative reading trajectory model, children from low-SES
backgrounds entering school with less proficient skills would be at a distinct disadvantage in
relation to their more proficient peers. Effective intervention should of necessity be targeted at
reaching “at-risk” populations early in their educational trajectories. If rates of gain are relative
and proportional to initial endowment, strengthening the early reading skills of children from
low-SES families should lessen the academic gap by placing these children on an equal footing

18

with their more advantaged peers. As stated in the 1998 national report, Preventing Reading
Difficulties in Young Children (Snow, Burns & Griffin, 1998):
Primary prevention steps designed to reduce the number of children with inadequate
literacy-related knowledge (e.g., concepts of print, phonemic awareness, and receptive
vocabulary) at the onset of formal schooling would considerably reduce the number of
children with reading difficulties and, thereby, the magnitude of the problem currently
facing schools.
This sentiment continues to be echoed in more recent research published in the Journal of School
Psychology (Al Otaiba et al., 2008):
Additionally, converging research findings have demonstrated that reading trajectories,
which are established early in children’s school careers, are remarkably stable and
therefore difficult to change. Specifically, the success with which children acquire
foundational skills in two domains of early literacy – code-focused skills (letter
knowledge and phonemic awareness) and meaning-focused skills (oral language and
comprehension) - shapes their future reading development. A gap between weak and
strong reading achievement in these two domains begins early and subsequently widens
over the elementary years and becomes increasingly difficult to close, particularly after
third grade. Our increasing knowledge about the difficulty of remediating poor reading
achievement has led to a heightened focus on preventing reading difficulties by helping
children in first grade with established emergent literacy and reading readiness skills.

19

National Policy Implications
The Early Reading First (Title I, Part B, Subpart 2) and the Reading First (Title I, Part B,
Subpart 1) initiatives were created in 2002 as part of the No Child Left Behind Act (P.L. 107-110)
in an attempt to eliminate discrepancies in reading achievement between children from low-SES
families and their more advantaged peers. The Early Reading First initiative provides federal
monies to help early childhood centers that serve children from low-income families to more
effectively provide support for pre-school children’s early language and pre-reading skills. The
Reading First initiative is a billion dollar a year federal reading program providing federal grant
money to low-income, low-performing schools to support evidence-based reading interventions
designed to ensure all children learn to read proficiently by third grade. The goal of these
programs are laudable if, as stated above, future academic success can be predicted with
reasonable accuracy by third grade reading proficiency (Snow, Burns & Griffin, 1998), but early
evaluations of these initiatives have found that children are not making as many gains as hoped.
While the National Evaluation of Early Reading First: Final Report (Russell et al., 2007) found
that this program had a statistically significant positive effect (effect size of .34) on children’s
print and letter knowledge, there was no significant effect on phonological awareness or oral
language, both early reading skills strongly associated with later reading achievement (Adams,
1990; Anthony & Lonigan, 2004; Burgess, Hecht, & Lonigan, 2002; Lonigan, Burgess, &
Anthony, 2000). Further, the Reading First Impact Study: Interim Report (Gamse, Bloom,
Kemple, Jacob, 2008) indicated significant increases in instructional time spent on the five
essential components of reading instruction, but failed to show any increases in reading
achievement. These findings are consistent with earlier research showing that children with weak
oral language and impoverished home environments are less likely to respond to early literacy

20

instruction (Al Otaiba et al., 2008; Al Otaiba & Fuchs, 2002; Lonigan, 2003; Torgeson, 2000;
Torgeson et al., 1999). While these programs are fairly new and more time or some adjustments
may be needed to see the expected gains, it is still unclear if providing early intervention to place
these children on an equal footing with their more advantaged peers is sufficient in and of itself
to close the gap in reading proficiency long-term. It is clear that more needs to be known about
the reading development of children from low-SES families in order to design the most effective
interventions for all children.
These federally funded interventions are consistent with a cumulative reading trajectory
model (or Matthew Effect). An intensive focus on improving the early reading skills of children
from low-SES backgrounds is expected to place these children on an equal footing with their
more advantaged peers and to therefore improve the reading proficiency of these children longterm. However, if factors other than initial endowment are related to the slower development for
children from low-SES families, then early intervention alone may not be sufficient to effectively
address the reading deficits typically found in this population. These federal programs are an
attempt to address the continued reading difficulties of children from low-SES families, but more
needs to be discovered about the reading development of these children. One of the questions
that needs to be answered is what trajectories best characterize the reading development of
children from low-SES families and is the Matthew Effect the most accurate representation of the
reading development of all children, especially those from low-SES backgrounds?
Many children who enter school with lower reading skills develop proficient reading
without any educational interventions (Aarnouste, Van Leeuwe, Voeten & Oud, 2001; McCoach,
O’Connell, Reis, & Levitt, 2006; Phillips, Norris, Osmond, & Maynard, 2002; Scarborough,
1998; Shaywitz et al., 1995; Spira, Bracken, & Fischel, 2005), while a significant number of

21

children fail to develop reading proficiency despite receiving effective evidence-based
interventions (Al Otaiba & Fuchs, 2002; Dally, 2006; Nelson, Benner, & Gonzalez, 2003;
Rabiner, Coie, & the Conduct Problems Prevention Research Group, 2000; Torgesen, Alexander,
Wagner, Rashotte, Voeller, & Conway, 2001). Further, the presence of a Matthew Effect for
reading development is much less established than many researchers contend and very few
studies have been conducted with populations containing a high proportion of children from lowSES families. While there is some research to support the validity of a Matthew Effect in reading
(Aram, 2005; Butler et al., 1985; Cunningham & Stanovich, 1997; Durham et al., 2007; Juel,
1988; Morris, Bloodgood, & Perney, 2003; Share et al., 1984), there are an equal number of
studies that show children with lower initial reading skills following a cumulative trajectory
(Aarnouste et al., 2001; McCoach et al., 2006; Phillips et al., 2002; Scarborough, 1998; Shaywitz
et al., 1995: Spira, Bracken, & Fischel, 2005) and even others suggesting that children’s reading
development may follow many different trajectories (Aunola, Leskinen, Onatsu-Arvilommi &
Nurmi, 2002; Leppanen, Niemi, Aunola, & Nurmi, 2004; Lerkkanen, Rasku-Puttonen, Aunola,
Nurmi, 2004; Parrila, Aunola, Leskinen, Nurmi, & Kirby, 2005). The divergent findings of this
research makes it essential to identify the specific developmental trajectories followed by all
children and especially those from low-SES backgrounds. This additional research is needed in
order to create the most effective educational interventions to improve the reading proficiency of
children from low-SES families.

Trajectories of Reading Development
There has been some limited support for a Matthew Effect or cumulative model of
reading. Significant correlations between early and later reading achievement have been found in

22

several studies (Aram, 2005; Cunningham & Stanovich, 1997; Durham et al., 2007 Juel, 1988);
however many of these studies have limited generalizability because of small and in some cases
biased samples. The Aram (2005) study had a sample of only 38 children and the Cunningham
and Stanovich (1997) study had an even smaller sample of 27 children. Juel’s (1988) reporting of
an .88 probability that a child would remain a poor reader at the end of fourth grade, if a child
were a poor reader at the end of first grade has been cited often (413 citations according to Web
of Science Citation Index and 598 citations according to Google Scholar). Although, these
finding seem quite compelling, the size and nature of the sample used in this study raises a
number of significant concerns. The sample was drawn from one school, in a low-SES, urban,
Southern community with a highly transient, unique population (military families) with only 54
participants. Further, the high attrition rate in this study, especially considering the much higher
attrition rate among the average to good readers (34%) when compared to poor readers (17%)
and the largely minority population (74%) certainly raises questions about the validity of these
findings. The study conducted by Durham and colleagues reported findings for 502 children, but
275 of the 502 children in this study were previously diagnosed with a speech and language
impairment significantly biasing the results. And, although the results are reported for a sample
of 502 children the measures were only obtained for 33% of the sample and multiple imputation
was conducted to complete the dataset. The limitations of these studies restrict their
generalizability to the larger population. There are however, other studies showing a correlation
between earlier and later reading achievement (e.g., Butler et al., 1985; Morris, Bloodgood, &
Perney, 2003; Share et al., 1984), but there have been questions about the usefulness of
regression analyses to provide empirical support for a Matthew Effect in reading (Bast &
Reitsma, 1997; Phillips et al., 2002).

23

Highlighting the lack of clear guidelines for the empirical investigation of the cumulative
model of reading development (Matthew Effect), Bast and Reitsma (1997) suggest that two
assumptions must be considered in order to provide appropriate empirical support. First, there
must be evidence of increasing individual variation in reading development characterized by a
stable rank ordering of individuals and an increase of performance differences across time.
Second, this increasing variation in reading achievement must be caused by developmentally
limited, reciprocal causal relationships between reading and other cognitive skills, attitudes or
behaviors. Regression analyses can only provide empirical evidence of stable rank ordering and
does not allow investigation of the relationships that may lead to this outcome.
In an attempt to provide the empirical validation they found lacking, Bast and Reitsma
(1998) conducted a study of 235 Dutch children, which provided some limited support for the
Matthew Effect. A reciprocal causal relationship between reading development and vocabulary
development was found. Word recognition skills measured in first grade were associated with
more positive attitudes toward reading. Children with more positive attitudes toward reading and
a good word recognition level tended to read more frequently during leisure time. The frequency
of reading outside of school was associated with the level of reading growth of students at the
end of second grade. However this pattern of interrelationships did not lead to the increasing
individual differences for reading comprehension predicted by the Matthew Effect. Although a
Matthew Effect (i.e., stable rank order with increasing individual differences) was found for word
recognition no such effect was found for other reading related skills. Also, individual differences
at first grade did not predict differences in second or third grade indicating that factors that
determine initial skill level do not necessarily also determine the continued progress of students
in reading.

24

However a replication study with a very similar sample of approximately 500 students
from a stratified sample of Dutch schools (Aarnouste et al., 2001) found no Matthew Effect for
any reading skill. They followed students’ progress in decoding efficiency, reading
comprehension and vocabulary from the end of first grade to the end of sixth grade. The results
indicated that students who started with less skill actually made more overall progress in all three
skill areas than did students with average or high beginning skills. The progress of the low
performers greatly exceeded the progress of the other two groups from fall to spring, showing the
group of poor readers clearly benefits the most from instruction. A number of other studies have
also not supported this relationship between early reading skills and later achievement
trajectories consistent with the Matthew Effect model (McCoach et al., 2006; Phillips et al., 2002;
Scarborough, 1998; Shaywitz et al., 1995: Spira, Bracken, & Fischel, 2005)
A study of a heterogeneous sample of 414 children followed from kindergarten thru sixth
grade showed no fan-spread effect for reading development. In fact, children with poorer reading
ability in kindergarten showed greater improvement over the course of the study relative to
children with better reading ability (Shaywitz et al., 1995). Another study of 64 students tracked
from the end of second grade until the end of eighth grade showed that forty-two percent of
children identified with a significant delay in reading ability at the end of second grade (1.5 SD
below the mean reading score) had improved sufficiently by the end of eighth grade to no longer
be considered reading disabled (Scarborough, 1998). A Canadian study of a homogenous sample
of 187 students from first to sixth grade showed students who began first grade as poor readers
had a .53 probability of becoming an average reader by sixth grade. Further in second grade the
results showed a convergence of scores in the average range with a number of below average
readers improving and a large percentage of above-average readers dropping into the average

25

category (Phillips et al., 2002). A study followed 288 children from kindergarten through the end
of fourth grade. Half of the sample scored within the average range in reading at the end of first
grade. Although the other 146 children scored in the lowest 30th percentile in reading at the end
of first grade, a full 30% of these children were within the average range in reading skill by the
end of fourth grade. Further, reading achievement at the end of first grade (.42) was only a
moderate indicator of reading level at the end of fourth grade (Spira, Bracken, & Fischel, 2005).
A more recent study of 8,089 students within 657 schools showed a high correlation among four
reading assessments from the beginning of kindergarten until the end of first grade. This finding
indicates a high stability over time in the rank order of students in terms of their relative reading
skills, but initial status was negatively correlated with reading skill growth in both kindergarten
and first grade. Children who entered school with less reading skills experienced more growth
than students with higher initial skills. Further, children who made slower growth in kindergarten
made greater growth in first grade. Children who entered kindergarten with the lowest skills did
not completely close the gap by the end of first grade. It is unclear whether these children would
continue to close the gap if the study had been extended (McCoach et al., 2006).
While finding similar accelerated growth for students who began as poor readers, a study
conducted in Helsinki, Finland measuring the relationship between self-concept and the growth
in reading skills across first grade for 105 students also found three different trajectories. One
group of 24 students who had poor entry reading skills and a low self-concept exhibited a
positive trend for self-concept and a negative trend for reading skills. The second group of 37
students who began school with a high level of reading skills and a high self-concept
experienced a positive trend for self-concept and a negative trend for reading skills. The final
group of 44 students who began school with poor entry reading skills, but a high self-concept

26

showed a positive trend for reading skills and self-concept (Aunola et al., 2002). Another study
of 196 Finnish children followed from the beginning of preschool until the end of first grade
provided evidence of a cumulative trajectory of reading development in preschool but a
compensatory trajectory in first grade, with the poorer readers making greater progress than their
more skilled peers. Their findings also indicated that the skills related to reading growth changed
from preschool to first grade. While phonological awareness was associated with better reading
skills measured at the beginning of preschool, the growth of reading skills in preschool was
predicted by letter knowledge and number sense and negatively related to listening
comprehension. However, the growth of reading skills in first grade was predicted by listening
comprehension and negatively associated with phonological awareness and number sense.
Further a cluster analysis of growth trajectories found three distinct trajectories of reading
development: a group of 71 who began school with proficient skills and continued to make
significant progress; a group of 113 who began with poor skills but who made rapid progress
during first grade and a final group of 11 who began with poor skills and continued to struggle
throughout the study (Leppanen et al., 2004).
Finally, using cluster analysis a study of 90 children from six elementary classes in 4
schools in Finland revealed three groups of readers: poor readers (poor word reading/poor
comprehension), technical readers (good word reading/poor comprehension) and competent
readers (good word reading/good comprehension). Only 31% of the students, all from the
competent reader group, showed stability in reading skill across all measurement occasions. The
remaining 69% showed a number of different developmental trajectories across time, only 3%
showing the consistent regression, which would be predicted by the Matthew Effect (Lerkkanen
et al., 2004). Although the majority of the studies supporting a compensatory development of

27

reading have been conducted in languages with a more transparent orthography (1) than English,
a recent study comparing the results of similar analyses using samples of both English speaking
and Finnish speaking students also found support for a compensatory model of reading
development for English speaking students (Parrila et al., 2005).
There is clearly substantial diversity in the findings of research examining the reading
trajectories of children entering school with poor reading skills. While this inconsistency could
be the result of sample selection, the choice of measures used or limitations in the methodology,
as argued by Parrila and colleagues (2005) and Bast and Reitsma (1997), it is also possible these
differences could be the result of some unmeasured sample characteristic moderating the
relationship between instruction and achievement. There is obviously less convergence in
research concerning the presence of a Matthew Effect for children who enter school with lower
initial reading skills. Clearly some research has found that students who enter school with poor
reading skills manage to benefit from classroom instruction, while others struggle and continue
to fall behind. More recent research seems to indicate that there are many different trajectories of
reading development. This research needs to be replicated for American school children and it is
important to determine whether and if so, how the reading development of children low-SES
families differs from that of their peers from higher SES families. Further, if differences are
found it is essential to ascertain what factors are related to these differences in order to design
and implement the most effective interventions to address deficits in reading development for
this population. Early intervention to address the discrepancy in early reading skill is certainly

In transparent orthographies, the mappings from letters to sounds are consistent. In more
opaque orthographies, like English, the same grapheme or graphemes may represent different
phonemes in different words (and vice-versa). For example in English, “ove” represents different
phonemes in the words, cove, love, and move. For more information see Joshi and Aaron (2006).
1

28

important, but it may not be sufficient to completely ameliorate the differences often found in the
reading development of children from low-SES families.

SES and Behavior Problems
One such factor may be behavior problems. The same demographic factors related to
lower early reading skills are also associated with increased problem behaviors, which may
hinder the effectiveness of classroom instruction (Carr, Taylor, & Robinson, 1991; DuPaul et al.,
2004; Gunter & Coutinho, 1997; Hamre & Pianta, 2001; Kochenderfer & Ladd, 1996; Lynch &
Cicchetti, 1997). The survey of 2,000 children drawn from the 1993 NHES (Zill et al., 1995),
which showed that children from low-SES families enter kindergarten with lower early reading
skills, also showed that many of these same children have significantly higher levels of problem
behavior at school entry. For instance children of mothers with less than a high school diploma
are more likely to be restless and fidgety (43% compared to 27%), to have a short attention span
(38% compared to 21%) and to have temper tantrums (36% compared to 21%) than their peers
with mothers with at least a high school diploma. The same differences are also found between
children from poor and non-poor families: restlessness (36% compared to 26%), short attention
span (32% compared to 20%), and temper tantrums (28% compared to 21%). This relationship
between factors leading to low early reading skills and behavior problems is also found in the
NCES (2000) study. Children from poor families were less likely to show prosocial behavior
(60% compared to 69%), and more likely to show antisocial behavior (15% compared to 10%),
and to be inattentive (47% compared to 32%) than their peers from non-poor families. Further
these children showed less task persistence (59% compared to 73%) and less eagerness to learn
(62% to 76%) when compared to their peers.

29

Reading Achievement and Behavior Problems
Behavior problems have been variously defined as inattention, hyperactivity, impulsivity,
aggression, defiance, disruptive behavior, lack of self-regulation, task avoidance, delinquency,
externalizing behavior or socioemotional impairment. These behaviors are also associated with
the psychiatric diagnoses of Attention-Deficit-Hyperactivity Disorder (ADHD), Oppositional
Defiant Disorder (ODD) and Conduct Disorder (CD) and eligibility for special education as
Emotional Behavior Disordered (also known as Emotionally Impaired). Regardless of how they
have been defined, behavior problems have consistently been associated with lower reading
achievement (Arnold, Goldston, Walsh, Reboussin, Daniel, Hickman et al., 2005; Hinshaw,
1992; McGee, Share, Moffitt, Williams & Silva, 1988; Reid, Gonzalez, Nordness, Trout &
Epstein, 2004; Trout, Nordness, Pierce, Epstein, 2003) with anywhere from 50% to 85% of
children with reading difficulties also being described as having behavior problems (Frick,
Kamphaus, Lahey, Loeber, Christ, Hart, & Tannebaum, 1991; Greenbaum, Dedrick, Freidman,
Kutash, Brown, Lardieri et al. 1996; Nelson et al., 2004). Horn and Packard (1985) conducted a
meta-analysis of 58 studies that reported correlations between behavior problems in kindergarten
or first grade and reading achievement later in elementary school, which indicated that the
average correlation between behavior problems and later reading achievement was .48
(inattention = .63; externalizing = .45; internalizing = .59 and social skills = .44). In fact,
inattention was more highly correlated with later reading achievement than either language
ability (.52) or IQ (.53). The results of another meta-analysis of 26 studies showed that children
with behavior problems had significantly lower reading achievement than their peers with an
average effect size of -.61 (Reid, Gonzalez, Nordness, Trout, & Epstein, 2004). Wentzel (1993)
showed that aggressive and non-compliant behavior in sixth and seventh grade was negatively

30

correlated with grade point average (-.55), scores on a standardized achievement test (-.35) and
with academic behavior (e.g., interest in schoolwork, independent learning and concern with
evaluation; -.49). Another study indicated that attention problems (-.38), delinquent behavior (.31) and aggressive behavior (-.28) were all negatively correlated with reading achievement for
adolescents between the ages of 11 and 17 (Barriga, Doran, Newell, Morrison, Barbetti &
Robbins, 2002). And Velting and Whitehurst (1997) found that hyperactivity and inattention
were negatively related to reading achievement in first grade (-.28) for a group of children from
low-SES families. Numerous other studies have linked early behavior problems to continuing
academic difficulties and eventual school dropout (Alexander & Entwisle, 1988; Birch & Ladd,
1997; Entwisle & Hayduk, 1988; Kochenderfer & Ladd, 1996; Lynch & Cicchetti, 1997; Pianta,
Sternberg & Rollins, 1995).
While the relationship between behavior problems and academic achievement has been
well documented (Hinshaw, 1992; McGee et al. 1988; Reid et al., 2004; Trout et al., 2003), the
exact nature of the relationship has remained unclear. There are currently four accepted causal
models to explain the nature of this relationship (for a review see Dionne, 2005; Hinshaw, 1992;
Mandel, 1997; Spira & Fischel, 2005). The first causal model suggests that underachievement
and behavior problems are not causally related, but rather some common, unmeasured
developmental antecedent may cause both academic underachievement and behavior problems,
and the two may have no direct effect on each other. Several developmental antecedent
explanations have been offered: auditory comprehension (Benner, Beaudoin, Kinder, & Mooney,
2005); Attention-Deficit-Hyperactivity Disorder (ADHD; Clark, Prior, Kinsella, 2002; Saudino
& Plomin, 2007); a common underlying genetic factor (Willcutt & Pennington, 2000); or a
common underlying environmental factor (Hinshaw, 1992; Trzesniewski, Moffitt, Taylor,

31

Maughan, & Caspi, 2006). The second causal model suggests that underachievement leads to
behavior problems. A history of academic failure may lead to frustration, low self-esteem, and
negative attitudes toward learning leading toward task-avoidant and acting out behaviors
(Arnold, 1997; Fleming, Harachi, Cortes, Abbott, & Catalano, 2004; McGee & Share, 1988;
McIntosh, Horner, Chard, Boland, & Good, 2006). The third causal model suggests that behavior
problems lead to underachievement. Off-task and disruptive behaviors decrease the student’s
ability to attend to and benefit from instruction (Dally, 2006; Deater-Deckard, Mullineaux,
Petrill, Thompson, 2009; Rabiner, Coie, & the Conduct Problems Prevention Research Group,
2000; Torgesen, Alexander, Wagner, Rashotte, Voeller, & Conway, 2001). The fourth and final
causal model suggests that a reciprocal relationship exists between reading and behavior
problems. In this model, regardless of the primacy of either causal factor, academic failure is
expected to lead to increasing academic disengagement thus exacerbating future academic and
behavior difficulties (McGee, Prior, Williams, Smart & Sanson, 2002; Onatsu-Arvilommi &
Nurmi, 2000; Rowe & Rowe, 1992; Stipek & Miles, 2008; Trzesniewski et al., 2006).
The above studies provide evidence both supporting and contradicting each of the four
proposed causal models. The multitude of studies examining this relationship have used samples
of varying ages and backgrounds, widely divergent definitions of behavior problems and
academic achievement and a variety of methodological approaches (Hinshaw, 1992; Spira &
Fischel, 2005) making it quite likely that all of these models are correct to some degree for
specific populations. Disentangling the association between reading difficulties and behavior
problems is often complicated by the complex, multicausal nature of these frequently comorbid
difficulties, and providing conclusive evidence in support of only one causal model is often
problematic (Hinshaw, 1992). With that in mind the goal of this study is not to invalidate any of

32

the above causal models explaining the relationship between academic achievement and
behavior problems, but rather to develop a better understanding of the reading development of
children from low-SES backgrounds by examining the role of behavior problems in that
development. There is evidence that early behavior problems that predate school entry or that
occur before the child has experienced a history of learning failure are associated with slower
growth in reading skills (Deater-Deckard et al., 2009; Rabiner, Coie et al., 2000; Smart et al.,
2001). There is also evidence that children with behavior problems have more difficulty
effectively engaging in classroom instruction (Baker, Clark, Maier, & Viger, 2008; DuPaul &
Stoner, 1994; Hamre & Pianta, 2001; Mantzicopoulos, 2005; Rothbart & Bates, 1998) and often
do not benefit from evidence-based reading interventions (Al Otaiba & Fuchs, 2002; Nelson,
Benner, & Gonzalez, 2003; Rabiner, Malone et al., 2004; Torgeson, Wagner, Rashotte, Rose,
Conway, & Garvan, 1999). This is not to say that the comorbid behavior problems of children
from low-SES families are the cause of their reading difficulties, but it certainly points to the
need for a closer examination of this relationship.
Many children from low-SES families enter school with increased behavior problems
(NCES, 2000; Zill et al., 1995) and as mentioned previously there is evidence that behavior
problems are related to slower reading development. The exact nature of this relationship is
unclear, but active engagement during classroom instruction is critically important for academic
success, particularly in the area of reading (McGee & Share, 1988; Rowe & Rowe, 1992;
Warner-Rogers, Taylor, Taylor & Sandberg, 2000). Children with higher parent ratings of
effortful control (measuring attention focusing, inhibitory control, low intensity pleasure and
perceptual sensitivity) showed greater growth in reading skills (.20) than children who were rated
lower in effortful control (Deater-Deckard, Mullineaux, Petrill, & Thompson, 2009). Behavior

33

problems can lead to downward deflections in achievement across time (Jimerson, Egeland, &
Teo, 1999) and a lack of behavior problems is associated with improvement in reading after
encountering difficulties in first grade (Spira, Bracken, & Fischel, 2005). A Finnish study
indicated that children in preschool who were rated as less task-oriented, more ego-defensive and
more socially dependent were more likely to be poor readers at the end of second grade
(Poskiparta, Niemi, Lepola, Ahtola, & Laine, 2003). An Australian study demonstrated the
persistence of reading difficulties from second to sixth was significantly related to early behavior
problems (Smart, Prior, Sanson, & Oberklaid, 2001). Rabiner, Coie and colleagues (2000)
demonstrated that inattention measured in early Kindergarten was negatively associated with
reading achievement in later Kindergarten (-.29) and first grade (-.29) when controlling for IQ
and earlier reading skills. Additionally, teacher ratings of attention in second grade were
associated with fifth grade reading (-.10). There was some evidence of a reciprocal relationship
with Kindergarten reading being negatively associated with first grade ratings of inattention
(-.18). However, when additional analyses were conducted to determine the nature of the causal
relationship between reading and inattention, they only found a significant causal pathway
between inattention and reading achievement with inattention associated with decreases in
reading achievement from -.52 to -.86. Yet there is also evidence of a reciprocal relationship. A
Finnish study (Onatsu-Arvilommi & Nurmi, 2000) showed that high levels of task-avoidant
behavior measured before school entry predicted lower levels of reading skills across first grade,
both from Time 1 to Time 2 (-.24) and from Time 2 to Time 3 (-.19). In turn, a low level of
reading skills at school entry predicted a high level of task-avoidant behavior from Time 1 to
Time 2 (-.18), but not from Time 2 to Time 3 (-.07). Another Finnish study indicated that
children with lower reading skills had higher rates of externalizing and internalizing problems

34

across preschool and first grade. Additionally both externalizing and internalizing behavior were
related to substantially lower levels of reading performance at the end of first grade (Halonen,
Aunola, Ahonen, & Nurmi, 2006). Stipek and Miles (2008) demonstrated that aggression had a
negative effect on achievement through reductions in task engagement associated with
heightened teacher conflict related to the student’s heightened level of aggression, but they also
found that lower achievement showed a weak association with increased aggression. Finally,
another American study (Trzesniewski et al., 2006) found that earlier behavior problems were
associated with lower levels of reading achievement at age 7, but they also found that early
reading skills were associated with a higher rate of behavior problems between ages 5 and 7.
Clearly, the presence of early behavior problems is associated with slower reading development,
which may in turn create a reciprocal relationship between reading development and behavior
problems, with continued academic failure leading to increased behavior problems.
Behavior problems often emerge before school entry and can be associated with later
academic achievement. Arnold (1997) demonstrated the correlation between disruptive behavior
and early academic skills was -.37 at age 3 and grew to -.84 by age 6. Students who exhibit early
behavior problems often lack the necessary skills to meet the academic and behavioral demands
of the typical classroom (DuPaul & Stoner, 1994, Rothbart & Bates, 1998). Early behavior
problems are strongly associated with self-regulation difficulties (Howse, Calkins, Anastopoulos,
Keane, & Shelton, 2003; Eisenberg et al., 2000), which means the individual has difficulty
focusing or shifting attention, demonstrating persistence on tasks, activating or inhibiting
behavior, and responding adaptively to novel situations (Rothbart & Bates, 1998). These deficits
in self-regulation have a profound impact on the classroom behavior of these students. Rather
than appropriately engaging in academic tasks, they are typically involved in other disruptive

