( . _
.
.u
.
I. .
!
x ..

 

 

 
 

.x. I E..E.F...
.14. r 3.. 1., .. -3

 

an”! n .
“mefi,
c :r. 2. ‘

.49.}..95

         

gig:

 

       

\

QM:

fink F
:93?» .

  

e .2? it;
.. .M. 5.: Sun.“
4‘ ‘ ”‘57 5-,

  

2 LIBRARY
200 <3 Michigan State
University

This is to certify that the
dissertation entitled

PREDICTORS OF READING ACHIEVEMENT IN A
POPULATION OF SCHOOL-AGED CHILDREN WITH
PARENT AND SCHOOL REPORTED ATTENTION-
DEFICIT/HYPERACTIVITY DISORDER

presented by

JULIA A. OGG

has been accepted towards fulfillment
of the requirements for the

Counseling, Educational
Psychology. and Special
Ph.D. degree in Education

 

 

 

 

 

Wrofessor’s Signature

Augst 5, 2008

 

MSU is an aflirmative-action, equal-opportunity employer

_.—.—._.—.—.-.-._.—.-.-.-‘_,_.-.-.-.-¢—-----—.—.-a—.-.-.—.-.—-—-.-—.—-.—

._ _‘-<—.—-.-.-.-._.-.—.-._._l

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DateDue indd

PREDICTORS OF READING ACHIEVEMENT IN A POPULATION OF SCHOOL-
AGED CHILDREN WITH PARENT AND SCHOOL REPORTED ATTENTION -
DEFICIT/HYPERACTIVITY DISORDER
By

Julia A. Ogg

A DISSERTATION

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

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

2008

ABSRACT

PREDICTORS OF READING ACHIEVEMENT IN A POPULATION OF SCHOOL-
AGED CHILDREN WITH PARENT AND SCHOOL REPORTED ADHD

By

Julia A. Ogg
Attention-Deficit/Hyperactivity Disorder (ADHD) is the most prevalent childhood
disorder (National Institutes of Health, 2000), and is also one of the most chronic health
problems for school-aged children (Pastor & Reuben, 2002). Underachievement is
experienced by as many as 80% of children with ADHD (DuPaul & Stoner, 2003). The
current study examined reading achievement over time, among children eligible for
special education, both with and without a parent and school report of ADHD, using
latent growth modeling. The impact of demographic, student, and intervention variables
on reading growth was examined. Findings suggest that schools play an important role
above and beyond demographic characteristics of students when'it comes to improving
reading growth across a number of outcome measures. While externalizing behavior and
stimulant medication status did not appear to have a significant impact on reading growth
in the present study, positive student behaviors (e. g. completing homework on time) were
related to growth on both a skill-based measure and on reading grades. Both academic
and behavioral intervention variables were important factors in improving reading growth
over time. The manner in which outcomes are measured is important and has
implications for how growth is interpreted, as student reading grades were influenced
differently than the skill-based measures. This study highlights intervention targets

amenable to change in the school setting.

ACKNOWLEDGEMENTS

Although I typically refer to this paper as “my dissertation”, in truth it is the
combined effort of a number of wonderful individuals. I am grateful to those who have
supported this effort in one way or another.

I would like to start by thanking my excellent advisor, Dr. John Carlson. John’s
passion for children and for training school psychologists is always reflected in his
actions, and it is my goal to have as much passion for my own work and to share it so
willingly. I have appreciated his guidance, as well as the independence he allowed me
throughout this project. John’s encouragement and belief in me has helped me reach
goals that I never thought possible.

I also would like to thank Dr. Sara Bolt for all of her support, especially in getting
this project off the ground. Her support in helping develop the initial idea and accessing
the SEELS data is much appreciated. I greatly respect Sara, and her advice and
willingness to answer all of my questions has been very influential in my development.

I will never forget all of the words of wisdom I received from Dr. Jean Baker,
who was an original member of my committee. She had the gift of empowerment and I
appreciated her support in this project and across my training. She is deeply missed.

I have appreciated the insight that Dr. Magen has offered to this project and to my
training. I highly value his perspective and am grateful to his commitment to

multidisciplinary work.

iii

I also want to thank Dr. John Kosciulek for his willingness to serve on my
committee. I asked him to join my committee because I thoroughly enjoyed taking his
course, and have appreciated his kind encouragement in his work with me.

I would also like to thank my friends whose Support through this process I have
deeply appreciated. A little humor goes a long way when glued to the computer. Several
in particular, deserve special thanks: Jessie Kruer, Tara Brinkman, and Jake Mathiason.

I believe that I am extremely blessed through the support I receive from my
family. My parents have sacrificed to ensure that I had an outstanding education, and for
this I cannot thank them enough. Their constant encouragement and love has kept me
going through several rough patches. My husband, Jake, also deserves many thanks. His
support and sacrifices have also been extensive. His willingness and contentment in
letting me work in my “lair” on many weekends and evenings is greatly appreciated. I
would also like to thank Fran and Elliott for their support throughout my educational
career. Their generosity in purchasing all of my books speaks to their investment in my
education, and to the amazing people that they are. I would like to thank all of my family
for their support, encouragement, and love throughout this process.

Finally, I have appreciated the financial support for this project I have been

provided through the AERA grants program and the MSU School Psychology program.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
Chapter 1 ............................................................................................................................. 1
Outcomes for Children Diagnosed with ADHD ............................................................. 2
Predictors of Academic Achievement ............................................................................. 4
Current Study .................................................................................................................. 5
Chapter 2 ............................................................................................................................. 8
E vidence-Based Treatment for ADHD ............................................................................ 8
Predictors of Academic Outcomes ................................................................................ l3
ADHD Models ............................................................................................................... 16
School Based Service Provision for ADHD .................................................................. 21
School Based Interventions ........................................................................................... 32
Current Study ................................................................................................................ 35
Chapter 3 ........................................................................................................................... 38
Data Set ......................................................................................................................... 38
Criteria for Selecting Sample ........................................................................................ 40
Instruments .................................................................................................................... 42
Variables to be Included in the Analysis ....................................................................... 43
Group Variables .................................... 44
Independent Variables .................................................................................................. 44
Demographic Variables ................................................................................................ 48
Dependent Variable ...................................................................................................... 49
Data Preparation and Screening .................................................................................. 52
Analytic Techniques ...................................................................................................... 55
Chapter 4 ........................................................................................................................... 61
Descriptive Statistics ..................................................................................................... 61
Research Question 1 ..................................................................................................... 62
Research Question 2 ..................................................................................................... 64
Chapter 5 ........................................................................................................................... 73
Mean Reading Growth .................................................................................................. 73
Modeling ofReading Growth ............. 76
Outcome Measures ........................................................................................................ 85
Group Differences ......................................................................................................... 86
Limitations .................................................................................................................... 87
Appendices ........................................................................................................................ 92

...................................... 109
Bibliography .............................................................................

vi

LIST OF TABLES

Table 1. Measurement Model Confirmatory Factor Analysis ................................. 93

Table 2. Descriptive Characteristics of Sample at wave 1 without Imputed

Data ................................................................................................... 94
Table 3. Descriptive Characteristics of Sample at Wave 1 with Imputed

Data ................................................................................................... 95
Table 4. Correlation Matrix ........................................................................ 96
Table 5. Reading Achievement Across Groups .................................................. 97
Table 6. Impact of Covariates on Oral Reading Fluency ...................................... 98
Table 7. Impact of Covariates on Letter Word Identification ................................. 99
Table 8. Impact of Covariates on Passage Comprehension .................................. 100
Table 9. Impact of Covariates on Reading Grades ............................................ 101

vii

LIST OF FIGURES

Figure]. Conceptual Model of ADHD (Rapport, Chung, Shore, & Isaacs,

2001) ............................................... _ ......................................... 102
Figure 2. Disinhibition Model (Barkely, 1997) ................................................ .103
Figure 3. Model of Academic Enablers (DiPerna, Volpe & Elliott, 2002) ................ 104
Figure 4. Model of Dual Developmental Pathways (Rapport, Scanlon & Denney, 1999)
........................................................................................................... 105
Figure 5. Model of ADHD & Academic Enablers (Volpe et a1., 2006) .................... 106
Figure 6. Hypothesized Latent Growth Model of Reading Achievement .................. 107
Figure 7. Means of ORF Over Time and Across Groups .................................... 108

viii

Chapter 1
Introduction

The Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR;
American Psychiatric Association, 2000) reports that Attention-Deficit! Hyperactivity
Disorder (ADHD) effects 3-7% of the school-age population. In school-aged children,
ADHD is the most common reason for a mental health referral (Barkley, 1998).
Additionally, ADHD and related externalizing behavior accounts for the most referrals to
school professionals (Phelps, Brown & Power, 2001). One possible reason for the high
number of school-based referrals for ADHD is that academic underachievement is often
considered one of the primary long-term outcomes of ADHD (DuPaul & Stoner, 2003).
Despite the significant presence and detrimental impact that ADHD may have in schools,
DuPaul and Eckert (1997) report that in comparison to other treatments for ADHD,
school-based treatment is relatively understudied.

Academic and behavioral difficulties associated with ADHD often result in the
need for specialized services and interventions within the school setting. One way that
these services may be provided is through special education. Special education represents
an important treatment option for children diagnosed with ADHD, by providing services
within the school setting. In fact, the recognition within special education law of ADHD
as a disability has been reported to account for one of the recent changes in the treatment
of children with ADHD (Smith, Barkley & Shapiro, 2006). Children with ADHD are
frequently eligible for special education under a number of different service categories.
Some recent estimates suggest that as many as half of children in special education are

diagnosed with ADHD (Bussing, Zima, Perwien, Belin & Widawski, 1998; Schnoes,

Reid & Marder, 2006). However, while special education services may be more
accessible to children with ADHD within the school setting and a large number of
students receive services, relatively little research has addressed the impact of school-
based service delivery on academic outcomes for children diagnosed with ADHD.
Additionally, the authors of a report by the President’s Commission on Excellence in
Special Education (2002) indicate that too often special education has an emphasis on the
process versus outcomes. The authors of this report also suggest that the current system
does not place enough emphasis on the consistent use of evidence-based practices.
Outcomes for Children Diagnosed with ADHD

Understanding the relationship between school-based services and academic
achievement is important, as outcomes associated with a diagnosis of ADHD can be very
negative across several areas of functioning. The manifestation of the three core
symptoms of ADHD: inattention and/or hyperactivity and impulsivity may make school
challenging for children with ADHD. In fact, Forness and Kavale (2001) suggest that the
core symptoms of ADHD, “seem very often to be at the very center of what school
requires of a child” (p. 224). DuPaul and Stoner (2003) cite academic underachievement
as one of the most common correlates of ADHD with up to 80% of children with ADHD
exhibiting academic performance problems. Academic difficulties in this population is
demonstrated in a number of ways, including studies demonstrating that children with
ADHD experience significantly more academic difficulty than children without ADHD
(i.e., Deshazo Barry, Lymon & Klinger, 2002). Additionally, eighty-four percent of
parents of children with ADHD reported that this disorder had impacted their child’s

academic or behavioral outcomes in school (LeFever, Villers, Morrow & Vaughn, 2002).

A number of academic difficulties are associated with ADHD, including
underachievement as indicated by low and erratic grades and lower than expected
achievement and intelligence test scores [American Academy of Child and Adolescent
Psychiatry (AACAP), 1997; DuPaul & Stoner, 2003]. It has been hypothesized that these
difficulties may be caused by a number of factors including, gaps in what a student has
learned, difficulty sustaining attention, lack of attending to lectures or group-work, and
difficulty with organization and study skills (AACAP). The AACAP also suggested that
it can be difficult for students to take tests with core symptoms of inattention and
impulsivity, and DuPaul and Stoner also cite poor test performance as a potential
manifestation of ADHD. Finally, it has been suggested that a failure to turn in homework
can lead to these negative outcomes (AACAP). This is particularly relevant when it
comes to reading, as 25—40% of individuals with either ADHD or RD also meet criteria
for the other disorder (Wilcutt et al., 2007).

ADHD and related academic underachievement are perVasive and present a set of
chronic challenges. Children with ADHD do not “outgrow” the symptoms associated
with disorder, as 70-80% of children with ADHD continue to exhibit significant deficits
in inattention and impulsivity compared to same—age peers (DuPaul & Stoner, 2003).
Half of these children also continue to exhibit symptoms into adulthood. There are a
myriad of problems in adolescence that can accompany an ADHD diagnosis. The drop-
out rate for children diagnosed with ADHD may be as high as 32% (Smith et al., 2006).
The AACAP also suggests that difficulties associated with ADHD can result in higher
than expected placement in special courses, suspension, and expulsion. The extant

literature also reports that aggression and noncompliance can occur in adolescence

(DuPaul & Stoner). Students with ADHD and aggressive and oppositional behaviors are
at greater risk for later delinquency and a diagnosis of Conduct Disorder (CD; Smith et
al.). Additionally, for those who do develop comorbid CD, higher rates of substance use
and abuse has been reported (Smith, et al.). Other outcomes in adolescents with ADHD
that have been reported in the literature include higher rates of teen pregnancy in some
studies, and an increased rate of automobile accidents and speeding citations (Smith et
al.).

There is less data available for outcomes in adult populations with this disorder.
However, there is evidence for interpersonal difficulties, criminal convictions, and higher
than expected substance use disorders (Smith et al., 2006). In addition, within adults the
rate of completing a university degree program is much lower in the population of
individuals with ADHD compared to those without (review DuPaul & Stoner, 2003). It
seems clear that both the short and long-term impact of this ADHD is associated with a
greater level of risk for negative outcomes than for individuals without this disorder.
Predictors of Academic Achievement

Given the negative trajectory often associated with ADHD, it is essential to
identify ways to alter this developmental trajectory and improve outcomes for children
diagnosed with ADHD. DuPaul and Weyandt (2006) suggest a need for increased focus
on improving outcomes for this population, versus the more common approach of
reducing or eliminating problem behaviors. Therefore, a systematic research approach is
needed to identify variables that predict academic achievement within this population, as
well as how these variables may differ from students without ADHD. The extant

literature has examined primarily how ADHD symptoms and cognitive mediators relate

to academic outcomes, however, a large portion of variance is unaccounted for in
predicting academic outcomes within this population (DuPaul, Volpe, Jitendra, Lute,
Lorah & Gruber, 2004). The role of additional variables in predicting academic
outcomes is yet to be well understood.

Identifying predictors of achievement for students with ADHD can inform school
professionals in meeting demands of increasing accountability. Emphasis on
accountability is driven largely by the No Child Left Behind Act of 2001 (2002; NCLB)
and the Individuals with Disabilities Education Improvement Act of 2004 (IDEA). While
NCLB is focused on achievement at the level of the school, IDEA is focused on
individual students, particularly those with disabilities. Within both laws, a means for
accomplishing accountability is by using “scientific research” to informing intervention
practices. However, DuPaul and Eckert (1997) report that research that examines
interventions implemented specifically for ADHD in the school setting is sparse.
Additionally, for significant portion of children diagnosed with ADHD, special education
is an important route for service delivery. However, it has been suggested that there is
little research that examines this area for children diagnosed with ADHD (Fomess &
Kavale, 2001) and special education has not consistently focused on the delivery of
evidence-based practice (President’s Commission on Excellence in Special Education,
2002). Further, DuPaul and Stoner (2003) suggest that there is currently ambiguity when
it comes to what components should be included in an educational plan for special
education students who are diagnosed with ADHD.

Current Study

The current study examines variables that relate to positive academic outcomes
for children with ADHD within a special education population. This study can help
address questions of concern to school professionals who seek to effectively provide
services to promote academic success. Additional knowledge of important predictors of
academic success and the relationships between the variables can be utilized for
prevention, assessment, and intervention purposes. Additionally, this study is intended to
provide a beginning point to further understand the relationship between special
education and student outcomes within the ADHD population.

The purpose of the proposed study is to examine several predictors of academic
outcomes. This study will utilize data from a large, nationally representative data set of
children who are currently receiving special education services (Special Education
Elementary Longitudinal Study, 2003). Specifically, the current study will examine the
relationships between predictors of reading growth within a population of special
education students with a parent and school reported diagnosis 0f ADHD, as well as
comparing these relationships to children who do not have a reported diagnosis of
ADHD. This study will build upon the extant literature in several ways. First, the nature
of the sample adds several distinctions. This study would add to the extant literature by
the use of a large, national sample of children. The longitudinal nature of this sample
will also be an addition to previous studies, which have recommended the need for
longitudinal data (DuPaul et al., 2004) in examining academic outcomes. By focusing on
the special education population, the focus is on the children who are often most
educationally impacted by their ADHD. Secondly, the current study will provide

valuable insight into interventions as typically delivered within the school setting, which

can be compared to the impact of interventions within a highly controlled setting. One
difficulty with the current literature is that most outcomes studies for ADHD have been
conducted in a very controlled manner (DuPaul & Weyandt, 2006). This study instead is
based on reported interventions from real classrooms. It is important to further
understand the relationships between variables within the “real world” school setting.
The following chapters describe the relevant literature related to this topic and
the methodology that is employed to answer the questions of interest. Chapter Two
provides a review of the literature that addresses in more detail, models and predictor
variables that have been identified in the extant literature to help delineate predictors of
academic achievement. Chapter Three describes the methodology employed to answer
the research questions. This chapter also focuses on methodological issues related to
using a large data set, as well as describing the analytic strategies. Chapter Four provides
the results from the analyses of interest, and Chapter Five provides a discussion of the
conclusions that can be drawn from these results, as well as recOmmendations for future

directions.