35

non-academic behavior (e.g., noncompliance, aggression, inattention, impulsivity, hyperactivity,
arguing, and rule breaking; Abikoff et al., 2002; Junod et al., 2006; Waller, 2006). These
difficulties preclude the active involvement in classroom tasks that facilitate learning for these
students.
Students who exhibit early behavior problems typically have lower rates of academic
engagement (DuPaul et al., 2004). This is unfortunate because student engagement is one of the
best predictors of academic achievement (Fredericks, Blumenfeld, & Paris, 2004; Greenwood,
Horton, & Utley, 2002). In one study, 44% of elementary students and 75% of middle school
students who reported higher levels of engagement had higher levels of academic achievement
(Klem & Connell, 2004). Students who are academically engaged are better able to benefit from
classroom instruction and have higher rates of time on task (Blair et al., 2004; Karweit, 1989).
Further, engaged students are afforded additional opportunities to participate in academic tasks
(DiPerna, Volpe, & Elliot, 2002) and have more supportive relationships with teachers (Good &
Brophy, 1994). These factors consistently predict higher achievement (Gest, Welsh, &
Domitrovich, 2005; Leach & Dolan, 1985). Conversely, lack of engagement is associated with a
number of negative outcomes such as higher disciplinary rates, absenteeism, and school dropout
(Croninger & Lee, 2001; Finn, 1989; Finn & Rock, 1997).
Further children with behavior problems often develop negative, conflictual relationships
with teachers, which further hinders their academic performance (Birch & Ladd, 1997; Ladd &
Burgess, 1999; Pianta, Steinberg, & Rollins, 1995). The social interactions of these children tend
to be abrasive and uncooperative (Dishion et al., 1995; Patterson et al., 1992). Further, they
typically exhibit more inappropriate, disruptive behavior and are less likely to comply with
teacher requests (DeMartini-Scully, Bray, & Kehle, 2000). Teachers find interacting with these

36

students to be aversive (Carey, 1998; Gunter & Coutinho, 1997) and respond to them with less
support and more punishment than other children (Brophy & Evertson, 1981; Little & Hudson,
1998). The majority of the interactions these students have with teachers are negative or neutral
(Jack et al., 1996). These students rarely receive praise for appropriate behavior (Shores et al.,
1993), and teachers are more likely to respond to their inappropriate behavior with aversive
consequences (Wehby et al., 1993). Van Acker, Grant and Henry (1996) showed that students
with behavior problems received little attention for appropriate behavior whereas they often
received reprimands for inappropriate behavior. Brendgen and colleagues (2006) demonstrated
that 15% of the students in their study, primarily boys who displayed antisocial and inattentive
behavior, experienced psychological abuse from the teacher in the form of verbal attacks on the
student’s character or ability (e.g., name calling, yelling, or public ridicule) or as acts of neglect
(e.g., ignoring). In a recent publication Jenson and colleagues (2004) described classrooms as a
“sea of negativity” for students with behavior problems.
The relationship that students form with teachers is an important source of stress or
support within the classroom environment and can have a significant influence on the students’
school performance. Students who have a close relationship with their teacher, characterized by
warmth, affection, and open communication, have higher rates of classroom participation, school
liking, cooperation in the classroom and self-directed behavior. In contrast, students who have
more conflictual relationships with their teacher, marked by a pattern of discordant relationships
through negative interactions and cognitions, have lower rates of these positive classroom
behaviors but higher rates of attention problems, classroom misconduct, aggression, school
avoidance, and disciplinary issues (Birch & Ladd, 1997; Ladd & Burgess, 1999; Ladd &
Burgess, 2001; Mantzicopoulos, 2005; Murray & Murray, 2004; Pianta and Steinberg, 1992;

37

Pianta, Steinberg, & Rollins, 1995). Also a student’s ability to gain academic assistance through
appropriate classroom interactions is strongly related to the student’s emotional security with the
teacher (Lynch & Cicchetti, 1997; Pallas et al., 1987; Pianta, 1992; Pianta & Nimetz, 1991) so
students with more negative teacher-student relationships are less likely to be able to effectively
use the teacher as a source of academic support (Newman, 1990; Newman & Schwager, 1993).
Teachers are also likely to modify their interactions with students with disruptive
behavior problems, influencing the quality and quantity of instruction received by these students
(Carr, Taylor, & Robinson, 1991; Good & Brophy, 1994; Gunter & Coutinho, 1997; Shores,
Gunter, & Jack, 1993; Walker et al., 1998; Wehby et al., 1998). In 1991, Carr and colleagues
conducted a small study examining teacher interactions with four problem students (students
with a history of disruptive behavior problems) compared to their interactions with four more
average students with no history of disruptive behavior. They found that teachers tend to alter
their method of teaching when interacting with children perceived to have disruptive behavior
problems. Teachers had a median number of 377 (range of 295 to 506) intervals of instructional
behavior for average students compared to a median number of 226 (range of 69 to 323) intervals
of instructional behavior for problem students. Teachers presented a median number of 147
(range 88 to 205) task commands to average students compared to a median number of 61 (range
of 0 to 115) task commands to problem students. The findings also demonstrated that the content
of instruction offered by the teachers differed for problem students when compared to average
students. Teachers presented all eight possible tasks to average students but only 2 out of 11
teachers presented all eight tasks to the problem student; in fact only 16.9% of the total
instruction offered to problem students would be considered challenging (Carr, Taylor, &
Robinson, 1991). The findings cannot be considered conclusive because of the small sample size.

38

However these findings do suggest that teachers alter their interactions with students with
disruptive behavior problems in significant ways.
Students who enter school with behavior problems are at a distinct disadvantage. They
lack the necessary self-regulatory skills to effectively engage in classroom instruction; they often
have more negative, less supportive relationships with teachers and they can receive lower
quality instruction. It is quite possible the persistence of reading difficulties for children from
low-SES families is related to the strong association between reading difficulties, behavior
problems and disadvantaged circumstances. These children who enter with much lower skills
and are therefore in much greater need for quality instruction may be less able to benefit from the
instruction received (Clark, Baker, Grant, 2007; Gettinger, 1985; Gettinger & Seibert, 2002). A
study evaluating the effectiveness of three instructional approaches for improving the reading
skills of young children with weak phonological skills found that 24% of the children did not
show significant improvement even using the most effective instructional approaches. Further
analyses to identify specific individual factors that predicted response to the reading
interventions indicated the most consistently important variables were rapid naming, home
background and classroom behavior ratings (Torgeson et al., 1999). In a review of similar
research studies Al Otaiba and Fuchs (2002) found that anywhere from 8% to 80% of children
who were at risk for reading difficulties did not benefit from generally effective reading
interventions. Attention was listed among the factors related to non-response in 7 out of 9 of the
studies reviewed. A meta-analysis of 30 studies (Nelson, Benner, & Gonzalez, 2003) identifying
learner characteristics that influence the treatment effectiveness of early literacy interventions
supported these earlier findings. This meta-analysis revealed that the presence of behavior
problems was moderately correlated (standardized effect size = .46) with the failure of these

39

interventions. Further the strength of this relationship was second only to rhyming (.53) and
rapid naming (.51) making behavior problems a stronger predictor of persistent reading
difficulties than phonological awareness (.42), alphabetic principle (.35), memory (.31) and IQ
(.26). Finally, a study by Rabiner, Malone and colleagues (2004) confirmed a relationship
between the persistence of reading difficulties and behavior problems. Children from
disadvantaged backgrounds were provided a phonics-based, mastery-oriented reading
intervention designed to promote the skills of students with low school readiness. The impact of
the tutoring on reading achievement was greatest for children with low externalizing behavior
problems. For children whose behavior problems approached clinically significant levels, the
impact of tutoring on reading achievement was minimal. The reading achievement for students
who were poor early readers with comorbid behavior problems was significantly worse than any
other groups (students with behavior problems and no reading difficulties and students with
reading difficulties but without behavior problems) even for students receiving a reading
intervention.
These findings may highlight the need to learn more about the relationship between lowSES and slower reading development. Most of the current focus on addressing the deficits in
reading proficiency for this population is related to a belief in a Matthew Effect, for example the
federally supported reading intervention programs, Reading First and Early Reading First.
However, if behavior problems are associated with slower reading development and if the
presence of behavior problems is associated with the lower effectiveness of evidence-based
literacy interventions, as many of the studies discussed above seem to suggest, then early reading
intervention alone may not be sufficient to ameliorate the reading deficits commonly observed
among children from low-SES families. Early evaluations of these federally funded programs

40

may also point to the need for more information about the relationship between low-SES,
behavior and reading development. Although the Reading First Impact Study: Interim Report
(Gamse, Bloom, Kemple, & Jacob, 2008) indicated significant increases in instructional time
spent on the five essential components of reading instruction, it failed to show any increases in
reading achievement. While there may be many reasons for these findings, it is interesting to
note that the average academically engaged time in the observed classrooms was only 46.9% for
the first graders and 49.7% for the second graders, which is considerably lower than what would
be expected. Prior research indicates that the typical level of task engagement in general
education classrooms ranges from 75% to 85% (Rich & Ross, 1989; Walker & Severson, 1990).
Children may link this lower task engagement to the higher rate of behavior problems
experienced from low-SES backgrounds. The substantial evidence indicating the detrimental
influence of behavior problems on the effectiveness of instruction, especially considering the
influence of comorbid behavior problems on reading development, underscores the critical need
for a better understanding of the environmental factors related to both the reading development
and the behavioral adjustment of children from low-SES families.

Summary and Hypotheses:
Reading development is a critical skill for participation in American society, yet children
from low-SES families have historically had significantly lower reading achievement when
compared with their peers (Bowey, 1995; Lonigan et al., 1998; Lee, Grigg & Donahue, 2007;
Molfese, Modglin, & Molfese, 2003; Whitehurst, 1997). Concerns over these achievement
deficits have led to an increased focus on providing early intervention to improve the early
reading skills of these children, with the expectation that this approach would considerably

41

reduce the number of children with reading difficulties (Burns, Griffin, & Snow, 1999; Entwisle
& Alexander, 1993; McCardle, Scarborough, & Catts, 2001; Shonkoff & Phillips, 2000; Snow,
Burns & Griffin, 1998; Storch & Whitehurst, 2002; Torgeson, 2002). These recommendations
are based upon the belief in a Matthew Effect for reading, in which the lower early reading skills
of children from low-SES families at school entry is expected to create a cumulative
disadvantage in reading achievement over time. However, there has only been limited support
for a Matthew Effect in reading (Aram, 2005; Cunningham & Stanovich, 1997; Juel, 1988;
Durham et al., 2007) and several of these studies have significant limitations. Other research has
supported a compensatory model of reading development in which the children, who enter school
with lower skills, actually show the greatest rate of growth and may follow multiple trajectories
of reading development (Aarnouste et al., 2001; Aunola et al., 2002; Leppanen et al., 2004;
Lerkkanen et al., 2004; Parrilla et al., 2005; Phillips et al., 2002; Shaywitz et al., 1995; Spira,
Bracken & Fischel, 2005). The increased focus on providing early intervention to increase the
early reading skills of children from low-SES families makes sense but the focus on addressing
deficits in early reading skills may be too narrow. It is also important to consider other factors
that may be influencing reading development
Hypothesis 1: Children will follow multiple trajectories of reading development. Some
children from low-SES families, who enter school with lower initial reading skills, may
follow a cumulative trajectory of reading development, consistent with the Matthew
Effect. However, it is expected that the majority of these children will follow a
cumulative trajectory of reading development through kindergarten, but show a
compensatory developmental trajectory in later grades.

42

Yet merely identifying the reading trajectories of children from low-SES families is
insufficient to support the creation of effective interventions for these children, because this
narrow focus on reading development ignores the larger context of development within a lowSES family. Children from low-SES families are also more likely to enter school with increased
behavior problems (NCES, 2000; Zill et al., 1995) and these comorbid behavior problems are
likely to decrease the effectiveness of classroom instruction in reading and specific interventions
for reading deficits (Al Otaiba & Fuchs, 2002; Barriga et al, 2002: Deater-Deckard et al., 2009;
Rabiner et al., 2004; Reid et al., 2004; Spira, Bracken & Fischel, 2005; Torgeson et al., 1999;
Velting & Whitehurst, 1997).

Hypothesis 2: The presence of comorbid behavior difficulties is expected to be
associated with distinct reading trajectories for children from low-SES families.
1) Children who enter school with low early reading skills, but no comorbid
behavior problems are expected to show a compensatory trajectory of reading
development.
2) Children who enter school with average early reading skills, but a comorbid
behavior problem are expected to show a slower rate of growth.
3) Children who enter school with both low early reading skills and behavior
problems are expected to show the slowest growth in reading and may follow a
cumulative trajectory.
4) Children who show decreases in behavior problems after school entry are also
likely to see a concomitant improvement in reading achievement; conversely the

43

later emergence of behavior problems is expected to be associated with declines
in reading growth trajectories.

44

CHAPTER 3
METHOD
Participants
The sample for this study was drawn from the Early Childhood Longitudinal StudyKindergarten Cohort (ECLS-K). This study, developed under the sponsorship of the United
States Department of Education National Center for Education Statistics (NCES), is following a
nationally representative cohort of children from kindergarten into high school. It provides a
comprehensive picture of children’s early family, neighborhood, and school experiences over
time. During the fall of the 1998-99 school year, the base year of data collection, a national
probability sample of 21,260 children was recruited from 1,277 kindergarten classrooms in about
800 public and 200 private schools. They were assessed at entry to kindergarten and in the spring
of their kindergarten year. Two more waves of data were collected in the fall and spring of the
1999-2000 school year when most, but not all, of the base year children were in first grade. The
fall-first grade data collection was limited to a 30 percent sub-sample of schools, but the springfirst grade data collection contained the full sample. A fifth wave of data was collected in the
spring of the 2001-02 school year and a sixth wave of data was collected in the spring of the
2003-2004 school year when most of the sampled children were in third grade and fifth grade
respectively. A sub-sample of 17,565 children was retained in the longitudinal kindergarten to
fifth grade sample.
Given the nature of longitudinal data collection, there is a large amount of missing data
for the outcome measures (8,696 children missing IRT scores and 10,306 children missing
teacher ratings of behavioral adjustment for at least one of five assessment data points). To
determine the nature of the missing data a series of comparisons were ran on the outcome
measures and various demographic variables. The students who were missing data had
45

significantly lower reading scores (see Table 1) and higher teacher rated behavior problems (see
Table 2) across all time points. These children were also more likely to come from families with
lower socioeconomic status and have parents with less education (see Table 3). Additionally,
proportionally more single parent, non-English speaking and minority families were excluded.
The nature of the missing data precluded conducting techniques (e.g., multiple imputation, hot
deck imputation, etc.) to replace the missing data (Little & Rubin, 1987). It is also important to
note that because the students that were retained for further analysis were significantly different
in many respects from those that were excluded because of missing data, the resulting analytic
sample is not representative of American school children. The final sample to be used in this
study consists of 6,698 children (see Table 4 for demographic information).

Measures
Child Outcomes
Reading Achievement. One-on-one, untimed direct child assessments, taking between 50
to 96 minutes, were administered using both hard-copy instruments and computer-assisted
personal interviewing (CAPI). Prior to each assessment period, assessors (anywhere from 102 to
343 in number depending on the assessment period) and field supervisors (anywhere from 39 to
112 in number) were trained through 8 hours of home study focused on study design, field
procedures and computer keyboard skills followed by several in-person training sessions. These
in-person training sessions included an overview of study activities to date, interactive lectures
based on the direct child assessments and the parent interview, practice parent interviews in pairs
using role-play scripts, practice direct child assessments using role-play scripts, direct child
assessment pre-certification exercises on each form of the direct child assessments, training on
appropriate use of standardized procedures, techniques for parent refusal avoidance, and

46

strategies for building rapport with children. All assessors and field supervisors were required to
complete a two-step certification process before data collection began to ensure all direct child
assessments were completed in a standardized manner. The first step of the certification process
was the completion of written exercises on each level form of each of the assessment domains.
Assessors completed written exercises following an interactive lecture and individual practice
with each form of an assessment domain. Each exercise required the trainees to score an
assortment of possible responses to certain assessment items from the level form that was most
recently discussed. Assessors who did not achieve a passing score were asked to attend an
additional training session to review the items and then retook the same exercises they had
previously failed. In the second stage of the certification process the assessors were observed
conducting a direct child assessment with children brought on site to the training session.
Trainers rated the assessors on skills such as rapport with the child, avoidance of coaching or use
of inappropriate probing, following proper administration procedures, and pacing. While the
assessor administered the assessment, an observer certified on the assessment simultaneously
coded the child’s answers to pre-selected open-ended questions. After the assessment was
completed, the answers recorded by the assessor were compared with those recorded by the
observer. Discrepancies in any of the recorded answers were included in the assessor’s overall
score on a certification form. Assessors who scored 85 percent or above were certified qualified
to administer the child assessments; those who failed to reach this standard were re-certified
following remedial training (Tourangeau et al., 2006).
The field staff was organized into work areas, with each data collection team consisting
of one field supervisor and two or more assessors. Continuous quality assurance procedures were
employed during all data collection activities, but with a particular focus on the assessments.

47

Field supervisors conducted two on-site observations of the child assessments for each assessor.
During these observations the assessor was evaluated on the rapport building skills, the
compliance with appropriate standardization procedures and the scoring of protocols. Feedback
was provided to the assessors on the strengths and weaknesses of their performance and, when
necessary, remedial training was provided in areas of weakness. The field supervisor, as a part of
the child assessment observations, completed “validation items,” which were at least one item
that both the observer and the assessor scored. The items that were scored by both the assessor
and observer had open-ended responses that called for interpretation on the part of the assessor to
determine whether a child’s response was correct. A measure of interrater reliability was
obtained by comparing the extent to which assessors and observers agreed on the scoring of
these “validation items.” The interrater reliability for these “validation items” was fairly high
with the lowest being .96.
Prior to direct child assessment during kindergarten and first grade, a language-screening
assessment, the Oral Language Development Scale (OLDS; Duncan & De Avila, 1998) was
administered to those children identified from their school records (or by their teacher if no
school records were available) as coming from a language minority background (meaning that
their primary home language was not English). This screening test was used to determine the
child’s proficiency in the English language. Children who passed the language screener received
the full English direct assessment battery. Children who did not pass an established cut score on
the language screener received a reduced version of the assessment battery, which did not
include a reading assessment. These students are not included in the analyses because they did
not have reading data available for the first assessment.

48

Reading assessments were individually administered at all five time points (i.e. fall &
spring kindergarten, fall & spring of first grade, spring of third and fifth grade). A team of
elementary education specialists developed the pool of test items. Test items were reviewed by
elementary school curriculum and content area specialists for appropriateness of content and
difficulty, and for relevance to the test framework. In addition, items were reviewed for issues
related to sensitivity to minority concerns. Items that passed these content, construct, and
sensitivity screenings were field tested before being administered to the participants. The validity
of the content in the ECLS-K item pools was established by comparing the results of the ECLSK with scores on the Woodcock-McGrew-Werder Mini-Battery of Achievement (MBA;
Woodcock, McGrew, & Werder, 1994) that was also administered during the field test. The
reliability between the MBA and the reading assessments were between .73 and .77 (no further
information was provided regarding field test procedures or sample sizes). At each assessment
period new items were chosen to extend the longitudinal scales and there were grade-appropriate
changes in content and format. For example, in the kindergarten and first-grade reading
assessment, children read short sentences. By fifth grade, the new passages were more complex
and more text was presented on a single page than had been the case in the third-grade reading
assessment.
A two-stage assessment approach was used; children’s responses from an 18-25 item
routing test were used to select the appropriate difficulty level of the second stage form. The
kindergarten and first grade reading assessment contained five proficiency levels reflecting a
progression of skills and knowledge: 1) identifying upper- and lower-case letters of the alphabet
by name; 2) associating letters with sounds at the beginning of words; 3) associating letters with
sounds at the end of words; 4) recognizing common “sight” words; and 5) reading words in

49

context. The third-grade reading assessment also included five proficiency levels: two retained
from the kindergarten and first-grade assessments, recognizing common “sight words” and
reading words in context; plus three new levels 1) making inferences using cues that were
directly stated with key words in text (literal inference); 2) identifying clues used to make
inferences (extrapolation), and using personal background knowledge combined with cues in a
sentence to understand use of homonyms; and 3) demonstrating understanding of author’s craft
and making connections between a problem in the narrative and similar life problems
(evaluation). The fifth-grade reading assessment included three items from the third-grade
assessment: making inferences using cues that were directly stated with key words in text (literal
inference); identifying clues used to make inferences (extrapolation), and using personal
background knowledge combined with cues in a sentence to understand use of homonyms; and
demonstrating understanding of author’s craft and making connections between a problem in the
narrative and similar life problems (evaluation) in addition to comprehension of biographical and
expository text (evaluating non-fiction).
Children’s performances on the reading assessments were calculated in a variety of ways;
number-right scores, Item Response Theory (IRT) scores, standardized t-scores, and criterionreferenced proficiency scores are provided. For the purposes of this study, the IRT scores and
criterion-referenced proficiency scores will be used to measure children’s progress in reading.
Number-right scores are counts of the raw number of items a child answered correctly. However,
due to the nature of the assessment methodology these scores are not comparable across grade
levels, making it difficult to make direct comparisons between the grade-levels. Standardized tscores are transformations of the IRT theta (ability) estimates, rescaled to a mean of 50 and
standard deviation of 10 using cross-sectional sample weights for each wave of data. These

50

scores provide a norm-referenced measurement of achievement relative to the population as a
whole and cannot be used to represent mastery of a particular set of skills. In contrast, IRT scores
use the overall pattern of right and wrong responses and the characteristics of each item to
estimate a child’s ability. IRT scores represent estimates of the number of items students would
have answered correctly at each point in time if they had taken all of the 186 questions in all of
the first- and second-stage reading forms administered. IRT scoring makes it possible to measure
the longitudinal gain in achievement over time, even though the assessments that are
administered are not identical at each point. The common items present in the routing test and in
overlapping second-stage forms allow the scores to be placed on the same scale, even as the twostage design adapts to children’s growth over time. Although IRT scale scores are useful in
identifying cross-sectional differences among subgroups in overall achievement level, and may
be used as longitudinal measures of overall growth, the gains made at different points on the
scale have qualitatively different interpretations. For example, children who made gains in
recognizing letters and letter sounds are learning very different skills than those who are making
the jump from reading words to reading sentences, although the gains in number of scale score
points may be the same. To allow for group comparison in growth for each specific skill area
secondary analyses will be performed on proficiency probability scores. These scores are
criterion-referenced measures of proficiency in specific skills derived from the overall IRT
model. They are useful as longitudinal measures of change because they show not only the extent
of gains but also where on the achievement scale the gains are taking place. Thus, they can
provide information on differences in skills being learned by different groups, as well as the
relationships with processes, both in and out of school, that correlate with learning specific skills.
The proficiency level at which the largest change is taking place is likely to be different for

51

children with different initial status, background, and school setting. Changes in proficiency
probabilities over time may be used to identify different developmental profiles of children from
low SES families and may lead to a better understanding of the reading development of children
with behavior problems.
The IRT scores were re-estimated in the longitudinal dataset using a process of vertical
equating; the kindergarten and first-grade responses were pooled with the third- and fifth-grade
data to stabilize the longitudinal estimates and the maximum values of the scale scores were
extended to include the more difficult items administered in the fifth-grade assessments. The
scale scores for each round of reading assessments were defined based on performance on all
tasks administered up to and including the fifth grade assessments allowing for the calculation of
meaningful estimates of gains over time. The IRT scale scores on the reading assessments range
from 0 to 186, the means and standard deviations are reported by grade level in Table 5. The
proficiency probability scores on the reading assessment range from 0 to 1, the means and
standard deviations are reported by grade and skill level in Table 6. The most appropriate
estimate of the reliability of the reading assessment is the reliability of the overall IRT ability
estimate, theta. This number is based on the variance of repeated estimates of theta, and applies
to all of the scores derived from the theta estimate, namely, the IRT scale scores, T-scores, and
proficiency probabilities. The reliability measured by estimates of theta was acceptable at all
assessment points (ranging from .91 in the fall of kindergarten to .96 in the spring of first grade).
Teacher Rating of Behavior Problems. Both teachers and parents provided rating of the
child’s behavioral adjustment using the Social Rating Scale (SRS), a measure adapted by the
ECLS-K from the Social Skills Rating Scale (SSRS; Gresham & Elliot, 1990). Only the teacher
ratings of behavioral adjustment are used in this study. Parent ratings and teacher ratings of

52

behavioral adjustment often capture very different types of behavior and there is often low interrater agreement between the two reports (Achenbach, McConaughy, & Howell, 1987; Johnston
& Murray, 2003; McDermott, 1993). Because the focus of this study is the influence of behavior
problems on reading development rather than the more global development of behavioral
adjustment it makes sense to focus exclusively on ratings of the child’s behavior in the
classroom. Children’s behavior in the classroom is more likely to directly influence reading
development than the children’s behavior at home or in the community.
Teachers rated individual student’s behavioral adjustment as part of a larger selfadministered questionnaire that was collected at every wave of data collection. Teachers used a
frequency scale (items rated from 1 (never) to 4 (very often) or N/O for no opportunity to
observe behavior) to report how often students exhibited certain social skills and behaviors. Five
scales were developed based on the teachers’ responses; three of the scales capture positive
aspects of children’s behavior (approaches to learning, self-control, & interpersonal skills) and
two scales represent problem behaviors (externalizing problem behaviors & internalizing
problem behaviors). The scale score is the mean rating on the items included in the scale and
scores were only computed if the student was rated on at least two-thirds of the items in that
scale. The scales are as follows:
Approaches to Learning scale measured behaviors that affect the ease with which
children can benefit from the learning environment. It included six items measuring the
child’s attentiveness, task persistence, eagerness to learn, learning independence,
flexibility and organization. This item was reverse coded so that higher numbers
represent lower behavioral adjustment.

53

Self Control scale included four items that indicate the child’s ability to control behavior
by respecting the property rights of others, controlling temper, accepting peer ideas for
group activities and responding appropriately to pressure from peers. This number was
reverse coded so that higher numbers represent lower behavioral adjustment.

Interpersonal Skills scale has five items that rate the child’s skill in forming and
maintaining friendships; getting along with people who are different; comforting or
helping other children; expressing feelings, ideas, and opinions in positive ways; and
showing sensitivity to the feelings of others. This number was reverse coded so that
higher numbers represent lower behavioral adjustment.

Externalizing Problem Behaviors scale has five items that rate the frequency of acting out
behaviors such as arguing, fighting, getting angry, acting impulsively, and disturbing
ongoing activities. Higher numbers on this scale represent lower behavioral adjustment.

Internalizing Problem Behaviors scale has four items that ask about the apparent
presence of anxiety, loneliness, low self-esteem and sadness. Higher numbers on this
scale represent lower behavioral adjustment.
The ranges, means, standard deviations and split-half reliabilities for each scale are presented by
assessment period in Table 7. The scales were totaled to produce a single number representing
overall behavioral adjustment, with higher numbers representing a higher teacher rating of
behavior problems.

54

Family Measures.
Parents or guardians, who resided with the sample child, answered questions using
computer assisted telephone interviewing (CATI) or computer assisted personal interviewing if
the parents did not have a phone. Interviews lasted approximately 44 minutes and included
approximately 330 questions covering children’s school experiences, childcare, family
characteristics, home environment, parenting behavior and household income. The parent
interview was conducted primarily in English, but provisions were made to interview parents
who spoke other languages. The questionnaire was translated into Spanish, Chinese, Lakota, and
Hmong languages; bilingual interviewers were trained to conduct the parent interview in the
parent’s primary language. The child’s mother was the respondent in 81 percent of the cases and
the child’s father in 8 percent. Other adults completed the parent interview in 11 percent of the
cases (typically grandparents of the sample child). Information from the parent interview was
used to derive the information necessary to determine socioeconomic status.
Socioeconomic Status. Two primary approaches were used to measure socioeconomic
status. The family’s SES was measured using a composite variable computed at the household
level that was derived by the U.S. Department of Education, which is derived from a logarithm
of five variables measuring family income, mother’s education, father’s education, mother’s
occupational prestige, and father’s occupational prestige (see Tourangeau, Nord, Lê, Pollack, &
Atkins-Burnett, 2006 for details). This composite has been computed for each year of data
collection and is provided as both a continuous variable and a categorical variable. The
continuous variable ranges from -4.75 to 2.75 with the higher values representing higher SES,
mean = .16, SD = .77. The SES composite is the average of up to five measures, each of which
was standardized to have a mean of 0 and a standard deviation of 1, hence the negative values.