Chapter 2
Literature Review

Children diagnosed with ADHD frequently experience academic difficulty. This
chapter will outline factors that have been proposed to ameliorate academic
underachievement. A review of the empirically supported treatments for ADHD and the
impact of each of these treatment modalities on academic outcomes will be provided. A
second portion of the literature review will address research that has examined predictors
and models that attempt to account for academic achievement within this population.
This will be followed by a review of how interventions are provided to children within
the school setting, with a focus on special education services. Finally, the current study
will be discussed and the research questions will be outlined.
E vidence-Based Treatment for ADHD

It is important to understand best practices in the care of children with ADHD, by
examining what treatments are empirically supported. Empirically supported treatments
for children diagnosed with ADHD include stimulant medication, behavioral
modification at home or school, and the combination of these two treatments (Pelham &
Gnagy, 1999). These treatments also represent those that are most well-studied (DuPaul
& Weyandt, 2006). In the context of the current study, it is important to examine the
impact of empirically supported treatments on academic outcomes within the literature.

Stimulants, or methylphenidate, are currently the most commonly used treatment
for ADHD and have been for at least the last 30 years (Pelham, et al., 2000). Rowland,
Lesesne, and Abramowitz (2002) reported that over 1.5 million children were taking

methylphenidate. Studies on the effects of stimulants have suggested that their impact

varies across symptoms with the strongest effects on attention, distractibility, impulsivity,
and observable social and classroom behavior (Brown, 2005). However, Brown reports
that stimulants have demonstrated less efficacy in the area of academic achievement, and
that their effect in this domain is “modest” (pg. 121).. As the current study is specifically
interested in the impact on academic functioning, this area will be examined in more
depth.

A recent review of ADHD treatment studies examined the impact of
pharmacological intervention on academic outcomes in children diagnosed with ADHD
(Ryan, Reid, Epstein, Ellis, & Evans, 2005). The authors cited a number of problems
with the current literature on this topic that can make it difficult to draw strong
conclusions. These include limited information on students, few studies with
adolescents, few types of medications studied, inadequate dependent measures, and short-
term interventions. An additional concern cited by the authors was that of those studies
that did report this information only a small number were conduCted within typical
settings for students with ADHD, including general education, resource, or self—contained
classrooms. Instead a majority of these studies were conducted in more restrictive
settings, such as university settings, psychiatric hospitals, and other clinical settings. The
results of the review indicated that medication intervention (83% methylphenidate)
resulted in a small to moderate benefit (mean ES = 0.327) in the academic areas.
Specifically 60% of the studies that reviewed led to an improvement in academic
performance, while 7% led to a negative outcome, 19% indicated mixed results, and the
final 14% reported no difference in academic performance. The results led to the largest

effect size in the subject history (0.74), followed by writing (0.45), reading (0.38), and

math (0.18), with a small negative impact in spelling (-0.05). In sum, the results of this
review suggested that medication does appear to have a moderate benefit on academic
performance.

While most research has suggested that stimulant medications have been shown to
be more effective than behavioral treatments, this is likely dependent on how the
medication is delivered and on what outcomes are measured. In the largest study to date
on ADHD, the Multimodal Treatment Study of ADHD (MTA), carefully titrated
psychostimulant treatment was more effective in reducing symptoms of ADHD than
routine community treatment (which may include treatment with psychostimulants) in the
sample within the MTA study. This was the case when multiple outcomes such as
academic, social, and family variables, were taken in to consideration (Jensen et al.,
2001). Research that specifically examines the impact of carefully controlled
psychostimulants on academic outcomes is important.

In addition to psychostimulants, behavioral treatments are empirically supported
for treating ADHD. According to the American Psychological Association Division 12
(Clinical Psychology), behavioral parent training and behavioral classroom modification
in the classroom are the only well established psychosocial treatments for ADHD
(Ollendick & King, 2004). However, reviews of the literature related to this topic
suggests that academic interventions, and to a lesser extent, social interventions have also
demonstrated efficacy for improving school outcomes associated with ADHD (DuPaul &
Eckert, 1997; DuPaul & Stoner, 2003; DuPaul & Weyandt, 2006).

One of the most comprehensive empirical references for school-based

interventions for ADHD is a meta-analysis conducted by DuPaul and Eckert (1997).

10

These authors suggested that compared to other treatment modalities for ADHD, like
stimulant medication, there has been relatively little research that examined school-based
interventions for children with ADHD. Additionally, less is known about the impact of
various school-based interventions. The authors reviewed 63 studies with 637
participants conducted between 1971 and 1995. The review addressed two primary
research questions: 1.) do school-based interventions significantly change target
behavior? 2.) does the type of intervention (contingency management, academic, or
cognitive-behavioral) influence the impact of the intervention? The study also examined
moderators (e. g. setting) from these studies that may influence outcomes. In reference to
the first research question, the analysis of extant literature provided support that school.-
based interventions significantly change target behavior. The impact varied depending
on the target behavior. The effect size for all interventions (contingency management,
academic intervention, and cognitive-behavioral intervention) was more significant on
behavior than on academic or clinic test performance outcomes; DuPaul and Eckert
concluded that school—based interventions are effective in reducing ADHD-related
behaviors. The effect size for behavioral outcomes was in the moderate to high range.
Although school-based interventions are less effective in improving academic
performance, this meta-analysis also revealed that these interventions can also have an
impact in this area. The effect size for academic outcomes was in the low range.

The second research question addressed in this meta-analysis concerned the
relative efficacy of several intervention types (contingency management, academic,
cognitive behavioral). Their analysis indicated that contingency management and

academic interventions were more effective than cognitive-behavior interventions in

11

improving classroom behavior, such as attention and disruptive behavior. There were no
significant differences between the contingency management and academic intervention
techniques for this purpose. This finding indicated that academic interventions offered a
viable intervention strategy. To explain this finding, DuPaul and Eckert postulated that
academic intervention may provide a replacement behavior (academic functioning) for
the disruptive behavior.

In terms of addressing potential moderators, DuPaul and Eckert (1997) found that
in some studies, interventions delivered within general education were less effective than
those delivered within special education in terms of behavioral outcomes. However, this
was only the case for the within-subject studies included within their review. There were
no differences for the mediating effect of setting across the other study types and
academic outcomes. The authors concluded that the latter result was promising given
that many children with ADHD spent a majority of time in general education.

DuPaul and Weyandt (2006) examined behavioral, academic, and social
interventions for ADHD. Behavior interventions presented the strongest evidence of
effectiveness. Several types of behavior treatments were discussed, including antecedent-
based strategies (e. g. modification of assignments), consequent-based strategies (e. g.
token economies), and self-management strategies (e. g. self-monitoring). The authors
also discussed academic interventions in this review, and suggested that these strategies
are important as they focus not only on reducing inappropriate behavior, but also on
improving positive outcomes. Academic interventions included changes in teacher
instruction, as well as peer- and computer-mediated interventions. These strategies have

demonstrated some efficacy with children diagnosed with ADHD, however, there is a

12

need for additional research examining the impact of these interventions. Lastly, social
relationship interventions are discussed. DuPaul & Weyandt suggested that additional
research in this area is needed, as typical social skills groups have not led to sustainable
improvements in social functioning. There is preliminary support for more
comprehensive social interventions, however, there is limited data at the present time. In
a recent review of evidence based psychosocial interventions for ADHD, Pelham and
Fabiano (2008) report that clinic-based social skills training have little impact, and are
not currently considered an evidence-based intervention for ADHD. These authors did
indicate that intensive behavioral interventions implemented in peer recreational settings,
have established efficacy for children with ADHD.

DuPaul & Weyandt (2006) suggested that both behavior and academic
intervention types have sufficient support within the empirical literature to support their
use. There is a relative void of research on school-based social relationship interventions
for children with ADHD. Although these interventions have shown less efficacy than the
behavioral and academic interventions, their use is promising within the literature. Also
implicated in this review was the need for multiple intervention targets. Many children
experience difficulties across several areas, such as in social and academic tasks, and
multiple intervention approaches are necessary.

Predictors of Academic Outcomes

In addition to research that has evaluated the empirical evidence for various
treatment modalities for ADHD, there is also research that has examined various
predictors of academic achievement. Unlike interventions, predictor is a more general

term and includes any variable that is related to a particular outcome. This could include

13

variables such as gender and classroom placement. Predictors of academic success are
also important to consider as they relate to outcomes and also may mediate the effects of
intervention efforts. A consideration of variables that predict positive academic
outcomes for children with ADHD is essential to improve outcomes.

The extant literature has demonstrated that both ADHD symptoms (i.e.
inattention, hyperactivity, impulsivity) and cognitive variables predict academic
underachievement (DuPaul et al., 2004). For example, studies have reported that
symptoms of ADHD have a negative linear relationship with academic outcomes, with
more excessive ADHD symptoms related to more negative academic outcomes (DeShazo
Barry et al., 2002). Of the core symptoms, inattention is the most strongly tied to
negative academic outcomes (Barkley, 2003). DuPaul et al. (2004) suggest that in
addition to symptoms of ADHD, cognitive variables have often shown to play a role in
predicting academic outcomes in the literature. For example, DeShazo Barry and
colleagues (2002) reported that ADHD symptoms and cognitive variables accounted for
15% to 28% of academic outcomes across studies.

In addition to a focus in the extant literature primarily on cognitive and core
symptoms as predicting academic outcomes, DuPaul et al., (2004) delineated some other
limitations of the current literature on this topic. For example, many existing studies in
this area focused only on one academic area (i.e. reading or math), they primarily used
norm-referenced standardized measures of achievement, and often only examined how
prediction within an ADHD population versus comparing the adequacy of predictors
across ADHD and non-ADHD populations. To address the need to identify additional

predictors, and current limitations with the existing literature, DuPaul et a]. examined

14

predictor variables using hierarchical regression analyses. Specifically, they examined
the predictive value of teacher ratings of ADHD symptoms, academic skills, achievement
related behavior, social skills, direct classroom observations, ethnicity, and SES. Their
analysis considered a model across two groups (ADHD and non-ADHD), two academic
areas (reading and math), and two outcome measures (standardized assessment and report
card grades). The analysis revealed that once ethnicity and SES was taken into account,
only a few variables significantly predicted academic outcomes. Specifically, for both
groups, teacher perceptions of academic skills was the strongest predictor of academic
achievement. Additionally, academic enablers, inattentive symptoms, and observed off-
task behaviors seemed to play a role in predicting outcomes. The predictors varied across
groups (control vs. ADHD), academic area (math vs. reading), and criterion measure
(report card grades vs. standardized assessments). Specifically, reading was more
adequately predicted with their model than math. For children with ADHD, the model
(including Ethnicity, SES, and teacher rating of reading skills) predicted 35% of the
variance on a standardized measure of reading. For report card grades, the model
(including Ethnicity, SES, teacher rating of reading skills, and teacher rating of academic
enablers) explained only 24% of the total variance. Within the control group, there were
differences in terms of what variables were included in the significant model, as well as
for the predictive value of the model. The predictive value of the model in the non—
ADHD population was stronger for the standardized criterion (R2=0.55) and the reading
report card grade (R2=0.32).

The results of this study suggest that several factors predict reading outcomes, but

that these vary across groups, academic areas, and outcome measures. Additionally, this

15

study is the first to indicate the importance of the academic enablers (DiPema & Elliott,
2000) for the ADHD population, as these played a significant role in the model predicting
report card grades in reading for children in the ADHD group. This was true even after
their ADHD symptoms have been accounted for in report card grades. The teacher rating
of academic enablers was not included in the significant model for a standardized
measure of reading achievement. The authors suggested that this may indicate that
academic enablers may play less of a role in skill attainment, but may be important
variables in how teachers grade students. Future research on this topic is needed. An
additional implication of this study was the importance of teachers’ perceptions of a
child’s academic skills. This emphasizes the need to not only reduce behaviors associated
with ADHD, but also to target the academic skills of these children. The results support
the need for future research to consider academic skills and enablers, rather than just
behavioral symptoms, in the assessment and implementation of interventions within this
population.
ADHD Models

In addition to identifying predictor variables, it is also important to understand the
impact of variables within a larger framework. A theoretical framework is essential for
integrating empirical data and developing working hypotheses about the relationships
between variables. The development of such models is essential to informing treatment
practices, as they provide a framework for developing testable hypotheses that can guide
intervention research. Models can help researchers and clinicians understand variables

that are hypothesized to predict positive outcomes. These models can outline various

16

frameworks, including biological, psychological, and social variables that are
hypothesized to impact development or outcomes.

Within the ADHD literature, many models have been developed to help explain
the nature and development of ADHD. An etiological model can be helpful in
understanding causal factors that are hypothesized to underlie the development of
ADHD. For example, an etiological conceptual model articulated by Rapport, Chung,
Shore and Isaacs (2001) suggested an underlying neurobiological system is responsible
for the core features of ADHD which include hyperactivity, impulsivity and inattention.
In this model, the authors hypothesized that the peripheral features of ADHD (e.g., poor
social skills, academic difficulty and internalizing problems) are a consequence of
dysfunction associated with the core features of ADHD. This model can be seen in
Figure 1. The authors of this model also outlined the hypothesized impact of different
interventions. For example, as seen in Figure 1, they hypothesize that methylphenidate
acts on the neurobiological substrates underlying ADHD, which in turn acts on the
symptomotology associated with ADHD. This model provides a comprehensive picture
of how the symptoms of ADHD are hypothesized to relate to one another, and how
modes of intervention may target different areas within this model.

Another prominent model for describing the dysfunction associated with ADHD
is Barkley’s Disinhibition Model (1997; 2003), as shown in Figure 2. This model
suggests that the executive function of behavioral inhibition influences four other
executive functions, including: deficits in nonverbal working memory; internalization of
speech; self-regulation of affect, motivation, and arousal; and impaired reconstitution. In

sum, it is hypothesized that these deficits across the executive functions impact the motor

17

actions. In particular, such deficits make it difficult to plan and execute actions in a goal-
directed and systematic manner. Barkley hypothesized that these deficits lead to the
behavioral manifestation of ADHD of impulsivity and hyperactivity. While models
developed by Rapport et a1. (2001) and Barkley (1997; 2003) have been influential in
delineating better understandings of ADHD, models that are more specifically focused on
academic achievement are helpful in understanding the relationship between ADHD and
scholastic outcomes, and are of particular interest in the present study.

There are a number of theoretical and empirical models of academic achievement
within the extant literature. The majority of these models have not been specifically
assessed in the context of an ADHD population, yet provide valuable insight into
thinking about the types of variables that may best predict academic achievement.
Academic achievement models consistently report student, instructional, and
environmental variables as impacting academic achievement (DiPerna, Volpe, & Elliott,
2002). Much recent work in this area has focused specifically on student variables.
Rationale for studying student factors is that these variables have been shown to have the
most significant impact on student outcomes (Wang, Haertel & Walberg, 1993; DiPerna
et al., 2002). Within the school psychology literature, some of this research has focused
on “Academic Enablers”, which can be defined as “attitudes and behaviors that allow a
student to participate in and ultimately benefit from academic instruction in the
classroom” (DiPerna & Elliott, 2002, pg. 294). These include factors of motivation,
interpersonal skills, engagement, and study skills. A model that describes the

relationship between these variables is shown in Figure 3.

l 8

That model of “Academic Enablers” was evaluated based on data from 394
students (DiPerna et al., 2002). The sample was split into two groups, primary (K-2) and
intermediate (3-6). The constructs of interest were measured using the Academic
Competence Evaluation Scales (DiPerna & Elliott, 2000; ACES). Results of evaluation
of this model indicated that there were similarities in model fit between the primary and
intermediate samples. The most significant predictor across models was prior
achievement, which accounted for a majority of the variance in current reading
achievement. Motivation and engagement were the academic enablers that accounted for
the most variance in the model. The magnitude of the effect size for these variables was
in the large to moderate range. These authors suggested that research on academic
enablers is valuable, as these are variables that are amenable to intervention within the
school setting (DiPerna et al., 2002). It is further suggested that evaluation of models,
such as the one in this study, are important to provide frameworks of testable hypotheses
for future intervention research.

Another model that provides valuable insight into academic outcomes,
specifically related to the ADHD population, is one developed by Fergusson and
colleagues (Fergusson & Horwood, 1995; Fergusson, Horwood & Lynskey 1993) and
replicated and expanded by Rapport, Scanlon, and Denney (1999). The expanded model
focused on specific student variables that impact scholastic achievement. This model, the
dual developmental pathways model, is shown in Figure 4. The authors hypothesized
that there is both a cognitive and behavioral pathway that predict scholastic achievement
in children exhibiting symptoms of ADHD. Specifically, they suggested that classroom

performance mediates a behavioral pathway to scholastic achievement, and that vigilance

19

and memory mediate a cognitive pathway to scholastic achievement, which was defined
in this study as scores on the Stanford Achievement Test. Findings indicated that the
dual pathway model fit well with the data. Specifically, the data indicated that both
behavioral and cognitive pathways mediate the relationship between IQ and ADHD with
scholastic achievement. Overall, the findings indicate that this model accounts for 77% of
the total variance. Interestingly, the direct impact of ADHD on scholastic achievement
was smaller in this study than found in studies that did not examine the role of mediators
between ADHD and achievement (standardized regression coefficient: -.07). However,
there was a strong, significant relationship between ADHD and classroom performance
(standardized regression coefficient: 0.33) and vigilance (standardized regression
coefficient: -0.20). These findings indicated that both cognitive and behavior mediating
variables largely account for the relationship between ADHD and scholastic
achievement.

Recently, another model that further delineates the relationship between
classroom behaviors, ADHD, and academic achievement was described. This model is
more specific than Rapport et al.’s (1999) model in defining behaviors that relate to
academic outcomes. Volpe and colleagues (2006) built upon a model of academic
achievement developed by DiPerna, Volpe and Elliott (2002) and examined child
variables in the classroom and how these mediate the relationship between ADHD and
academic achievement. This model is shown in Figure 5. The authors of this model
purported that ADHD has an impact on prior achievement, motivation, and interpersonal
skills. Motivation is also impacted by prior achievement and interpersonal skills.

Motivation, in turn, acts indirectly on current achievement, through its impact on study

20

skills and engagement. The impact of prior achievement on current achievement is also
hypothesized. In evaluating this model, the authors found that the model accounted for
30% of the total variance in reading achievement, with all pathways being significant
predictors, except for engagement. These findings sirggest that while these academic
enablers account for some of the variance in predicting academic outcomes, additional
predictors remain.

There are a number of variables and models in the extant literature that explain a
portion of the variance associated with academic outcomes for children diagnosed with
ADHD. These include cognitive variables, ADHD symptoms, classroom behaviors,
academic skills, academic enablers, and others. However, there is still unexplained
variance in fully accounting for academic outcomes. Additionally, there is a need to
further understand how intervention strategies influence academic outcomes. The school
serves as an important context for considering both academic outcomes and intervention.
Understanding how interventions may be provided within the school setting, is therefore
an important compliment to understanding predictors of academic achievement, as it
sheds light on how and where these predictors may be influenced.

School Based Service Provision for ADHD

Because academic underachievement and other negative school outcomes are
often so salient for many children with ADHD, the school is often an important context
for intervention. In fact, it has been reported that there have been recent changes in the
treatment of ADHD in children, and that these changes are based in part on the
recognition within special education law of ADHD as a disability (Smith et al., 2006).

An additional reason why an understanding these services provided in the school setting

21

is essential is based on data that suggests that the school often serves as the primary
services provider to children. For example, there is evidence that a majority of children
with mental health needs do not receive services (Hoagwood & Johnson, 2003). In fact,
it has been suggested that only six to twenty-five percent of children receive mental
health services. Additionally, for a majority of children, the school is often the only
service provider of such services (Hoagwood & Johnson). The next section will further
explore what these services may look like for children diagnosed with ADHD.

Within the school setting, children with ADHD may receive both formal and
informal services and these services can be widely variable. One avenue of services for
children with ADHD is through special education. Historically, children with ADHD did
not receive special education specifically as it related to this disorder. These children
could and often were eligible under other special education categories, however, there
was not a category that included ADHD until 1991 when ADHD was included as a
subcategory in the Other Health Impairment (OHI) label. However, a diagnosis of
ADHD is not sufficient for a student to be eligible for special education services. Within
the law, special education is reserved for children where ADHD is not only present, but
also where there is a need for special education services.

While it is clear that not all children with ADHD are eligible for special
education, it is the case that many children with ADHD do enter the special education
system. Children who are exhibiting the highest level of behavior or academic difficulties
within the schools may often be referred for special education. A recent survey based on
a national sample suggested that over 11% of children with ADHD, without learning

disabilities, receive special education services. Additionally, among children with both

22

ADHD and LD, almost 65% were receiving special education (Pastor & Reuben, 2002).

Starting in 1991, three routes for services were established for children with ADHD,

which have been summarized by DuPaul and Stoner (2003).

Children with ADHD could qualify under a special education eligibility category
other than OHI (e. g. Specific Learning Disability (LD), Emotionally Disturbance
(ED), Mental Retardation (MR)) if they met the eligibility criteria for that
category. Certain categories are more common for children with ADHD. Forness
and Kavale (2001) suggest that children with ADHD are often found in the
Emotional Disturbance (ED) and LD categories, and account for over forty
percent of children within these groups.

Children with ADHD could also qualify under the eligibility category of Other
Health Impaired (OHI). In 2002, approximately 7% of children in special
education were qualified under the OHI category. This number of students
qualified under this category has increased significantly from 1992 to 2002 (US.
Department of Education, 2005) a result that has been attributed to the inclusion
of ADHD in this category since 1991 (Fomess & Kavale, 2001). The specific
definition that makes this category appropriate for children with ADHD is that the
other health condition results in “having limited strength, vitality, or alertness,
including a heightened alertness to environmental stimuli, which results in limited
alertness with respect to the educational environment” (Department of Education,
2006). To become eligible in this category, a child must also undergo a

comprehensive evaluation by a multidisciplinary team, which includes a

23

physician. In addition, they must demonstrate a need for special education

services.

0 Although not a special education decision, a child could qualify for services
within general education under Section 504 of the Rehabilitation Act. Unlike
special education, there is not funding attached to this service. Fomees and
Kavale (2001) suggest that the accommodations provided through this service are
likely to be less intensive than those received within special education.

Given the focus on reviewing children with ADHD receiving special education services,
only the first two areas of eligibility will be discussed within this review.

Recent research has examined what special education category children with
ADHD are categorized under, as well as providing demographic and service delivery
information on this topic. A study by Schnoes and colleagues (2006) suggested that
children with ADHD are found across a number of special education categories. The
sample in this study was drawn from the Special Education Elementary Longitudinal
Study (SEELS), a nationally representative sample of elementary-aged students in special
education, and includes data from parent report, teacher report, school report, and through
direct assessment. The sample used for that study included children with the special
education eligibility labels of LD, ED, Speech/Language Impairment (SLI), OHI, and
MR. The rationale for selecting these categories, as described by Schnoes and
colleagues, was due to the fact that these categories, particularly LD, ED, and OHI, are
most likely to include children diagnosed with ADHD, based on prior research in this
area. From this sample, only children where both parents and the school agreed that the

child had ADHD were included. For the comparison group, children where both parents

24

and school report that they do not have ADHD were included. The total sample included
1419 children.

The prevalence of ADHD was reported for each disability category. The OH 1
category had the highest percentage of children with ADHD with approximately 66%,
followed by ED (58%), MR (21%), LD (20%), and SLI (5%). In contrast to looking at
the prevalence of ADHD within each disability category, this study also examined the
disability categories reported for the ADHD population. Within the population of
children with ADHD eligible for special education services, LD was the most common
diagnosis with approximately half (50%) of children with ADHD in special education
being eligible for services under the LD disability category. The percentage of children
in the sample eligible under the LD label was similar for children with and without
ADHD. Over 17% of children with ADHD in special education were eligible under the
OHI category, and this percentage was much higher than for non ADHD children in
special education (2%). A similar pattern was found for the ED category, where over
13% of children with ADHD were placed, compared to approximately 3% of children
without ADHD in special education. In the SL1 category, fewer children in the ADHD
group were eligible for these services (7%) than in the non-ADHD group (35%). Within
the MR category, 3 similar percentage of students with ADHD (12%) and without ADHD
( 12%) were found.

In terms of both where a high percentage of children with ADHD are placed, as
well as categories in which ADHD are more likely to be placed compared to all students
in special education, three categories stand out. These include the areas of LD, ED, and

OHI.

25

ADHD & Learning Disabilities. ADHD and learning disabilities commonly
overlap. This is logical given the fact that academic difficulties are a peripheral feature
associated with ADHD. The overlap is also significant in the context of special
education, as learning disabilities represent almost half of the children who are qualified
for special education (US. Department of Education, 2005). Estimates for the overlap
between ADHD and LD have ranged from 7 to 92 percent. However, Fomess and
Kavale (2001) suggested that the most appropriate comorbidity of these two diagnoses is
likely between 10 and 20 percent. In a review of the literature, DuPaul and Stoner (2003)
suggested that children with ADHD are three to four times more likely than their peers to
exhibit learning disabilities. Additionally, the percentage of children with LDs that are
diagnosed with ADHD is even higher. Children with learning disabilities are seven times
more likely to exhibit significant symptoms of ADHD. Clearly, these disorders and
symptoms of these disorders often co-occur.

In one of the largest studies to date that focused on children with Attention Deficit
Disorder (ADD; the DSM-III precursor to ADHD) and a LD, Pastor and Reuben (2002)
explored the relationship between these two diagnostic categories. Their study was based
on the National Center for Health Statistics Survey from 1997-1998 and included more
than 8600 children, that were representative of the US. population. The results of their
study suggested that 2.6 million of children age 6-11 years were diagnosed with ADD or
LD. Within these groups they found that 3% were diagnosed only with ADD and 4
percent with only LD. Additionally, there was another 4% with both ADD and LD. This
underscores the idea that some but not all children with ADHD will experience learning

difficulties. In this sample, seven percent of school-aged children had ADD, which is

26

consistent with the high end of the 3-7% of school-aged children reported in the DSM-
IV-TR (APA, 2000). The fact that more than one-half of these children had a LD in
addition to their ADD is important to consider.

Pastor and Reuben (2002) also reported on the service utilization of children with
ADD and a LD. Children with both LD and ADD had highest rates of medication use
and mental health services, followed secondly by children with ADHD. Pastor and
Reuben (2002) also examined the service utilization in the school setting. Children with
sole diagnosis of LD were five times more likely to be participating in special education
than children with ADD only. The addition of LD to ADD may make a child more likely
to have services both within the community and school setting.

There have been a number of other studies that have examined this relationship
between ADHD and learning disabilities. DuPaul and Stoner (2003) provided several
relevant conclusions based on their examination of these studies. First, they suggested
that these studies provide support for a significant correlation between academic
underachievement and symptoms of ADHD. They also suggested that this relationship is
strongest in elementary children. Additionally, they suggested that of the externalizing
behavior disorders the link between learning disabilities and ADHD is the strongest.
Despite the fact that this relationship is supported in the literature, there are several
methodological issues that make it difficult to draw conclusions about the overlap
between ADHD and LD.

A particularly common learning disability for children with ADHD is a reading
disability. In their review of the literature, Wilcutt and colleagues (2007) suggest that

from 25—40% of individuals with either a reading disability or ADHD also meet the

27

criteria for the other disorder. This is significant as approximately 18% of the US.
population of children will experience reading problems in the early school years
(National Reading Panel Progress Report, 2000). To explain this high rate of
comorbidity, previous research has revealed that both reading disabilities and ADHD
have a significant genetic component, and that children with a relative with ADHD or a
reading disability are 6-8 times more likely to have the disorder than children who do not
have a relative with the disorder (review Wilcutt et al.).

A better understanding of the pathways that mediate the correlation between
reading disabilities and ADHD is important. A recent study in preschool-aged children
was important in shedding light on the developmental pathways through their use of
young children (Wilcutt et a1.) Wilcutt and colleagues found correlations between ratings
of ADHD and reading measures in preschool children. Consistent with previous
research, their study also revealed that the reading difficulties were more highly linked to
difficulties with attention, as opposed to hyperactive and impulsive symptomatology.
The authors concluded that this finding provides evidence against the idea that one
disorder is the consequence of the other. Additionally, Wilcutt et al. found that
inattention was not more strongly associated with any particular type of pre-reading task.
In other words, they did not find one specific reading skill that was more highly
correlated with ADHD.

Another area that is relevant to special educations service delivery for children
with ADHD and LD is the recent changes in special education law that have changed the
requirements for a child to be eligible for a specific learning disability. The Individuals

with Disabilities Education Improvement Act of 2004 has changed the requirements for

28

determining children eligible for special education (U. S. Department of Education,
2006). The federal law now states that the discrepancy formula does not have to be used
to identify children with learning disabilities, and that alternative methods may be used.
A popular alternative for determining eligibility for special education is the response to
intervention model [National Association of State Directors of Special Education
(NASDSE), 2006]. This model uses a dual discrepancy model which relies on both
learning rate and performance that is discrepant from peers. While there are several ways
in which this model is conceptualized, service provision is provided largely outside of
special education and when eligibility is considered it may be based on a lack of response
to high quality intervention (NASDSE). Little is known regarding the implications of
this change in the law specifically for children diagnosed with ADHD. Further research
is needed to better understand the implications of this change on academic outcomes for
students.

ADHD & ED. Compared to the amount of research that'examines the link
between ADHD and learning disabilities, Schnoes et a1. (2006) reported that there is less
research that has focused specifically on the relationship between ADHD and ED. This
may be due to the fact that ED is not a psychiatric diagnosis, and is solely limited to the
educational classification systems used within school settings. Schnoes and colleagues
(2006) suggested that one reason children with ADHD are likely to receive services
under the ED category because of the high rates of comorbidity between ADHD and
other psychiatric diagnoses.

For children in special education programs for ED, there have been widely

varying rates of ADHD reported. In their review of studies, Fomess and Kavale (2001)

29

suggested that in the few studies that examine this overlap, estimates averaged out to
approximately 43 percent of children with an ED eligibility with a diagnosis of ADHD.
Other reviews have suggested the prevalence rates for the overlap for these disorders
range from 25-44% of children in the ED category as having a diagnosis of ADHD
(Schnoes et al., 2006). Within a recent study on this topic, almost sixty percent of the
children qualified under the ED category were diagnosed with ADHD. After the OHI
category, this was the highest percent in any special education category for children with
ADHD in special education. As another way to examine the overlap, this study also
reported that of the children with ADHD in special education almost 14% were eligible
under the ED label. This was the third most frequent category after LD and OHI
(Schnoes et al.).

The distinction between ADHD and ED is more ambiguous than that between
ADHD and LD. For example, social difficulty is a commonly reported peripheral
symptom of ADHD (DSM-IV—TR) and the “inability to build or maintain interpersonal
relationships with peers and teacher” (Department of Education, 2006) is also a
characteristic that can make one eligible as a student with an ED.

It seems clear that a number of children with ADHD receive special education
services through the LD and ED categories. As such, children with ADHD may often
exhibit behaviors within the school setting beyond the core symptoms of hyperactivity,
impulsivity, and inattention. This has important implications for both diagnosis and
treatment. For example, it appears that children with ADHD and 3 LD are more likely
than children with just ADHD to receive services, particularly in the school setting. In

addition to receiving special education services through the LD or ED label, children with

30

ADHD may receive services specifically related to their ADHD diagnosis under the OHI
label.

ADHD & OHI. In addition to qualifying under another eligibility category, a
second route children with ADHD are often qualified to receive special education
services is through the OHI category. This is the category that is most specific to ADHD.
DuPaul and Stoner (2003) suggested the OHI route is the most common way for children
with ADHD to be classified. This eligibility requires that children not only have a
diagnosis of ADHD, but also that these symptoms significantly impact their performance
within an educational setting. DuPaul and Stoner suggested that if these guidelines are
interpreted loosely, that many children with ADHD would qualify for special education
services under the OHI label. DuPaul and Stoner also suggested that currently guidelines
related to OHI are ambiguous, and require an outside diagnosis, a demonstrated difficulty
in school, and a need for special education services. However, the school is left without
specific procedures for determining eligibility.

It has been reported that there is very little data regarding the number of children
with ADHD in the OHI category, as this category includes a heterogeneous group of
children with a number of other health issues. However, several studies have attempted
to answer questions related to the prevalence of ADHD within OHI. Fomess and Kavale
(2001) estimated the prevalence of ADHD in OHI by examining the yearly increases in
the number of students eligible for services under the OHI label after ADHD was allowed
as a diagnosis in this category beginning in 1991. This methodology yielded an estimate
of approximately 40 percent of children in the OHI category as having ADHD. This

estimate is low compared with a recent study (Schnoes et al., 2006) where an estimate

31

was obtained through the analysis of a nationally representative data set and examined
the number of children eligible under OHI with home and school reported ADHD.
Results of this analysis suggested over 65% of children in the OHI group are diagnosed
with ADHD. In fact, this is the highest percentage across any eligibility category in this
study. Within the group of children with ADHD children in special education almost
18% qualified under the OHI label and this number was much higher for this group than
for the special education population without ADHD.

Summary. While an understanding of the category itself where children with
ADHD are placed does not provide an understanding of the specific intervention services
provided to these children within the school setting, it can help uncover the scope of
services that children with ADHD are provided in the school. It can also speak to the
number of children with ADHD who are having significant academic, behavioral, and
emotional dysfunction, as a large number of children with ADHD are represented in
special education. Another benefit to reviewing this data is that it can be linked to the
literature regarding the specific kinds of interventions and accommodations received by
children in special education. This can shed light on the types of interventions that
children with ADHD are receiving within the school setting.

School Based Interventions

Beyond having knowledge of the diagnostic categories under which children with
ADHD are often eligible, it is important to understand the actual services and
interventions received by these children. Children with ADHD in special education may
not always receive services specific to this disorder. In fact, DuPaul and Stoner (2003)

suggested a need to better understand how school-based services for children with ADHD

32

differ from other disabilities. One study that examined the care that children diagnosed
with ADHD who were placed in special education, found that only half of these children
were receiving services specifically for this disorder (Bussing et al., 1998). Additionally,
it has been reported that little research has examined the extent to which special
education and general education teachers implement interventions for students with
ADHD (Schnoes et al., 2006).