55

For ease of interpretation this scale was converted to a continuous positive scale by adding 4.84
to each value of the original SES continuous scale, mean = 5.00, SD = .77. A categorical variable
representing the quintile for the value of the SES continuous variable (see Table 4 for the
percentages) was also used to indicate SES and ranges from 1 (representing the lowest SES
category) to 5 (representing the highest SES category).
During the parent interview information was gathered about the family income as well as
the current occupation and the highest educational level achieved by both parents. Information
about the family income was collected in the spring of each assessment year. It ranged from 0 to
a million dollars with only 200 people having more than $200,000 and 880 having less than
$10,000; the median income was $45,000 (25th quartile = $24,000, 75th quartile = $70,000) and
the mean was $55,618 with a standard deviation of 57,616. Approximately 2,000 families (or
16.7% of the initial kindergarten sample) were below the 1998 Federal poverty threshold.
Parents’ occupation was recoded to reflect the average of the 1989 General Social Survey (GSS;
Nakao & Treas, 1992) prestige score. This 22-position scale ranks occupations according to their
relative prestige (see Tourangeau et al, 2006 for a description of the aggregated categories that
were used for coding occupation in the ECLS-K). Information is provided for each parent
separately; this scale ranges from 29.6 to 75.5, with higher codes representing more prestigious
occupations (see Table 8 for the frequency of occupations found in the initial kindergarten
sample). The highest level of education is provided individually for each parent and as a
composite indicating the highest educational level achieved for any parent or guardian currently
residing with the child (see Table 4).

56

Data Analysis
Initially, a split-plot ANOVA was completed to determine if there were mean differences
across levels of SES, consistent with previous research, on IRT reading scores and teacher
behavior ratings. After this analysis, the hypotheses were tested.
The first hypothesis predicted that there would be multiple trajectories of reading
development and that although some children from low-SES backgrounds would follow a
cumulative trajectory of reading development, the majority would follow a compensatory
trajectory. Bast and Reitsma (1997) argue that evidence for a cumulative model must consist of
stable rank ordering and increasing individual differences in reading achievement over time.
Two analytical models have been proposed to provide this evidence (Aunola, Leskinen, OnatsuArvilommi, & Nurmi, 2002; Bast & Reitsma, 1997,1998). The first analytical model, the simplex
model, is particularly well suited to longitudinal series in which there is occasion-to-occasion
transmission, that is to say that the observation at time 2 depends on the observation at time 1,
and in turn the observation at time 3 depends on the observation at time 2, and so forth. Simplex
modeling (Jöreskog, 1979) is a special case of structural equation modeling, in which a factor or
construct is estimated across several measurements by using only one indicator variable for each
measurement. The advantage of a simplex model is that it allows for other sources of variance
than those responsible for initial performance differences to be incorporated into the model. This
model represents the expectation that different factors affect the inter-individual variance at
different times. The second analytical model, latent growth curve modeling (Muthén & Khoo,
1998), allows for the analysis of associations between initial level and growth over time. Within
this model, two latent factors, level and shape, represent dimensions of individual differences in
growth over time. These common factors are used to describe the values of the status and growth

57

of individuals on a trait at all occasions. Using this type of analysis allows the examination of the
association between the initial level and the growth in reading skills across time. Since both
models represent different components of the reading trajectory, both models were used to test
the first research hypothesis.
To examine the stability and variance of reading performance from kindergarten through
fifth grade, a simplex model with the continuous SES composite as a covariate (see Figure 1)
was estimated to represent the relationship between reading development at subsequent time
points. A cumulative model of the reading development of children from low-SES backgrounds
would be represented by a significant relationship between low-SES and lower entry scores, high
stability in reading skills at each time point and increasing variance in assessment points over
time. This finding would provide evidence that the lower entry skills of children from low-SES
backgrounds is predictive of slower growth in reading development over time in comparison
with their more advantaged peers. A compensatory model of the reading development of children
from low-SES backgrounds would be represented by decreasing variance in reading skills across
time despite evidence demonstrating the lower entry skills of children from low-SES
backgrounds.
A latent growth curve model with the continuous SES composite as a covariate was
estimated to explore the association between the level of entry reading skill and the growth of
those skills (i.e. developmental trend) over the five measurement occasions (see Figure 2),
however the fit indices that this model did not fit the data well. Considering previous research
concerning the nature of reading growth across time, with growth in reading skills being
different within each time frame (Bast & Reitsma, 1998; Leppanen, 2004), it was decided that a
piecewise growth model would better fit the data. A piecewise growth model is a special case of

58

general growth curve modeling, which allows for the separate estimation of growth trajectories at
different phases of development (Li et al., 2001; Muthén & Muthén, 1998). Using this type of
analysis not only allowed the examination of the association between the initial level and the
growth in reading skills at different developmental periods differently but it also allowed the
association between the separate growth components to be examined separately. A piecewise
growth curve model with the continuous SES composite as a covariate was estimated to explore
the association between the level of entry reading skill and the growth of those skills (i.e.
developmental trend) over the five measurement occasions, from the beginning to the end of
kindergarten, from the end of kindergarten to the end of first grade, from the end of first grade to
the end of third grade and from the end of third grade to the end of fifth grade (see Figure 3). It
was expected that SES would be positively associated with initial scores; children from low-SES
backgrounds should enter school with much lower reading ability on average than their more
advantaged peers. A cumulative model of the reading development of children from low-SES
backgrounds would be represented by a significant positive relationship between the level and
slope component of the model, with lower early reading skills (level) predicting slower growth.
A compensatory model of the reading development of these children would be represented by a
significant negative relationship between the level and slope component of the model, indicating
that children with lower entry skills grow faster to compensate for their slower start.
The second hypothesis predicted that the socioemotional development of children from
low-SES families would be associated with their reading development. Several analyses were
conducted to test this hypothesis. To examine possible multi-construct associations between
reading development and teacher rated behavior, a multi-construct simplex model was estimated
with SES as a covariate. First, a simplex model for behavior model with the continuous SES

59

composite as a covariate (Figure 4) was estimated without any relationship to reading
development and then a multivariate simplex model was estimated by adding statistically
significant cross-construct paths between reading and behavior to the model (Figure 5). It was
expected that low SES would predict both lower initial reading skills and higher ratings of
behavior problems. Further, it was expected that higher ratings of behavior problems should be
associated with lower reading achievement over time.
To further explore the relationship between the reading development and teacher ratings
of behavior across time, a multivariate piecewise growth curve model was estimated. This
analysis allows the examination of possible multi-construct associations between the level and
shape components of the IRT reading scores and the teacher rating of behavior. To conduct this
analysis, a univariate latent growth curve model for behavior model with the continuous SES
composite as a covariate was estimated without any relationship to reading development (see
Figure 6), however the fit indices that this model did not fit the data well. Considering that
reading growth showed a different rate of growth within each time frame, it was decided that a
piecewise growth model would better fit the data. So a piecewise growth model for behavior
with the continuous SES composite as a covariate was estimated without any relationship to
reading development (see Figure 7). It was expected that children from low-SES families would
have significantly higher ratings of behavior problems at school entry (level) and would show
more stable or increasing ratings of behavior problems over time (shape) when compared to their
more advantaged peers. Once the univariate analysis for behavior ratings was completed, the
multivariate piecewise growth curve analysis was conducted with the continuous SES composite
as a covariate (Figure 8). It was expected that there would be a significant relationship between
the level and shape components of behavior ratings and the shape component of reading

60

development with higher ratings of behavior problems predicting lower reading achievement
even when controlling for entry levels of behavior and reading skill.
To determine the number and types of reading trajectories found in this sample, a
clustering-by-cases analysis for longitudinal data (Aunola, Leskinen, Onatsu-Arvilommi, &
Nurmi, 2002; Bergman, Eklund, & Magnusson, 1991; Bergman & Magnusson, 1997) was
completed. To carry out this analysis, a piecewise growth curve analysis was carried out.
Standardized scores for the latent level and shape components of reading achievement were
computed based on the results from the piecewise growth curve analysis. These new variables
were used as the criteria variables to group students into homogenous subgroups using analysis
of clustering by cases. The methodological process outlined by Aunola and colleagues (2002)
and recommended by Bergman and Magnusson (1991) was used to analyze clustering by cases.
First a hierarchical cluster analysis was conducted, selecting the squared Euclidian distance as a
similarity measure and using Ward’s (1963) methods to form initial clusters without restricting
their number. This analysis provided a dendrogram based on the distance of the clusters. This
dendrogram was inspected to determine the number of meaningful clusters in the sample. To
verify the final solution, alternate solutions of different numbers of clusters were examined in
terms of conceptual clarity and the number of students in each cluster. The appropriate number
of clusters was determined and then the final clusters were formed using SPSS quick cluster
analysis. This quick cluster procedure selects initial cluster centers according to MacQueen’s kmeans clustering methods. After this, the algorithm assigns each case to the nearest cluster
center. Several analyses were conducted until the solution stabilized, in these additional analyses
the earlier solutions were stored and used as initial centers in the next run. It was expected that
several reading trajectories would be found.

61

An examination of the group differences at the different measurement points was
conducted for the clusters identified based on their initial reading skill and growth trajectories.
Since IRT scale scores, which were used to derive groups, are good at measuring overall growth
but not in determining the specific area of growth (see discussion in Measures section), the
proficiency probability scores were used to allow for a descriptive group comparison in growth
for each specific skill area. Finally, a statistical comparison of differences in SES and behavior
ratings between the clusters was completed using a Kendall’s tau-c analysis.

62

CHAPTER 4
RESULTS
The results of the analyses are presented in several steps. First, the results of the analyses
to determine if there were mean differences across levels of SES, consistent with previous
research, on IRT reading scores and teacher behavior ratings are presented. Second, hypotheses
regarding the two research questions of the present study are tested. Finally, the number and
types of reading trajectories found in this sample through the use of a clustering-by-cases
analysis is presented with a discussion of the relationship between the different trajectories of
reading development and two independent variables, teacher rating of behavior and SES.
A split-plot ANOVA was conducted using time as the within factor and categorical SES
as the between factor. The reading scores of children from lower SES families were lower than
their peers from higher SES families, F(4, 5219) = 18.67, p < .001, Eta squared = .04 and there
was a significant SES by time interaction F(4, 5219) = 10.12, p < .001, Eta squared = .02. A
trend analysis suggests that there was both a linear and quadratic trend for time and a linear trend
for SES. Post hoc comparisons using the Tukey HSD test indicated that the mean IRT scores
(see Table 9 for means and standard deviations) were different between the lowest SES quintile
and all other SES quintiles except the fourth SES quintile, p < .01 for each comparison. The only
other group that showed a statistically significant difference from the other SES quintiles was the
highest SES group, which had IRT scores that were significantly higher than all but the second
SES quintile, p < .01 for each comparison. There was not a difference between the IRT scores of
the remaining SES quintiles. The behavior ratings of children from lower SES families indicated a
higher rate of behavior problems than their peers from higher SES families, F(4, 5219) = 4.94, p =

63

.001, Eta Squared = .002, however there was no significant interaction between time and SES
F(4, 5219) = 1.20, p = .26. A trend analysis suggests a linear and quadratic trend for time and a
linear and cubic trend for SES. Post hoc comparisons using the Tukey HSD test indicated that the
mean behavior ratings (see Table 10 for means and standard deviations) were only significantly
different between the first SES quintile and the fifth SES quintile, p < .001, there were no other
differences in behavior ratings between the other SES quintiles.

Simplex Model for Reading
The first hypothesis predicted that some children from low-SES families, who entered
school with lower initial reading skills, might follow a cumulative trajectory of reading
development, consistent with the Matthew Effect. However, it was expected that the majority of
these children would follow a cumulative trajectory of reading development through kindergarten,
but show a compensatory developmental trajectory in later grades. Two analytical models have
been proposed to provide evidence of a cumulative developmental trajectory, the simplex model
and the latent growth curve model (Aunola, Leskinen, Onatsu-Arvilommi, & Nurmi, 2002; Bast
& Reitsma, 1997, 1998). The simplex model (see Figure 1) allows an examination of the stability
and variance of reading performance across five measurement occasions. In this model, the latent
skill variables consisted of one indicator, the observed reading skill variable (the IRT reading
score) for that measurement occasion. The residual variances of the observed reading skill
variables were constrained to be equal across time to identify the model. The fit indices indicated
a good fit (Chi-Square = 98.5, df = 5, p < .001; CFI = .996; TLI = .99; RMSEA = .058, 90%
Confidence Interval .049 .to 069; SRMR = .01 and Hoelter’s Index = 618). Inspection of
64

modification indices suggested that estimating the covariance between the residuals of the latent
reading construct at Time 1 and Time 5 (standardized estimate = .10, t = 5.83, p < .001), between
Time 3 and Time 5 (standardized estimate = .14, t = 7.69, p < .001), and between Time 2 and
Time 4 (standardized estimate = -.05, t= -2.44, p = .02) would improve the fit of the model.
After the simplex model was specified this way, the path between SES and Reading at Time 2
was no longer statistically significant, so it was removed from the model. After these model
changes, the fit of the model was excellent (Chi-Square = 27.4, df = 3; CFI = 1.0; TLI = 1.0;
RMSEA = .038, 90% Confidence Interval .026 to .052; SRMR = .005 and Hoelter’s Index =
1,567). The results are presented in Table 11.
The results indicated high stability in reading skill development with at least 50% of the
reading skill development at any time point being explained by the development in reading skill at
the previous time point. As might be expected given that the time between assessments was so
short relative to other adjacent time points, the strongest relationship between reading at adjacent
time points was between reading at the beginning of kindergarten and the end of kindergarten (Rsquared = .79). The lowest stability between reading skills at adjacent points was between the
end of first grade and the end of third grade (R-squared = .46), which, although a weaker
relationship, is still relatively stable.
In addition, the results showed that the variance of the latent reading skill constructs
(Table 11) showed increasing variance from the beginning of kindergarten until the end of third
grade (86.43 at the beginning of kindergarten to 508.59 at the end of third grade). There was a
slight decrease in variance from the end of third grade until the end of fifth grade (from 508.59 to
412.77). The evidence of stable rank order and increasing variance from the beginning of
65

kindergarten until the end of third grade suggests the presence of a cumulative growth trajectory
for this time segment (Bast & Reitsma, 1997). However, the slight decrease in variance from the
end of third grade to the end of fifth grade suggests a more compensatory trajectory for this time
segment.
There was a moderate relationship between socioeconomic status and initial reading skill
(standardized estimate .37, t = 30.09, p < .001) and a small, but statistically significant
relationship between SES and reading skill development at all the remaining time points except
the end of kindergarten. These results show that children from lower SES families are more likely
to enter kindergarten with lower reading skills and that SES continues to be negatively associated
with reading development across time, though to a lesser degree.
The residual variances were all statistically significant indicating that there were other
factors associated with reading skills at each point than those contained in this model. However,
because the reading skills at the previous time point explained a considerable amount of variance
in the reading skills at the adjacent time points, the residual variances are relatively small except
for the residual variance at the first time point. This finding shows that SES explains only a small
portion of the variation in initial reading skill. This model suggests the presence of both
compensatory and cumulative trends in the growth of the measured reading skills across time.
This model suggests that reading performance was stable across elementary school with children
who entered school with lower IRT reading scores remaining behind their peers in reading skills
across elementary school. Additionally, there was increasing variance in reading scores from the
beginning of kindergarten suggesting that there was a widening gap between the students with the
highest IRT reading scores and those with the lowest scores. However, there were also
66

indications that the relative stability of reading scores decreased across elementary school and the
gap in reading performance appeared to narrow somewhat between the end of third and the end
of fifth grade.

Growth Curve Analysis for Reading
The first hypothesis predicted that some children from low-SES families, who entered
school with lower initial reading skills, might follow a cumulative trajectory of reading
development, consistent with the Matthew Effect. However, it was expected that the majority of
these children would follow a cumulative trajectory of reading development through kindergarten,
but show a compensatory developmental trajectory in later grades. The latent growth curve
model (Muthén & Khoo, 1998), allows for the analysis of associations between initial level and
growth over time. Within this model, two latent factors, level and shape, represent dimensions of
individual differences in growth over time. These common factors are used to describe the values
of the status and growth of individuals on a trait at all occasions. A latent growth curve analysis
with the continuous SES composite as a covariate (see Figure 2) was conducted to examine the
association between initial reading level and growth in reading skills over time. In this model, two
latent factors, level and shape, represent dimensions of individual differences in growth over time.
All loadings for the first factor, initial status, are fixed at 1. For the second factor, change over
time (overall change; shape parameter), the first loading is fixed at 0 and the last loading is fixed at
1. The intermediate loadings of the shape factor are free parameters. To institute the assumption
of measurement invariance across measurement occasions, the error variances of the five
measurement occasions are constrained to be equal over time. The fit indices indicated a poor fit
67

(Chi-Square = 7092, df = 14, p < .001; CFI = .69; TLI = .67; RMSEA = .303, 90% Confidence
Interval .297 .to .309; SRMR = .39 and Hoelter’s Index =19).
Because the nature of reading growth is expected to be different within each time frame
(Bast & Reitsma, 1998; Leppanen, 2004), it was determined that a piecewise growth model may
better reflect the reading growth across time. A piecewise growth model is a special case of
general growth curve modeling which allows for unique growth trajectories at different phases of
development (Li et al., 2001; Muthén & Muthén, 1998). This model (see Figure 3) consisted of
four components: one initial status factor (level) and three shape-growth factors: 1) a linear
growth factor from the beginning of kindergarten to the end of first grade (Growth I); 2) a linear
growth factor from the end of first grade to end of third grade (Growth II); and 3) a linear growth
factor from the end of third grade to the end of fifth grade (Growth III). The level factor consisted
of the five observed reading skill variables (Time 1 thru Time 5) with each of the loadings fixed at
1. To measure Growth I the loadings of the observed reading skill variables were set to 0, 1, 3, 3,
3, respectively. To measure Growth II the loadings of the observed reading skill variables were
set to 0, 0, 1, 4, 4, respectively. To measure Growth III the loadings of the observed reading skill
variables were set to 3, 3, 3, 6, 9, respectively. The fit indices indicated an excellent fit (ChiSquare = 7.38, df = 2, p = .03; CFI = 1.0; TLI = 1.0; RMSEA =.022, 90% Confidence Interval
.007 .to 04; SRMR = .005 and Hoelter’s Index = 4,462). The results are presented in Table 12.
The intercepts for the latent shape factors (Table 12) were positive and significant for all
three growth periods, indicating steady growth in reading skills across all time periods. The
greatest rate of growth was found from the beginning of kindergarten until the end of first grade,
with a mean growth rate of 8.57 (se = .35) in the latent reading skill variable, followed closely by
68

the growth rate from the end of third grade to the end of fifth grade with a mean growth rate of
8.17 (se = .24). The slowest rate of growth in the latent reading skill variable was between the
end of first grade to the end of third grade with a mean growth rate of 4.46 (se = .45). There was
also considerable variance in the Level component (148.36) and the variances of two of the
growth components (Growth I, 15.14 and Growth II, 13.17) indicating that there were
considerable interindividual differences in the initial reading skills (Level) and the growth in
reading skills from the beginning of kindergarten to the end of third grade. The variance in the
third growth component was considerably smaller (1.05) indicating there were considerably less
interindividual differences in the growth of reading skills during the period from the end of third
to the end of fifth grade.
The covariance between latent Level and Growth I was positive and statistically
significant (standardized coefficient = .34, t = 8.29, p < .001), the covariance between latent Level
and Growth II was not statistically significant (standardized coefficient = .11, t = 1.85, p = .06)
and the covariance between latent Level and Growth III was negative and statistically significant
(standardized coefficient = -.72, t = -35.23, p < .001). The results also show that covariance
between the shape factor for Growth I and the shape factor for Growth II is negative and
statistically significant (standardized coefficient = -.25, t = -8.67, p <.001). The covariance
between the shape factors for Growth I and Growth III is also negative and statistically
significant (standardized coefficient = -.34, t = -2.91, p = .004) as is the case for the covariance
between the shape factors for Growth II and Growth III (standardized coefficient = -.43, t = 4.29, p < .001). These results indicate higher initial reading skills are associated with greater
growth during the kindergarten year, but they have no significant association with growth in the
69

measured reading skills from the end of first to the end of third grade. The most interesting
finding is that higher initial reading skills are associated with significantly slower growth during
the period from the end of third grade to the end of fifth grade. This finding suggests that at least
some children entering school with lower entry skills, who follow a cumulative developmental
trajectory through kindergarten, are beginning to close the gap with their peers and following a
more compensatory trajectory from the end of first grade to the end of fifth grade.
The results also showed that SES had a positive and statistically significant association
with the latent level factor (standardized coefficient = .34, t = 17.05, p < .001), the shape factor
for Growth I (standardized coefficient = .16, t = 9.71, p < .001) and the shape factor for Growth
II (standardized coefficient = .23, t = 9.28, p < .001). However, SES had a negative and
statistically significant relationship with the shape factor for Growth III (standardized coefficient
= -.18, t = -3.38, p =.001). This finding suggests that higher SES is associated with higher entry
level reading skills and higher rates of growth in reading skill from the beginning of kindergarten
until the end of the third grade. The negative and significant association between SES and reading
development from the end of third grade to the end of fifth grade suggests that children from
lower SES families are experiencing compensatory growth during this period.
The residual variances of Level (146.21, t = 11.91, p < .001) and the variances of the three
growth components (Growth I, 20.93, t = 23.90, p < .001; Growth II, 19.69, t = 8.22, p < .001;
Growth III, 2.52, t = 2.12, p = .03) were all statistically significant, indicating that there was a
significant amount of variance in the initial entry skills and the growth in skills from the beginning
of kindergarten that remain unexplained by this model. Again this model suggests the presence of
both compensatory and cumulative trends in the growth of the measured reading skills across
70

time. There was a statistically significant positive association between level and the development
of reading skills from the beginning of kindergarten to the end of first grade suggesting that
children who enter school with lower IRT reading skills are experiencing slower growth in reading
skills during this time when compared to their peers who enter school with higher IRT reading
scores. However, there was no relationship between level and the growth in reading from the end
of kindergarten to the end of third grade and a significant negative association between level and
growth in reading from third to fifth grade suggesting that those children who entered school with
lower skills experienced a period of compensatory growth in reading skills. Further, there were
statistically significant negative associations between the growth slopes, suggesting that
individuals who experienced rapid growth during one period of time showed slower growth
during the subsequent time period.

Simplex Model for Reading and Behavior
The second hypothesis predicted that the socioemotional development of children from
low-SES families is associated with their reading development. Several analyses were conducted
to test this hypothesis. To examine possible multi-construct associations between reading skill
and behavior adjustment trajectories across time, a multivariate simplex model was constructed
with SES as a covariate. To construct such a model it was necessary to first create a simplex
model for behavior (Figure 6). This model was estimated without any relationship to reading
development to examine the stability and variance of behavioral ratings across the five
measurement occasions. In this model, the latent skill variables consisted of one indicator, the
observed behavior rating variable for that measurement occasion. The residual variances of the
71

observed behavior rating variables were constrained to be equal across time for model
identification. The fit indices indicated a good fit (Chi-Square = 58.55, df = 5; CFI = .95; TLI =
.86; RMSEA = .04, 90% Confidence Interval .034 .to 055; SRMR = .01 and Hoelter’s Index
=1,039). Inspection of modification indices suggested that estimating the covariance between the
residuals of the latent behavior construct at Time 1 and Time 5 (standardized estimate .17, t =
3.65, p < .001) and between Time 1 and Time 4 (standardized estimate .25, t = 5.32, p < .001)
would improve the fit of the model. After these specifications, the path between SES and
Behavior at Time 2 and the path between SES and Time 4 were no longer statistically significant,
so they were removed from the model. After these model changes, the fit of the model improved
(Chi-Square = 30.65, df = 5; CFI = .98; TLI = .93; RMSEA = .031, 90% Confidence Interval
.021 .to 041; SRMR =.016; Hoelter Index = 1,984). The results are presented in Table 13.
The results indicated high stability in behavior ratings from Time 1 to Time 2 (Rsquared= .53), but less so between other time points (R-squared from 10.9 to 20.3), although all
pathways were statistically significant. Additionally the results showed that the variance of the
behavior ratings increased quite dramatically from Time 1 to Time 2, decreased from Time 2 to
Time 3 and remained fairly stable from Time 3 to Time 5. Socioeconomic status showed a small
but statistically significant relationship with behavior ratings at the beginning of kindergarten and
at the end of first and third grade. These results show that children from lower SES families are
somewhat more likely to demonstrate more behavior problems than their peers from higher SES
families. The residual variances were all moderate to large and statistically significant (see Table
13), indicating that there are other factors associated with the behavior ratings at each point than
those contained in this model. This finding shows that SES explains only a small percentage of
72

the variance in the behavior ratings at each time point.
A multivariate simplex model was conducted to determine if there were multi-construct
associations between the latent variables of reading and behavior across the five time points.
Simplex models for reading and behavior (see Figures 1 and 4) without any relationships between
them were estimated separately (results presented in Table 11 and 13), and then the latent
behavior variables were regressed on the latent reading variables (see Figure 5). The fit indices
indicated a good fit (Chi-Square = 243.98, df = 29; CFI = .99; TLI = .98; RMSEA = .037, 90%
Confidence Interval .033 .to 041; SRMR = .031 and Hoelter’s Index = 958). The results are
presented in Table 14.
There were only slight changes in the parameter estimates from the separate simplex
models for reading and behavior and the multivariate simplex model. SES continued to have a
small but statistically significant association with both reading and behavior. There was a positive
and statistically significant association between behavior and reading but it was extremely small
(the largest relationship was found between the behavior rating at the end of first grade and the
reading score at the end of third grade of .12, while the remaining associations were even smaller
from .03 to .04). These results indicate that higher rates of behavior problems were related to less
proficient reading, but the small size of this coefficient suggests that this relationship has little
practical significance. Earlier reading proficiency still remains the best predictor of reading
proficiency at later grades and SES has a stronger association with reading scores than behavior
ratings at all time points with the exception of the end of kindergarten. Further, there were only
slight decreases in the residual variances of the reading scores at the four time points with the
addition of behavior ratings to the model, indicating that regressing the latent behavior variables
73

on the latent reading variables did little to increase the explanatory power of the model. The
results of these analyses do not appear to support the hypothesis that behavior problems have a
strong association with the reading development of children from low-SES families.

Growth Curve Model For Reading and Behavior
To further examine the hypothesis that the socioemotional development of children from
low-SES families is associated with their reading development, a multivariate latent growth curve
model was estimated. This model allowed for the examination of the multi-construct associations
between the level and shape components of the reading skill and behavioral rating variables. To
conduct this analysis, it was necessary to first create a univariate latent growth for behavior
ratings (see Figure 6). This model was conducted to examine the association between the initial
behavior rating level and change in behavior ratings over time, without any relationship to reading
development. Within this model, two latent factors, level and shape, represent dimensions of
individual differences in change over time. All loadings for the first factor, initial status, are fixed
at 1. For the second factor, change over time (overall change; shape parameter), the first loading is
fixed at 0 and the last loading is fixed at 1. The intermediate loadings of the shape factor are free
parameters. To institute the assumption of measurement invariance across measurement
occasions, the error variances of the five measurement occasions are constrained to be equal over
time. The fit indices indicated a poor fit (Chi-Square = 210.84, df = 14; CFI = .82; TLI = .81;
RMSEA = .051, 90% Confidence Interval .045 to .057; SRMR = .12 and Hoelter’s Index =617).
Because the nature of the behavior ratings is also likely to be different within each time
frame and to be consistent with the model estimated for reading growth, a piecewise growth curve
74

model was constructed for the behavior ratings at the five time points (Figure 7).This model
consisted of four components: one initial status factor (Level) and three shape-growth factors: 1)
a linear growth factor from the beginning of kindergarten to the end of kindergarten (Growth I); 2)
a linear growth factor from the end of first grade to the end of third grade (Growth II); and 3) a
linear growth factor from the end of third grade to the end of fifth grade (Growth III). The level
factor consisted of the five observed behavior rating variables (Time 1 thru Time 5) with each of
the loadings fixed at 1. To measure Growth I the loadings of the observed reading skill variables
were set to 0, -1, 1, 1, 1, respectively. To measure Growth II the loadings of the observed reading
skill variables were set to 0, 0, 1, 2, 2, respectively. To measure Growth III the loadings of the
observed reading skill variables were set to 1, 1, 1, 2, 3, respectively. The fit indices indicated an
excellent fit (Chi-Square = 8.35, df = 2; CFI = .99; TLI = .96; RMSEA = .024, 90% Confidence
Interval .009 to 042; SRMR = .007 and Hoelter’s Index =3,942). The results are presented in
Table 15.
The intercepts for the latent shape factors (Table 15) were statistically non-significant for
all three growth periods indicating there was no changes in behavior ratings across the
measurement period. The variance of Level (.56) and the variance of the first growth component
(Growth I; .15) were statistically significant indicating that there were interindividual differences
in the initial behavior ratings (Level) and the first growth component. The variance for the second
and third growth components (Growth II, .02 and Growth III, .01) were extremely small
indicating that there was little interindividual difference in behavior ratings at these time points.
The residual variances of Level (1.14, t = 7.091, p < .001) and Growth I (.40, t = 2.41, p = .02)
were statistically significant, indicating that there was a significant amount of variance in the
75

initial entry behavior ratings and the change in behavior ratings across kindergarten that remain
unexplained by this model.
The covariance between latent Level and Growth I (standardized coefficient = -.56, t = 3.82, p < .001) and latent Level and Growth III (standardized coefficient = -.55, t = -3.34, p <
.001) were negative and statistically significant. However, the covariance between latent Level
and Growth II was not statistically significant. There was also no statistically significant
association between any of latent growth factors. These findings indicate that children who are
entering school with higher levels of behavior problems are showing improvements in behavior
during the kindergarten year and between the end of third grade to the end of fifth grade. There
was also a positive and statistically significant association between SES and the latent level factor
of behavior ratings (standardized coefficient = .13, t = 4.18, p < .001), but no statistically
significant association between SES and any of the growth factors. So although children from
low-SES families are more likely to exhibit behavior problems upon school entry, the frequency
of behavior problems changes over time does not appear to be related to SES.
A multivariate piecewise growth curve (Figure 8) was conducted to determine if there
would be multi-construct associations between the level and shape components of the reading
and behavior variables. To examine this potential relationship, the piecewise growth curve models
for reading and behavior (see Figures 3 and 7) were first estimated separately without any
relationships between them (results presented in Table 12 and 15), and then covariances between
the latent level and shape components of the reading and behavior ratings were estimated. The fit
indices indicated an excellent fit (Chi-Square = 26.71, df = 13, p = .03; CFI = 1.0; TLI = 1.0;
RMSEA = .014, 90% Confidence Interval .006 .to 021; SRMR = .005 and Hoelter’s Index =
76

4,598). The results are presented in Table 16.
Again, there were only slight changes in the parameter estimates from the separate
piecewise growth curve models for reading and behavior and the multivariate piecewise growth
curve model. The results also showed that the covariance between the latent Level factor for
reading skill and the latent Level factor for behavior was positive and significant (standardized
coefficient = .10, t = 2.38, p = .02). There was also a positive and significant association between
the latent Level factor for behavior and the first growth shape for reading (standardized
coefficient = .16, t = 4.15, p < .001) and between the latent Level factor for reading and the first
growth shape for behavior (standardized coefficient = .16, t = 2.20, p = .03), There were no other
significant associations between the latent Level factors for behavior ratings and reading scores
and any of the other growth shapes for either behavior or reading. The associations between any
of the growth factors for reading scores and behavior ratings were non-significant. Further, there
were no decreases in the residual variances of the level and growth components for the reading
scores with the addition of behavior ratings to the model, indicating that this addition did little to
increase the explanatory power of the model.

Clustering by Cases Analyses
To determine the number and types of reading trajectories a clustering-by-cases analysis
for longitudinal data (Aunola, Leskinen, Onatsu-Arvilommi, & Nurmi, 2002; Bergman, Eklund, &
Magnusson, 1991; Bergman & Magnusson, 1997) was completed. To carry out this analysis, the
standardized scores for the latent level and three latent growth components of reading skills
piecewise growth curve analysis for reading were used as the criteria variables to group pupils
77

into homogenous subgroups using analysis of clustering by cases. The methodological process
outlined by Aunola and colleagues (2002) and recommended by Bergman and Magnusson (1991)
was used to analyze clustering by cases. Initially, a hierarchical cluster analysis was conducted,
selecting the squared Euclidian distance as a similarity measure and using Ward’s (1963) methods
to form initial clusters without restricting their number. Hierarchical cluster analysis is most
appropriate for small sample sizes of 100 to 200. So to use this approach to determine the
number of meaningful clusters in a considerably larger sample, 20 random samples of 200 were
extracted from the larger dataset. Hierarchical cluster analyses were conducted on these samples
to determine the number of meaningful clusters for each of these smaller samples. These analyses
provided a dendrogram based on the distance of the clusters. The dendrogram was inspected to
determine the number of meaningful clusters in each smaller sample. To verify these solutions,
the solutions of different numbers of clusters in terms of conceptual clarity and the number of
students in each cluster were examined. An examination of the separate cluster solutions
suggested that there were six meaningful clusters represented in this data sample. This
information was used to form the final clusters using SPSS quick cluster analysis. This quick
cluster procedure selects initial cluster centers according to MacQueen’s k-means clustering
methods. After this, the algorithm assigns each case to the nearest cluster center. Several analyses
are conducted until the solution stabilizes, in these additional analyses the earlier solutions are
stored and used as initial centers in the next run.
The clustering-by-cases analysis indicated that there were six meaningful clusters
represented in the data (see Figure 9). Cluster 1 consisted of 35 children (less than 1% of the
sample). These children entered school with the highest reading skills (IRT reading score = 98.27)
78

and continued to remain the strongest readers at all measurement occasions. These children
showed little growth (change in mean IRT scores = 9.95) across the kindergarten year, moderate
but steady growth from the end of kindergarten to the end of third grade (change in mean IRT
scores was 24.24 from end of kindergarten to end of first grade and 26.39 from the end of first to
the end of third grade) and reduced growth from the end of third to the end of fifth grade (change
in mean IRT scores = 12.43). These children entered kindergarten with a number of proficient
reading skills. According to the proficiency scores (see Table 17) these children demonstrated
mastery of the letter recognition, beginning and ending sounds and sight word components of the
reading assessment upon their entry into kindergarten. Most of their growth in reading skills from
the end of kindergarten to the end of fifth grade was in the higher order skills related to
comprehension such as literal inference, extrapolation, evaluation and evaluating non-fiction. By
the end of fifth grade, the children in this cluster have demonstrated mastery in almost all areas of
proficiency except for evaluation (.88) and evaluating non-fiction (.57).
Cluster 2 consisted of 64 children (1.2% of the sample). These children enter school with
the second highest reading skills (IRT reading score = 68.02) and continue to remain among the
strongest readers at all measurement occasions. The children in this cluster show the fastest
growth of all clusters across the kindergarten year (change in mean IRT scores = 22.77), moderate
but steady growth from the end of kindergarten to the end of third grade (change in mean IRT
scores was 28.96 from end of kindergarten to end of first grade and 29.07 from the end of first to
the end of third grade) and reduced growth from the end of third to the end of fifth grade (change
in mean IRT scores = 13.97). Although these children enter school with IRT reading scores 30
points behind the first cluster by the end of fifth grade their IRT scores are only 8.5 points
79

behind this cluster’s fifth grade IRT score, indicating that the gap in reading skills between the
two groups is closing. The children in this cluster entered kindergarten with a number of
proficient reading skills. According to the proficiency scores (see Table 17) these children
demonstrated mastery of the letter recognition, beginning and ending sounds and sight word
components of the reading assessment upon their entry into kindergarten. Children in this cluster
show much lower proficiency in understanding words in context and literal inference than
children in the first cluster (.44 compared to .88 for words in context and .07 compared to .46 for
literal inference). Most of their growth in reading skills from the end of kindergarten to the end of
fifth grade is also in the higher order skills related to comprehension such as literal inference,
extrapolation, evaluation and evaluating non-fiction. However, children in this cluster show
slower growth in these areas and demonstrate mastery on these proficiency scores later in their
reading developmental trajectory when compared to the first cluster. By the end of fifth grade,
the children in this cluster have demonstrated mastery in almost all areas of proficiency except
for evaluation (.76) and evaluating non-fiction (.33), but again children in this cluster are showing
less proficient skills in these areas than their peers in the first cluster (.88 and .57 respectively).
Cluster 3 consisted of 384 children (7% of the sample). These children enter school with
reading skills (mean IRT Reading score = 46.15) over 20 points behind the children in Cluster 2
(mean IRT reading score = 68.02) but by the end of fifth grade these children’s mean IRT reading
score (164.20) has surpassed their peers in Cluster 2 (mean IRT reading score = 162.79) and their
mean IRT reading score is only 7 points behind their peers in Cluster 1 (mean IRT reading score
= 171.28). These children show the second fastest growth (change in mean IRT score is 22.77)
across the kindergarten year, moderate but rapid growth from the end of kindergarten to the end
80

of third grade (change in mean IRT score from end of kindergarten to end of first grade is 41.23
and change in mean IRT score from the end of first to the end of third grade is 49.43) and reduced
growth from the end of third to the end of fifth grade (change in mean IRT score is 14.37). When
compared to Cluster 1 and Cluster 2, these children enter kindergarten with much less proficient
reading skills. According to the proficiency scores (see Table 17) these children only demonstrate
mastery of the letter recognition and beginning sounds components of the reading assessment
upon their entry into kindergarten. Children in this cluster show lower proficiency in recognizing
ending sounds (.80 compared to 1.0 for Cluster 1 and .98 for Cluster 2); recognizing sight words
(.21 compared to 1.0 for Cluster 1 and .87 for Cluster 2) and understanding words in context (.05
compared to .88 for Cluster 1 and .44 for Cluster 2). Although the children in this cluster
demonstrate delayed growth in literal inference, extrapolation, evaluation and evaluating nonfiction when compared to children in Cluster 2, by the end of fifth grade they have reached a
comparable level in these areas to children in Cluster 2.
Cluster 4 consisted of 1184 children (21.6% of the sample). These children enter school
with reading skills (mean IRT Reading score = 36.71) less than 10 points behind the children in
Cluster 3 (mean IRT reading score = 46.15) and they follow a parallel growth trajectory with
their peers in Cluster 3, remaining 7 points behind in their IRT scores at the end of fifth grade.
These children show slow growth (change in mean IRT score = 12.20) in their mean IRT score
across the kindergarten year, but rapid growth from the end of kindergarten to the end of third
grade (change in mean IRT score from end of kindergarten to end of first grade is 41.23 and 49.43
from the end of first to the end of third grade) and reduced growth from the end of third to the
end of fifth grade (change in mean IRT score is 17.29). These children enter kindergarten with
81

much less proficient reading skills. According to the proficiency scores (see Table 17) these
children only demonstrate mastery of the letter recognition component of the reading assessment
upon their entry into kindergarten. Children in this cluster show lower proficiency in recognizing
beginning sounds (.71 compared to .95 for Cluster 3); recognizing ending sounds (.41 compared
to .80 for Cluster 3) and recognizing sight words (.02 compared to .21 for Cluster 2). In spite of
their rapid growth in mean IRT score from the end of kindergarten until the end of third grade,
these children continue to lag behind their peers in the first three clusters, in their proficiency in
evaluation and evaluating non-fiction.
Cluster 5 consisted of 2200 children (40.1% of the sample). These children enter school
with considerably lower reading skills (mean IRT Reading score = 28.80) than the other clusters.
Even though they show rapid growth from the end of kindergarten to the end of third grade
(change in mean IRT scores is 89.45) this cluster of children is the first to fall further behind their
peers, although only slightly (1.85 points lower at the end of fifth than at the end of first). These
children show slow growth (change in mean IRT score = 11.67) in their mean IRT score across
the kindergarten year, but rapid growth from the end of kindergarten to the end of third grade
(change in mean IRT score from end of kindergarten to end of first grade is 32.74 and 56.71 from
the end of first to the end of third grade) and reduced growth from the end of third to the end of
fifth grade (change in mean IRT score = 17.18). These children enter kindergarten with much less
proficient reading skills than the other clusters. According to the proficiency scores (see Table
17) these children do not demonstrate mastery of any of the proficiency components upon their
entry into kindergarten. Children in this cluster show lower proficiency in letter recognition (.81
compared to .98 for Cluster 4); recognizing beginning sounds (.25 compared to .71 for Cluster 4)
82

and recognizing ending sounds (.08 compared to .41 for Cluster 4). By the end of third grade
these children have caught up to their peers in the remaining clusters in all the basic reading
proficiencies measured (letter recognition, beginning and ending sounds, sight words and words in
context), however they continue to lag behind their peers in their proficiency in extrapolation,
evaluation and evaluating non-fiction.
Cluster 6 consisted of 1687 children (30.7l% of the sample). These children enter school
with lowest reading skills of all the clusters (mean IRT Reading score = 22.78) and though there
is only small difference between the entry skills of Cluster 5 and Cluster 6 (difference in initial
mean IRT scores is 5.42) by the end of fifth grade this difference has grown to a difference of
more than 22 points. These children show the slowest growth of any cluster (change in mean IRT
scores = 9.82) in their mean IRT score across the kindergarten year, and moderate growth from
the end of kindergarten to the end of first grade (change in mean IRT score = 24.18). The children
in this cluster show their largest growth from the end of first grade to the end of third grade
(change in mean IRT scores = 42.35) and they show the greatest growth of any cluster from the
end of third to the end of fifth grade (change in mean IRT score = 25.86). However, these children
continue to remain behind their peers in a number of skills. These children enter kindergarten with
the least proficient reading skills of any cluster. According to the proficiency scores (see Table
17) these children enter school with very low reading skills. Children in this cluster show lower
proficiency in letter recognition (.39 compared to .81 for Cluster 5); recognizing beginning sounds
(.05 compared to .25 for Cluster 5) and recognizing ending sounds (.01 compared to .05 for
Cluster 5). These children reach mastery level in their proficiency of the letter recognition
component of the reading skill test by the end of kindergarten, but it takes them until the end of
83

third grade before they reach mastery level proficiency in the recognizing sight words and reading
words in context components of the reading skill test. The children in this cluster also show
substantial deficits in literal inference, extrapolation, evaluation and evaluating non-fiction
components of the reading test in comparison to their peers in the other clusters.

Cluster membership by categorical SES and behavior
Differences in cluster membership by categorical SES and behavior ratings were evaluated
using Kendall’s tau-c. The continuous behavior ratings variables were averaged across the five
time points and then recoded into a categorical variable by dividing the scores into quartiles. The
lowest quartile included children with behavior ratings from10.27 to 14.18, the second quartile
ranged from 14.19 to 14.92, the third quartile from 14.93 to 15.64 and the highest quartile from
15.65 to 18.06. Kendall’s tau-c is a measure of ordinal association and is calculated as the excess
of concordant over discordant pairs multiplied by a term representing an adjustment for the size
of the table. The value of Kendall’s tau-c varies from -1 to 1, with 1 or -1 indicating a perfect
relationship and 0 indicating no relationship. Cluster membership was significantly associated
with categorical SES, Kendall’s tau-c (20) = .30, p < .001 and categorical behavior ratings,
Kendall’s tau-c (15) = .12, p < .001. To determine how these factors were jointly related to
cluster membership, first a nested table was created looking at the association between cluster
membership and categorical SES by behavior ratings. This analysis indicated that there were some
slight differences in the relationship between the categorical SES variable and cluster membership
by behavior ratings. Kendall’s tau-c ranged from .33 for those children with the highest ratings of
behavior problems to .26 for those children with the lowest ratings of behavior problems.
84

However all relationships remained significant with a p-value of less than .001. A second nested
table was created looking at the association between cluster membership and behavior ratings by
categorical SES. This analysis also indicated that there were some slight differences in the
relationship between behavior ratings and cluster membership by categorical SES. Kendall’s tau-c
ranged from .11 for those children from the families within the first quintile of SES to .05 for
those children living in families within the fifth quintile of SES. However all relationships
remained significant with a p-values ranging from less than .001 to .03.

85

CHAPTER 5
DISCUSSION
Consistent with previous research (Lee, Grigg, & Donahue, 2007; Lonigan, Burgess,
Anthony, & Barker, 1998; Molfese, Modglin, & Molfese, 2003; Whitehurst, 1997), children
from low-SES families were on average behind their peers in reading development at all
measured time points. Children from the lowest SES quintile families entered school with lower
reading skills than their peers from higher SES families; they were on average 2.7 to 10.2 points
behind their peers from higher SES families at school entry. This difference grew across the
grade levels to as much as a 10.6 to 30.5 point difference at the end of third grade. The
significance of these numbers is unclear seeing that these IRT scores are not easily comparable
to school-based skill assessment, but children from the lowest SES quintile were behind their
peers in all the reading subskills measures within this study. Also consistent with previous
research (NCES, 2000; Zill et al., 1995), children from lower SES families were on average
more likely to exhibit a higher rate of behavior problems at school entry but this difference was
considerably smaller and did tend to narrow across elementary school.

Reading Developmental Trajectories
There is a mix of research findings regarding the nature of the reading development of
children who enter school with lower initial reading skills (Aarnouste et al., 2001; Cunningham
& Stanovich, 1997; Juel, 1988; Phillips et al., 2002), with results showing both cumulative and
compensatory trajectories for reading development. However, some of this previous research has
shown that children who enter school with lower initial reading skills typically follow a
cumulative trajectory of reading development through kindergarten, but show a compensatory

86

developmental trajectory in later grades (Aarnouste et al., 2001; Aunola et al., 2002; Leppanen et
al., 2004; Lerkkanen et al., 2004; McCoach et al., 2006; Parrila et al., 2005). The findings of this
study are consistent with the findings of this earlier research. According to Bast and Reitsma
(1997) evidence of cumulative trajectory consists of three components of developmental change:
the stability of interindividual differences across time, increases in interindividual variance
across time and the association between initial skill level and its growth across time. The results
of this study indicated a high stability in reading scores (an association of .79) between the
beginning and end of kindergarten with an increasing variance (from 86.43 to 163.70) showing a
fan spread effect expected from cumulative development. There was also a positive association
between initial reading level and the growth in reading skill across the kindergarten year,
indicating that children who enter kindergarten with higher letter recognition, better phonemic
awareness, and a greater knowledge of sight words show the fastest growth in IRT reading scores
across kindergarten. However, consistent with earlier research (Aarnouste et al., 2001; Aunola et
al., 2002; Leppanen et al., 2004; Lerkkanen et al., 2004; McCoach et al., 2006; Parrila et al.,
2005), there was evidence of a compensatory developmental trajectory in later grades. There was
no evidence of a relationship between initial level of reading skills and the growth of reading
skills from the end of first to the end of third grade and a significant negative association
between initial level of reading and the growth in reading skill from the end of third grade to the
end of fifth grade, indicating that children who entered school with lower proficiency in the
previously mentioned reading related skills on average experienced greater growth in their IRT
reading scores during this period of time. Further, growth in reading development from the
beginning to the end of kindergarten is significantly negatively associated with growth in reading
skills from the end of kindergarten to end of fifth grade, indicating that even though children

87

with higher initial reading skills are experiencing a period of more rapid growth across
kindergarten they are showing slower growth than their peers with lower initial reading skills
from the end of kindergarten to the end of fifth grade.
These findings call into question the assertion that reading skill at school entry
exclusively shapes the future of reading development, and that the gap between weak and strong
reading achievement widens over the elementary years, becoming increasingly difficult to close.
It is true there was a fairly high stability in reading skill, suggesting that relative position in
achievement remained stable across elementary school and that children who entered school with
more proficient reading skills on average remained more proficient than their less skilled peers.
The relationships between reading scores at adjacent time points ranged from.79 between the
beginning and the end of kindergarten to .46 between third grade and fifth grade. However, while
there was a strong likelihood that children who entered kindergarten with less reading
proficiency would end kindergarten behind their more proficient peers, this likelihood was
substantially reduced by fifth grade. It appears as if some of these children with lower reading
skills at school entry were able to make substantial gains, closing the initial gap between them
and their more proficient peers. This finding was further supported by data showing that
although there was a relationship between initial reading proficiency and growth across
kindergarten, there was no relationship between initial proficiency and growth from first to third
grade. Even more importantly the data suggest that children who were less proficient at school
entry are actually experiencing faster growth in reading skill from third to fifth. This finding
would seem to cast doubt on the assertion that closing the gap in reading proficiency after third
grade is extremely difficult. While this is only one study and further replication is necessary, the
results do raise doubts about the presence of a Matthew Effect for children entering school with

88

lower reading proficiency. These results are consistent with earlier research (Aarnouste et al.,
2001; Aunola et al., 2002; Leppanen et al., 2004; Lerkkanen et al., 2004; McCoach et al., 2006;
Parrila et al., 2005) suggesting that these children follow multiple trajectories and these
trajectories may be a mix of both cumulative and compensatory growth.

Reading development and SES
It was hypothesized that the majority of children from low-SES families, who are more
likely to enter school with lower initial reading skills (Aram, 2005; Bowey, 1995; Christian,
Morrison, & Bryant, 1998; Hood, Conlan, & Andrews, 2008; Korat, Klein, & Segal-Drori, 2006;
Spira, Bracken, & Fischel, 2005; Walker, Greenwood, Hart & Carta, 1994; Zill, Collins, West, &
Hausken, 1995), would follow a cumulative trajectory of reading development through
kindergarten, but show a more compensatory developmental trajectory in later grades. As
expected, on average children from lower SES families in this sample did enter school with
lower initial reading skills as evidenced by lower letter recognition, phonemic awareness,
knowledge of sight words and reading words in context. However, there was only a moderate
relationship between SES and initial reading skills (.37 in the univariate simplex model and .34
in the univariate piecewise growth curve model) suggesting that other factors not measured in the
models are associated with lower initial reading skills upon entry to kindergarten. This moderate
relationship may be the result of using a global measure of SES rather than looking at more
proximal indicators of reading skill development that are frequently associated with low-SES.
Low-SES has been associated with home language environment, parental support, cognitive
ability, parental expectations and family instability which have all been related to lower reading
skills (Eamon, 2002; Gershoff et al., 2007; Votruba-Drzal, 2006), but the relationship between

89

SES and these factors are is not perfect. There is considerable variation in the types of academic
and emotional support that low-SES families are able to provide their children. It may be wise in
future studies to identify specific factors that underlie this relationship between SES and reading
skill development, especially considering that these factors are often more likely to be addressed
by effective interventions. It is usually very difficult to change the family income, parental
occupational status or parental education level to any great degree, but there are a number of
proven interventions that have been shown to improve parental emotional support and to increase
the quality of the home language environment (Justice & Ezell, 2000; Nicholson, Berthelsen,
Williams, & Abad, 2010; Van der Heyden, Snyder, Broussard, & Ramsdell, 2008; WebsterStratton & Herman, 2008)
As expected low-SES was associated with lower IRT scores upon school entry, however
this study provided mixed results concerning the pattern of reading development for children
from low-SES families. There was a continued significant association between SES and the
development of reading skills even after controlling for initial reading skills at school entry.
Although the association was small, on average children from higher SES families received
higher IRT reading scores at each point excepting the end of kindergarten when no significant
association was found. Further, higher SES was associated with faster growth in IRT reading
scores from the beginning of kindergarten to the end of third grade. However, the results did
suggest that children from low-SES families are experiencing a period of faster growth in
reading scores than their peers from higher SES families from the end of third grade to the end of
fifth grade. These findings seem to suggest that many children from low-SES families may still
experience slower growth in reading skills independently of their lower reading scores at school
entry. Further although there was some support for a compensatory growth in reading scores

90

from the end of third grade to the end of fifth grade for children from lower SES families, on
average this rate of growth was insufficient to overcome the gap in skills that widened from the
beginning of first grade to the end of third grade.
The nature of this study did not allow the long-term value of early reading intervention to
be evaluated, seeing that this data was not included in the study. It may have been that early
intervention may have boosted the reading skills of lower SES children allowing a higher
percentage of them to close the gap between their more proficient peers. However, the findings
suggesting an association between low-SES and the slower development of reading skills across
elementary school, even when controlling for initial reading proficiency, highlights the necessity
for an even deeper exploration into the relationship between low-SES and reading development.
Early intervention is clearly an important first step in addressing the deficits in reading skills
experienced by many children from low-SES families, but these findings suggest it may not be
enough. It is important to delve deeper into the achievement of students from low-SES families
to determine what other factors, other than SES and early achievement, are influencing the
reading development of children from low-SES families. There are many potential explanations
for this slower reading development that were not explored within this study such as school
effects, summer effects, and the emotional aspects of schooling and school-based relationships
among others.