One study that has examined this question was conducted Reid, Maag, Vasa and
Wright (1994). This study reported on the use of interventions and accommodations for
children in general and special education settings. Their findings indicated that special
education teachers reported using behavior modification, consultation, and one-on-one
teaching significantly more than general education teachers. Both groups of teachers
reported using modified seating equally. For general education teachers, the most
common modification was a change in seating, while for special educators behavior
modification was the most common. Teachers in special educatiOn also used the
strategies at a much higher rate than teachers within general education. The
correspondence of evidence-based practice used by special education teachers in this
study is promising.

A recent evaluation of children with ADHD receiving special education services
is particularly helpful in providing information on the types of services that children with
ADHD receive (Schnoes et al., 2006). Like children in special education without ADHD,
the majority of children with ADHD (63%) who were eligible for special education
received most of their education within the general education classroom. Just over a

quarter of children with ADHD (28%) eligible for special education received 80% or

33

more of their instructional time within the general education classroom, which was a
significant difference from the non-ADHD group where 44% of the students received
80% or more of their instructional time within the general education classroom. This
study found that two-thirds of children with ADHD (68%) received at least one form of
nonacademic service both within and outside of the school setting, the most common of
which was a behavior management program with over 37% of children receiving this
service. This study also found that many of the children with ADHD in special education
also received academic services, the most common of which was monitoring of progress
by the special education teacher. Seventy-two percent of these children received this
service. Almost half (49.3%) of the children received tutoring by the special education
teacher. Additionally, close to half (45%) of the children received academic services
from a teacher aide and just over 40% received learning strategies or study skills
instruction. In addition to interventions, children with ADHD also received a number of
educational accommodations. For example, in this study it was reported that 38% of
children had modified grading standards. Over 80% had some form of testing
accommodation (e. g. more time, modified tests, etc.). There were also accommodations
to instruction and assignments, with extended time for assignments being the most
common with over 71% of children receiving this service.

The Schnoes et a1. (2006) study offers valuable insights into a population of
children that is not well studied. However, Schnoes et a1. do not report on the academic
outcomes of this population. Given an emphasis on accountability within schools, it is
increasingly important to understand the relationships between various interventions,

particularly those administered through special education, and children’s academic

34

outcomes. A focus on academic outcomes in particularly important with the ADHD
population in order to move away from an emphasis on simply reducing negative
behaviors, to improving outcomes. Further research on the relationships between various
child and intervention variables and academic outcomes is warranted.

Current Study

Currently many children with ADHD are provided with services within special
education, where the relationship between interventions and academic outcomes is
unclear. The focus of the current study will be on academic outcomes in reading, as
reading outcomes represent an essential developmental task for children, and is related to
overall school achievement and adjustment (DuPaul & Stoner, 2003). Additionally,
ADHD in the early years may predispose children to reading difficulties (Barkley, 2003)
making it a particularly important academic area for study. The purpose is to develop a
better understanding of the factors that relate to positive outcomes for children with
ADHD, with an emphasis on variables that are amenable to change in the school setting.
Also important to this study is an understanding of how these predictors are similar and
distinct from other children exhibiting significant school dysfunction, but without a
reported diagnosis of ADHD. The following research questions will be addressed:

An initial research question examines mean differences in reading over time and
across groups. The purpose of this research question is to determine if there is growth
over time and whether the change over time for children with a reported diagnosis of
ADHD is similar to the change over time for children without a reported diagnosis of
ADHD.

1.) Are there mean differences over time in reading achievement?

35

a. Are there significant mean differences in achievement over time across the
ADHD and non-ADHD groups?

b. How do these means differ from children with and without ADHD in a
special education population?

c. Are there interactions between Time and ADHD Status?

It is hypothesized that there will be mean differences between time one and time
two, with a significantly greater mean at time two. It is further hypothesized that there
will be mean differences between the groups with the non-ADHD population performing
significantly higher than the ADHD population. It is also hypothesized that there will be
an interaction between these relationships with the non-ADHD population showing a
steeper rate of growth in reading performance.

A second set of questions examines how these variables predict academic
outcomes in the context of other variables using latent growth modeling. This allows for
the examination of the larger picture of relationships between the variables and is
especially appropriate for non-experimental data (Keith, 1999). Additionally, the
generation of models is useful in creating a framework for hypothesis testing (DiPerna et
al., 2002). The factors can be tested not only in isolation, but also in how they impact the
overall model. The rationale for the variables included in this model is based on prior
research regarding variables that are hypothesized to impact academic performance.

2.) What factors relate to initial reading skills and growth over time?
a. How do the introduction of demographic, behavioral, or intervention
variables impact the initial reading level?

i. Standardized assessments?

36

ii. Curriculum Based Assessments?
iii. Reading Grades?
b. How do the introduction of demographic, behavioral, or intervention
variables impact the change over time in reading?
i. Standardized Assessments?
ii. Curriculum Based Assessments?
iii. Reading Grades?
c. How does the impact of the predictor variables differ between the ADHD
and no-ADHD groups?
i. Standardized Assessments?
ii. Curriculum Based Assessments?
iii. Reading Grades?

The hypothesized model is outlined in Figure 6. It is hypothesized that there is
linear growth over time in terms of the reading outcomes. It is also hypothesized that
there will be differences in predictive value across outcome measures. Based on previous
research comparing outcomes across grades and standardized measures (DuPaul et al.,
2004) it is hypothesized that a stronger relationship between predictors and standardized
assessments will be determined. Research in this area has not examined outcomes using
curriculum based assessment (CBA) in the past, and the impact of these variables on
CBAs is unknown. Further description of hypotheses related to this model can be found

in Chapter Three.

37

Chapter 3
Methods

The relationships between various predictors of academic achievement are
examined within a parent and school reported ADHD and non-ADHD sample. The
following sections will describe the data set, sample, instruments, variables, and analytic
techniques for this study. The analytic techniques are broken into down .into two sections
based on research questions. As described in the previous section, the first set of research
questions explores mean differences over time and across groups using analysis of
variance (ANOVA). A second set of research questions utilize latent growth modeling to
evaluate a proposed model of reading growth
Data Set

In the present study data from the Special Education Elementary Longitudinal
Study (SEELS; 2003) will be utilized. The SEELS is intended to provide the first
national representation of longitudinal outcomes in special education students and
includes data collected from parents, teachers, schools, and directly from the student.
This SEELS includes a nationally representative, stratified, random sample. This data set
was collected between 1999 and 2005. The sample was drawn at two levels: through
local education agencies (LEAs) and individual students. At the LEA level, a sample was
drawn from 245 districts and from 32 state-supported special schools. The LEAs were
stratified by size (student enrollment), geographic region, and district/community wealth.
The participating LEAs provided rosters of students in special education. The original
eligible student sample included 11,512 children between the age of 6 and 12 and at least

in the first grade at the start of the study, during the 1999-2000 school year. Parents of

38

these 1 1,512 students were contacted and data was originally collected from a total of
9,824 these students, representing a 85.3% response rate from the total eligible total
sample. Of the non-respondents, there were three primary reasons cited for nonresponse:
refusal (n=455), language barrier (n=156), and no response provided (n=l,077) (SRI
International, 1999). Data was collected in three waves over five academic years. Wave
one was collected during the summer of 2000 and during the 2000-01 school year. Wave
two was gathered during the 2001 -02 school year and wave three was collected during
the 2003-04 school year.

As mentioned above, the SEELS sample was stratified by region (Northeast,
Southeast, Midwest, and West), LEA Size (Very Large, Large, Medium & Small), and by
LEA/Community Wealth (high, medium, low & very low). The region was defined by
those criteria used by the Department of Commerce, the Bureau of Economic Analysis,
and the National Assessment of Educational Progress (SRI International, 1999). LEA
. Size categories were defined by the following criteria. Very Large designates schools
with estimated enrollment greater than 17,411 in grades one through seven. Large
describes school with estimated enrollment from 4,707 to 17,411 in grades one through
seven. Medium includes schools with estimated enrollment between 1,548 to 4,706 in
grades one through seven. Small describes schools with estimated enrollment between
10 and 1,547 in grades one through seven. The LEA/Community Wealth was
categorized using the Orshansky index, which is defined as the proportion of the student
population living below the federal definition of poverty (SRI International) This
variable was broken into the following categories: High (0-12% Orshanksy), Medium

(13-34% Orshansky), Low (35-45% Orshansky), and Very Low (over 45% Orshansky).

39

This data set is appropriate for the current study for several reasons, including that
there is little research that has examined the relationships between special education
services and ADHD, despite the high number of children with ADHD in this population.
The use of a nationally representative data set also allows for examination of the
variables of concern within a diverse population of students. Additionally, previous
research on school-based interventions has cited the use of cross-sectional samples as a
limitation (Volpe et al., 2006), and the longitudinal nature of the SEELS adds to the
extant literature.

Criteria for Selecting Sample

The present study includes a mixed sample of children who have a parent and
school reported diagnosis of ADHD and those who do not have ADHD reported by
either. A requirement for a diagnosis of ADHD is that the dysfunction is exhibited in two
or more settings (APA, 2000). Therefore, only children where both parents and school
report the presence of ADHD at Wave one were selected to represent the ADHD sample.
Parents were asked the following question, “With what physical, sensory, learning, or
other disabilities or problems has {child} been diagnosed?” For parents who do not
mention ADHD, they are specifically asked, “has {child} been diagnosed with attention
deficit disorder or attention deficit/hyperactivity disorder? These are sometimes called
ADD or ADHD.” A variable that was created that combines both of the responses to
these questions will be used. The school report was based on responses provided school
staff, to a question that provided them with a list of disabilities and they were asked to

identify all of a student’s disabilities. Children with a parent and school report of ADHD

40

will be referred to in this paper as children with ADHD, and those with a no from both
parent and school will be referred to as no-ADHD.

Schnoes and colleagues (2006) suggest this methodology allows for the
identification of students whom the school considers to have ADHD, which is an
important factor in understanding school-based service delivery. Additionally, Schnoes
et al. report that this methodology has been used by the United States Centers for Disease
Control and Prevention (Pastor & Reuben, 2002) and other researchers (e. g., Redden et
al., 2003) to identify children with ADHD. Pastor and Reuben (2002) recommend
examining populations of school-aged children to help reduce inaccuracies in parents’
reporting. Additional support for this method for selecting the current sample is provided
by the results found by Schnoes et a1. Their analysis of the SEELS Wave one data
revealed that the demographics of the population of ADHD students, identified through
parent and school report, were consistent with ADHD demographic data reported in the
literature for male to female ratios and medication status.

The sample includes all children with a special education eligibility status under
the following categories: Other Health Impairment (OHI), Learning Disabilities (LD),
Emotional Disturbance (ED), and Speech and Language Impaired (SLI). Previous
research, including a study with this sample of children (Schnoes etal., 2006) indicates
that these eligibility categories are most common among children with ADHD in special
education. Although previous research has also found a high percentage of children with
a diagnosis of Mental Retardation (MR; 20.6%) having a diagnosis of ADHD, these
children will be excluded from the current study. The rationale for this exclusion is that

cognitive variables have been implicated in influencing academic achievement, and

41

excluding children with MR is hoped to control as much as possible the inclusion of
children with extremely low cognitive abilities. Children who are eligible under the OHI,
LD, ED, and SLI categories will be excluded if they are also classified with the MR label.
Additional exclusions will include children with eligibility in the Autism, Developmental
Delay, Traumatic Brain Injury, Visual Impairment, Hearing Impairment, and Multiple
Disabilities categories, as defined by report from the school on a child’s eligibility status
in each of these areas. This is consistent with previous studies that have excluded these
disabilities from studies that examine factors that relate to academic achievement within
an ADHD population (DuPaul et al., 2004). No other exclusionary criteria will be
applied. The original sample size for the study was 1029.

Instruments

For the SEELS data set, data was collected from several sources including the
parent/guardian, teacher, school, and directly from the student (SRI International, 1999).
The variables of interest in this study come from several different instruments. A brief
description of each of the data collection tools is outlined below.

Parent/Guardian Interview. The parent survey was conducted over the phone or
through a questionnaire sent through the mail. This survey gathered information
including student characteristics, household characteristics, family involvement,
academic progress, non-school activities, personal/social information, physical health,
and school satisfaction.

Teacher Survey. A survey was conducted with teachers who provide language
arts instruction to the SEELS student sample. This survey collected information

regarding the language arts setting, instruction practices, assessment practices, behavior

42

and discipline data, accommodations/ modifications, support provision, student
performance, and teacher background.

Survey of Student’s School Program. This survey was completed by school
personnel who were familiar with a student’s school program and collected information
about the school program, transitions, special education services, state and district tests,
accommodations/modifications, support provision, performance/support, and parental
involvement.

School Characteristics Survey. The School Characteristics Survey was completed
by a school coordinator, principal, or a designee that could provide information about the
schools. This survey provided general information about the schools that participated in
this study. For example, it surveyed information regarding the school and community,
students, staff, programs, special education policies/practices, and parent involvement.

Direct Student Assessment. Direct assessments were conducted with each child in
the SEELS sample. Direct assessment procedures collected information regarding
academics, including the Woodcock-Johnson Research Edition (Woodcock, McGrew &
Mather, 2001) and curriculum-based measures of oral reading fluency. Additionally,
information was collected on the student’s self concept, attitudes, and friendships.

Transcripts. Transcripts were collected for students once they reached the
secondary level. These were used to provide information on the courses that students
take.

Variables to be Included in the Analysis
Variables for the current study include those taken from the Parent/Guardian

Interview, the Teacher Survey, the School Programs Survey, the School Characteristics

43

Survey, and the Direct Student Assessment. No datum from the Transcripts was used
within the current study. The variables that will be used in this study include:
0 Group Variable
o ADHD Status
0 Predictor Variables
o Teacher-Rated Student Behaviors
o Stimulant Medication Use
0 Behavioral Interventions
0 Academic Interventions
0 Student Grade
0 Gender
0 Ethnicity
0 Household Income
0 Mother’s Education
0 Reading Goals
0 Functional Skills
0 Dependent Variable
0 Academic Achievement
These variables will be described in detail below.
Group Variables
The variables used to define the ADHD versus non-ADHD group have been
described in the Sample section.

Independent Variables

Teacher-rated student behaviors. Within the SEELS Teacher Survey is a set of 24
behaviors rated by the child’s language arts teacher. This variable contains items about
many types of student behaviors, and includes items such as, “student easily transitions
activities”, “student fights with others”, “student joins group activities”, and “student has
low self esteem”. The items are rated on a three point scale with options of never (I),
sometimes (2) and very often (3). Based on previous research on predictors of academic
outcomes in children with ADHD, of interest in this study were both behaviors associated
with ADHD, as well as behaviors associated with enabling academic success. The
behaviors associated with ADHD were expanded to be more generally described as
externalizing behaviors for several reasons. First, it was thought that this description may
encompass behaviors that were not only relevant for the ADHD group, but also for the
non-ADHD group, because it provided a more general description of behavior.
Additionally, because there were a limited number of items specifically related to ADHD,
this was expanded to include a range of externalizing behavior in order to include a
greater number of items. Therefore, based on the items that were asked of the teachers,
two distinct groups of items were created. The first group consisted of six items thought
to represent externalizing behavior. Construction of this scale included a consideration of
available items that appeared to fit within this construct. Items in the original analysis of

9, “ 99 66

this scale included: “student argues with others , student fights with others , student
responds appropriately when hit or pushed”, “student avoids trouble”, “student controls
temper”, and “student acts impulsively”. Items that were phrased positive (e. g. “student

responds appropriately when hit or pushed”) were reversed, so all items on this scale

where phrased negatively.

45

A second set of nine factors was hypothesized to represent “student” behaviors, or
behaviors that would help a student to be successful academically. An existing scale of
student behaviors (“Academic Enablers”), the Academic Competence Evaluation Scales
(DiPerna & Elliott, 2000), was considered for variables to include on the student scale, as
well as a consideration of the items available that appeared to fit with this concept. Items

9’ 66

in the original analysis of this scale included: “student keeps at task until done , student
does things on own”, “student performs up to ability”, “student completes homework on
time”, “student follows directions”, “student asks for what she/he needs”, “student easily
transitions activities”, “student gets easily distracted”, and “student communicates
thoughts/ideas”. Only one item from this original set of variables needed to be reversed
(e. g. “student gets easily distracted”) to make all items on this scale phrased positively.

A confirmatory factor analysis (CFA) of the items within the current study sample
has been conducted. The purpose of such an analysis is to define sets of variable that
reliably measure behaviors of interest to the study. The CFA was conducted with these
data using LISREL 8.80 (Joreskog & Sorbom, 2006). The analysis originally included 15
variables that were hypothesized to compose an “Extemalizing” and a “Student” scale.
Based on modification indices, changes were made to the loadings of the original items.
This included removal of five items, two from the Extemalizing scale; three from the
Student subscale. The final factors and their standardized and unstandardized
coefficients are shown in Table 1. The final model fit the data well. This is based on
criteria for good fit recommended by Kline (2005). These indices include the model Chi-

square, the Root Mean Squared Error of Approximation (RMSEA), the Comparative Fit

Index (CPI), and the Goodness of Fit Index (GFI). Fit indices for the data included a

46

non-significant Chi-Square value of 39.65 (p=.55). Additionally, the RMSEA was equal
to 0.02 (Cl=0-.032), NFI=1.00, CFI=1.00, and GFI=0.99.