Other factors to be considered
Previous journalistic work and numerous research studies have consistently shown that
there are substantial differences in the quality of low-SES schools when compared to higher SES
schools. Jonathan Kozol (1991) provides a poignant description of some of the harsh conditions

91

faced by students in low-SES schools in his book, Savage Inequalities, Children in America’s
Schools. He describes low-SES schools awash in sewage, lacking qualified teachers and
adequate school-related supplies and providing substantially inferior educational opportunities.
While it is almost certain that the schools he describes are not typical of most low-SES schools,
previous research does provide evidence to indicate that low-SES schools are on average
deficient when compared to their higher SES counterparts. Classic work by Jean Anyon (1981)
discusses the divergent manner in which knowledge is defined in schools across the SES
spectrum. She found that knowledge was conceptualized as “fragmented facts,” “practical,”
“mechanical,” and “procedural” in nature in low-SES schools in comparison to higher SES
schools, which treated knowledge as more “conceptual” and more open to discovery, reason
and/or logic. Anyon (1981) found that these divergent conceptualizations of knowledge led to
differences in curriculum, differences that could arguably lead to differences in academic growth
across the years of schooling.
More recent research in this area supports the contention that low-SES schools are often
inferior to their higher SES counterparts. Previous research has shown that low-SES schools on
average have smaller libraries (Guice, Allington, Johnston, Baker, & Michaelson, 1994) and
fewer and older computers than higher SES schools (Becker & Sterling, 1987). Haycock (2000)
found that children in high poverty schools were less likely to be taught by teachers who are
fully certified and who themselves perform less well on external assessment. Tabors and Snow
(2001) found that on average teachers of students in low-SES schools have lower expectations of
their students and Barone (2002) discovered that teachers in low-SES schools typically use more
narrow instructional strategies when teaching reading. They tended to use less extended text in
their instruction, rather choosing to focus the majority of the instruction on letter recognition and

92

the sounds of the letters. This finding was consistent with earlier work completed by Duke in
2000, which compared the print environment and experiences offered in low- and high-SES
schools.
While there was a wide diversity in the types of literacy experiences offered in low-SES
first grade classrooms, Duke (2000) found substantial differences in the experiences offered in
low-SES classrooms in comparison to high-SES classrooms. On average, low-SES classrooms
contained fewer literacy materials and made less use of the materials that they did possess. First,
students in low-SES classrooms encountered less environmental print and experienced fewer
references to that print. Further it was found that the references that were given in low-SES
classrooms were typically less integrated with print resources and the topic of study in the class
at the time. For example, Duke (2000) describes one high-SES classroom in which many pieces
of environmental classroom print related to the topic of horses were displayed, including poems
about horses (some written by students), a large display of horse jokes, a list of the types of
horses, a class-composed informational text about horses, and student-made labels on various
horse-related items, and accompanying this environmental print there were a large number of
books about horses. While Duke (2000) does not indicate if this horse-related literacy material
was integrated with classroom instruction on horses, it is quite likely that this was the case. This
same rich literacy environment found in this high-SES classroom was rarely found in low-SES
classrooms observed in this study. In addition, on average low-SES classroom libraries contained
fewer magazines and books than higher-SES classrooms, with as much as a 40 percent mean
difference between the two settings. This number is even more alarming considering that lowSES schools often had three to four more students in their classrooms than did higher SES
schools. Duke (2000) also found that children in low-SES schools often had fewer opportunities

93

to access their classroom libraries and in general the proportion of the library displayed was
much less in low-SES schools when compared to higher SES schools. She also found that the
available print was less integrated into classroom instruction. Finally, Duke (2000) found
substantial differences in student agency with print between low- and high-SES schools.
Students in low-SES schools had fewer choices in their assigned reading material, had less
choice in the types of written activities they engaged in and a much narrower audience available
for their written activities. Duke (2000) contends that this less rich literacy environment in lowSES schools may be related to the slower growth in reading skills for low-SES students.
Other research has shown that the make-up of the student population of the school is also
associated with the academic achievement of low-SES students. The aggregated or mean school
SES has been shown to be independently associated with student achievement beyond that of
individual student backgrounds, with schools with a higher percentage of children from low-SES
families having lower average scores on standardized tests than their higher-SES counterparts
(Kahlenberg, 2001; Rumberger & Palardy, 2005; Sirin, 2005; Willms, 1999). In some cases the
correlation between school SES and student achievement has been found to be even stronger
than the association with individual SES and student achievement (Sirin, 2005). Logan and
Petscher (2010) have furthered this research using latent profile analysis to show that not only
the proportion of students from low-SES families, but also the proportion of minority students
and the proportion of English Language Learners (ELL) are associated with poorer academic
performance. Their analyses found four clusters that they called the “Low-Risk” group, the
“Average-Risk” group, the “Poverty-Risk” group and the “Language-Risk” group. Their findings
showed that there were differences in the initial scores on the Dynamic indicators of basic early
literacy (DIBELS) oral reading fluency (ORF) (5th edition; Good, Kaminski, Smith, Laimon, &

94

Dill, 2001) by group membership at first, second and third grades. Further they found differences
in growth rates in words correct per minute (WCPM) as measured by the DIBELS ORF between
the four groups with the “Poverty-Risk” group and the “Language-Risk” group in all three grades
showing slower growth. This finding was affirmed by Borman and Dowling (2010), who found
that the racial/ethnic and social class composition of a student’s school was 1.75 times more
important than the student’s individual SES or minority status when predicting educational
achievement.
Additionally the nature of this study also did not provide the opportunity to control for
“summer effects,” or the educational disadvantage experienced by children from lower SES
families when compared to their peers from higher SES families over the summer months. On
average low-SES parents are less likely to provide the same level of educational enrichment in
the home environment as their higher SES counterparts. The same disadvantages experienced by
children from low-SES families that are often associated with their lower reading skills upon
school entry are also often associated with a greater decline of reading skills over the summer
months. A number of studies have shown that children from lower SES families experience a
greater decline in reading skills over the summer when compared to their peers from higher SES
families (Alexander, Entwisle, & Olson, 2007; Cooper, Nye, Charlton, Lindsay, & Greathouse,
1996; Downey, von Hippel, Broh, 2004; McCoach et al., 2006; Ready, 2010). In fact, while
children from low-SES families typically experience declines in reading skills, children from
higher SES families often seem to continue to develop new skills or to strengthen their already
existing skills (Alexander, Entwisle, & Olson, 2007; Cooper et al., 1996; McCoach et al., 2006;
Ready, 2010). Downey and colleagues (2004) found that children from higher SES families
experienced a relative gain of .07 points per month in kindergarten and .05 points per month in

95

first grade on a measure of reading skills (letter recognition, phonemic awareness and sight word
recognition) in comparison to their peers from lower SES families, but the relative gain increased
to .16 points per month during the summer months. Further the negative effects of this “summer
effect” seem to increase with increases in the students’ grade level and the complexity of the
academic material (Alexander, Entwisle & Olson, 2007; Cooper et al., 1996). While the findings
of this study do suggest that children from low-SES families are making slower growth in
reading skills during elementary school, it is impossible to determine how much of this widening
of the gap is occurring during the summer months. If the gap between the reading skills of lowand higher SES families is occurring during the summer months, this may be an area that would
benefit from targeted intervention for these children.
Finally, the emotional aspects of schooling and school-based relationships cannot be
ignored and were not included in this study. There is an emerging consensus that the emotional
aspects of the classroom are predictive of academic achievement above and beyond the nature
and quality of the instructional practices provided (Bransford, Brown, & Cocking, 1999;
Cameron et al., 2005; Curby, Rimm-Kaufman, & Ponitz, 2009; Eccles & Gootman, 2002; Hamre
& Pianta, 2007). Teachers’ sensitivity and emotional support have been found to be associated
with achievement gains for first-graders (Hamre & Pianta, 2005). Further students who have
more supportive teachers are much more likely to demonstrate behaviors that support learning
(e.g., motivation, self-regulation, engagement and autonomy) and that are frequently associated
with higher academic achievement (Crosnoe, Johnson, & Elder, 2004; Greenberg, Weissberg,
O’Brien, Zins, Fredericks, Resnick, et al., 2003; Gregory & Weinstein, 2004; Pianta,LaParo,
Payne, Cox, & Bradley, 2002; Rimm-Kaufman et al., 2005; Roeser, Eccles, & Sameroff, 2000;
Wentzel, 2002; Zins, Bloodworth, Weissberg, & Walberg, 2004). In more recent research, Pianta

96

and colleagues (Pianta, Belsky, Vandergrift, Houts, & Morrison, 2008) found that the warmth of
teacher-child interactions, as well as the adults’ skill in detecting and responding to the students’
individual needs, was a consistent predictor of reading growth from first to third grade. This is
especially relevant to the findings in this study seeing that Kirby and colleagues (Curby, RimmKauffman & Ponitz, 2009) found that the students with low initial skills benefited even more
from classrooms with a high level of emotional support than did their higher achieving peers.
Finally, Hauser-Cram and colleagues (Hauser-Cram, Durand, & Warfield, 2007) showed that
children’s early attitudes toward school predicted fifth-grade literacy skills even when
controlling for initial skills prior to school entry. It seems as if children who initially hold more
ambivalent or negative attitudes toward school also experience slower growth in achievement
across the elementary school years. Although it is important to recognize that other factors than
initial reading skills may be associated with the reading development of children from low-SES
families, it is significantly more important to determine what these other factors may be and how
their interaction may influence the development of reading across elementary school as well as
the effectiveness of intervention for these students.

Clustering-by-Cases analysis
Consistent with expectations, there were multiple trajectories of reading development; the
clustering-by-cases analysis suggested the presence of six different reading development
trajectories. There was a mix of cumulative, compensatory and parallel trajectories.
Approximately 8% of this sample of children demonstrated a compensatory trajectory and
although beginning school with IRT scores that were 30 to 50 points behind their peers with the
most proficient reading skills at school entry, they were able to close the gap and in some cases

97

surpass these peers by fifth grade. The majority of the sample (approximately 60%) follows a
somewhat parallel trajectory, maintaining their relative position in the hierarchy of reading skill
development. However, despite following a largely parallel course these students do narrow the
gap in reading skills with their highest achieving peers. There is an approximately 70 point
difference in IRT scores at school entry between the highest scoring cluster and the next to
lowest scoring cluster compared to only approximately 25 points difference at the end of fifth
grade. The most concerning findings were the approximately 30% of the sample that appear to
follow a cumulative trajectory. This group of students enters school with the lowest IRT reading
scores of any group and shows the slowest growth in reading scores through elementary school.
Further, although their IRT scores are only approximately 5 points behind the majority of their
peers in the next highest group at school entry by the end of fifth grade the gap has widened to
22 points and they continue to remain 46 points behind their highest scoring peers.
These findings further raise questions regarding the strong assertion that reading skill at
school entry exclusively shapes the future of reading development, and that the gap between
weak and strong reading achievement widens over the elementary years, becoming increasingly
difficult to close. The data gathered in this study seem to suggest otherwise. The majority of the
children in this sample (approximately 60%) followed a more parallel trajectory. Although these
students did not catch up with their more proficient peers, they did appear to close the gap in
reading proficiency across their years in elementary school. Only 30% of the sample appears to
follow the more compensatory trajectory consistent with a Matthew Effect. Further, the results
do seem to call into question the assertion that intervention after third grade is unlikely to
provide any substantial benefits. Even though it was insufficient to close the gap with their
98

higher achieving peers and the improvement that was seen was in primarily in basic reading skills
(i.e., beginning and ending sounds, sight words and words in context) rather than more higher
order skills (i.e., literal inference, extrapolation, evaluation and evaluating non-fiction), this group
of children following this more cumulative trajectory experienced a period of more rapid growth
after third grade. Although these findings do cast doubt on the validity of the Matthew Effect
model, they do not necessarily negate the benefit of early, intensive intervention. The nature of
this study precludes determining the relative benefit of early intervention and its ability to alter
later reading trajectories, however it is still unclear if an intensive early intervention alone would
be sufficient to meet the needs of these students, especially those following the more cumulative
trajectory of reading development.

Cluster membership relationship with SES and behavior
The limited nature of this study precludes substantial knowledge about the characteristics
of these clusters; however, the analyses that were conducted suggest that cluster membership is
associated with both SES and the level of teacher rated behavior problems. Children from lowSES families were significantly more likely to follow a cumulative trajectory of reading
development and were predominantly grouped in the clusters with the slowest reading growth
(58% of the 1st SES quintile and 35% of the 2nd quintile were in the 6th cluster and 32% and 42%
respectively were in the 5th cluster). There was also a relationship between a higher rate of
teacher rated behavior problems and cluster membership, with the children with a higher rate of
behavior problems being overrepresented in the lowest reading group (35% of the children in the
6th cluster were identified as having the highest rate of behavior problems compared to only 22%
99

of children with the lowest rate of behavior problems). However, it should be noted this
relationship was not as strong as the relationship between SES and reading cluster. Yet it was
interesting to note that the relationship between behavior ratings and reading cluster was stronger
for children from the lower SES quintiles in comparison to children from the highest SES
quintiles. This finding suggests that a higher rate of behavior problems may be more related to
slower growth in reading skills for those children from the lowest SES families than for their more
advantaged peers. This finding makes sense in light of research that has shown a relationship
between multiple risk factors and poorer academic performance (Deater-Deckard, Dodge, Bates,
& Pettit, 1998; Greenberg, Speltz, DeKlyen, & Jones, 2001; Jones, Forehand, Brody, &
Armistead, 2002; Rutter, 1979; Sameroff, 2000; Williams, Anderson, McGee, & Silva, 1990)
and seeing that many of these children enter school significantly behind their peers in reading
proficiency.
Although many of the findings of this study suggest that the Matthew Effect model may
not be representative of the majority of students, the data does show that children from lower
SES families are more likely to follow a cumulative trajectory of reading development. Almost
60% of children from families in the lowest quintile of SES and 35% of the children from families
in the second lowest quintile of SES were in the sixth cluster and followed a cumulative
trajectory. Further 32% of children from families in the lowest SES quintile and 58% of children
from the second lowest quintile were in the fifth cluster, and although following a somewhat
parallel trajectory, they remained the furthest behind their peers, showed the slowest growth for
a group with a parallel trajectory and experienced much of their growth in basic reading skills.
These findings indicate that 90 % of children from families in the lowest SES quintile and 77% of

100

children from families in the second lowest SES quintile were in the lowest two reading clusters,
trajectories of reading development, which left them well behind their higher achieving peers. This
finding was inconsistent with expectations and is highly intriguing. Many of the children from
other socioeconomic backgrounds who entered school with lower initial reading skills followed a
relatively parallel trajectory of development and were able to close the gap in reading skills, at
least to some degree. A small number of these children even demonstrated a compensatory
trajectory, closing the gap in reading proficiency between themselves and their higher achieving
peers. These findings would seem to suggest that something more than lower initial reading skills
are related to the slower growth in reading development for these children from lower SES
families.
This finding is highly pertinent because the two federal initiatives, Early Reading First
and Reading First, have been derived from a belief in a Matthew Effect. These programs have
been initiated to address this widening gap in the reading development of children from the
lowest SES families and their more proficient peers by providing early, intensive reading
interventions in an attempt to close the gap in reading proficiency early. It is the expectation that
closing this gap early will positively influence long-term development and ameliorate the
widening gap in reading proficiency between children from the lowest SES families and their
higher achieving peers. Although these programs have not been in place long enough to truly
evaluate their relative effectiveness, several cautions need to be raised. First, the findings of this
study seem to indicate that children from low-SES families still experience slower growth even
when controlling for their initial skills. There may be many reasons for this slower growth, some
of which have been discussed above, poorer schools, summer effects, other family factors to
name a few. These findings may suggest that early reading intervention alone may not be
101

sufficient to address the needs of these children. Further previous research has found that
students with weak oral language skills and impoverished home environments are less likely to
benefit from early literacy instruction (Al Otaiba et al., 2008; Al Otaiba & Fuchs, 2002; Lonigan,
2003; Torgeson, 2000; Torgeson et al., 1999). While it is unclear if this lower effectiveness is
related to some facet of the reading instruction or some other unmeasured correlate of these
factors, it does highlight the need for a deeper, more thorough understanding of the relationship
between reading development and low-SES. Finally, although the findings of this study
regarding the relationship between behavior ratings and reading development were much weaker
than expected, there was a small relationship found between behavior rating and cluster
membership. Students who followed a more cumulative trajectory of reading development were
more likely to be rated as having a higher rate of behavior problems than their peers who
followed a more parallel or compensatory trajectory. Additionally, the data suggest that this
relationship was stronger for children from lower SES families.

Reading development and behavior
It was also hypothesized that behavior problems would be associated with the slower
reading development of children from low-SES families. Previous research has shown that
children from low-SES families are more likely to have behavior problems in school (NCES,
2000; Zill et al., 1995) and that these behavior problems may be associated with slower
development of reading skills (Arnold et al., 2005; Hinshaw, 1992; McGee et al., 1988; Reid et
al., 2004; Trout et al., 2003). The findings of this study do suggest a relationship between lowSES and a higher rate of behavior problems, but the association, although statistically significant,
was small. In the univariate simplex model a significant association was only found between

102

SES and three time points (beginning of kindergarten, .12; end of first grade, .12; and the end of
fifth grade, .09) and there was only a significant relationship between SES and the initial level of
behavior problems. These results suggest that children from lower SES families are somewhat
more likely to demonstrate behavior problems than their peers from higher SES families, but the
findings of this study provided little support for a relationship between a higher rate of behavior
problems and the slower reading development of children from low-SES families. The addition of
behavior ratings to the univariate models of reading development did very little to add
explanatory power to the model. Further, while there were a few statistically significant
relationships found between a higher rate of teacher rated behavior problems and slower reading
development, these relationships were small and unlikely to offer much practical significance.
According to the findings of this study, earlier reading proficiency still remains the best predictor
of reading proficiency at later grades and SES has a stronger association with reading scores than
behavior ratings at all time points with the exception of the end of kindergarten.
Although the findings of this study do not support the hypothesis that a higher rate of
behavior problems are associated with the slower reading development of children from low-SES
families, it is quite possible that this relationship exists. Previous research has shown that a
higher rate of behavior problems is associated with lower reading achievement (Arnold et al.,
2005; Hinshaw, 1992; McGee et al., 1988; Reid et al., 2004; Trout et al., 2003) and that a higher
rate of behavior problems in kindergarten and first grade is associated with slower reading
growth across the remainder of elementary school (Horn & Packard, 1985; Rabiner et al., 2000;
Reid et al., 2004; Velting & Whitehurst, 1997). Further, children with a higher rate of behavior
problems are more likely to have lower academic engagement (Abikoff et al., 2002; DuPaul et

103

al., 2004; Junod et al., 2006; Waller, 2006), more conflictual relationships with teachers (Birch &
Ladd, 1997; Brendgen et al., 2006; DeMartini-Scully, Bray, & Kehle, 2000; Dishion et al., 1995;
Ladd & Burgess, 1999; Shores et al., 1993) and less academic support than their peers with
better classroom behavior (Carr, Taylor, & Robinson, 1991; Good & Brophy, 1994; Gunter &
Coutinho, 1997; Shores, Gunter, & Jack, 1993; Walker et al., 1998; Wehby et al., 1998). These
factors should conceivably be associated with lower academic achievement for these children
(Dally, 2006; Deater-Deckard et al., 2009; DiPerna, Volpe, & Elliot, 2005; McGee & Share,
1988; Rabiner, Coie et al., 2000; Velting & Whitehurst, 1997; Warner-Rogers et al., 2000;
Wentzel, 1993). Finally, a number of studies have demonstrated that children with behavior
problems are less likely to benefit from effective reading interventions (Nelson, Benner, &
Gonzalez, 2003; Otaiba & Fuchs, 2002; Rabiner et al., 2004: Torgeson et al., 1999). Considering
the findings of previous research, it is quite likely that the lack of findings supporting a
relationship between higher behavior problems and slower reading development is more the
result of the way in which behavior was measured in this study, rather than a lack of such a
relationship. Behavioral ratings were gathered from teachers at each time point using an
adaptation of the Social Skills Rating Scale (SSRS; Gresham & Elliot, 1990). This behavior
measure provided a global rating of the teacher’s perception of the student’s approach to
learning, self-control, interpersonal skills, externalizing problem behaviors and internalizing
problem behaviors. While this measure did provide some information about the student’s overall
behavior it may not have been sensitive enough to capture the types and frequencies of behavior
that may be related to the slower development of reading skills. Additionally, the long span of
time between ratings of behavior may have obscured the subtle changes in weekly behavior that
could have been associated with changes in reading growth. However, it could be that a more

104

sensitive measure would not have yielded substantially different findings. As mentioned earlier
the relationship between behavior and reading is still somewhat uncertain with several different
relationships being found (Dionne, 2005; Hinshaw, 1992; Mandel, 1997; Spira & Fischel, 2005).
Clearly additional research is needed to clarify this question further.

Limitations of this Study
While this study adds to the current knowledge base about the reading development of
children from low-SES families, it also has a number of limitations. First, this study is
considering SES as a composite variable. While the use of a composite measurement of SES
represents SES better than any single construct alone with each construct predicting unique
variance in children’s developmental outcomes (Bradley & Corwyn, 2002; Conger & Donnellan,
2007), it does not allow for the consideration of the more proximal effects of SES on the
student’s development (see Luthar & Zigler, 1991; McLloyd, 1998; Sameroff, Seifer, Baldwin,
& Baldwin, 1993 for a more detailed discussion). The findings of this study suggest that children
from low-SES families are more likely to enter school with fewer reading skills than their peers
from higher SES families and are more likely to experience slower growth of reading skills
across elementary school. However, other studies have shown that the proximal factors
associated with lower SES, such as quality of emotional support, level of cognitive stimulation,
and the experience of material hardship, largely mediate the relationship between SES and these
associated outcomes (Eamon, 2002; Gershoff et al., 2007; Votruba-Drzal, 2006). Further, these
more proximal processes are easier to address with effective intervention through parent
education or the provision of cognitively supporting enrichment activities for the children, than
family income, job status and parental education level.

105

Second, the nature of the measurement of classroom behavior may not have been
sensitive enough to detect any relationship between behavior and reading development.
Behavioral ratings were gathered from teachers at each time point using an adaptation of the
Social Skills Rating Scale (SSRS; Gresham & Elliot, 1990). This behavior measure provided a
global rating of the teacher’s perception of the student’s approach to learning, self-control,
interpersonal skills, externalizing problem behaviors and internalizing problem behaviors. While
this measure did provide some information about the student’s overall behavior it relied on the
teachers’ overall impression of the student and may not have captured the types and frequencies
of behavior that may be related to the slower development of reading skills. Additionally, the
long span of time between ratings of behavior may have obscured the subtle changes in weekly
behavior that could have been associated with changes in reading growth. Finally, the children
from the most disadvantaged circumstances, and likely the highest rates of severe behavior
problems,were underrepresented in this study.
Third, this study did not consider the classroom or the school effects on children’s
development. Another study looking at this sample of data (McCoach et al. 2006) found that
most of the variance between schools could be explained by the differences in the catchment area
that the school drew their children from. The growth rate of academic skills was comparable
across schools, but schools in advantaged areas tended to draw larger populations of advantaged
children who entered school with higher academic scores, making the school appear to be of a
higher quality. However, these school and classroom effects may be significant, previous
research has shown that approximately 10%-30% of the variance in student achievement lies
between schools and another 12%-20% of variance in student achievement lies between
classrooms within schools (Kyriakides, Campbell, & Gagatsis, 2000; Raudenbush & Byrk, 2002;

106

Rowan, Correnti, & Miller, 2002). As discussed above, previous research has also shown that the
nature of schools and literacy instruction in classrooms can be substantially different (Barone,
2002; Borman & Dowling, 2010; Duke, 2000; Haycock, 2000; Logan & Petscher, 2010; Sirin,
2005; Tabors & Snow, 2001) and these differences may be related to the findings of this study.
Fourth, the nature of the data precludes the measurement of “summer effects.” Previous
studies (Alexander, Entwisle, & Olson, 2007; Cooper, Nye, Charlton, Lindsay, & Greathouse,
1996; Downey, von Hippel, & Broh, 2004; McCoach et al., 2006) have indicated that children
from low-SES suffer greater skill loss over the summer break than their peers from higher SES
families. It appears that the children from the lowest SES families are entering school behind
their peers in reading skills and then remaining significantly behind them in reading skills across
elementary school. However, the design of the study precludes the partitioning of the growth in
IRT reading scores during school from the growth or decline of reading skills across the summer
months, when most children are away from school. For many children this “summer effect” may
be the difference between closing the gap in IRT reading scores with their more skilled peers and
remaining significantly behind them. For the children from the lowest SES families, the summer
may be a critical period that may provide an opportunity for the implementation of intensive
interventions to address the widening gap in IRT reading scores across the elementary school
years.
Finally, the significant amount of missing data limits the generalizability of this study,
especially considering the disproportionate amount of data missing from those children from
extremely disadvantaged families. While it is impossible to be sure, it is highly likely that these
children from the most disadvantaged families would show the slowest growth in reading
development across elementary school and is likely to be a group in the greatest need of effective

107

interventions. On average these families are more likely to experience a greater number of risk
factors associated with slower academic growth (Deater-Deckard, Dodge, Bates, & Pettit, 1998;
Greenberg, Speltz, DeKlyen, & Jones, 2001; Jones, Forehand, Brody, & Armistead, 2002;
Rutter, 1979; Sameroff, 2000; Williams, Anderson, McGee, & Silva, 1990). Additionally, these
more disadvantaged families are more likely than higher SES families to experience high levels
of family instability, which often leads to frequent school absences and a higher than average
number of school transfers (Ackerman, Brown, D’Erama, & Izard, 2002; Capaldi & Patterson,
1991; Eckenrode, Rowe, Laird, & Braithwaite, 1995; Seccombe, 2000; Seltzer, 2000; White &
Rogers, 2000). These transitions and absences are likely to further inhibit these children’s
academic development (Ready, 2010).

Implications of findings for practice and future research
The implications of this study for practice are restricted because of the inherent
limitations of the study and because of the narrow focus of the research questions; however, this
study should serve as a foundation for future research. First, this study supported earlier research
indicating a relationship between low-SES and lower reading skills at school entry, but the
moderate relationship did suggest that there were other unmeasured factors associated with these
lower initial reading skills. Other studies have found that more proximal factors that are often
associated with lower SES mediate this relationship (Eamon, 2002; Gershoff et al., 2007;
Votruba-Drzal, 2006) and these factors are more likely to respond to effective intervention.
Second, there were mixed findings regarding the types of reading developmental
trajectories that children from low-SES families followed. A large percentage of children from
low-SES families followed a cumulative trajectory of reading development, entering school with

108

the lowest IRT reading scores and showing a much slower growth rate than their peers from
higher SES families. However, not all children who entered with the lowest initial IRT reading
scores demonstrated this slower growth rate and some were able to catch up or surpass their
peers who began school with more proficient reading skills. It would be beneficial to take a
closer look at the factors other than SES and behavior ratings that were associated with
membership in the lowest reading cluster to determine if there were specific factors that were
associated with either a more cumulative or compensatory trajectory for children from low-SES
families. It may also be informative to determine if there is an overlap between the different
trajectories found in this study and the different reading profiles found by earlier research
(Pierce, Katzir, Wolf & Noam, 2007; Riddle-Buly & Valencia, 2002; Valencia & Riddle-Buly,
2004). Further, even when controlling for initial IRT reading scores, lower SES was associated
with slower reading growth across elementary school. However, the nature of this study
precluded a separate analysis of school and classroom effects and the separation of growth
during periods when students were receiving instruction in school and the growth or decline
occurring during the summer months. Future research will need to be designed to control for
school and classroom effects and the “summer effects” in reading growth across elementary
school. Finally, this study provided little support regarding the association between higher rates
of behavior problems and slower reading development for children from low-SES families.
However there is reason to believe this relationship may exist and may be stronger than the
findings of this study suggest. Future studies should attempt to collect more detailed data about
behavior during reading instruction, including data gathered through regular classroom
observations.

109

APPENDICES

110

APPENDIX A

TABLES

111

Table 1. Comparison of Reading IRT Scores by Grade Level for Included and Excluded Samples.

Included
Sample
Mean (SD)

Excluded
Sample
Mean (SD)

Independent Samples t-test
(df)

Fall Kindergarten

30.4 (10.0)

24.9 (14.0)

-29.0 (15818), p < .001**

Spring Kindergarten

42.3 (13.4)

36.8 (17.4)

-23.1 (16548), p < .001**

Fall First-grade

50.0 (17.3)

43.6 (20.9)

-12.0 (4710), p < .001**

Spring First-grade

74.8 (21.3)

67.1 (26.0)

-20.9 (16018), p < .001**

Spring Third-grade

122.4 (23.3)

112.1 (30.4)

-23.0 (14048), p < .001**

Spring Fifth-grade

143.1 (21.3)

133.4 (27.1)

-21.1 ((8271), p < .001**

112

Table 2. Comparison of Teacher-Rated SRS Scores by Assessment Period for Included and
Excluded Samples.

Fall Kindergarten
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior

Included
Sample
Mean (SD)

Excluded
Sample
Mean (SD)

Independent
Samples t-test (df)

2.7 (.84)
2.7 (.87)
2.7 (.80)
1.6 (.59)
1.5 (.50)

2.7 (.78)
2.7 (.81)
2.6 (.75)
1.7 (.65)
1.6 (.53)

-4.30 (13839), p < .001**
0.15 (13482), p = .88
-2.26 (13493), p = .02*
10.14 (15046), p < .001**
7.64 (14624), p < .001**

2.7 (.95)
2.6 (.96)
2.6 (.92)
1.6 (.60)
1.5 (.48)

2.6 (.88)
2.6 (.90)
2.6 (.86)
1.7 (.66)
1.6 (.53)

-2.10 (13643), p =. 04*
0.98 (13714), p = .33
1.50 (13621), p = .14
9.64 (15193), p < .001**
9.20 (15022), p < .001**

2.7 (.90)
2.6 (.93)
2.6 (.90)
1.6 (.61)
1.6 (.49)

2.6 (.85)
2.7 (.87)
2.6 (.85)
1.7 (.66)
1.6 (.53)

-4.53 (14013), p < .001**
1.61 (13848), p = .25
0.69 (13818), p = .50
8.86 (14680), p < .001**
7.65 (14585), p < .001**

Spring Kindergarten
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior
Spring First-Grade
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior

113

Table 2 (cont’d)

Spring Third-Grade

Included
Sample Mean
(SD)

Excluded
Sample
Mean (SD)

Independent
Samples t-test (df)

Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior

2.7 (.87)
2.7 (.93)
2.6 (.88)
1.7 (.59)
1.6 (.51)

2.7 (.84)
2.6 (.90)
2.6 (.86)
1.7 (.62)
1.7 (.56)

-3.60 (10923), p < .001**
-0.72 (10791), p = .47
-0.55 (10703), p =. 58
4.68 (10291), p < .001**
6.57 (10023), p < .001**

2.7 (.88)
2.7 (.94)
2.6 (.87)
1.6 (.56)
1.6 (.53)

2.7 (.84)
2.7 (.91)
2.7 (.82)
1.7 (.61)
1.7 (.55)

-1.20 (9018), p = .24
0.49 (8769), p = .62
0.82 (8749), p = .41
4.14 (8053), p < .001**
4.40 (8191), p < .001**

Spring Fifth-Grade
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior

114

Table 3. Sociodemographic Information Comparison Between Excluded and Included Samples.
Included
Sample
6,698

Excluded Sample

Characteristic
Total
Child Gender
Male
Female

3,309 (49.4%)
3,389 (50.6%)

5,676 (52.2%)
5,180 (47.7%)

Child race/ethnicity
White, Non-Hispanic
African American
Hispanic
Asian/Pacific Islander
Native American
More than one race
Unknown

4,534 (67.7%)
701 (10.5%)
752 (11.3%)
407 (2.9%)
113 (1.7%)
187 (2.8%)
4 (.1%)

5,357 (49.3%)
1,793 (16.5%)
2,310 (21.3%)
909 (8.4%)
203 (1.9%)
261 (2.4%)
34 (.3%)

Highest parent level of education
Less than high school
High school graduate
Vocational/technical
Some college
College graduate
Masters
Ph.D./Professional
Unknown

316 (4.7%)
1,534 (22.9%)
358 (5.3%)
1,861 (27.8%)
1,600 (23.9%)
563 (8.4%)
364 (5.4%)
102 (1.5%)

Chi square (df)

1,323 (12.2%)
2,779 (25.6%)
586 (5.4%)
2,615 (24.1%)
1,849 (17%)
640 (5.9%)
394 (3.6%)
681 (6.3%)

Family Type
Dual Parent
Single Parent
Other
Unknown
Home Language
English
Non-English
Unknown
Socioeconomic Status
First Quintile
Second Quintile
Third Quintile
Fourth Quintile
Fifth Quintile
Unknown

10,867
20.6 (2), p < .001**

660 (8), p < .001**

446 (8), p < .001**

161 (4), p < .001**
5,282 (78.9%)
1,092 (16.3%)
93 (1.4%)
231 (3.4%)

6,285 (57.8%)
2,120 (19.5%)
192 (1.8%)
2,270 (20.9%)
496 (2), p < .001**

6,143 (91.7%)
446 (6.7%)
109 (1.6%)

8,240 (75.8%)
1,848 (17%)
779 (7.2%)
470 (4), p < .001**

709 (10.6%)
1,213 (18.1%)
1,371 (20.5%)
1,550 (23.1%)
1,753 (26.2%)
102 (1.5%)

115

2,330 (21.4%)
1,990 (18.3%)
1,975 (18.2%)
1,945 (17.9%)
1,946 (17.9%)
681 (6.3%)

Table 4. Sociodemographic Information for Final Sample Used in this Study.
_________Sector_______
Public
Private

Characteristic

Total

Total
Region
Northeast
Midwest
South
West

6,698

5,138 (76.7%)

1,560 (23.3%)

1,336 (19.9%)
1,968 (29.4%)
2,105 (31.4%)
1,289 (19.2%)

1,012 (19.7%)
1,396 (27.2%)
1,795 (34.9%)
935 (18.2%)

324 (20.8%)
572 (36.7%)
310 (19.9%)
354 (22.7%)

Type of locale
Large & Mid-size city
Suburbs & Large Town
Small Town & Rural

2,211 (33%)
2,592 (38.7%)
1,895 (28.3%)

1,377 (26.8%)
2,103 (40.9%)
1,658 (32.3%)

834 (53.5%)
489 (31.3%)
237 (15.2%)

Child race/ethnicity
White, Non-Hispanic
African American
Hispanic
Asian/Pacific Islander
Native American
More than one race

4,534 (67.7%)
701 (10.5%)
752 (11.3%)
407 (2.9%)
113 (1.7%)
187 (2.8%)

3,356 (65.3%)
639 (12.4%)
585 (11.4%)
312 (6.1%)
99 (1.9%)
145 (2.8%)

1,178 (75.5%)
62 (4%)
167 (10.7%)
95 (6.1%)
14 (.9%)
42 (2.7%)

Child Gender
Male
Female

3,309 (49.4%)
3,389 (50.6%)

2,537 (49.4%)
2,601 (50.6%)

772 (49.5%)
788 (50.5%)

Highest parent level of education
Less than high school
High school graduate
Vocational/technical
Some college
College graduate
Masters
Ph.D./Professional
Unknown

316 (4.7%)
1,534 (22.9%)
358 (5.3%)
1,861 (27.8%)
1,600 (23.9%)
563 (8.4%)
364 (5.4%)
102 (1.5%)

307 (5.9%)
1,378 (26.8%)
300 (5.8%)
1,464 (28.5%)
1,062 (20.7%)
353 (6.9%)
193 (3.8%)
81 (1.6%)

9 (.6%)
156 (10%)
58 (3.7%)
397 (25.4%)
538 (34.5%)
210 (13.5%)
171 (11%)
21 (1.3%)

Socioeconomic Status
First Quintile
Second Quintile
Third Quintile
Fourth Quintile
Fifth Quintile
Unknown

709 (10.6%)
1,213 (18.1%)
1,371 (20.5%)
1,550 (23.1%)
1,753 (26.2%)
102 (1.5%)

690 (13.4%)
1,077 (21%)
1,107 (21.5%)
1,114 (21.7%)
1,069 (20.8%)
81 (1.6%)

19 (1.2%)
136 (8.7%)
264 (16.9%)
436 (27.9%)
684 (43.8%)
21 (1.3%)

116

Table 5. Means and Standard Deviations for Reading Assessments by Grade Level.

Mean

Standard
Deviation

Fall Kindergarten

30.4

10.0

Spring Kindergarten

42.3

13.4

Fall First-grade

50.0

17.3

Spring First-grade

74.8

21.3

Spring Third-grade

122.4

23.3

Spring Fifth-grade

143.1

21.3

117

Table 6. Means and Standard Deviations for Proficiency Probabilities by Grade and Skill Level.

Fall Kindergarten
Letter Recognition
Beginning Sounds
Ending Sounds
Sight Words
Words in Context
Literal Inference
Extrapolation
Evaluation
Evaluating Non-fiction

Mean

Standard
Deviation

.73
.34
.19
.03
.02
.00
.00
.00
.00

.31
.33
.27
.13
.08
.04
.01
.01
.00

.95
.75
.56
.17
.07
.01
.00
.00
.00

.12
.29
.33
.27
.16
.07
.02
.02
.00

.98
.87
.73
.32
.14
.03
.00
.00
.00

.24
.30
.35
.40
.32
.24
.21
.21
.20

Spring Kindergarten
Letter Recognition
Beginning Sounds
Ending Sounds
Sight Words
Words in Context
Literal Inference
Extrapolation
Evaluation
Evaluating Non-Fiction
Fall First-Grade
Letter Recognition
Beginning Sounds
Ending Sounds
Sight Words
Words in Context
Literal Inference
Extrapolation
Evaluation
Evaluating Non-Fiction

118

Table 6 (Cont’d)

Spring First-Grade
Letter Recognition
Beginning Sounds
Ending Sounds
Sight Words
Words in Context
Literal Inference
Extrapolation
Evaluation
Evaluating Non-fiction

Mean

Standard
Deviation

1.0
.99
.96
.82
.52
.19
.04
.04
.00

.02
.07
.12
.27
.31
.24
.12
.06
.00

1.0
1.0
1.0
.98
.94
.75
.50
.28
.01

.00
.00
.01
.04
.12
.26
.37
.21
.05

1.0
1.0
1.0
1.0
.98
.91
.79
.50
.09

.00
.00
.00
.01
.05
.15
.30
.26
.20

Spring Third-Grade
Letter Recognition
Beginning Sounds
Ending Sounds
Sight Words
Words in Context
Literal Inference
Extrapolation
Evaluation
Evaluating Non-Fiction
Spring Fifth-Grade
Letter Recognition
Beginning Sounds
Ending Sounds
Sight Words
Words in Context
Literal Inference
Extrapolation
Evaluation
Evaluating Non-Fiction

119

Table 7. Means and Standard Deviations for Teacher-rated SRS by Assessment Period.

Fall Kindergarten
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior

Mean

Standard
Deviation

Split half
Reliability

2.7
2.7
2.7
1.6
1.5

.84
.87
.80
.59
.50

.89
.79
.89
.90
.80

2.7
2.6
2.6
1.6
1.5

.95
.96
.92
.60
.48

.89
.80
.89
.90
.78

2.7
2.6
2.6
1.6
1.6

.90
.93
.90
.61
.49

(NR)
(NR)
(NR)
(NR)
(NR)

2.7
2.7
2.6
1.7
1.6

.87
.93
.88
.59
.51

.91
.79
.89
.89
.76

2.7
2.7
2.6
1.6
1.6

.88
.94
.87
.56
.53

.91
.79
.88
.89
.77

Spring Kindergarten
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior
Spring First-Grade
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior
Spring Third-Grade
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior
Spring Fifth-Grade
Approaches to Learning
Self-Control
Interpersonal
Externalizing Behavior
Internalizing Behavior

120

Table 8. GSS Occupational Codes by Parent for Initial Kindergarten Sample.
GSS Code

Occupation

Father

Mother

29.6
33.42
34.95
35.63
35.78
35.92
37.67
38.18
39.18
39.2
48.69
52.54
53.5
57.83
59
61.56
62.87
63.43
64.89
72.1

Handler, Equip, Cleaner, Helpers, Labor
Production Working Occupation
Service Occupations
Agriculture, Forestry, Fishing Occupations
Marketing & Sales Occupation
Transportation, Material Moving
Precision Production Occupation
Administrative Support, Including Clerk
Mechanics & Repairs
Construction & Extractive Occupations
Technologists, Except Health
Writers, Artists, Entertainers, Athletes
Executive, Admin, Managerial Occupation
Health Technologists & Technicians
Social Scientist/Workers, Lawyers
Registered Nurses, Pharmacists
Natural Scientists & Mathematicians
Teacher, except Postsecondary
Engineers, Surveyors, & Architects
Teachers, College; Postsecondary
Counselors; Librarians
Physicians, Dentists, Veterinarians

199 (1.5%)
920 (6.8%)
861 (6.3%)
339 (2.5%)
852 (6.3%)
696 (5.1%)
269 (2.0%)
531 (3.9%)
798 (5.9%)
1,076 (7.9%)
380 (2.8%)
112 (.8%)
1,607 (11.8%)
60 (.4%)
217 (1.6%)
75 (.6%)
129 (.9%)
142 (1.0%)
354 (2.6%)

93 (.7%)
486 (3.6%)
1,909 (14%)
44 (.3%)
901 (6.6%)
106 (.8%)
53 (.4%)
2,297 (16.9%)
22 (.2%)
24 (.2%)
171 (1.3%)
171 (1.3%)
1,063 (7.8%)
252 (1.9%)
176 (1.3%)
448 (3.3%)
52 (.4%)
575 (4.2%)
39 (.3%)

101 (.7%)
235 (1.7%)

112 (.8%)
85 (.6%)

75.5

121

Table 9. Reading IRT Scores by SES Quintile & Measurement Point
Beg. of K
SES
Quintile
1st
2nd
3rd
4th
5th

Mean (SD)

1st
2nd
3rd
4th
5th

35.74 (9.75)
38.76 (11.72)
41.24 (12.13)
43.95(12.34)
47.60 (15.24)

1st
2nd
3rd
4th
5th

62.35 (17.20)
68.87 (18.93)
72.70 (18.82)
78.11 (20.08)
83.88 (22.03)

1st
2nd
3rd
4th
5th

104.10 (23.18)
114.73 (22.84)
120.85 (22.13)
127.01 (20.35)
134.64 (19.26)

1st
2nd
3rd
4th
5th

126.96 (22.36)
135.78 (21.52)
142.03 (19.35)
147.41 (17.53)
154.42 (16.26)

24.85 (5.46)
27.53 (8.50)
29.56 (8.80)
31.55 (11.81)
35.09 (10.14)

End of K

1st Grade

3rd Grade

5th Grade

122

Table 10. Behavior Ratings by SES Quintile & Measurement Point
Beg. of K
SES
Quintile
1st
2nd
3rd
4th
5th

Mean (SD)

1st
2nd
3rd
4th
5th

14.43 (2.03)
14.64 (2.10)
14.77 (2.11)
14.88 (2.11)
14.77 (2.12)

1st
2nd
3rd
4th
5th

14.44 (2.14)
14.68 (2.00)
14.80 (2.01)
14.94 (2.14)
15.01 (2.17)

1st
2nd
3rd
4th
5th

14.39 (2.10)
14.73 (2.03)
14.63 (2.11)
14.91 (2.13)
14.89 (2.13)

1st
2nd
3rd
4th
5th

14.38 (1.99)
14.69 (2.03)
14.68 (2.09)
14.84 (2.15)
15.00 (2.23)

14.65 (1.95)
14.95 (2.01)
15.02 (1.95)
15.03 (1.93)
15.11 (2.03)

End of K

1st Grade

4th Grade

5th Grade

123

Table 11. Univariate Simplex Model for Reading with SES.
Parameter Estimates
Read T2  Read T1
Read T3  Read T2
Read T4  Read T3
Read T5  Read T4
Read T1  SES
Read T2  SES
Read T3  SES
Read T4  SES
Read T5  SES
Variance (Latent Read T1)
Variance (Latent Read T2)
Variance (Latent Read T3)
Variance (Latent Read T4)
Variance (Latent Read T5)
Residuals (Read T1 with Read T5)
Residuals (Read T2 with Read T4)
Residuals (Read T3 with Read T5)
Residual for Latent Read T1
Residual for Latent Read T2
Residual for Latent Read T3
Residual for Latent Read T4
Residual for Latent Read T5

Unstandardized
1.21** (.02)
1.21** (.02)
.74** (.01)
.69**(.01)
4.85** (.18)
n.s. (dropped from model)
2.74** (.27)
5.36** (.32)
2.79** (.24)
86.43
163.70
428.55
508.59
412.77
9.71** (1.69)
-4.46* (1.82)
18.06** (2.42)
80.50** (1.97)
35.18** (2.69)
163.31** (4.18)
232.48** (4.99)
109.69** (2.94)

124

Standardized
.89** (.01)
.76** (.01)
.68** (.01)
.77** (.01)
.37** (.01)
n.s. (dropped from model)
.10** (.01)
.17** (.01)
.10** (.01)
1.0
1.0
1.0
1.0
1.0
.10** (.02)
-.05* (.02)
.14** (.02)
.87** (.01)
.21** (.02)
.37** (.01)
.45** (.01)
.26** (.01)

Table 12. Univariate Piecewise Growth Curve Model for Reading with SES.
Parameter Estimates
Level with Read Growth I
Level with Read Growth II
Level with Read Growth III
Read Growth I with Read Growth II
Read Growth I with Read Growth III
Read Growth II with Read Growth III
Level  SES
Read Growth I  SES
Read Growth II  SES
Read Growth III  SES
Intercept (Level)
Intercept (Read Growth I)
Intercept (Read Growth II)
Intercept (Read Growth III)
Residual for Read T1
Residual for Read T2
Residual for Read T3
Residual for Read T4
Residual for Read T5
Residual for Level
Residual for Read Growth I
Residual for Read Growth II
Residual for Read Growth III

Unstandardized
18.54** (1.67)
5.87 (3.56)
-13.75** (3.68)
-5.07**(.83)
-2.44** (.43)
-3.03** (1.13)
6.05** (.31)
1.04** (.11)
1.44** (.14)
-.39** (.07)
-9.29** (1.00)
8.57** (.35)
4.46** (.45)
8.17** (.24)
3.43* (1.40)
33.62** (1.34)
72.22** (4.53)
64.86** (9.40)
67.71** (16.90)
146.21** (12.27)
20.93** (.88)
19.69** (2.40)
2.52* (1.19)

125

Standardized
.34** (.04)
.11 (.06)
-.72** (.02)
-.25** (03)
-.34* (.12)
-.43** (.10)
.34** (.02)
.16** (.02)
.23** (.03)
-.18** (.05)
-.72** (.08)
1.85** (.09)
.98** (.12)
5.60** (1.16)
.03* (.01)
.19** (.01)
.16** (.01)
.12** (.02)
.16** (.04)
.88** (.01)
.97** (.01)
.95** (.01)
.97** (.02)

Table 13. Univariate Simplex Model for Behavior with SES.
Parameter Estimates
Beh T2  Beh T1
Beh T3  Beh T2
Beh T4  Beh T3
Beh T5  Beh T4
Beh T1  SES
Beh T2  SES
Beh T3  SES
Beh T4  SES
Beh T5  SES
Variance (Latent Beh T1)
Variance (Latent Beh T2)
Variance (Latent Beh T3)
Variance (Latent Beh T4)
Variance (Latent Beh T5)
Residuals (Beh T1 with Beh T5)
Residuals (Beh T1 with Beh T4)
Residual for Latent Beh T1
Residual for Latent Beh T2
Residual for Latent Beh T3
Residual for Latent Beh T4
Residual for Latent Beh T5

Unstandardized
1.07** (.11)
.24** (.03)
.44** (.06)
.46**(.06)
.17** (.03)
n.s. (dropped from model)
.19** (.04)
n.s. (dropped from model)
.14** (.06)
.54
1.41
.47
.56
.55
.19** (.06)
.27** (.06)
2.77** (.13)
2.08** (.12)
3.17** (.14)
3.09** (.13)
3.07** (.14)

126

Standardized
.73** (.03)
.33** (.03)
.42** (.05)
.45** (.05)
.12** (.02)
n.s. (dropped from model)
.12** (.02)
n.s. (dropped from model)
.09** (.02)
1.0
1.0
1.0
1.0
1.0
.17** (.05)
.25* (.05)
.71** (.03)
.46** (.02)
.71** (.03)
.69** (.03)
.68** (.03)

Table 14. Multivariate Simplex Model for Reading and Behavior with SES.
Parameter Estimates
Read T2  Read T1
Read T3  Read T2
Read T4  Read T3
Read T5  Read T4
Beh T2  Beh T1
Beh T3  Beh T2
Beh T4  Beh T3
Beh T5  Beh T4
Read T2  Beh T1
Read T3  Beh T2
Read T4  Beh T3
Read T5  Beh T4
Read T1  SES
Read T2  SES
Read T3  SES
Read T4  SES
Read T5  SES
Beh T1  SES
Beh T2  SES
Beh T3  SES
Beh T4  SES
Beh T5  SES
Variance (Latent Read T1)
Variance (Latent Read T2)
Variance (Latent Read T3)
Variance (Latent Read T4)
Variance (Latent Read T5)
Variance (Latent Beh T1)
Variance (Latent Beh T2)
Variance (Latent Beh T3)
Variance (Latent Beh T4)
Variance (Latent Beh T5)

Unstandardized
1.10** (.01)
1.14** (.02)
.71** (.01)
.68**(.01)
1.10** (.10)
.25** (.03)
.49** (.07)
.47** (.05)
.57** (.15)
.35* (.16)
2.47** (.41)
.75** (.22)
4.89** (.18)
n.s. (dropped from model)
3.12** (.27)
5.11** (.33)
2.89** (.23)
.17** (.03)
n.s. (dropped from model)
.19** (.04)
n.s. (dropped from model)
.14** (.04)
100.61
178.67
444.95
524.97
429.09
.50
1.35
.45
.57
.52

127

Standardized
.83** (.00)
.72** (.01)
.65** (.01)
.75** (.01)
.74** (.03)
.35** (.03)
.46** (.05)
.47** (.05)
.04** (.01)
.03* (.01)
.12** (.02)
.04** (.01)
.36** (.01)
n.s. (dropped from model)
.11** (.01)
.16** (.01)
.10** (.01)
.12** (.02)
n.s. (dropped from model)
.12** (.02)
n.s. (dropped from model)
.08** (.02)
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0

Table 14. (cont’d.)
Parameter Estimates
Residual (Read T1 with Read T5)
Residual (Read T2 with Read T4)
Residual (Read T3 with Read T5)
Residual (Beh T1 with Beh T5)
Residual (Beh T1 with Beh T4)
Residual for Latent Read T1
Residual for Latent Read T2
Residual for Latent Read T3
Residual for Latent Read T4
Residual for Latent Read T5
Residual for Latent Beh T1
Residual for Latent Beh T2
Residual for Latent Beh T3
Residual for Latent Beh T4
Residual for Latent Beh T5

Unstandardized
9.37** (1.62)
-3.79* (1.74)
19.09** (2.33)
.17** (.05)
.26** (.06)
88.77** (1.72)
54.57** (1.12)
182.80** (3.55)
238.53** (4.88)
122.06** (2.42)
2.81** (.12)
2.15** (.12)
3.26** (.14)
3.10** (.12)
3.12** (.13)

128

Standardized
.09** (.02)
-.03* (.02)
.13** (.02)
.17** (.05)
.25** (.05)
.87** (.01)
.30** (.01)
.41** (.01)
.46** (.01)
.29** (.01)
.73** (.03)
.48** (.03)
.73**(.03)
.69** (.02)
.69** (.02)

Table 15. Univariate Piecewise Growth Curve Model for Behavior with SES.
Parameter Estimates
Level with Beh Growth I
Level with Beh Growth II
Level with Beh Growth III
Beh Growth I with Beh Growth II
Beh Growth I with Beh Growth III
Beh Growth II with Beh Growth III
Level  SES
Beh Growth I  SES
Beh Growth II  SES
Beh Growth III  SES
Intercept (Level)
Intercept (Beh Growth I)
Intercept (Beh Growth II)
Intercept (Beh Growth III)
Residual for Beh T1
Residual for Beh T2
Residual for Beh T3
Residual for Beh T4
Residual for Beh T5
Residual for Level
Residual for Beh Growth I
Residual for Beh Growth II
Residual for Beh Growth III

Unstandardized
-.38** (.10)
.08 n.s. (.13)
-.20 n.s. (.15)
-.04 n.s. (.16)
-.01 n.s. (.09)
-.05 n.s. (.16)
.19** (.04)
.06 n.s. (.04)
-.06 n.s. (.05)
.02 n.s. (.05)
14.25** (.14)
-.01 n.s. (.13)
-.08 n.s. (.18)
.03 n.s. (.14)
3.07** (.10)
2.44** (.20)
3.83** (.11)
3.77** (.11)
3.81** (.21)
1.14** (.16)
.40* (.17)
.13 n.s. (.24)
.12 n.s. (.17)

129

Standardized
-.56** (.14)
.22 n.s. (.34)
-.55** (.16)
-.16 n.s. (.54)
-.05 n.s. (.44)
-.40 n.s. (.73)
.13** (.03)
.07 n.s. (.05)
-.12 n.s. (.16)
.05 n.s. (.09)
13.25** (.94)
-.02 n.s. (.20)
-.22 n.s. (.54)
.09 n.s. (.40)
.78** (.02)
.55** (.04)
.86** (.02)
.84** (.02)
.84** (.04)
.98** (.01)
1.0** (.01)
.99** (.04)
1.0** (.01)

Table 16. Multivariate Growth Curve Model for Reading and Behavior with SES.
Parameter Estimates
Read Level with Read Growth I
Read Level with Read Growth II
Read Level with Read Growth III
Beh Level with Beh Growth I
Beh Level with Beh Growth II
Beh Level with Beh Growth III
Read Growth I with Read Growth II
Read Growth I with Read Growth III
Read Growth II with Read Growth III
Beh Growth I with Beh Growth II
Beh Growth I with Beh Growth III
Beh Growth II with Beh Growth III
Read Level  SES
Read Growth I  SES
Read Growth II  SES
Read Growth III  SES
Beh Level  SES
Beh Growth I  SES
Beh Growth II  SES
Beh Growth III  SES
Read Level with Beh Level
Read Level with Beh Growth I
Read Level with Beh Growth II
Read Level with Beh Growth III
Beh Level with Read Growth I
Beh Level with Read Growth II
Beh Level with Read Growth III
Read Growth I with Beh Growth I
Read Growth II with Beh Growth I
Read Growth III with Beh Growth I
Read Growth I with Beh Growth II
Read Growth II with Beh Growth II
Read Growth III with Beh Growth II
Read Growth I with Beh Growth III
Read Growth II with Beh Growth III
Read Growth III with Beh Growth III

Unstandardized
18.59** (1.67)
5.89 n.s. (3.56)
-13.36** (3.66)
-.37** (.10)
.07 n.s. (.13)
-.18 n.s. (.15)
-5.11** (.82)
-2.43** (.43)
-3.05* (1.13)
-.04 n.s. (.16)
-.01 n.s. (.09)
-.05 n.s. (.16)
6.05** (.31)
1.04** (.11)
1.44** (.14)
-.39** (.07)
.19** (.04)
.06 n.s. (.04)
-.06 n.s. (.05)
.02 n.s. (.04)
1.27 * (.53)
1.16* (.47)
-1.19 n.s. (.65)
.37 n.s. (.50)
.79** (.19)
.40 n.s. (.23)
-.19 n.s. (.13)
.27 n.s. (.17)
.13 n.s. (.21)
-.15 n.s. (.11)
.02 n.s. (.23)
.14 n.s. (.29)
.04 n.s. (.16)
-.27 n.s. (.18)
-.01 n.s. (.23)
.04 n.s. (.12)

130

Standardized
.34** (.04)
.11** (.06)
-.72** (.02)
-.56** (.15)
.17 n.s. (.31)
-.53** (.18)
-.25** (.03)
-.34* (.12)
-.44** (.10)
-.16 n.s. (.52)
-.05 n.s. (.48)
-.40 n.s. (.75)
.35** (.02)
.16** (.02)
.23** (.03)
-.18** (.06)
.13** (.03)
.07 n.s. (.05)
-.11 n.s. (.14)
.05 n.s. (.10)
.10* (.04)
.16* (.07)
-.26 n.s. (.27)
.10 n.s. (.15)
.16** (.04)
.09 n.s. (.05)
-.12 n.s. (.08)
.10 n.s. (.06)
.05 n.s. (.08)
-.16 n.s. (.13)
.01 n.s. (.13)
.08 n.s. (.19)
.06 n.s. (.28)
-.18 n.s. (.18)
-.01 n.s. (.15)
.08 n.s. (.26)

Table 16. (cont’d)
Parameter Estimates
Intercept (Read Level)
Intercept (Read Growth I)
Intercept (Read Growth II)
Intercept (Read Growth III)
Intercept (Beh Level)
Intercept (Beh Growth I)
Intercept (Beh Growth II)
Intercept (Beh Growth III)
Residual for Read T1
Residual for Read T2
Residual for Read T3
Residual for Read T4
Residual for Read T5
Residual for Read Level
Residual for Read Growth I
Residual for Read Growth II
Residual for Read Growth III
Residual for Beh T1
Residual for Beh T2
Residual for Beh T3
Residual for Beh T4
Residual for Beh T5
Residual for Beh Level
Residual for Beh Growth I
Residual for Beh Growth II
Residual for Beh Growth III

Unstandardized
-9.29** (1.0)
8.57** (.35)
4.46** (.45)
8.17** (.24)
14.25** (.14)
-.01 n.s. (.13)
-.08 n.s. (.18)
.03 n.s. (.14)
3.58* (1.40)
33.48** (1.34)
72.78** (4.50)
63.51** (9.25)
70.17** (16.67)
144.86** (12.20)
20.87** (.87)
19.93** (2.38)
2.40* (1.19)
3.05** (.10)
2.46** (.20)
3.84** (.11)
3.76** (.11)
3.84** (.21)
1.12** (.16)
.39* (.17)
.14 n.s. (.24)
.11 n.s. (.17)

131

Standardized
-.72** (.08)
1.85** (.09)
.97** (.12)
5.19** (1.24)
13.38** (.96)
-.02 n.s. (.20)
-.20 n.s. (.50)
.09 n.s. (.43)
.04* (.01)
.18** (.01)
.16** (.01)
.12** (.02)
.16** (.04)
.88** (.01)
.97** (.01)
.95** (.01)
.97** (.02)
.78** (.02)
.55** (.04)
.86** (.02)
.84** (.02)
.85** (.04)
.98** (.01)
1.0** (.01)
.99** (.03)
1.0** (.01)

Table 17. Comparison of IRT Reading Scores and Proficiency Scores by Cluster.
Cluster 1
Time
Point
Beg K
End K
First
Third
Fifth
Cluster 2
Time
Point
Beg K
End K
First
Third
Fifth

IRT
Score

Letter
Recog

Beg
Sounds

End
Sounds

Sight
Words

Words in
Context

Literal
Inferen

Extrap

Eval

98.27
(11.99)
108.22
(13.83)
132.46
(14.94)
158.85
(12.3)
171.28
(9.08)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.88 (.01)

.46 (.03)

.08 (.02)

.08 (.01)