The reliability of this scale within the current sample has also been assessed using
SPSS 15.0. This was conducted by examining the reliability of each of the subscales
(Student Behavior & Extemalizing Behavior). It has been suggested that a reliability
coefficient greater than 0.7 is considered “adequate”, above 0.8 is considered “very
good”, and above 0.9 is considered “excellent” (Kline, 2005). The Cronbach’s alpha
coefficient for this sample for the externalizing subscale was a=.802. The Cronbach’s
alpha coefficient for the student subscale was a=.841. Both reliabilities appear to be
adequate for the current study.

Stimulant medication. This variable is a dichotomous (yes/no) created variable
based on parent-report of various stimulant medication use. The created variable is
described as that the student is taking one or more stimulant medications.

Behavioral intervention. DuPaul & Weyandt (2006) described several types of
behavioral interventions. These included, antecedent, consequent, and self-management
approaches. Within the current sample, there are several variables that fit into these
approaches. The data are based on report from the language arts teacher and refer to
intervention approaches implemented within that classroom. These data are dichotomous
(yes/no) variables that indicate whether or not the intervention approach was
implemented. These include responses to the question “the following were provided to
the student to help in this class”. Specific items selected were: behavior management
program, shorter or different assignments, and more frequent feedback. These variables

will be summed for analysis in the current study.

47

Academic intervention. DuPaul and Weyandt (2006) describe three types of
academic interventions that have been empirically studied for improving outcomes for
children with ADHD: modifications to teacher instruction, peer-mediated strategies, and
computer assisted instruction. These three categories of intervention will compose the
academic intervention variable. They are based on report from the language arts teacher,
who is instructed to report these for their use in the language arts instructional setting.
These include responses to the question “the following were provided to the student to
help in this class”. Responses were provided in a dichotomous (yes/no) format. These
include, tutoring by an adult, whether the child has received peer tutoring, and whether
computer software for student’s unique needs will be used, or computer software
designed for students with disabilities, or the use of computers for activities not allowed
for other students. Additionally, a created variable that describes the teachers
individualization in teaching practices for this student will be included, which is based on
several summed items, and includes a scale from 2 (indicating no modifications) to 8
(often modified). These variables will be summed for analysis.

Demographic Variables

Several variables will be included in the analyses to control for their impact on
academic achievement. These include grade, gender, ethnicity, household income,
mother’s education, reading goals, and functional skills. Grade, household income (16
categories), and mother’s education (4 categories) are considered continuous variables.
Based on preliminary analyses suggesting that approximately 74% of the sample consists
of white children, Ethnicity was coded as a dichotomous variable of either white or non-

white. Ethnicity is an important variable to consider in the context of the current sample,

48

as previous research using the National Longitudinal Transition Study-Two found that
students with disabilities, who were classified as non-white, obtained lower academic
achievement scores than did white children with disabilities (Wagner, Newman, Cameto,
Levine & Gozales, 2006).

Student’s goals for reading achievement were also considered. This is a variable
that asked teachers the primary reading goal for their students. The response choices
included: (1) reading at grade level, (2) improving reading skills, (3) developing
functional reading skills, (4) building pre-reading skills, and (5) no goals for reading
achievement. Thus lower scores on this measure are more desirable.

The students’ “functional cognitive skills”, as rated by the child’s parent, was also
included. This included ratings from 1-16 that indicated how high a student’s functional
cognitive skills were with 1 being low, 6 being medium, and 11 being high. Parents were
asked to rate their child’s ability to complete skills that are used in daily activity on a four
point scale from “not at all well” to “very well”. The specific skills that parents were
asked to rate included, reading and understanding common signs, telling time on a clock
with hands, counting change, and looking up telephone numbers and using the telephone.
Wagner and colleagues (2006) suggest that these skills are called “functional cognitive
skills” because of the requirement to use one’s cognitive ability to “read, count, and
calculate” (pg. 25). Additionally, Wagner et al. found that these skills were strongly
linked to a child’s academic performance.

Dependent Variables
Academic achievement. Academic achievement in reading is the outcome of

interest in this study. Three different types of achievement variables will be examined by

49

running the model separately for each type: grades in language arts, standardized
assessment in reading, and curriculum-based assessment. Previous research on academic
achievement in students with ADHD has utilized both grades and standardized
assessments as a measure of academic achievement (DuPaul et al., 2004; Volpe et al.,
2006). In some cases, these different criterion have resulted in different outcomes, which
provides the rationale for running models separately with each of these outcome
measures. Previous research has suggested the use of CBAs as an outcome measure or
predictor, as they may be a more direct measure of academic skills (DuPaul et al., 2004).
The variable to define grades that will be used in the current study is the teacher reported
grade in language arts, which can range from 1 to 4, with 1 indicating “Mostly As/As and
Bs”, a 2 indicating “Mostly Bs/Bs and Cs”, a 3 indicating “Mostly Cs/Cs and Ds”, and a
4 indicating “Mostly Ds and/or Fs”.

The variables used to describe a standardized measure of reading achievement is
composed of the standard score from the Letter-Word Identification and the Passage
Comprehensions subtests from the Woodcock Johnson-Third Edition (WJ-III) research
edition (Woodcock, McGrew & Mather, 2001). The Letter —Word Identification subtest
requires a student to read isolated letters and words of increasing difficulty. The items
are arranged in order of difficulty, beginning with the easiest items. The Passage
Comprehension subtest requires an individual to read a short passage and to identify a
missing word. The correct word choice or choices fit the context of the passage. The
items are arranged in order of difficulty, beginning with the easiest items. For both the
Letter-Word Identification and Passage Comprehension measures, the W score will be

used. A W score is a transformation of the Rasch ability scale (Woodcock, McGrew &

50

Mather). The W scale is centered at 500, which is the approximate performance of a 10
year old. Reliability and validity evidence for the use of the WJ-III subtest is strong and
is outlined in the manual. The median reliability coefficient alpha for the Letter-Word
Identification subtest was .94, and for the Passage Comprehension subtest is .88. Test-
retest reliabilities are higher for the Letter-Word Identification subtest and are above .90
for all age groups within a one year or less retest interval. For the Passage
Comprehension subtest all values were above .80. As far as validity of the WJ-III,
growth curves and confirmatory factor analysis support its use as a measure of
achievement.

Finally, the CBA measure utilized in this study is the Oral Reading Fluency
measure from the Standard Reading Passages (Marston & Deno, 1986). This requires a
child to read a short passage in a set amount of time. This is a measure of the rate, or
fluency, at which a student can progress through a passage. The words read correctly
score will be used. This is calculated by subtracting the number of errors from the
number of words read. Two passages were read at each wave, and the average words
read correctly across these two passages at each wave will be used. The technical
evidence for CBM has been developed by a number of authors. For example, Demo and
colleagues (1982) found that ORF correlates highly with standardized measures of
reading achievement with validity coefficients ranging from .73 to .93. Since that time
the validity of CBM has been further established to relate to a number of standardized
measures, include high-stakes testing. As far as reliability, this has been assessed in a
number of studies on CBM. Marston (1989) summarizes a number of studies on the

reliability of CBM, and reported reliability coefficients ranging from .82 to .97, with

51

most above .90. He also reports that parallel form estimates ranged from .84 to .96, with
most above .90. Inter-rater reliability estimates were .99. This evidence supports their
use in the present study.

A table of correlations between the variables Was created and is shown in Table 4.
Data Preparation and Screening

Missing data. Tabachnick and Fidell (2001) suggest that if only a small number
of data are missing from a large data set and that they are rrrissing in a random pattern,
the problem is likely not serious. However, based on the large sample and longitudinal
nature of this study, some of the variables in the current study have a large number of
missing data. The missing data was imputed, using the multiple imputation. Specifically
the expectation maximization method was used to impute the data using LISREL 8.80.
The variables of interest both with and without data imputed can be seen in Tables 2-3.

Assumptions. Several assumptions were assessed prior to the analyses. These
include sample size, multicollinearity and singularity, outliers, normality, linearity, and
homoscedasticity (Tabachnick & Fidell, 2001; Kline, 2005). The methods to check for
these assumptions are described below.

Sample Size. Tabachnick and Fidell (2001) recommend the following equation

for determining requirements for sample size. They suggest that N> 50 + 8m,

where m=the number of independent variables. The final sample size in the

present study of 1019 (ADHD=389; No-ADHD=630) is well above this criteria.

Multicollinearity & Singularity. Multicollinearity occurs when the variables are

highly correlated or above r>.85 (Kline, 2005). A correlation matrix of all

independent variables can be seen in Table 4. None of the variables were highly

52

correlated in the present study. Singularity is defined as occurring when one
independent variable is the amalgamation of other independent variables. In the
current study, this was not problematic.

Outliers. Because many statistical analyses are sensitive to outliers, screening for
univariate and multivariate outliers was conducted. Univariate outliers are very
large standardized scores (Tabachnick & Fidell, 2001). Tabachnick and Fidell
suggest that univariate outliers can be defined as standardized residual values
outside of the range of -3.29 to 3.29. In a large sample, these authors suggest that
a few scores in excess of these values are expected. Univariate outliers can also
be examined graphically using a histogram, where case or cases that are
unattached to the rest of the distribution as a means for identifying an outlier.
Based on this method, several cases were identified that exceeded the criterion
(i.e., above 3.29 or below-3.29), however, just those cases that had a value outside
this criterion on two or more of the variables were deleted. Variables with just
one value above or below the criterion were kept in the analysis. This meant that
six cases were deleted, and eight cases remained with standardized residual values
outside of the -3.29 and 3.29 range. Multivariate outliers are cases with an
unusual group of scores (Tabachnick & Fidell). To find multivariate outliers,
Mahalanobis distance was calculated for each case. Tabachnick and Fidell
suggest that the criterion for multivariate outliers is a significant Mahalobonis
Distance at p<.001. As such, with 12 dependent variables, four cases falling
outside of the critical value (32.91) were deleted. The final sample consisted of

1019 students.

53

Normality. Tests of univariate normality and multivariate normality were
examined through SPSS. Multivariate normality is defined by Tabachnick and
Fidell (2001) as “the assumption that each variable and all linear combinations of
the variable are normally distributed” (p. 70). Using the Komogorov-Smirnov
statistic, the only variable that was normal was the ORF variable at time 1. For the
remainder of the variables this value was not significant, suggesting violation of
the normality assumption. However, this can be common in larger samples
(Pallant, 2007). Skewness and kurtosis are both components of normality and
were assessed. A value of 0 would suggest a normal distribution (Pallant). The
skewness value furthest away from 0 was -l.258 , and the kurtosis furthest from 0
was 3.655. Kline (2005) suggests that skewness values greater than 3.0 and
kurtosis values greater than 10.0 are problematic. Therefore, these are not
problematic in the current study. Also the shape of the distribution can be
examined using histograms, Normal Q-Q plots, and Detrended Normal Q-Q plots.
Examination of these plots revealed distributions falling within the normal
parameters. Based on these analyses, there is some evidence of non-normality
within the current data set.

Linearity & Homoscedacity. Homoscedacity and linearity are components of
normality. Linearity is the assumption that the dependent variable scores should
have a linear relationship with the residuals (Tabachnick & Fidell, 2001).
Homescedacity refers to the idea that the variance of the residuals around the
dependent variables should be the same for all predicted scores (Tabachnick &

Fidell, 2001). Linearity and homoscedacity were examined using scatterplots

54

between pairs of dependent variables for each analysis and the plots showed some

evidence of violations of these assumptions. For this analysis, the file was also

split for analysis of non-ADHD and ADHD groups.
Analytic Techniques

The present study utilized several statistical techniques to answer the research
questions. Descriptive statistics were calculated for each of the measures of academic
achievement (ORF, grades, standardized measures), as well as for the other variables of
interest. Latent growth modeling and ANOVA were the primary methods for data
analysis to delineate the relationships between various predictors of academic
achievement within a special education population. Data preparation and analyses were
conducted using SPSS 15.0, LISREL 8.80, and MPlus 4.21.

Demographic Analyses

Before conducting the analyses of interest, demographic information about the
groups is described. The means and standard deviations of the independent and outcome
variables are presented. Additionally, the differences between the ADHD and no-ADHD
groups on these variables were assessed through independent samples t-test for each
variable. These are shown in Table 2 and Table 3. The correlations for the variables are
provided in Table 4.

Analysis of Variance

The initial research question examines the mean differences within the
populations of interest in this study. A Mixed Between-Within Design analysis of
variance (ANOVA) examines both between group (ADHD vs. non—ADHD) and over

repeated measures (Wave 1 - 3). This technique examines mean differences in academic

55

outcomes across ADHD vs. non-ADHD groups and over time. Additionally, the analysis
was run separately across the four different types of academic achievement measures:
language arts grades, standardized assessments of letter-word identification, standardized
assessment of comprehension, and a curriculum-based assessment of oral reading
fluency. In addition to examining the main effects of this analysis, interaction effects
were investigated. This analysis was conducted using SPSS 15.0 statistical software.

Latent Growth Modeling

The complexity of the relationship between the variables of interest provides
rationale for using latent growth modeling (LGM). LGM is a method of analysis within
the framework of structural equation modeling (SEM). Specifically, a level and shape
model was analyzed. In order to best understand the hypothesized relationships between
ADHD, student classroom behaviors, school based intervention, and academic outcomes,
the model shown in Figure 6 was analyzed quantitatively using LGM. This method
allows for the evaluation of the adequacy of this theoretical model in explaining the data
from the SEELS, or from a national sample of children who are receiving special
education services. One rationale for using SEM is that it is an especially appropriate
analysis technique for non-experimental data such as the SEELS (Keith, 1999).
Additionally, SEM also offers the advantage of describing relationships free from
measurement error, as this is something that is specifically estimated (Tabachnick &
Fidell, 2001). It also allows one to examine the relationship between the variables in the
context of other variables of interest.

The hypothesized model can be seen in Figure 6, and is a representation of the

hypothesized relationships between the variables. Latent variables, or unobserved

56

variables, are represented in the diagram as ovals. Measured variables are represented as
rectangles. The lines between the variables represent the hypothesized relationships.
Lines with one arrow indicate a hypothesized direct relationship between variables.
Lines with a double arrow indicate an unanalyzed relationship. Rather, these variables
are hypothesized to be correlated. Two arrows, one in each direction between two
variables indicates a reciprocal or non-recursive relationship. The absence of a line
indicates that no relationship has been hypothesized.

Model specification. Model specification refers to using available theory and
research to develop a hypothesized model or relationships between variables. This
hypothesized model for the current study is shown in Figure 6. The primary variables of
interest will include manifest, or observed, variables. These include Demographic
Variables (Student Grade, Gender, Ethnicity, Household Income, Mother’s Education,
Reading Goals & Functional Skills), Student Variables (Total Extemalizing behavior,
Total Student behavior, Stimulant Medication Status), Intervention Variables (Behavioral
Intervention & Academic Intervention), and Reading Outcomes (Reading grades, Letter-
Word Identification standard score, Passage Comprehension standard score & ORF
average score).

In addition to the manifest variables, the model will also include two latent
variables. A model utilizing latent change analysis will be assessed. Specifically, a level
and shape model would be assessed, which provides information on the correlates and
predictors of change (Raykov & Marcoulides, 2006). The shape factor is a latent variable
that provides information on the overall ability change (Raykov & Marcoulides, 2006).

By looking at the relationships between the predictor variables and the shape factor,

57

information about the predictor variables and their relationship with overall change can
be assessed. Several pieces of information can be gathered by examining the quantitative
results of such a model. If the shape factor is positive, indicating that there is a positive
mean change, and the relationships between the shape factor and other predictor variables
are positive, individuals with high values are those who improve the most, while those
with lower values show less improvement. If the relationships are negative, individuals
with high values on the predictors are those who decline the most. Additionally, using a
level and shape model across populations allows for the exploration of differences in
variability and means of the starting level and change over time across the groups, as well
as whether the demographics, behavior, and intervention variables of interest in the
current study have equal predictive power in the change over time in reading across the
groups (Raykov & Marcoulides, 2006).

The inclusion of the variables in the current study is based on both theoretical and
empirical conceptions. The primary variables of interest can be grouped into three types:
demographic variables, intervention variables, and behavior variables. These groupings
are an effort to simplify the model and do not have a bearing on the analysis.

Model estimation. MPlus 4.21 (Muthén & Muthe’n, 2006) statistical software will
be utilized to quantitatively examine the hypothesized model. A raw data file will be
used for this analysis. While a multiple group model was initially proposed, differences
between the ADHD and non—ADHD groups were found, and subsequent analyses
included both single and two group analyses. Robust maximum likelihood (MLR)
estimation will be used for the continuous variables to estimate the model parameters.

Based on violations of normality, linearity, and homoscedacity found within the current

58

data set when the normality assumption was assessed, there is support for using this
robust method. Estimation of the categorical outcome variable (reading grades) will be
estimated using Unweighted Least Squares estimation.