Eval
NonFiction
.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.93 (.01)

.63 (.03)

.20 (.03)

.13 (.01)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.98 (.00)

.88 (.02)

.66 (.05)

.35 (.03)

.01 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.97 (.00)

.71 (.03)

.19 (.05)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.88 (.02)

.57 (.06)

IRT
Score

Letter
Recog

Beg
Sounds

End
Sounds

Sight
Words

Words in
Context

Literal
Inferen

Extrap

Eval

68.02
(8.26)
90.79
(13.70)
119.75
(13.20)
148.82
(14.08)
162.79
(13.06)

1.0 (.00)

1.0 (.00)

.98 (.00)

.87 (.01)

.44 (.02)

.07 (.01)

.00 (.00)

.01 (.00)

Eval
NonFiction
.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.80 (.01)

.36 (.03)

.05 (.01)

.06 (.01)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.96 (.00)

.77 (.02)

.41 (.03)

.22 (.02)

.01 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.96 (.01)

.90 (.02)

.56 (.03)

.07 (.02)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.01)

.97 (.01)

.76 (.02)

.33 (.04)

132

Table 17. (cont’d)
Cluster 3
Time
Point
Beg K
End K
First
Third
Fifth
Cluster 4
Time
Point
Beg K
End K
First
Third
Fifth

IRT
Score

Letter
Recog

Beg
Sounds

End
Sounds

Sight
Words

Words in
Context

Literal
Inferen

Extrap

Eval

46.15
(4.07)
65.98
(12.18)
110.49
(14.40)
149.83
(13.11)
164.20
(11.54)

1.0 (.00)

.95 (.00)

.80 (.00)

.21 (.01)

.05 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

Eval
NonFiction
.00 (.00)

1.0 (.00)

.99 (.00)

.97 (.00)

.76 (.01)

.39 (.01)

.08 (.01)

.00 (.00)

.01 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.93 (.00)

.65 (.01)

.24 (.01)

.15 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.97 (.00)

.92 (.01)

.58 (.01)

.06 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.98 (.00)

.78 (.01)

.33 (.02)

IRT
Score

Letter
Recog

Beg
Sounds

End
Sounds

Sight
Words

Words in
Context

Literal
Inferen

Extrap

Eval

36.71
(3.52)
48.91
(6.37)
90.14
(13.89)
139.57
(13.48)
156.86
(13.18)

.98 (.00)

.71 (.01)

.41 (.01)

.02 (.01)

.01 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

Eval
NonFiction
.00 (.00)

1.0 (.00)

.96 (.00)

.84 (.00)

.31 (.00)

.10 (.00)

.01 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.98 (.00)

.79 (.00)

.34 (.01)

.05 (.01)

.01 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.93 (.00)

.80 (.01)

.43 (.01)

.02 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.98 (.00)

.96 (.00)

.68 (.01)

.16 (.01)

133

Table 17. (cont’d)
Cluster 5
Time
Point
Beg K
End K
First
Third
Fifth
Cluster 6
Time
Point
Beg K
End K
First
Third
Fifth

IRT
Score

Letter
Recog

Beg
Sounds

End
Sounds

Sight
Words

Words in
Context

Literal
Inferen

Extrap

Eval

28.80
(3.18)
40.47
(5.50)
73.21
(12.00)
129.92
(15.04)
147.10
(15.84)

.81 (.00)

.25 (.00)

.08 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

Eval
NonFiction
.00 (.00)

.99 (.00)

.82 (.00)

.58 (.01)

.09 (.00)

.03 (.01)

.00 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.89 (.00)

.53 (.01)

.13 (.00)

.01 (.00)

.02 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.97 (.00)

.84 (.00)

.56 (.01)

.29 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.99 (.00)

.95 (.00)

.87 (.00)

.54 (.00)

.07 (.00)

IRT
Score
22.78
(3.02)
32.60
(5.69)
56.78
(11.23)
99.13
(16.55)
124.99
(19.17)

Letter
Recog
.39 (.01)

Beg
Sounds
.05 (.00)

End
Sounds
.01 (.00)

Sight
Words
.00 (.00)

Words in
Context
.00 (.00)

Literal
Inferen
.00 (.00)

Extrap

Eval

.00 (.00)

.00 (.00)

Eval
Fiction
.00 (.00)

.88 (.00)

.48 (.01)

.24 (.01)

.02 (.00)

.01 (.01)

.00 (.00)

.00 (.00)

.00 (.00)

.00 (.00)

1.0 (.00)

.96 (.00)

.89 (.00)

.58 (.01)

.23 (.00)

.03 (.00)

.00 (.00)

.01 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.98 (.00)

.85 (.00)

.49 (.01)

.13 (.00)

.10 (.00)

.00 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

1.0 (.00)

.96 (.00)

.80 (.01)

.54 (.01)

.29 (.00)

.01 (.00)

134

APPENDIX B

FIGURES

135

Figure 1. Univariate Simplex Model for Reading Development.

Fall K

Spring K

Spring 1st

Spring 3rd

Spring 5th

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

SES

136

Figure 2. Univariate Latent Growth Curve Model for Reading Development.

Reading
Time 1

Reading
Time 2

Reading
Time 3

Level

Reading
Time 4

Slope

SES

137

Reading
Time 5

Figure 3. Univariate Piecewise Growth Curve Model for Reading Development.

Reading
Time 1

Level

Reading
Time 2

Reading
Time 3

Growth I

Reading
Time 4

Growth II

SES

138

Reading
Time 5

Growth III

Figure 4. Univariate Simplex Model for Behavior Ratings.

Fall K

Spring K

Spring 1st

Spring 3rd

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

Spring 5th

Behavior
Ratings

SES
Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

139

Behavior
Ratings

Behavior
Ratings

Figure 5. Multivariate Simplex Model for Reading and Behavior.
Fall K

Spring K

Spring 1st

Spring 3rd

Spring 5th

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Reading
Skills

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

SES

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

Behavior
Ratings

Fall K

Spring K

Spring 1st

Spring 3rd

140

Behavior
Ratings

Spring 5th

Figure 6. Univariate Latent Growth Curve Model for Behavior Ratings.

Behavior
Time 1

Behavior
Time 2

Behavior
Time 3

Level

Behavior
Time 4

Slope

SES

141

Behavior
Time 5

Figure 7. Univariate Piecewise Growth Curve Model for Behavior Ratings.

Behavior
Time 1

Level

Behavior
Time 2

Behavior
Time 3

Growth I

Behavior
Time 4

Growth II

SES

142

Behavior
Time 5

Growth III

Figure 8. Multivariate Piecewise Growth Curve Model for Reading & Behavior.
Behavior
Time 1

Level

Behavior
Time 2

Behavior
Time 3

Growth I

Behavior
Time 4

Growth II

Behavior
Time 5

Growth III

SES

Reading
Time 1

Reading
Time 2

Reading
Time 3

143

Reading
Time 4

Reading
Time 5

Figure 9. Comparison of Reading Trajectories by Clusters.

* For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this
dissertation.

144

References

145

REFERENCES
Aarnouste, C., Van Leeuwe, J., Voeten, M., & Oud, H. (2001). Development of decoding,
reading comprehension, vocabulary and spelling during the elementary school years.
Reading and Writing: An Interdisciplinary Journal, 14, 61-89.
Aber, L., Bennett, N., Conley, D., & Li, J. (1997). The effects of poverty on child health and
development. Annual Review of Public Health, 18, 463–483.
Abikoff, H. B., Jensen, P. S., Arnold, L. E., Hoza, B., Hechtman, L., Pollack, S., et al. (2002).
Observed classroom behavior of children with ADHD: Relationship to gender and
comorbidity. Journal of Abnormal Child Psychopathology, 30, 349– 360.
Adams, M. J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA:
MIT Press.
Alexander, K. L., & Entwisle, D.R. (1988). Achievement in the first 2 years of school: Patterns
and processes. Monographs of the Society for the Research in Child Development, 53 (2,
(Serial No. 218).
Al Otaiba, S.L., Connor, C., Lane, H., Kosanovich, M.L., Schatschneider, C., Dyrlund, A.K.,
Miller, M.S., & Wright, T.L. (2008). Reading First kindergarten classroom instruction
and students’ growth in phonological awareness and letter naming- decoding fluency.
Journal of School Psychology, 46, 281-314.
Al Otaiba, S. & Fuchs, D. (2002). Characteristics of children who are unresponsive to early
literacy intervention: A review of the literature. Remedial and Special Education, 23,
300-316.
Anthony, J. L. & Lonigan, C.J. (2004). The nature of phonological awareness: Converging
evidence from four studies of preschool and early grade school children. Journal of
Educational Psychology, 96, 43-55.
Anyon, J. (1981). Social class and school knowledge. Curriculum Inquiry, 11, 3-42.
Aram, D. (2005). Continuity in children's literacy achievements: A longitudinal perspective from
kindergarten to school. First Language, 25, 259-289.
Arnold, D. H. (1997). Co-Occurrence of externalizing behavior problems and emergent
academic difficulties in young high-risk boys: A preliminary evaluation of patterns and
mechanisms. Journal of Applied Developmental Psychology, 18, 317-330.
Arnold, E. M., Goldston, D. B., Walsh, A. K., Reboussin, B. A., Daniel, S. S., Hickman, E., et
al. (2005). Severity of emotional and behavioral problems among poor and typical
readers. Journal of Abnormal Child Psychology, 33, 205-217.

146

Aunola, K., Leskinen, E., Onatsu-Arvilommi, T., & Nurmi, J.E. (2002). Three methods for
studying developmental change: A case of reading skills and self-concept. British Journal
of Educational Psychology, 72, 343-364.
Baker, J.A., Clark, T.P., Maier, K., Viger, S. (2008). The differential influence of instructional
context on the academic engagement of students with behavior problems. Teaching and
Teacher Education, 24, 1876-1883.
Barriga, A. Q., Doran, J.W., Newell, S.B., Morrison, E.M., Barbetti, V., & Robbins, B.D.
(2002).Relationships between problem behaviors and academic achievement in
adolescents: The unique role of attention problems. Journal of Emotional and
Behavioral Disorders, 10, 233-240.
Barone, D. (2002). Literacy teaching and learning in two kindergarten classrooms in a school
labeled at-risk. The Elementary School Journal, 102, 415-441.
Bast, J. & Reitsma, P. (1997). Matthew Effects in reading: A comparison of latent growth curve
models and simplex models with structured means. Multivariate Behavioral Research,
32, 135-167.
Bast, J. & Reitsma, P. (1998). Analyzing the development of individual differences in terms of
Matthew Effects in reading: Results from a Dutch Longitudinal Study. Developmental
Psychology, 34, 1373-1399.
Becker, H. & Sterling, C.W. (1987). Equity in school computer use: National data and neglected
considerations. Journal of Educational Computing Research, 3, 289-311.
Benasich, A. A. & Brooks-Gunn, J. (1996). Maternal attitudes and knowledge of child-rearing:
Associations with family and child outcomes. Child Development, 67, 1186-1205.
Benner, G.J., Beaudoin, K., Kinder,D. & Mooney, P. (2005). The relationship between the
beginning reading skills and social adjustment of a general sample of elementary aged
children. Education and Treatment of Children, 28, 250-264.
Benzies, K., Keown, L., & Magill-Evans, J. (2009). Immediate and sustained effects of parenting
on physical aggression in Canadian children aged 6 years and younger. The Canadian
Journal of Psychiatry, 54, 55-64.
Bergman, L. R., Eklund, G., &Magnusson, D. (1991). Studying individual development:
Problems and methods. In D. Magnusson, L. R. Bergman, G. Rudinger & B. Torestad
(Eds.), Problems and methods in longitudinal research: Stability and change (pp. 1–27).
Cambridge: Cambridge University Press.
Bergman, L.R. & Magnusson, D. (1997). A person-oriented approach in research on
developmental psychopathology. Development & Psychopathology, 9, 291-319.

147

Birch, S.H. & Ladd, G.W. (1997). The teacher-child relationship and children’s early school
adjustment. Journal of School Psychology, 35, 61-79.
Blair, K.A., Denham, S.A., Kochanoff, A. & Whipple, B. (2004). Playing it cool: Temperament,
emotional regulation, and social behavior in preschoolers. Journal of School Psychology,
42, 419-433.
Blechman, E.A., & McEnroe, M.J. (1985). Effective family problem solving. Child
Development, 56, 429–437.
Borman, G.D. & Dowling, M. (2010). Schools and inequality: A multilevel analysis of
Coleman’s Equality of Educational Opportunity Data. Teachers College Record, 112,
1201-1246.
Bowey, J. A. (1995). Socioeconomic status differences in preschool phonological sensitivity and
first-grade reading achievement. Journal of Educational Psychology, 87, 476-487.
Bradley, R. H. & Corwyn, R.F. (2002). Socioeconomic status and child development. Annual
Review of Psychology, 53, 371-399.
Bradley, R. H., Corwyn, R.F., Burchinal, M., McAdoo, H.P., & Coll, C.G. (2001). The home
environments of children in the United States part II: Relations with behavioral
development through age 13. Child Development, 72, 1868-1886.
Bransford, J. D., Brown, A. L., & Cocking, R. R. (Eds.). (1999). How people learn:Brain, mind,
experience, and school. Washington, DC: National Academy Press.
Brendgen, M., Wanner, B., & Vitaro, F. (2006). Verbal abuse by the teacher and child
adjustment from Kindergarten through grade six. Pediatrics, 117, 1585-1598.
Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and
design. Cambridge, MA: Harvard University Press.
Bronfenbrenner, U. (1999). Environments in developmental perspective: Theoretical and
operational models. In S. L. Friedman & T. D.Wachs (Eds.), Measuring environment
across the life span: Emerging methods and concepts (pp. 3–28).Washington, DC:
American Psychological Association.
Brooks-Gunn, J. & Duncan, G.J. (1997). The effects of poverty on children. Future of Children,
7, 55-71.
Brooks-Gunn, J., Klebanov, P.K., & Liaw, F. (1995). The learning, physical, and emotional
environment of the home in the context of poverty: The Infant Health and Development
Program. Children and Youth Services Review, 17, 251-276.

148

Brophy, J. & Evertson, C.M. (1981). Student Characteristics and Teaching. New York:
Longman, Inc.
Burgess, S. R., Hecht, S.A., & Lonigan, C.J. (2002). Relations of home literacy environment
(HLE) to the development of reading-related abilities: A one-year longitudinal study.
Reading Research Quarterly, 37, 408-426.
Burns, M.S., Griffin, P. & Snow, C.E. (1999). Starting Out Right: A Guide to Promoting
Children’s Reading Success.
Butler, S. R., Marsh, H.W., Sheppard, M.J., & Sheppard, J.L. (1985). Seven-year longitudinal
study of the early prediction of reading achievement. Journal of Educational Psychology,
77, 349-361.
Cameron, C. E., Connor, C. M., & Morrison, F. J. (2005). Effects of variation in teacher
organization on classroom functioning. Journal of School Psychology, 43, 61–85.
Carr, E. G., Taylor, J. C., & Robinson, S. (1991). The effects of severe behavior problems in
children on the teaching behavior of adults. Journal of Applied Behavior Analysis, 24,
523-535.
Chen, E., Matthews, K. A., & Boyce, T. (2002). Socioeconomic status differences in health:
What are the implications for children? Psychological Bulletin, 128, 295– 329.
Christian, K., Morrison, F.J., & Bryant, F.B. (1998). Predicting kindergarten academic skills:
Interactions among child care, maternal education, and family literacy environments.
Early Childhood Research Quarterly, 13, 501-521.
Clark, C., Prior, M., & Kinsella, G. (2002). The relationship between executive function abilities,
adaptive behaviour, and academic achievement in children with externalizing behaviour
problems. Journal of Child Psychology and Psychiatry, 43, 785-796.
Clark, T.P., Baker, J.A., Grant, S. (2007). Time on Task: Explaining the relationship
between externalizing disorders and academic performance. Unpublished manuscript.
Conger, R.D. (1995). Unemployment. In D. Levinson (Ed.), Encyclopedia of Marriage and the
Family (pp. 731–735). New York: MacMillan.
Conger, R. D. & Conger, K.J. (2002). Resilience in Midwestern families: selected findings from
the first decade of a prospective, longitudinal study. Journal of Marriage and Family, 64,
361–373.
Conger, R.D. & Donnellan, M.B. (2007). An interactionist perspective on the socioeconomic
context of human development. Annual Review of Psychology, 58, 175-199.

149

Conger R.D., Wallace, L.E., Sun, Y., Simons, R.L., McLoyd, V.C., Brody, G. (2002).
Economic pressure in African American families: a replication and extension of the
family stress model. Developmental Psychology, 38, 179–93.
Cooksey, E. C., Menaghan, E.G.,, & Jekielek, S.M. (1997). Life-course effects of work and
family circumstances on children. Social Forces, 76, 637-665.
Côté, S.M., Boivin, M., Nagin, D.S., Japel, C., Xu, Q., Zoccolillo, M., Junger, M., & Tremblay,
R.E. (2007). The role of maternal education and nonmaternal care services in the
prevention of children’s physical aggression problems. Archives of General Psychiatry,
64, 1305-1312.
Croninger, R. G., & Lee, V. E. (2001). Social capital and dropping out of school: Benefits to atrisk students of teachers' support and guidance. Teachers College Record, 103, 548-581.
Crooks, D. (1995). American children at risk: Poverty and its consequences for children’s health,
growth, and school achievement. Yearbook of Physical Anthropology, 38, 57-86.
Crosnoe, R., Johnson, M. K., & Elder, G. (2004). Intergenerational bonding in school: The
behavioral and contextual correlates of student-teacher relationships. Sociology of
Education, 77, 60–81.
Culp, A., Culp, R., Blankemeyer, M., & Passmark, L. (1998). Parent Education Home Visitation
Program: Adolescent and nonadolescent mother comparison after six months of
intervention. Infant Mental Health Journal, 19, 111–123.
Cunningham, A. E. & Stanovich, K.E. (1997). Early reading acquisition and its relation to
reading experience and ability 10 years later. Developmental Psychology, 33, 934-945.
Curby, T.W., Rimm-Kaufman, S.E., & Ponitz, C.C. (2009). Teacher-child interactions and
children’s achievement trajectories across kindergarten and first grade. Journal of
Educational Psychology, 912-925.
Dally, K. (2006). The influence of phonological processing and inattentive behavior on reading
acquisition. Journal of Educational Psychology, 98, 420-437.
Deater-Deckard, K., Mullineaux, P.Y., Petrill, S.A., & Thompson, L.A. (2009). Effortful
control, surgency, and reading skills in middle childhood. Reading and Writing, 22,
107-116.
DeMartini-Scully, D., Bray, M.A., & Kehle, T.J. (2000). A packaged intervention to reduce
disruptive behaviors in general education students. Psychology in the Schools, 37,
149-156.

150

Dichtelmiller, M., Meisels, S., Plunkett, J., Bozynski, M., Claflin, C., & Mangelsdorf, S. (1992).
The relationship of parental knowledge to the development of extremely low birth weight
infants. Journal of Early Intervention, 16, 210–220.
Dickinson, D.K., McCabe, A., Anastasopoulos, L., Peisner-Feinberg, E.S., & Poe, M.D. (2003).
The comprehensive language approach to early literacy: The interrelationships among
vocabulary, phonological sensitivity, and print knowledge among preschool-aged
children. Journal of Educational Psychology, 95, 465-481.
Dionne, G. (2005). Language development and aggressive behavior. In R. E. Tremblay, W. M.
Hartup, & J. Archer (Eds.), Developmental origins of aggression. New York: Guilford
Publications.
DiPerna, J. C., Volpe, R. J., & Elliott, S. N. (2002). A model of academic enablers and
elementary reading/language arts achievement School Psychology Review, 31,
298-312.
Duke, N.K. (2000). For rich it’s richer: Print experiences and environments offered to children in
very low- and very high-socioeconomic status first-grade classrooms. American
Educational Research Journal, 37, 441-478.
Duncan, G. J. & Brooks-Gunn, J. (1997). Income effects across the life span: Integration and
interpretation. In G. J. Duncan & J. Brooks-Gunn (Eds.), Consequences of growing up
poor (Vol. Russell Sage Foundation, pp. 596-610). New York.
Duncan, G. J. Magnuson, K.A. (2003). Off with Hollingshead: Socioeconomic resources,
parenting, and child development. In M. H. Bornstein & R. H. Bradley (Ed.),
Socioeconomic status, parenting, and child development (pp. 83-106). Mahwah, NJ:
Lawrence Erlbaum.
Duncan, S.E. & De Avila, E.A. (1998). preLAS 2000 Cue Picture Book English Form C,
CTB/McGraw-Hill Companies, Inc.
DuPaul, G. J., & Stoner, G. (1994). ADHD in the schools: Assessment and intervention
strategies. New York: Guilford Press.
DuPaul, G. J., Volpe, R. J., Jitendra, A. K., Lutz, J. G., Lorah, K. S., & Gruber, R. (2004).
Elementary school students with AD/HD: Predictors of academic achievement. Journal
of School Psychology, 42, 285–301.
Durham, R. E., Farkas, G., Hammer, C.S., Tomblin, J.B., & Catts, H.W. (2007). Kindergarten
oral language skill: A key variable in the intergenerational transmission of
socioeconomic status. Research in Social Stratification and Mobility, 25, 294-305.
Eccles, J., & Gootman, J. A. (2002). Community programs to promote youth development.
Washington, DC: National Academies Press.

151

Entwisle, D. R. Alexander, K.L. (1993). Entry into schools: The beginning school transition and
educational stratification in the United States. Annual Review of Sociology, 19, 401-423.
Entwisle, D.R., Alexander, K.L., Olson, L.S. (1994). The gender gap in math: Its possible origins
in neighborhood effects. American Sociological Review, 59, 822-838.
Entwisle, D.R. & Hayduk, L.A. (1988). Lasting effects of elementary school. Sociology of
Education, 61, 147-159.
Evans, G. W. (2004). The environment of childhood poverty. American Psychologist, 59, 77-92.
Federal Interagency Forum on Child and Family Statistics. (2005). America's children: Key
indicators of well-being, 2005. Washington, DC: U.S. Government Printing Office.
Finn, J. D. (1989). Withdrawing from school. Review of Educational Research, 59, 117– 142.
Finn, J.D. & Rock, D.A. (1997). Academic success among students at risk for school failure.
Journal of Applied Psychology, 82, 221-234.
Finnie, R. & Meng, R. (2005). The importance of functional literacy: Reading and math skills
and labour market outcomes of high school drop-outs. Ottawa, Canada: Business and
Labour Market Analysis Division.
Fleming, C. B., Harachi, T. W., Cortes, R. C., Abbott, R. D., & Catalano, R. F. (2004). Level and
change in reading scores and attention problems during elementary school as predictors
of problem behavior in middle school. Journal of Emotional and Behavioral Disorders,
12, 130-144.
Foorman, B.R., Anthony, J., Seals, L., & Mouzaki, A. (2002). Language development and
emergent literacy in preschool. Seminars in Pediatric Neurology, 9, 173-184.
Fredericks, J.A., Bumenfeld, P.C., & Paris, A.H. (2004). School engagement: Potential of the
concept, state of the evidence. Review of Educational Research, 74, 59-109.
Frick, P., Kamphaus, R.W., Lahey, B.B., Loeber, R., Christ, M.G., Hart, E., & Tannenbaum,
L.E. (1991). Academic underachievement and the disruptive behavior disorders. Journal
of Consulting and Clinical Psychology, 59, 289-294.
Gamse, B. C., Bloom, H.S., Kemple, J.J., Jacob, R.T. (2008). Reading First Impact Study:
Interim Report (NCEE 2008-4016). Retrieved October 18, 2008 from
http://ies.ed.gov/ncee/pubs/20084016/index.asp.
Gershoff, E. (2003). Low income and the development of America's kindergarteners (Research
Brief No.4). New York: National Center for Children in Poverty.

152

Gest, S.D., Welsh, J.A., & Domitrovich, C.E. (2005). Behavioral predictors of changes in social
relatedness and liking school in elementary years. Journal of School Psychology, 43, 281301.
Gettinger, M. (1985). Time allocated and time spent relative to time needed for learning as
determinants of achievement. Journal of Educational Psychology, 77, 3-11.
Gettinger, M. & Seibert, J.K. (2002). Best practices in increasing academic learning time. In A.
Thomas & J. Grimes (Eds.), Best practices in school psychology-III (pp. 943-954).
Washington, DC: National Association of School Psychologists.
Good, T. L. & Brophy, J. E. (1994). Looking in classrooms (6th ed.). New York: Harper Collins.
Goodenow, J. J. & Collins, W. A. (1990). Development according to parents : The nature,
sources, and consequences of parents’ ideas. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Greenbaum, P. E., Dedrick, R. F., Freidman, R. M., Kutash, K., Brown, E. C., Lardieri, S. P., et
al. (1996). National Adolescent and Child Treatment Study (NACTS): Outcomes for
children with serious emotional and behavioral disturbance. Journal of Emotional and
Behavioral Disorders, 4, 130-146.
Greenberg, M. T., Weissberg, R. P., O’Brien, M. U., Zins, J. E., Fredericks, L., Resnik, H., et al.
(2003). Enhancing school-based prevention and youth development through coordinated
social, emotional, and academic learning. American Psychologist, 58, 466–474.
Greenwood, C.R., Horton, B.T., & Utley, C.A. (2002). Academic engagement: Current
perspectives on research and practice. School Psychology Review, 31, 328-349.
Gregory, A., & Weinstein, R. S. (2004). Connection and regulation at home and in school:
Predicting growth in achievement for adolescents. Journal of Adolescent
Research, 19, 405–427.
Guice, S., Allington, R., Johnston, P., Baker, K., & Michaelson, N. (1994). Access? Books,
children and literature-based curriculum in schools (Rep. No. 1.13). State University of
New York, Albany: National Research Center on Literature, Teaching and Learning.
Gunter, P L., & Coutinho, M. J. (1997). Negative reinforcement in classrooms: What we're
beginning to learn. Teacher Education and Special Education, 20, 249-264.
Guo, G. & Harris, K.M. (2000). The mechanisms mediating the effects of poverty on children's
intellectual development. Demography, 37, 431-447.
Halonen, A., Aunola, K., Ahonen, T., & Nurmi, J. (2006). The role of learning to read in the
development of problem behavior: A cross-lagged longitudinal study. British Journal of
Educational Psychology, 76, 517-534.

153

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, 625-638.
Hamre, B. K., & Pianta, R. C. (2005). Can instructional and emotional support in the first grade
classroom make a difference for children at risk of school failure? Child Development,
76, 949–967.
Hamre, B. K., & Pianta, R.C. (2007). Learning opportunities in preschool and early elementary
classrooms. In R. Pianta, M. Cox, & K. Snow (Eds.), School reading and the transition to
kindergarten in the era of accountability (pp. 49–84). Baltimore: Paul H. Brookes.
Han, W. J. (2008). Shift work and child behavioral outcomes. Work, Employment & Society, 22,
67-87.
Hanson, T. L., McLanahan, S., & Thomson, E. (1997). Economic resources, parental practices
and children's well-being. In G. J. D. J. Brooks-Gunn (Ed.), Consequences of Growing
Up Poor. New York: Russell Sage Foundation.
Hart, B. & Risley, T. (1995). Meaningful differences in the everyday experiences of young
American children. Baltimore: Brookes Publishing.
Hauser-Cram, P., Durand, T.M., & Warfield, M.E. (2007). Early feelings about school and later
academic outcomes of children with special needs living in poverty. Early Childhood
Research Quarterly, 22, 161-172.
Haveman, R. & Wolfe, B. (1994). Succeeding generations: On the effects of investments in
children. New York: Russell Sage Foundation.
Haycock, K. (2000). No more settling for less. Thinking K–16, 4, 3–16.
Hinshaw, S.P. (1992). Externalizing Behavior Problems and Academic Underachievement in
Childhood and Adolescence: Causal Relationships and Underlying Mechanism.
Psychological Bulletin, 111, 127-155.
Honig, A., & Wittmer, D. (1991). Socialization and discipline for infants and young children.
Early Child Development and Care, 66, 65–73.
Hood, M., Conlan, E., & Andrews, G. (2008). Preschool home literacy practices and
children’s literacy development: A longitudinal analysis. Journal of Educational
Psychology, 100, 252-271.
Horn, W.F., & Packard, T. (1985). Early identification of learning problems: A metaanalysis. Journal of Educational Psychology, 77, 597–607.

154

Howse, R., Calkins, S., Anastopoulos, A., Keane, S., & Shelton, T. (2003). Regulatory
contributors to children's academic achievement. Early Education and Development, 14,
101−119.
Jimerson, S. Egeland, B., & Teo, A. (1999). A longitudinal study of achievement trajectories:
Factors associated with change. Journal of Educational Psychology, 91, 116-126.
Jöreskog, K.G. (1979). Statistical estimation of structural models in longitudinal development
investigations. In J.R. Nesselroade & P.B. Baltes (Eds.), Longitudinal research in the
study of behavior and development (pp. 303-351). New York: Academic Press.
Joshi, P. & Bogen, K. (2005). Nonstandard schedules and young children's behavioral outcomes
among working low-income families. Journal of Marriage and the Family, 69, 139-156.
Juel, C. (1988). Learning to read and write: A longitudinal study of 54 children from first
through fourth grades. Journal of Educational Psychology, 80, 443-447.
Justice, L.M. & Ezell, H.K. (2000). Enhancing children’s print and word awareness through
home-based parent intervention. American Journal of Speech-Language Pathology, 9,
257-269.
Kahlenberg, R. (2001). All together now: Creating middle-class schools through public school
choices. Washington, DC: Brookings Institution.
Karweit, N. (1989). Time and learning: A review. In R.E. Slavin (Ed.), School and classroom
organization (pp.69-95). Hillsdale, NJ: Lawrence Erlbaum.
Kinnunen, U., Gerris, J., & Vermulst, A. (1996). Work experiences and family functioning
among employed fathers with children of school age. Family Relations, 45, 449-455.
Klem, A.M. & Connell, J.P. (2004). Relationships matter: Linking teacher support to student
engagement and achievement. The Journal of School Health, 74, 262-273.
Kochenderfer, B.J. & Ladd, G.W. (1996). Peer victimization: Cause or consequence of school
maladjustment? Child Development, 67, 1305-1317.
Kohn, M.L. (1959). Social class and the exercise of parental authority. American Sociological
Review, 24, 352-366.
Kohn, M.L. (1963). Social class and parent-child relationships: An interpretation. American
Journal of Sociology, 68, 471-480.
Kohn, M.L. (1977). Class and conformity: A study of values (2nd ed.). Chicago: University of
Chicago Press.

155

Kohn, M.L. (1979). The effects of social class on parental values and practices. In D. Reiss &
H.A. Hoffman (Eds.), The American Family: Dying or Developing, (pp. 45-68). New
York: Plenum Press.
Kolen, M. J. & Brennan, R. L. (1995). Test equating: Methods and Practices. New York:
Springer-Verlag.
Korat, O., Klein, P., & Segal-Drori, O. (2007). Maternal mediation in book reading, home
literacy environment, and children's emergent literacy: a comparison between two social
groups. Reading and Writing, 20, 361-398.
Kozol, J. (1991). Savage inequalities: Children in America’s schools. New York: Harper Collins.
Kutner, M., Greenberg, E., Jin, Y., Boyle, B., Hsu, Y., & Dunleavy, E. (2007). Literacy in
Everyday Life: Results from the 2003 National Assessment of Adult Literacy (NCES
2007-480): Washington, DC: National Center for Education Statistics.
Ladd, G. W. & Burgess, K. B. (1999). Charting the relationship trajectories of aggressive,
withdrawn, and aggressive/withdrawn children during early grade school. Child
Development, 70, 910–929.
Ladd, G.W. & Burgess, K.B. (2001). Do relational risks and protective factors moderate the
linkages between childhood aggression and early psychological and school adjustment?
Child Development, 72, 1579-1601.
Laosa, L.M. (1982). School, occupation, culture, and family: The impact of schooling on the
parent–child relationship. Journal of Educational Psychology, 74, 791–827.
Larson, C.P. (2007). Poverty during pregnancy: Its effects on child health outcomes. Paediatrics
and Child Health, 12, 673-677.
Leach, D., & Dolan, N. (1985). Helping teachers increase student engagement rate. Behavior
Modification, 9, 55– 71.
Lee, J., Grigg, W. & Donahue, P. (2007). The Nation’s Report Card: Reading 2007 (NCES
2007–496). National Center for Education Statistics, Institute of Education Sciences, U.S.
Department of Education, Washington, D.C.
Lee, V. E. & Burkam, D.T. (2002). Inequality at the starting gate: Social background differences
in achievement as children enter school. Washington, DC: Economic Policy Institute.
Leppanen, U., Niemi, P., Aunola, K., & Nurmi, J. (2004). Development of reading skills among
preschool and primary school pupils. Reading Research Quarterly, 39, 72-93.

156

Lerkkanen, M. K., Rasku-Puttonen, H., Aunola, K., & Nurmi, J.E. (2004). Reading performance
and its developmental trajectories during the first and the second grade. Learning and
Instruction, 14, 111-130.
Linver, M. R., Brooks-Gunn, J., & Kohen, D.E. (2002). Family processes as pathways from
income to young children's development. Developmental Psychology, 38, 719-734.
Little, R.J.A. & Rubin, D. B. (1987). Statistical analysis with missing data. New York: Wiley.
Logan, J.A.R & Petscher, Y. (2010). School profiles of at-risk student concentration: Differential
growth in oral reading fluency. Journal of School Psychology, 48,163-186.
Lonigan, C. J., Burgess, S.R. & Anthony, J.L. (2000). Development of emergent literacy and
early reading skills in preschool children: Evidence from a latent-variable longitudinal
study. Developmental Psychology, 36, 596-613.
Lonigan, C. J., Burgess, S.R., Anthony, J.L. & Baker, T.A. (1998). Development of phonological
sensitivity in two- to five-year-old children. Journal of Educational Psychology, 90, 294311.
Luster, T., Rhoades, K., & Haas, B. (1989). The relation between parental values and parenting
behavior: A test of the Kohn hypothesis. Journal of Marriage and the Family, 51,
139-147.
Luthar, S. & Zigler, E. (1991). Vulnerability and competence: A review of research on resilience
in childhood. American Journal of Orthopsychiatry, 61, 6–22.
Lynch, M. & Cicchetti, D. (1997). Children’s relationships with adults and peers: An
examination of elementary and junior high school students. Journal of School
Psychology, 35, 81-99.
Ma, C.T., Gee, L., Kushel, M.B. (2008). Associations between housing instability and food
insecurity with healthcare access in low-income children. Ambulatory Pediatrics, 8,
50-57.
Mandel, H. P. (1997). Conduct disorder and underachievement: Risk factors, assessment,
treatment, and prevention. New York: John Wiley & Sons.
Mantzicopoulos, P. (2005). Conflictual relationships between kindergarten children and their
teachers: Associations with child and classroom context variables. Journal of School
Psychology, 43, 425-442.
McCardle, P., Scarborough, H.S., & Catts, H.W. (2001). Predicting, explaining, and preventing
children's reading difficulties. Learning Disabilities Research & Practice, 16, 230-239.

157

McCoach, D. B., O'Connell, A.A., Reis, S.M., & Levitt, H.A. (2006). Growing readers: A
hierarchical linear model of children's reading growth during the first 2 years of school.
Journal of Educational Psychology, 98, 14-28.
McGee, R., Prior, M., Williams, S., Smart, D., & Sanson, A. (2002). The long-term significance
of teacher-rated hyperactivity and reading ability in childhood: Findings from two
longitudinal studies. Journal of Child Psychology and Psychiatry, 43, 1004–1017.
McGee, R. Share, D.L. (1988). Attention deficit disorder-hyperactivity and academic failure:
Which comes first and what should be treated? Journal of the American Academy of
Child and Adolescent Psychiatry, 27, 318-325.
McGee, R., Share, D., Moffitt, T. E., Williams, S., & Silva, P. A. (1988). Reading disability,
behaviour problems, and juvenile delinquency. In D. H. Saklofske & S. B. G. Eysenck
(Eds.), Individual differences in children and adolescents: International perspectives (pp.
158-172). London: Hodder & Stoughton.
McIntosh, K., Horner, R.H., Chard, D.J., Boland, J.B., & Good, R.H. (2006). The use of reading
and behavior screening measures to predict non-response to school-wide positive
behavior support: A longitudinal analysis. School Psychology Review, 35, 275-291.
McLoyd, V. C. (1998). Socioeconomic disadvantage and child development. American
Psychologist, 53, 185-204.
Menaghan, E. G., & Parcel, T. L. (1991). Determining children's home environments: The
impact of maternal characteristics and current occupational and family conditions.
Journal of Marriage and the Family, 53, 417-431.
Menaghan, E. G. & Parcel, T.L. (1995). Social sources of change in children's home
environments: The effects of parental occupational experiences and family conditions.
Journal of Marriage and the Family, 57, 69-84.
Merton, R. (1968). The Matthew effect in science. Science, 56-63.
Miller, S. (1988). Parents’ beliefs about children’s cognitive development. Child Development,
59, 259–285.
Molfese, V. J., Modglin, A. & Molfese, D.L. (2003). The role of environment in the
development of reading skills: A longitudinal study of preschool and school-age
measures. Journal of Learning Disabilities, 36, 59-67.
Morris, D., Bloodgood, J., & Perney, J. (2003). Kindergarten predictors of first and second grade
reading achievement. The Elementary School Journal, 104, 93-109.

158

Morrow, L. M., O'Connor, E. M., & Smith, J. K. (1990). Effects of a story reading
program on the literacy development of at-risk kindergarten children. Journal of
Reading Behavior, 22, 255-275.
Murray, C. & Murray, K.M. (2004). Child level correlates of teacher-student relationships: An
examination of demographic characteristics, academic orientations and behavioral
orientations. Psychology in the Schools, 41, 751-762.
Muthén, B.O. & Khoo, S.T. (1998). Longitudinal studies of achievement growth using latent
variable modeling. Learning and Individual Differences, 10, 73-102.
Neitzel, C. & Stright, A.D. (2004). Parenting behaviours during child problem solving: The roles
of child temperament, mother education and personality, and the problem-solving
context. International Journal of Behavioral Development, 28, 166-179.
Nelson, J.R., Benner, G.J., & Gonzalez, J. (2003). Learner characteristics that influence the
treatment effectiveness of early literacy interventions: A meta-analytic review. Learning
Disabilities Research & Practice, 18, 255-267.
Nelson, J. R., Benner, G. J., Lane, K., & Smith, B. W. (2004). Academic achievement of K-12
students with emotional and behavioral disorders. Exceptional Children, 71, 59-73.
Newman, R. S. (1990). Children's help seeking in the classroom: The role of motivational factors
and attitudes. Journal of Educational Psychology, 82, 71-80.
Newman, R.S. & Schwager, M.T. (1993). Students' perceptions of the teacher and classmates in
relation to reported help seeking in math class. The Elementary School Journal, 94, 3-17.
Nicholson, J.M., Berthelson, D., Williams, K.E., & Abad, V. (2010). National study of an early
parenting intervention: Implementation differences on parent and child. Prevention
Science, 11, 360-370.
Onatsu-Arvilommi, T., & Nurmi, J. E. (2000). The role of task-avoidant and task-focused
behaviors in the development of reading and mathematical skills during the first school
year: A cross-lagged longitudinal study. Journal of Educational Psychology, 92, 478-491.
Owens, P.L., Zodet, M.W., Berdahl, T., Dougherty, D., McCormick, M.C., & Simpson, L.A.
(2008). Annual report on health care for children and youth in the United States: Focus
on injury-related emergency department utilization and expenditures. Ambulatory
Pediatrics, 8, 219-240.
Pallas, A. M., Entwisle, D. R., Alexander, K. L., & Cadigan, D. (1987). Children who do
exceptionally well in first grade. Sociology of Education, 60, 257-271.

159

Parcel, T. L., & Menaghan, E. G. (1990). Maternal working conditions and children's verbal
facility: Studying the intergenerational transmission of inequality from mothers to young
children. Social Psychology Quarterly, 53, 132-147.
Parker, F. L., Piotrkowski, C. S., Horn, W., & Green, S. (1995). The challenge for Head Start:
Realizing its vision as a two generation program. In I. Sigel (Series Ed.) & S. Smith (Vol.
Ed.), Advances in applied developmental psychology: Two generation programs for
families in poverty: Vol. 9. (pp. 135–159). Norwood, NJ: Ablex.
Parrila, R., Aunola, K., Leskinen, E., Nurmi, J., & Kirby, J.R. (2005). Development of individual
differences in reading: Results from longitudinal studies in English and Finnish. Journal
of Educational Psychology, 97, 299-319.
Phillips, L. M., Norris, S.P., Osmond, W.C., Maynard, A.M. (2002). Relative reading
achievement: A longitudinal study of 187 children from first through sixth grades.
Journal of Educational Psychology, 94, 3-13.
Pianta, R. C. (1992). Beyond the parent: The role of other adults in children's lives: New
Directions in Child Development, Vol. 57. San Francisco: Jossey-Bass.
Pianta, R., Belsky, J., Vandergrift, N., Houts, R. M., & Morrison, F. J. (2008). Classroom effects
on children’s achievement trajectories in elementary school. American Educational
Research Journal, 45, 365–397.
Pianta, R. C., La Paro, K. M., Payne, C., Cox, M. J., & Bradley, R. (2002). The relation of
kindergarten classroom environment to teacher, family, and school characteristics and
child outcomes. Elementary School Journal, 102, 225–238.
Pianta, R. C. & Nimetz, S. L. (1991). Relationships between children and teachers:
Associations with classroom and home behavior. Journal of Applied Developmental
Psychology, 12, 379-393.
Pianta, R. C., & Steinberg, M. (1992). Teacher-child relationships and the process of adjusting to
school. New Directions for Child Development, 57, 61-80.
Pianta, R.C., Steinberg, M., & Rollins, K.B. (1995). The first two years of school: Teacher child
relationships and deflections in children's classroom adjustment. Development and
Psychopathology, 7, 295-312.
Pierce, M.E., Katzir, T., Wolf, M., & Noam, G.G. (2007). Clusters of second and third grade
dysfluent urban readers. Reading and Writing, 20, 885-907.
Piotrkowski, C. S., & Katz, M. H. (1982). Indirect socialization of children: The effects of
mothers' jobs on academic behaviors. Child Development, 53, 1520-1529.

160

Porr, C., Drummond, J., & Richter, S. (2006). Health literacy as an empowerment tool for
low-income mothers, Family & Community Health, 29, 328-355.
Poskiparta, E., Niemi, P., Lepola, J., Ahtola, A., & Laine, P. (2003). Motivational-emotional
vulnerability and difficulties in learning to read and spell. British Journal of Educational
Psychology, 73, 187-206.
Rabiner, D., Coie, J.D., & the Conduct Problems Prevention Research Group. (2000). Early
attention problems and children’s reading achievement: A longitudinal investigation.
Journal of the American Academy of Child and Adolescent Psychiatry, 39, 859–867.
Rabiner, D. L., Malone, P.S., & The Conduct Problems Prevention Research Group. (2004). The
impact of tutoring on early reading achievement for children with and without attention
problems. Journal of Abnormal Child Psychology, 32, 273-284.
Reich, S. (2005). What do mothers know? Maternal knowledge of child development. Infant
Mental Health Journal, 26, 143-156.
Reid, R., Gonzalez, J.E., Nordness. P.D., Trout, A., & Epstein, M.H. (2004). A meta-analysis of
the academic status of students with emotional/behavioral disturbance. Journal of Special
Education, 38, 130-143.
Rich, H., & Ross, S. (1989). Students’ time on learning tasks in special education. Exceptional
Children, 55, 508–515.
Riddle-Buly, M. & Valencia, S.W. (2002). Below the bar: Profiles of students who fail state
reading assessments. Educational Evaluation and Policy Analysis, 24, 219-239.
Rimm-Kaufman, S., La Paro, K., Pianta, R., & Downer, J. (2005). The contribution of classroom
setting and quality of instruction to children’s behavior in the kindergarten classroom.
Elementary School Journal, 105, 377–394.
Roeser, R. W., Eccles, J. S., & Sameroff, A. J. (2000). School as a context of early adolescents’
academic and social-emotional development: A summary of research findings.
Elementary School Journal, 100, 443–471.
Rogers, S. J., Parcel, T.L., & Menaghan, E.G. (1991). The effects of maternal working
conditions and mastery on child behavior problems: Studying the intergenerational
transmission of social control. Journal of Health and Social Behavior, 32, 145-164.
Rothbart, M. K., & Bates, J. E. (1998). Temperament. In W. Damon, & N. Eisenberg (Eds.),
Handbook of child psychology: Vol. 3. Social, emotional, and personality development
(5th ed.) (pp. 105– 176). New York: Wiley.

161

Rowe, K. J. & Rowe, K.S. (1992). The relationship between inattentiveness in the classroom and
reading achievement (Part A): Methodological issues. Journal of the American Academy
of Child and Adolescent Psychiatry, 31, 349-356.
Rumberger, R. W., & Palardy, G. J. (2005). Does segregation still matter? The impact of student
composition on academic achievement in high school. Teachers College Record, 107,
1999–2045.
Russell J., McCoy, A., Pistorino, C., Wilkinson, A., Burghardt, J., Clark, M., Ross C., Schochet,
P. & Swank, P. (2007). National Evaluation of Early Reading First: Final Report, U.S.
Department of Education, Institute of Education Sciences, Washington, DC: U.S.
Government Printing Office.
Sameroff, A., Seifer, R, Baldwin, A., & Baldwin, C. (1993). Stability of intelligence from
preschool to adolescence: The influence of social and family risk factors. Child
Development, 64, 80-97.
Sampson, R. J. (2003). The neighborhood context of well-being. Perspectives in Biology and
Medicine, 46, 53-64.
Saudino, K.J. & Plomin, R. (2007). Why are hyperactivity and academic achievement related?
Child Development, 78, 972-986.
Scarborough, H. S. (1998). Predicting the future achievement of second graders with reading
disabilities: Contributions of phonemic awareness, verbal memory, rapid naming and IQ.
Annals of Dyslexia, 48, 115-136.
Scholer, S.J., Hickson, G.B., & Ray, W.A. (1999). Sociodemographic factors identify U.S.
infants at high risk for injury mortality. Pediatrics, 103, 1183-1188.
Share, D. L., Jorm, A.F., Maclean, R., & Matthews, R. (1984). Sources of individual differences
in reading acquisition. Journal of Educational Psychology, 76, 1309-1324.
Shaywitz, B. A., Holford, T.R., Holohan, J.M., Fletcher, J.M., Stuebing, K.K., Francis, D.J., &
Shaywitz, S.E. (1995). A Matthew Effect for IQ but not for reading: Results from a
longitudinal study. Reading Research Quarterly, 30, 894-906.
Shonkoff, J. P. Phillips, D.A. (Eds.). (2000). From neurons to neighborhoods: The science of
early childhood development. Washington, DC: National Academy Press.
Shores, R.E., Gunter, P.L., & Jack, S.L. (1993). Classroom management strategies: Are they
setting events for coercion? Behavioral Disorders, 18, 92-102.
Sirin, S. R. (2005). Socioeconomic status and academic achievement: A meta-analytic review of
research. Review of Educational Research, 75, 417–453.

162

Smart, D., Prior, M., Sanson, A., & Oberklaid, F. (2001). Children with reading difficulties: A
six-year follow-up from early primary school to secondary school. Australian Journal of
Psychology, 53, 45-53.
Snow, C. E. (1991). The theoretical basis for relationships between language and literacy in
development. Journal of Research in Childhood Education, 6, 5-10.
Snow, C. E., Burns, S. M., & Griffin, P. (Eds.). (1998). Preventing reading difficulties in young
children. Washington, DC: National Academy Press.
Spira, E.G. & Fischel, J.E. (2005). The impact of preschool inattention, hyperactivity, and
impulsivity on social and academic development: A review. Journal of Child Psychology
and Psychiatry, 46, 755-773.
Spira, E. G., Bracken, S.S., & Fischel, J.E. (2005). Predicting improvement after first-grade
reading difficulties: The effects of oral language, emergent literacy and behavior skills.
Developmental Psychology, 41, 225-234.
Stanovich, K. E. (1986). Matthew Effects in reading: Some consequences of individual
differences in the acquisition of literacy. Reading Research Quarterly, 21, 360-407.
Stipek, D. & Miles, S. (2008). Effects of aggression on achievement: Does conflict with teacher
make it worse? Child Development, 79, 1721-1735.
Storch, S. A. Whitehurst, G.J. (2002). Oral language and code related precursors to reading:
Evidence from a longitudinal structural model. Developmental Psychology, 38, 934-947.
Strazdins, L., Clements, M.S., Korda, R.J., Broom, D.H., & D'Souza, R.M. (2006). Unsociable
work? Nonstandard work schedules, family relationships, and children's well-being.
Journal of Marriage and the Family, 68, 394-410.
Tabors, P. & Snow, C. (2001). Young bilingual children and early literacy development. In S.
Neuman & D. Dickinson (Eds.), Handbook of early literacy research (pp. 159-178).
New York: Guilford.
Tarabulsy, G. M., Bernier, A., Provost, M. A., Maranda, J., Larose, S., Moss, E., Larose, M., &
Tessier, R. (2005). Another look inside the gap: Ecological contributions to the
transmission of attachment in a sample of adolescent mother–infant dyads.
Developmental Psychology, 41, 212–224.
Torgesen, J. K. (1998). Catch them before they fall: Identification and assessment to prevent
reading failure in young children American Educator, 22, 1-8.
Torgesen, J. K. (2002). The prevention of reading difficulties. Journal of School Psychology, 40,
7-26.

163

Torgesen, J.K., Alexander, A.W., Wagner, R.K., Rashotte, C.A., Voeller, K.K.S., & Conway, T.
(2001). Intensive remedial instruction for children with severe reading disabilities:
Immediate and long-term outcomes from two instructional approaches. Journal of
Learning Disabilities, 34, 33–58.
Torgesen, J.K., Wagner, R.K., Rashotte, C.A., Rose, E., Lindamood, P., Conway, T., & Garvan,
C. (1999). Preventing reading failure in young children with phonological processing
disabilities: Group and individual responses to instruction. Journal of Educational
Psychology, 91, 579–593.
Trout, A. L., Nordness, P. D., Pierce, C. D., & Epstein, M. H. (2003). Research on the academic
status of children with emotional and behavioral disorders: A review of the literature
from 1961 to 2000. Journal of Emotional and Behavioral Disorders, 11, 198-210.
Trzesniewski, K. H., Moffitt, T.E., Caspi, A., Taylor, A., & Maughan, B. (2006). Revisiting the
association between reading achievement and antisocial behavior: New evidence of an
environmental explanation from a twin study. Child Development, 77, 72-88.
United States Department of Education. National Center for Education Statistics. (2000).
America's Kindergartners, NCES 2000-070, by Kristin Denton, Elvira GerminoHausken. Project Officer, Jerry West, Washington, D.C.
United States Department of Health and Human Services. (2005). Child Health USA 2005.
Washington, DC: Government Printing Office.
Van Acker, R., Grant, S.H., & Henry, D. (1996). Teacher and student behavior as a function of
risk for aggression. Education and Treatment of Children, 19, 316- 334.
Valencia, S.W., & Riddle-Buly, M. (2004). Behind test scores: What struggling readers really
need. Australian Journal of Language and Literacy, 27, 217-233.
Van Alstyne, D. (1929). The environment of three-year-old children: Factors related to
intelligence and vocabulary tests. New York: Columbia University Bureau of
Publications.
Van der Heyden, A.M., Snyder, P.A., Broussard, C., & Ramsdell, K. (2008). Measuring response
to early literacy intervention with preschoolers at risk. Topics in Early Childhood Special
Education, 27, 232-249.
Van Doesum, K.T.M., Hosman, C.M.H., Risken-Walraven, J.M., & Hoefnagels, C. (2007).
Correlates of depressed mothers’ sensitivity toward their infants: The role of maternal,
child, and contextual characteristics. Journal of the American Academy of Child and
Adolescent Psychiatry, 46, 747-756.

164

Velting, O. N. & Whitehurst, G.J. (1997). Inattention-hyperactivity and reading achievement in
children from low-income families: A longitudinal model. Journal of Abnormal Child
Psychology, 25, 321-331.
Vrijheid, M., Dolk, H., Stone, D., Alberman, E., & Scott, J.E. (2000). Socioeconomic
inequalities in risk of congenital anomaly. Archives of Disease in Childhood, 82,
349-352.
Walberg, H. J. Tsai, S. (1983). Matthew Effects in education. American Educational Research
Journal, 20, 359-373.
Walker, D., Greenwood, C., Hart, B., & Carta, J. (1994). Prediction of school outcomes based on
early language production and socioeconomic factors. Child Development, 65, 606-621.
Walker, H. M., Fomess, S. R., Kauffman, J. M., Epstein, M. H., Gresham, F M., Nelson, C. M.,
& Strain, P S. (1998). Macro-social validation: Referencing outcomes in behavioral
disorders to societal issues and problems. Behavioral Disorders, 24, 7-18.
Walker, H. M., & Severson, H. (1990). Systematic screening for behavior disorders (SSBD).
Longmont, CO: Sopris West.
Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of the
American Statistical Association, 58, 236-244.
Warner-Rogers, J., Taylor, A., Taylor, E., & Sandberg, S. (2000). Inattentive behavior in
childhood: Epidemiology and implications for development. Journal of Learning
Disabilities, 33, 520-536.
Webster-Stratton, C. & Herman, K.C. (2008). The impact of parent behavior-management
training on child depressive symptoms. Journal of Counseling Psychology, 55, 473-484.
Wehby, J. H., Symons, F. M., Canale, J., & Go, F. (1998). Teaching practices in classrooms for
students with emotional and behavioral disorders: Discrepancies between
recommendations and observations. Behavioral Disorders, 24, 52-57.
Wentzel, K.R. (1993). Does being good make the grade? Social behavior and academic
competence in middle school. Journal of Educational Psychology, 85, 357-364.
Wentzel, K. (2002). Are effective teachers like good parents? Teaching styles and student
adjustment in early adolescence. Child Development, 73, 287–301.
Whitehurst, G. J. (1997). Language processes in context: Language learning in children reared in
poverty. In L. B. Adamson & M. S. Romski (Eds.), Research on communication and
language disorders: Contribution to theories of language development (pp. 233-266).
Baltimore, MD: Brookes.

165

Whitehurst, G. J. & Lonigan, C. J. (1998). Child development and emergent literacy. Child
Development, 69, 848-872.
Whiteley, H. E., Smith, C.D., & Connors, L. (2007). Young children at risk of literacy
difficulties: Factors predicting recovery from risk following phonologically based
intervention. Journal of Research in Reading, 30, 249-269.
Willcutt, E.G. & Pennington, B.F. (2000). Psychiatric comorbidity in children and adolescents
with reading disability. Journal of Child Psychology and Psychiatry, 41, 1039-1048.
Willms, J. D. (1999). Quality and inequality in children’s literacy: The effects on families,
schools, and communities. In D. P. Keating & C.Hertzman (Eds.), Developmental health
and the wealth of nations: Social, biological, and educational dynamics (pp. 72–93).
New York: Guilford Press.
Wilson, J. B., Ellwood, D. T., & Brooks-Gunn, J. (1995). Welfare-to-work through the eyes of
children. In P. L. Chase-Lansdale & J. Brooks-Gunn (Eds.), Escape from poverty: What
makes the difference for children? (pp. 63 –86). New York: Cambridge University Press.
Woodcock, McGrew, & Werder (1994). Woodcock-McGrew-Werder Mini Battery of
Achievement. Itasca, IL: Riverside Publishing.
Yeh, S.S. (2003). An evaluation of two approaches for teaching phonemic awareness to children
in Head Start. Early Childhood Research Quarterly, 18, 513-529.
Zill, N., Collins, M., West, J., & Hausken, E.G. (1995). Approaching kindergarten: A look at
preschoolers in the United States. Washington, DC: Office of Educational Research and
Improvement, N.C.E.S., and N.H.E.S.
Zins, J. E., Bloodworth, M. R., Weissberg, R. P., & Walberg, H. (2004). The scientific base
linking social and emotional learning to school success. In J. E. Zins, R. P. Weissberg, M.
C. Wang, & H. J. Walberg (Eds.), Building academic success on social and emotional
learning: What does the research say? New York: Teachers College Press.

166