Weighting. Because the sample was stratified, weights were included in the
analysis. This takes into account the fact that there was not an equal probability of
selection. Asparouhov (2005) suggests that if this is not taken into account, the
parameter estimates may be biased. The SEELS data set has a weight for each
instrument, and they recommend using the weight for the instrument with the least
number of respondents. SRI (1999) report that the “student sampling weights are the
product of the LEA sampling weights and the inverse of the student sampling fraction”
(pg. 4-19). In the current study, the weight from the direct assessment at wave three was
used as an initial starting point, and then the sampling weight for those with data missing
was filled in from the direct assessment at wave one and two and from other instruments
as needed. Using Mplus 4.21, all of the models were analyzed taking this information
into account by using the SEELS weight.

Adequacy of Model. Several kinds of information regarding the adequacy of the
hypothesized model are available (Joreskog & Stirbom, 1996-2001). Several indices will
be used to determine the goodness of the fit of the model in the current study. The first
fit index that will be examined is the Comparative Fit Index. A value greater than 0.9 is
generally considered adequate and will be utilized for this the current study (Kline,
2005). The model Chi—square will also be reported. Ideally, a non—significant Chi—
square is indicative of a well fitting model. However, the chi-square value is sensitive to

sample size Kline (2005). For this reason, the chi-square will be reported, but will be

59

considered with the other data to determine model fit. Finally, the Root Mean Squared
Error of Approximation (RMSEA) will be reported. Kline reports that a value 5 0.05
suggest a close fit, while values between .05 and .08 indicate an adequate fit. However,
some authors have suggested that a more relaxed criteria is appropriate (Hu & Bentler,
1999), and recent studies have utilized a criteria of .10 as indicative of adequate fit
(Baker, Smolkowski, Katz, Fien, Seeley, Kame’enui et al., 2008). Therefore, the criteria
of .10 will be utilized for the present study.

When working with a two group model the likelihood ratio test (LRT) was
applied to test if restrictions are feasible. From this information, one can ascertain
whether restrictions are equal across groups, and thus obtain information on the similarity
and differences in the groups of interest. Because the MLR estimation is used a corrected
chi-square difference test (CSCD test) must be used. This formula is similar to the LRT
calculation, however, a correction factor generated by the MPlus 4.21 software is used to
‘ correct the values. The formula for this test requires you to divide the difference between
the restricted and full chi—square values (multiplied by their correction factors generated
by MPlus) by the difference between the degrees of freedom from the restricted to the
full model (also multiplied by the correction factor). This product is then multiplied by
the difference between the degrees of freedom without the correction factor (Raykov,

2007).

60

Chapter 4
Results

Descriptive Statistics

The ADHD and no-ADHD groups were compared across several characteristics
including grade, gender, household income, ethnicity, mother’s education, student
behavior, stimulant use, and disability status. This analysis was first run without the
missing datum imputed. The results of this unweighted analysis can be seen in Table 2.
A second analysis was run with all of the datum imputed and the results of this analysis
can be seen in Table 3. Independent samples t-tests between the ADHD and no-ADHD
groups were done to determine if there is a significant difference in the means scores
across the groups. With a majority of the variables, equal variance was not assumed.
The grade level, age, functional cognitive skills, reading goals, and acaderrric intervention
variables at wave one were not significantly different. This was true with both the
original data set and for the imputed data set. There were significant differences in terms
of gender, ethnicity, income, mother’s education, medication status, externalizing
behavior, student behavior, and behavioral intervention at wave one between the ADHD
and no-ADHD groups in both data sets. Additionally, all of the differences between
membership in the four disability groups that were included in this study (SED, LD, OHI,
SLI) were significant between the ADHD and no-ADHD groups. Given these differences
between the groups, the model was run as both a single group model (ADHD only) and
as a two group model.

Correlations between all the variables of interest were computed and are provided

in Table 4. The means and standard deviations for the outcome variables are shown in

61

Table 5. Additionally, an independent samples t-test was run between the ADHD and
no-ADHD groups comparing the scores. All of the scores were significantly different
between the groups with the ADHD group scoring consistently higher on all of the
outcome measures. The one exception was for ORF at wave three. The difference at this
point between the ADHD and no-ADHD group was not significant.

Research Question I

This set of research questions examined the mean differences over time in reading
for the children with a parent and school report of ADHD versus those without a parent
and school report of ADHD. The results are organized by the outcome measure.

Oral Reading Fluency. There was a significant effect for time, Wilks Lambda =
.27, F(2,1016)=1402.94, P<.0005, multivariate eta squared =.73. This indicates a large
effect size. Additionally, the main effect for group was significant F(l)=7.92, P=.005.
There was also a significant interaction effect, Wilks Lambda = .98, F(2,1016)=O.84,
P<.0005, multivariate eta squared =.02. This indicates that the change in scores over
time is different for the two groups. Figure 2 shows the interaction effect for ORF. While
the children in the sample with ADHD started at a higher rate for ORF, the students in the
no—ADHD had a faster rate of growth from time one to time three.

Letter Word Identification There was a significant effect for time, Wilks
Lambda = .28, F(2, 1016)=15.30, P<.0005, multivariate eta squared =.72. This indicates a
large effect size. The main effect for group was significant F( l )=l8.49, P<.0005,
multivariate eta squared =.02. This indicates a small effect size. The interaction effect
was not significant, Wilks Lambda = 1.00, F(2, 1016)=1.03, P=.357. This indicates that

the change in Letter Word Identification scores over time was not significantly different

62

for the ADHD and no-ADHD groups. In summary, the ADHD group had higher scores
on the LW task, and no significant differences in growth over time were noted between
groups.

Passage Comprehension. There was a significant effect for time, Wilks Lambda
= .51 l, F(2, 1016)=485.60, P<.0005, multivariate eta squared=.883. This indicates a
large effect size. The main effect for group was significant F(l)=12.33, P<.0005,
multivariate eta squared = .01. This indicates a small effect size. The interaction effect
was not significant, Wilks Lambda = 1.00, F(2, 1016)=1.327, P=.226. This indicates that
the change in Passage Comprehension scores over time was not significantly different
across the ADHD and no-ADHD groups. Overall, the ADHD group had higher scores on
the PC task, and no significant differences in growth over time were noted between
groups.

Reading Report Card Grades. There was a significant effect for time, Wilks
Lambda = .98, F(2, 1016)=8.04, P<.0005, multivariate eta squared =.02. This indicates a
small effect size. The main effect for group was significant F(l)=25.29, P<.0005, partial
eta squared: .02. This indicates a small effect size. The interaction effect was not
significant, Wilks Lambda 2 1.00, F(2, 1016)=0.555, P=.574. This indicates that the
change over time in terms of reading report card grades is similar across the ADHD and
no-ADHD groups. The ADHD group had higher grades, and no significant differences in
growth over time were noted between groups.

In sum, reading scores were higher in ADHD group versus no-ADHD group
across all four reading outcomes. Only on the ORF task was the growth over time

different between the ADHD and no-ADHD groups, with the ADHD group showing a

63

slower rate of growth. For LW, PC, and reading grades, there was no interaction in
scores across groups over time.
Research Question 2

The second set of research questions examine how the variables of interest predict
initial reading skills and growth over time using a latent growth model. To answer this
question, growth over time without the impact of the hypothesized predictors of change,
is assessed. Three types of information are examined from the results of this modeling.
This includes whether the model provides a good fit to the data, which determines
whether the resulting parameters can be trusted. Next one can assess whether the shape
(or in the special case of linear fit the slope values) is significant, and finally whether
there are individual differences in the initial value and change over time.

Next the covariates of interest are added to the model. The overall fit of the
model with the hypothesized predictor variables included is first assessed. Next the
variables that are significantly predictive of starting value and growth over time can be
examined. If the relationship between the predictors and the overall change is significant
and positive, this suggests that individuals with a high value on the covariate will
improve the most (Raykov & Marcoulides, 2006). For covariates that are significant and
negative, high values are associated with a decrease in the growth factor. The results for
the models are presented below organized by outcome measure.

Single Group Models

Oral Reading Fluency. The Level and Shape model provided the best fit to the

model for the Oral Reading Fluency outcome. The overall model provided a strong fit to

the data, demonstrated by a Chi-square value of 1.359 (p: 0.507). The RMSEA value

64

was 0.00 and CFI=1.00. The mean starting level and the rate of change from time one to
time three was significant. Additionally, the starting value was significantly different
across individuals, as was the slope.

When the predictor variables were added to the model the fit was strong: Chi-
square value of 12.73 (p: .548). The RMSEA value was 0.000 and CFI=1.00. The
impact of the predictor variables on the initial reading skills and reading growth over time
is shown in Table 6. Several variables emerged as significantly predictive of the level, or
initial reading skills, on the ORF measure. These included the child’s grade and gender,
which both had a positive impact. The functional skills, student behavior, and academic
intervention had a negative impact on the initial reading skills. In terms of the impact of
the covariates on the shape factor, or the growth over time, the child’s grades were
negatively associated. The child’s reading goal was also negatively associated, which
can be interpreted as a higher goal associated with a higher rate of growth (a lower score
on the goal variable is more desirable). The child’s student behavior and behavioral
interventions were related to positive growth over time.

Letter Word Identification. The Level and Shape model provided the best fit to
the LW data. The overall model provided an adequate fit to the data, demonstrated by a
Chi-square value of 9.238 (p: 0.009). Additionally, the RMSEA value was 0.096 and
CFI=0.98. The mean starting level and the rate of change from time one to time three
was significant. Additionally, the starting value was significantly different across
individuals, however, the slope was not.

When the predictor variables were added to the model the fit was adequate: Chi-

square value of 35.05 (p: .001 ). The RMSEA value was 0.062 and CFI=0.97. The

65

impact of the predictor variables on the initial reading skills and reading growth over time
is shown in Table 7. Several variables emerged as significantly predictive of the level, or
initial reading skills, on the LW measure. This included the child’s grade, which had a
positive impact. The child’s household income and reading goal also had a significant
positive impact. The functional skills, student behavior, and academic intervention had a
negative impact on the initial reading skills. In terms of the impact of the covariates on
the shape factor, or the growth over time, the child’s grades were negatively associated.
The child’s behavioral interventions were related to positive growth over time.

Passage Comprehension. The Level and Shape model provided the best fit to the
data. This model provided a good fit, indicated by a Chi-square value of 9.490 (p =.023).
The RMSEA was equal to 0.075 and the CFI=.90. The mean starting level and the rate of
change from time one to time three was significant. Additionally, the starting value was
significantly different across individuals, as was the slope.

When the predictor variables were added to the model the fit was as follows: a
Chi—square value of 49.008 (p: .000), an RMSEA value of 0.076 and a CFI=0.903. The
impact of the predictor variables on the initial reading skills and reading growth over time
is shown in Table 8. Several variables emerged as significantly predictive of the level, or
initial reading skills, on the PC measure. This included the child’s grade which had a
positive impact. The mother’s education also had a significant positive impact. The
child’s ethnicity, functional skills, and the academic intervention had a negative impact
on the initial reading skills. Two variables were also related significantly to the Shape

factor of the PC model. This included the grade, which had a negative association with

66

growth. The implementation of academic intervention also had a positive association
with growth.

Grades in Language Arts. The Level and Shape model provided a good fit to the
data, however, the Intercept and Slope Model provides a stronger fit, indicating linear
change over time. The overall fit of the Intercept and Slope model was strong, as
indicated by an insignificant Chi-square value of .91 l (p = 0.634). The RMSEA was
equal to .000 and the CFI=1.00. The mean starting level and the rate of change from time
one to time three was not significant. The starting value was significantly different
across individuals, while the slope was not.

When the predictor variables were added to the model the fit was as follows: a
Chi-square value of .471 (p: .493). The RMSEA value was 0.000 and CFI=1.00. These
values suggest a strong fit. The impact of the predictor variables on the initial reading
skills and reading growth over time is shown in Table 9. Several variables emerged as
significantly predictive of level, or initial reading skills. The child’s functional skills had
a significant positive impact. The child’s student behavior had a negative impact on the
starting value. In terms of the Slope factor, or change over time, mother’s education had
a negative impact. The child’s student behavior had a significant positive impact on
change over time.

Two Group Models

A two group model comparing the reported ADHD and no-ADHD group was also
analyzed. This type of model allows for a comparison of the model across groups
regarding initial status and change over time, as well as to see if the covariates of interest

are equally predictive of change over time in reading achievement across groups.

67

Oral Reading Fluency. By fitting the model to the data with no cross-group
constraints, the fit statistics are adequate: Chi-square value of 13.18 (p = 0.0104). The
RMSEA was equal to 0.067 and the CFI=.98. When the restriction regarding group
identity in terms of variances and covariances between the level and shape factors were
applied the fit was as follows: Chi-Square=16.322 (p=.0223), RMSEA=.051, and
CFI=.983. The resulting LRT was equal to 2.43 (df=3), which was not significant. This
suggests that there were no group differences in individual variability on initial starting
position and change in reading scores over time. Additionally, there are no group
differences in initial true starting position variability, as well as true change over time.
When the restriction of equal means in starting value and change over time across groups
is applied, the fit of the model is adequate. This includes a Chi-square value of 20.967 (p
= 0.0128). The RMSEA was equal to 0.051 and the CFI=.978. The LRT was not
significant (LRT=6.89; df=5), which implies that there are were not significant group
differences in terms of mean initial starting point and change over time.

When the predictors are added to the model for the ORF data without any
constraints, the model provides an adequate fit to the data: Chi-Square = 57.042
(p=.0010), RMSEA = 0.045, and CFI = .986. The predictors for the ADHD versus the
ADHD group were similar. For the ADHD group and gender were positively correlated
with starting value in reading, while functional skills, student behavior, and academic
intervention were negatively associated with the initial starting value. For children with
no report of ADHD, grade and gender were again positively predictive, and functional
skills and academic intervention were negatively predictive of initial reading skills.

Additionally, mother’s education was also positively predictive of initial reading skills.

68

In terms of growth over time, for the ADHD group, grade and reading goal had a
negative impact on growth. Student behavior and behavior intervention both had a
significant positive impact. For the no-ADHD group, grade had a negative impact. For
this group, the only other significant predictor was income, which had a positive impact
on reading growth. No other variables were significantly related to growth over time.

In order to test whether there were significant differences between the groups in
terms of the impact of the covariates, this restraint was applied and the resulting fit
statistics were as follows: Chi-Square = 91.656 (.0006), RMSEA=.039, and CFI=.981.
The resulting LRT (32.84, df=24) was not significant, indicating no significant
differences between the ADHD and no-ADHD groups in terms of how the initial reading
skills and growth over time was impacted by the covariates of interest.

Letter- Word Identification. By fitting the model to the data with no cross-group
constraints, the level and shape model provides a good fit to the data. The overall fit of
the Level Shape model was adequate, as indicated by an insignificant Chi-square value of
13.162 (p = 0.0105). The RMSEA was equal to 0.067 and the CFI=.992. When the
restriction regarding group identity in terms of variances and covariances between the
level and shape factors were applied the fit is as follows: Chi—square =19.792 (p =
0.0060). The RMSEA was equal to 0.060 and the CFI=.988. The resulting LRT is not
significant (6.64; df=3) indicating that there is not a significant difference between the
ADHD and no-ADHD groups in terms of individual variability in starting position and
change over time. Additionally, there were no group differences in initial true starting
position variability, as well as true change over time. When the restriction of equal means

in starting value and change over time across groups is applied, the fit is as follows: a

69

Chi-square value of 20.954 (p = 0.0129). The RMSEA was equal to 0.051 and the
CFI=.99. The LRT was not significant (LRT=7.91; df=5), which implies that there are
no significant group differences in terms of mean initial starting point and change over
time.

When the predictors are added to the model for the LW data without any
constraints, the model provides an adequate fit to the data: chi-square = 68.595
(p=.0003), RMSEA = 0.046, and CFI = .985. The predictors for the ADHD versus the
ADHD group were similar. For the ADHD group, grade and income were positively
correlated with starting value in reading, while functional skills, reading goals, and
academic intervention were negatively associated with the initial starting value. For
children with no report of ADHD, the same predictor was significant in the same
direction, except the impact of income was not significant. However, mother’s education
was significant on the initial starting value for the no-ADHD group. In terms of growth
over time, for the ADHD group, grade and ethnicity had a negative impact on growth.
Behavior intervention both had a significant positive impact. For the no-ADHD group,
grade had a negative impact. For this group, medication had a significant positive
impact. No other variables were significantly related to growth over time.

In order to test whether there were significant differences between the groups in
terms of the impact of the covariates, this restraint was applied and the resulting fit
statistics were as follows: Chi-Square = 94.353 (.001), RMSEA=.037, and CFI=.983.
The resulting LRT (25.09, df=23) was not significant, indicating no significant
differences between the ADHD and no-ADHD groups in terms of how the growth over

time was impacted by the covariates of interest.

70

Passage Comprehension. The Level and Shape model provides a good fit to the
data, when no cross group constraints are imposed. The overall fit of the Level and
Shape model was adequate, as indicated by a Chi-square value of 22.135 (p = 0.0011).
The RMSEA was equal to 0.073 and the CFI=.946. When the restriction regarding group
identity in terms of variances and covariances between the level and shape factors were
applied the fit is as follows: Chi-square value of 24.551 (p = 0.0035). The RMSEA was
equal to 0.058, and the CFI=.948. The resulting LRT was not significant (3.56; df=3)
indicating that there is not a significant differences between the ADHD and no-ADHD
groups in terms of individual variability in starting position and change over time on
passage comprehension. Additionally, there were no group differences in initial true
starting position variability, as well as true change over time. When the additional
restriction of equal means in starting value and change over time across groups is applied,
the fit of the model was adequate with a Chi-Square value of 27.236 (p = 0.0042). The
RMSEA was equal to 0.054, and the CFI=.945. The LRT was not significant
(LRT=5.67; df=5), which implies that there are not significant group differences in terms
of mean initial starting point and change over time.

When the predictors are added to the model for the PC data without any
constraints, the model provides an adequate fit to the data: Chi-Square = 107.553
(p=.0000), RMSEA = 0.064, and CFI = .937. The predictors for the ADHD versus the
ADHD group were similar. For the ADHD group, grade, mother’s education, and
student behavior were positively correlated with starting value in reading, while ethnicity,
functional skills, and academic intervention were negatively associated with the initial

starting value. For children with no report of ADHD, the same predictors were

71

significant in the same direction. In terms of growth over time, for the ADHD group,
grade had a negative impact on growth, and academic intervention had a positive impact.
For the no—ADHD group, only grade had a significant negative relationship with the
latent shape factor. No other variables were significantly related to growth over time. In
order to test whether there were significant differences between the groups in terms of the
impact of the covariate, this restraint was applied and the resulting fit statistics were as
follows: Chi-Square = 135.258 (p=.000), RMSEA=.051, and CFI=.932. The resulting
LRT (24.20, df=23) was not significant, indicating no significant differences between the
ADHD and no-ADHD groups in terms of how the growth over time was impacted by the
covariates of interest.

Grades in Language Arts. The chi-square difference test cannot be used with the
ULS estimation method (Muthén & Muthe’n, 2006). Therefore, only the model with the
predictors without cross—group constraints will be provided. When the predictors are
added to the model for the report card data without any constraints the model provides a
strong fit to the data: chi-square = 1.472 (df=3; p=.6887), RMSEA = 0.000, and CFI =
1.000. The predictors for the ADHD versus the ADHD group were similar. For the
ADHD group functional skills were positively correlated with starting value in reading,
while student behavior was negatively associated with the initial starting value. For
children with no report of ADHD, only student behavior was negatively associated with
the starting value. In terms of growth over time for the ADHD group, student behavior
had a positive impact on growth. Mother’s education had a significant negative impact.

For the no-ADHD group, only student behavior had a positive impact on growth.

72

Chapter 5

Discussion
Mean Reading Growth

The first research question asked whether or not there were mean differences over

time and across ADHD and no-ADHD groups on several measures of reading outcomes.
This research question also asked whether or not there was an interaction between time
and ADHD status. It was hypothesized that there would be mean differences over time.
This was true for all outcome measures. It was also hypothesized that there would be
mean differences between the groups with the no-ADHD population performing
significantly higher than the ADHD population. This hypothesis was incorrect, as the
ADHD group performed significantly higher on all four of the outcome measures. This
is despite no significant differences across the groups in terms of their goals for reading
and in terms of their functional cognitive skills. There were a number of demographic
differences between the groups. These included the ADHD group having more males,
more white students, higher household income, higher mother’s education, higher levels
of stimulant medication use, more externalizing behavior, and fewer student behaviors.
Additionally, the distributions of the special education disability categorizations across
the groups was dissimilar. The ADHD group included more students with a Severe
Emotional Disturbance and Other Health Impairment, while the no-ADHD group had a
greater percentage of students with a Learning Disability and a Speech and Language
Disability. Schnoes and colleagues (2006) also used the SEELS sample, and although
they relied on slightly different inclusion criteria their distributions of students with

ADHD in each category in this study was similar.

73

One explanation for the higher performance of the ADHD group is the fact that
the no-ADHD population included a significantly more students with learning
disabilities, a difficulty which is defined by academic underachievement. This is a
distinction of the current sample from most prior research on the topic, which compares
children with and without ADHD, regardless of their special education eligibility status.
Due to the increased comorbidity between ADHD and LD, an ADHD group typically has
more students with LD than the non-ADHD comparison group (ex. DeShazo Barry et al.,
2002). However, when the students were compared to other special education students,
this was not the case.

It was also hypothesized that there would be an interaction between these
relationships with the no-ADHD population showing a steeper rate of growth in reading
performance. This was untrue for the norm-referenced, standardized measures of reading
achievement (i.e. LW & PC) and for the reading grades. However, this hypothesis was
supported with the ORF data. The ADHD students started at a higher level, however, the
progress over time was less steep than for the no—ADHD group. This may be in part due
to the fact that curriculum-based assessment is designed to be sensitive to change than the
other measures of reading achievement in this study (Shinn, 1999). One explanation is
that because this assessment is sensitive to change, the ORF is picking up on a more
subtle pattern than the other norm-referenced measures. An alternative explanation is
that fluency is the only skill exhibiting this particular pattern, and is therefore an area of
particular dysfunction for children with ADHD. This explanation has not been supported

in previous research, which does identify reading as an area of difficulty for children with

74

ADHD (DeShazo Barry et al., 2002), but has identified reading comprehension as an area
of particular difficulty (Ghelani, Sidhu, Jain & Tannock, 2004).

In addition to exploring the interaction between the groups, it is also informative
to examine the actual increase in reading fluency, as measured ORF, over the course of
several years by the students in this sample. The time line of the SEELS direct
assessment included one year between wave one and wave two data collection, and two
years between wave two and three data collection (SR1, 1999). Deno and colleagues
(2001) contrasted growth rates in samples using typical instructional practices versus
effective instructional practices. They found that typical instructional practices for
students in special education yielded gains of less than 1 word per week in first grade,
and around 0.6 words per week in 2"d through 6‘h grade, while potential growth using
effective practices was 2 words per week in 1" and 1 word per week in 2nd through 6th
grades. When these standards are compared to the results in the current study, the
findings are grim. Although the data in the present study is aggregated across grades and
children , the ADHD group on average increased from 93 words per minute at wave one
(data collected between 2/01-5/01) to 108 words per minute at wave two (data collected
between 2/02-5/02). This is a gain of approximately 15 words over the course of one full
academic year. In addition, from wave two to three (data collected between 2/02—5/02)
the students gained approximately 19 words over two school years. These rates of growth
are much closer and below those described by Deno when typical and not effective
strategies are used, and suggest that the rate of growth for the children in this study are

well below their peers not eligible for special education.

75

To further elucidate why the rate of growth appeared less steep for children with
ADHD than for those with no report of ADHD in this study, it would be helpful to break
these results down by grade level in future studies. Previous research has suggested that
children with ADHD tend to experience increasing difficulty as they progress through
school due to increased demands (Barbaresi et al., 2007). This phenomenon may
underlie the findings in this study. Also interesting to consider would be how the rate of
growth compares to other children not in special education, including those with ADHD.
Previous research has revealed that the rate of growth for children in general education is
approximately twice that of children in special education (Deno et al., 2001). This
consideration makes it even more concerning that the children with parent and school
reported ADHD in this study progressed a slower than typical rate for children in special
education on a measure of ORF. Further exploration of the development of reading skills
over time for children with ADHD, particularly those eligible for special education, is
warranted.

Modeling of Reading Growth

An examination of these growth patterns within a more flexible modeling
technique was carried out in the current study to provide more insight into initial reading
level and reading growth over time. Specifically, the research questions asked about how
the introduction of demographic, student, and intervention variables impact the initial
reading level and growth in reading over time. Also of interest was how this varied
between outcome measures. Several interesting findings emerged from the second set of
research questions. A number of demographic, student, and intervention variables were

significantly predictive of initial reading level and change over time.

76

Demographics. While a number of variables related to the student’s initial
reading status, fewer of the demographic variables related to the rate of reading growth
over time. This is a promising finding, as it points to the important role of the school in
supporting student reading growth over and beyond the impact of demographic variables.
Those demographic variables that were significantly related to children’s initial reading
factor (with the direction in parentheses), included grade level (+), gender (+), household
income (+), functional skills (+ and -), reading goal (-), mother’s education (+), and
ethnicity (-). Only grade level and functional skills were significant across more than one
measure. Grade level was predictive of higher reading scores across three measures
(LW, PC, and ORF). This is logical, given that the score used for each of these measures
is not referenced to the child’s age or grade level and increases as skill increases.
Therefore, these results suggest that the higher the grade level, the higher the skill level-
However, more important to consider is the rate of growth over time, which is a strength
of a longitudinal analyses. This will be discussed in more depth in the next section.

Functional Skills had a less consistent relationship. They were significantly
related to the initial status on ORF, LW, and PC in the negative direction, while they
were significantly positively related to the initial status on grades. It would be expected
that these skills are important to reading both at the initial level and to growth over time,
and that higher functional skills would relate to increased scores. Previous research using
this same scale of functional cognitive scales, suggested that these were related to
achievement (Wagner et al., 2006). The functional skills were based on parent ratings
and in the present study there was consistency between parent’s perceptions of these

skills and teacher’s initial grades. However, there was a relationship in the opposite and

77

unexpected direction to the skill-based measures (ORF, LW & PC). Traditional,
standardized and norm-referenced cognitive assessments are correlated moderately with
academic achievement. For example, Sattler (2001) reports that across a number of
studies the WISC-III Full Scale IQ was correlated at.65 with the Reading Composite
from the WIAT, at .65 with the Reading from the WRAT—3, .53 with the Reading
Composite from the K-TEA, and at .48 with reading grades. A good possibility is that
the functional cognitive skills used in the current study are measuring something distinct
from skills assessed on a cognitive assessment, such as the WISC-III, as this variable has
previously related significantly to academic outcome measures in the expected direction.
This difference may underlie the unexpected finding in this study. Alternatively, the lack
of relationship in the expected direction of the functional skills and measures of academic
achievement, may suggest a distinction in this population of children from those used in
the standardization samples of the tests discussed above, that warrants further
exploration.

In regards to growth over time, several factors were related to growth across the
four measures. These included grade (-), mother’s education (-), and reading goal (-).
Grades were significantly related in the negative direction to the ORF, LW, and PC tasks.
Grade level did not negatively impact the reading grades growth. Previous research has
suggested that there is a decline in achievement over time for children with ADHD, due
to increased cognitive and organizational demands (Barbaresi et al., 2007). Previous
research has also demonstrated that on measures of ORF that the rate of growth decreases
over time (Hasbrouck & Tindal, 2006). In addition, the commonly described “Matthew

Effect”, which suggest that strong readers continue to get stronger, while poor readers

78

continue to struggle (Stanovich, 1986), may be applied here. Individuals who are poor
readers in early elementary tend to continue to be poor readers in later elementary and
beyond (Juel, 1988; Shaywitz, et al., 1992; Francis et al., 1996). This speaks to the need
for early intervention to improve growth over time. This contrasts with the traditional
special education identification model of reading disability (or any specific learning
disability), which required a “severe discrepancy” between ability and achievement,
which often is not present in children in early elementary grades and therefore may delay
prevention or early intervention efforts (NASDSE, 2006). This perpetuates a “wait to
fail” system; however, recent changes in the law no longer require this discrepancy
(IDEIA, 2004). This change creates opportunities for schools to implement alternative
methods for supporting children who are not achieving at grade-level that can be utilized
earlier.

Mother’s education also had a significant negative impact on the growth latent
factor for reading grades only. In other words, the higher the mother’s education, the
lower the rate of growth on reading grades. Mother’s education is one commonly used
factor to measure socioeconomic status. Socioeconomic status has been reported to play
a role in academic achievement (DuPaul et al., 2004), however, in the current study the
finding was in the opposite direction of that which would be expected. Follow-up to
determine distinction of the impact of mother’s education on grades versus skill-based
measures would be helpful. Having a higher reading goal only influenced the ORF
measure. This may again be related to the sensitivity of the ORF measure to assess
change. This also speaks to a larger point made in the literature, which calls for the

combination of CBM and specific and measureable goals. A concern that traditional IEP

79

goals are vague, not measureable, and not related to current achievement has been voiced
(Shinn, 2002).Research that has suggested that the use of CBM in writing (IEP) goals has
led to improved academic outcomes (Shinn). In fact, the effect size for the use of CBM
in writing goals, has exceeded .5 (Shinn). This study supports the notion that CBM may
be a helpful and valuable tool for measuring growth toward academic goals.

Student Variables. The impact of externalizing behavior, student behavior, and
medication status was considered under the heading of “student variables”. Higher
scores on the “student” behavior scale were significantly related to lower initial reading
scores for the ORF, LW, and reading grades. This was unexpected, as it would be
expected that these skills would relate to a higher or neutral initial score, as opposed to a
decrease. However, this same variable was linked to positive growth on reading grades
and ORF. A previous study comparing the impact of the “academic enablers” on grades
and skill measures found that these only impacted grades (DuPaul et al., 2004). The
results of this study suggests that improved student behavior, positively impacts teacher’s
ratings of their grades, and also relates to some skill-based measures. This finding raises
a number of questions. It might be expected that improved persistence, following of
directions, and homework completion would relate to improvements in skills on
standardized measures of reading achievement. However, this conclusion has not been
supported in previous research (DuPaul et al.), nor consistently in this study.
Alternatively, these types of behaviors do improve reading grades, which are given by the
teacher, who may be influenced by positive behavior on behalf of the student.
Unfortunately, this may not necessarily translate into improved skills. A unique finding

in the present study was that these skills also had a positive impact on one skill-based

80

measure. A previous study that explored the impact of academic enablers on academic
achievement did not use CBAs as an outcome measure (DuPaul et al., 2004). The finding
in the current study may highlight another area of distinction between traditional, nonn-
referenced measures of academic achievement and CBA. An alternative explanation is
that the impact of these academic enablers could be different for fluency, than for
measures of letter and word reading and reading comprehension. Follow-up to this
pattern could highlight not only potential intervention targets, but could also provide
insight into ways to monitor progress in this area.

Extemalizing behavior and medication status were not linked to improvements
across any of the outcome measures. This finding is not consistent with previous
research that links increased ADHD behaviors to decreased academic outcomes
(DeShazo Barry et al., 2002), and medication use to moderate improvements in reading
outcomes (Ryan et al., 2005). Possible explanations could have to do with lack of
variability on the externalizing measure, as the range of possible scores was small. It is
unclear why the impact of medication was not significant. Without information
regarding the dose or adherence to the medication, there are number of reasons why
medication was not significantly predictive, that are unable to be further explored within
the current study. The difference could also be related to the population of students
examined in this study, or those in special education, compared to previous studies on the
topic. For example, the sample used to detemrine a clear linear effect between ADHD
symptom severity and ADHD (DeShazzo Barry et al.) used a sample where only a
portion of the students with ADHD were in special education (42%) and the students all

met the diagnostic criteria for ADHD (versus parent and school report). While in a meta-

81

analysis that examined the impact of medication on academic achievement (Ryan et al.)
the sample included some students in general education, resource, or self-contained
classrooms, however, many of the participants were from university settings, psychiatric
hospitals, or other clinical settings. The students in the current study are likely those with
the most educational dysfunction, as they are all eligible for special education, and the
impact of medication or externalizing symptoms could be different for those students
than for other students without the same level of educational impact.

Intervention Variables. Academic intervention was associated with lower initial
reading scores on the ORF, LW, and PC measures. This is logical and suggests that those
students with the lowest reading scores were being provided with the most academic
support. Behavioral or academic intervention were related to improved rate of growth on
all three skill based measures, but not to reading grades.

Academic intervention was related to improvements on the PC measure. Some
authors have suggested that comprehension is an area of particular difficulty for children
with ADHD, as it represents a high-order process (Ghelani, Sidhu, Jain, & Tannock,
2004). In fact, Sattler (2001) suggests that passage comprehension requires the highest
level of functioning of all reading tasks. From the present findings, it can be
hypothesized that academic intervention efforts are needed to address this higher level
skill, but that behavioral intervention is sufficient to lead to improvements in sight word
recognition and reading fluency. In the context of Rapport and colleagues model (1999)
the trickle down effect described may be adequate to improve lower level academic
outcomes, but not for higher level peripheral features of ADHD. Further exploration is

important to highlight specific types of interventions that can best address deficit areas.

82

Behavioral intervention was significantly predictive of higher rates of growth on
the ORF and LW measures. This suggests that by improving behavior, there is an
academic benefit. While authors have criticized the focus in schools on only a reduction
in negative behaviors associated with ADHD versus an emphasis on improving positive
outcomes, such as grades, (DuPaul & Weyandt, 2006), this study does provide support
for increasing rates of growth in reading through the use of behavioral intervention.
While it is unknown if the behavior actually improved as a result of the behavioral
intervention, the findings in the current study indicate that reading fluency and
letter/word identification did improve.

Both academic and behavioral interventions have promise in improving academic
outcomes for children in special education. Previous research shows that academic
interventions have a positive impact on behavioral outcomes, as much as behavior
interventions do (DuPaul & Eckert, 1997). This speaks to the utility of academic
interventions in addressing student difficulties. The finding by DuPaul & Eckert calls
into question the model proposed by Rapport and colleagues (1999), which suggests that
treatment directed at peripheral symptoms of ADHD would only impact those areas, and
not have an impact on the core symptoms. Alternatively, DuPaul and Eckert suggested
that academic interventions may provide a replacement behavior for a display of the core
symptoms of ADHD. Additional research is needed to address the types of academic
interventions that yield the most positive benefit for students for both academic and
behavioral outcomes. In addition, this study suggests that different skills may be affected
differentially by the type of intervention employed. As discussed by DuPaul and

Weyandt (2006), interventions can be moderated through the teacher, other students, and

83

through technology. There are therefore a wide variety of strategies available that have
been empirically-supported for children with ADHD that can be applied to the specific
presenting problem.

Academic and behavioral interventions did not have a significant impact on
grades. Grades are not commonly used as an outcome measure in studies in this content
area. Intervention studies typically examine the performance of a specific skill.
However, it is important to consider that grades are perceived as an indicator of student
outcomes. DuPaul and colleagues (2004) did examine the impact of teacher perceptions
of academic skills, and found that these were predictive of report card grades in reading.
The DuPaul study implies that an increase in skill would result in a higher grade. As
found in the present study, academic intervention did increase scores in reading
comprehension, while behavioral intervention increased performance on the ORF and
word identification tasks. Yet these improvements were not also seen in reading grades.
It is possible that the student’s teachers did not perceive an improvement in skill.
Alternatively, this particular population of students may not have been as impacted by
their teacher’s perception of their skills as those in DuPaul and colleagues sample.
Report grades are often a primary way that student achievement is relayed to parents and
other individuals outside of the school system, and plays an important role in admissions
to programs and post-secondary opportunities. Grades are also often considered when
determining special education eligibility, as a measure of academic relevance. Yet in
this study they were not found to be sensitive to improvement garnered from
interventions in the same way that skills measures were. While it is challenging to

determine growth from grades, as there is significant diversity in how student grades are

84

determined, it is relevant to consider that grades serve as important indicators of
academic achievement for students.
Outcome Measures

Differences between outcome measures have arisen in this study. The difference
between direct skill measures and reading report card grades was particularly clear.
Specifically, the report card grades were unique when it came to functional skills, grade
level, mother’s education, student behavior, and intervention variables. As hypothesized
above, the fact that the functional skills relate positively to initial reading grades may
suggest more similarities between parent and teacher perceptions versus parent
perceptions of functional skills and reading skill. The student’s grades were also
impacted less by grade level than the skill based measure, suggesting that the Matthew
Effect is not portrayed through grades, despite this effect occurring on skill-based
measures within the same sample of students. As discussed previously, mother’s
education was predictive of report card grades in the unexpected direction, and follow-up
of these results is needed. Some of these findings replicate earlier findings. For example,
DuPaul and colleagues (2004) found that academic enablers related only to grades, but
not skill based measures. Another important and unique finding for grades was that skill
measures were influenced by intervention and the report card grades were not. The
implication for this finding is that outcome measures need to be appropriately matched to
goals. If a goal is to measure student progress in response to a particular intervention,
grades, and even traditionally written IEP goals, are not be an appropriate tool to do so.
A number of studies have shown that by data-based goals and monitoring improves

student achievement (Shinn, 2002), therefore schools are obligated to use tools that

85

facilitate this. Complexities relating to measuring growth using reading grades, may have
also played a role in the unique nature of these findings. For example, grades are not
standardized, are filtered through teacher perceptions, and many students in special
education receive modified grades. This speaks to the importance focusing on outcome
measures that appropriately measure the types of effects that are targeted through
intervention.
Group Difi‘erences

The final research question asked whether there were group differences in how
the predictor variables influenced initial reading skills and growth over time. There were
fewer group differences than expected, despite differences in the composition of the
groups on factors such as gender, ethnicity, income, mother’s education, stimulant
medication status, externalizing and student behavior, and behavioral intervention. There
were no differences across the ORF, LW, and PC measures in terms of individual
variability on the level or shape factor. There were also no differences in the mean of the
level and shape factors across the groups. Finally, there were no differences between the
groups in terms of the impact of covariates on the level and shape factors on the overall
model. At the predictor level, there were a few differences on the impact on the level and
shape factors across the groups. However, when the overall model was assessed to
examine differences between groups in terms of the impact of the covariates, no
significant differences were found. Previous research has documented differences
between children with and without a diagnosis of ADHD (ex. DuPaul et al., 2004),
however, there were not consistently significant outcome differences in this study. This

may suggest that the children’s status as a child eligible for special education may be

86

more salient than their ADHD status. For example, Deno and colleagues (2001) found
that general education students experienced growth on ORF at a rate twice that of their
special education peers. As discussed earlier, much of the literature on academic
achievement in students with ADHD has not specifically focused on students eligible for
special education (ex. DeShazo Barry et al., 2002; Ryan et al., 2005). Or in studies
focused on children with ADHD in special education, the emphasis has been on
demographic variables and not on achievement (Schnoes et al., 2006). Perhaps this
population, who is most academically impacted (and thus eligible for special education)
is distinct from other students with ADHD. Also, along the same lines, this finding may
not generalize to children with and without ADHD who are not eligible for special
education. Special education status may be an important are to consider when it comes to
exploring predictors of achievement for students with ADHD.
Limitations

This study has a number of limitations. These included limitations with the
available data, possible informant differences, generalizability of the sample, and the
amount of missing data. Given the use of a pre-existing data set, the variables that served
to operationalize the variables of interest were not always ideal. One benefit of using
data from a large and diverse sample is that it provides insight into what typical services
look like for students with disabilities within school systems. This can help us better
understand the current practices, yet limits the kinds of information that one has access
to. For example, the created externalizing behavior scale consisted of only four items.
This limits the variability in responses that can be provided, which makes it less sensitive

to differences between groups. A more extensive measure may have highlighted

87

differences not found in the present study. Another limitation in being limited in the
types of data available in an existing data set, is that there was no information available
beyond parent and teacher report of interventions that were in place. There was no
available information regarding the actual implementation of behavior or academic
intervention or for the integrity with which the stimulant medication was delivered. This
makes it challenging to assess whether the intervention itself was effective or ineffective
versus the extent to which it was carried out as being the influential factor. Another
challenge related to working with an existing data set is that number of data collection
points is limited. While three data points improves on cross-sectional research on this
topic, with three time points, there is less flexibility in modeling growth.

Another limitation to consider is the potential differences between parent and
teacher reports of behavior. For example, in the present study, the children’s functional
skills were reported by the parents. Unexpectedly, functional skills did not play a
significant role in student achievement. One possible explanation is that parent’s
reporting of functional skills is not consistent with teacher perceptions of a child’s ability
It is also possible that parents are not reliable reporters of their child’s functional abilities.
The reliance on parent and teacher report could also have impacted the report of who has
ADHD versus who does not in the study. The question does not ask if the child has a
verified diagnosis from a mental health provider, and some parents may describe their
child’s active behavior as ADHD, when in fact the child may not meet the diagnostic
criteria for the disorder.

Despite the large size of the data set, the restriction to children with both a parent

and school report of ADHD, resulted in a much smaller sample. Although there were not

88

strict inclusion criteria outside of parent and school report, it is felt that these children do
represent those that the school feels has ADHD. However, implying that these results
generalize to all children with ADHD is unwarranted, as no diagnostic procedures were
employed in this study. Additionally, the number and type of comorbid conditions in
these children are not known. The subtypes of ADHD were also unknown in the current
study. Previous research has suggested that there are differences between the subtypes in
terms of the reading skills, with children with the Predominantly Inattentive subtype of
ADHD, experiencing more academic difficulties than those with a diagnosis of ADHD,
Predominantly Hyperactive/Impulsive subtype (Wilcutt et al., 2007). More specific
understanding of the interplay between the ADHD subtypes and special education would
be useful to further determine how to differentiate services for children exhibiting these
distinct constellation of symptoms. Also important to consider as a limitation is the fact
that the ADHD status of the children was assessed at wave 1, but not at wave 2 and 3.
This means that there could have been children who were diagnosed with ADHD at wave
2 or 3, included in the control sample. Additionally, while ADHD is thought to be a
long-term disorder, it is possible that children who were diagnosed with this disorder at
wave 1, were no longer diagnosed at wave 2 or 3.

Additionally, given the longitudinal nature of the sample, there was a large
amount of missing data. For this reason, a large quantity of data was imputed, which
could have influenced the results of this study. However, examination of the missing
data sample versus the sample with the missing data imputed (Tables 2 & 3) reveals
small differences in means and standard deviations, supporting the use of the imputed

data.

89

Conclusions

Valuable information about how students achieve in schools can be obtained
through the use of large data sets. The sample sizes that are available in these data sets
would not be possible for many researchers to obtain, and thus they are extremely useful.
However, follow-up studies in more highly controlled conditions are warranted to further
ascertain the specifics regarding student achievement.

The findings from this study suggest that schools play an important role above
and beyond demographic characteristics of students when it comes to improving reading
skills across a number of outcome measures. This was demonstrated by several
demographic variables having a positive impact on initial reading skills, but less so when
it comes to growth over time. Additionally, while externalizing behavior did not appear
to have a significant impact on reading growth in the present study, positive student
behaviors were related to growth on both a skill-based measure (i.e. ORF) and on reading
grades. Both academic and behavioral intervention variables were important factors in
reading growth over time. However, this was only true on the skill-based measures and
not for reading grades. This study highlights several factors amenable to change within
the school setting that are important to academic outcomes for children with ADHD.
Additionally, these findings for children with parent and school reported ADHD were not
distinct from children with no report of ADHD in the sample, suggesting the fact that
these students are in special education is an important variable to consider when
designing interventions, as this was the common thread between both groups. While this
study and another using this sample (Schnoes et al., 2006), found that characteristics such

as gender distribution and medication status were similar to the general population of

90

individuals with ADHD, the results of this study were different in some ways from
previous studies on children with ADHD.. This may suggest that this sample of children
(those in special education) may be a distinct group from the overall population of
children with ADHD. Finally, the manner in which outcomes is measured is important
and has implications for how growth is interpreted.

Given the large number of students with ADHD that are impacted academically,
additional research needs to further delineate predictors of academic achievement in this
population. Of particular interest are those variables that are amenable to intervention
within the educational system. The results of this study suggest that variables that can be
altered within the school setting can play an important role in student outcomes,
including student behavior and intervention variables. The President’s Commission on
Excellence in Special Education (2002) suggests that our current system of special
education has not been always focused on the use “evidence-based practice”. As such,
studies have also revealed effect sizes below .5 for the impact of special education
(Kavale & Fomess, 1999) suggesting that change is needed. An emphasis on reforming
this system for one of the most frequent diagnoses for the school-aged population is

imperative to improve the negative trajectory often associated with ADHD.

91

APPENDICES

92

Appendix Table 1

Measurement Model Confirmatory Factor Analysis

 

 

 

 

Extemalizing Student

Item Estimate R2 Estimate R2
Impulsive 0.47 .65
Temper 0.38 .60
Trouble 0.48 .80
Responds 0.50 .78
Directions 0.33 .67
Performs 0.42 .77
Own 0.43 .70
Keeps 0.45 .74
Homework 0.40 .66
Communication 0.28 .49
Cronbach’s Alpha .802 .841
Chi Square (df; p) 39.65 (2 7;

.055)
NFI; GFI; CFI 1.00; 0.99;

1.00
RMSEA (CI) .02 (.00;

.032)

 

93

Appendix Table 2

Descriptive Characteristics of ADHD & no-ADHD without Imputed Data

 

 

 

 

ADHD No-ADHD
Variable N Mean Std. N Mean Std Sig
t
Grade 389 4.85 1.81 630 4.81 1.94 .746
a t*
Gender 389 1.20 0.39 630 1.38 0.48 .000
I*
Ethnicity 389 1.22 0.42 630 1.30 0.46 .008
C
Household Income 354 9.30 4.77 579 8.42 4.64 .005*
Mother’s Education 351 2.66 0.89 583 2.44 0.96 900*
e t*
Stimulant Status 389 0.70 0.46 630 0.01 0.10 .000
{31:
Extemalizing Behavior 334 8.15 1.92 544 6.64 2.18 .000
. . g t*
Behavror Intervention 341 1.43 1.05 534 0.10 0.93 .000
_ _ h t
Academic Intervention 335 5.56 2.27 523 5.45 2.47 .534
I {at
Student Behavior 333 12.71 2.45 550 13.98 2.79 .000
Functional SkillsJ 382 7.34 2.71 621 7.21 2.69 .482
Primary Goal for Reading 346 1.69 1.01 565 1.70 0.98 .827
1
Disability Status
8*
Serious Emotional Disturbance 389 0.24 0.43 630 0.08 0.28 .000
91¢
Learning Disability 389 0.44 0.50 630 0.60 0.49 .000
I*
Other Health Impaired 389 0.30 0.46 630 0.16 0.37 .000
Spch./Lang. Impairment 389 0.15 0.36 630 0.40 0.49 000t*

 

T = equal variances not assumed; a = (1=male: 2=female); b =(1=white: 2=nonwhite); c = (1=low.
.l6=high); d = (1=low-4=high) e = (0=no: l=yes); f: (higher=more negative); g = (range =0-3 ; higher
equals more intervention); h = (range =2-1 1; higher equals more intervention); 1 = (higher=more positive):
j = (l=low; 6=medium; l 1=high); k = (l=reading at grade level. . . 5: no reading goals): 1 = (0=no: l=yes);
*significant (p<.05)

94

Appendix Table 3

Descriptive Characteristics of ADHD & no-ADHD with lrnputed Data

 

 

 

 

ADHD No-ADHD
Variable N Mean Std. N Mean Std Sig
t
Grade 389 4.85 1.81 630 4.81 r .94 74"
a {31:
Gender 389 1.19 0.39 630 1.37 .48 .000
b t*
Ethnicity 389 1.22 0.42 630 1.30 .46 .008
C
Household Income 389 9.26 4.60 630 8.32 4.53 .001 *
Mother's Education 389 2.65 0.87 630 2.42 0.95 000*
C tale
Stimulant Status 389 0.70 0.46 630 0.01 .10 .000
t*
Extemalizing Behavior 389 8.12 1.84 630 6.64 2.07 .000
I*
Behavior Intervention 389 1.41 1.01 630 .99 .88 .000
t
Academic Intervention 389 5.54 2.17 630 5.39 2.34 .295
I I*
Student Behavior 389 12.74 2.32 630 13.99 2.65 .000
Functional SkillsJ 389 7.34 2.70 630 7.22 2.70 .465
k
Primary Goal for Reading 389 1.71 0.96 630 1.71 .94 .958
1
Disability Status
tilt
Serious Emotional Disturbance 389 0.24 0.43 630 0.08 0.28 .000
{:8
Learning Disability 389 0.44 0.50 630 0.60 0.49 .000
t*
Other Health Impaired 389 0.30 0.46 630 0.16 0.37 .000
{wk
Spch./Lang. Impairment 389 0.15 0.36 630 0.40 0.49 .000

 

T = equal variances not assumed: a = (1=male: 2=female); b =(1=white; 2=nonwhite); c = (1=low.
.16=high); d = (1=low-4=high) e = (0=no: l=yes); f: (higher=more negative); g = (range =0-3 : higher
equals more intervention); h = (range =2-1 1; higher equals more intervention); i = (higher=more positive):
j = (1=low; 6=medium: l 1=high); k =(1=reading at grade level. . . 5: no reading goals); 1 = (0=no: l=yes);
*significant (p<.05)

95

cove 2.85:»; 1. ”Gave 28%»? H_
o>§> 5:53.25 BEoowonnN— : o>§> =0:=o>.§:_ .0323 n: 2 o>§> .mow mE—omB uo_ ; 033 532.3 Eogauo : o>§> 5322 wEN=~E8xouw
: 0225 :6? 3:058:lo H _ u>a3 259m. uoEuo 2 9295 5:833 Means—cum n. u>§> oEooE Eonogguv ” E25203“ "LoocownN ”oomemfl .202

 

 

:me. omo. oNo.- ooo. **moN. for **N:.- **mnN.- **oo_. mmor omor N—

..inov. 2o. moor aim—N. aim—N. wmor their **omN.- 3.0:. for **hoN.- 2
*_o_. for swoo- ooo. ovo. ooo. mwor gov woo. woo. 1oz. o—
..31 _N.- **moN.- *mwor toomr 3o. Vmo. 3o. .1. _m_. for *3 _. Scot 0
INS. **3N. o3. 1.03.- So: wmo. .3..th **mmm.- *RVN. *NN_.- .532: w
EARN. **_o_. 31.3. act—N: mmo. ooo.- .INE: :v—Nr woo. :o. *Eomr N.
Vmo. omo. moor 50.. **N_ _. mmo. mmo. 3o. Nvor _ _o.- mmor o
**mNN.- *sz _ .- o _o. 53.2. * So: who.. oNo.- imov. imm _ .- Noo. :o. m
1%ri 1.32.- _mo.- .3..an **mNN.- film..- mNor Town. Immmr mNo. boo. v
.Ibm _. .Inon _. ooo. *som _ .- **o_N. mNo woo *iVmNr 1.50- ooor moor m
who: nio— _.- hNo. **mN _. **oo_ .- N _ o. No. Nmor oNo.- Rio: omo. N
3mm _. MNor woo. mnor 0N0. .25ri wmor *owo- Soot wNor moo. _

N _ _ _ o _ o w b o n v m N _

 

3:90 2030: QEQ< do C253 QEQ< 0: 5o ism—z coca—250

4 22¢ 528%.

96

Appendix Table 5

Reading Achievement Across Groups

 

 

 

ADHD Non-ADHD
(n=389) (n=630)
Variable Means SD Means SD Sig
Letter-Word Identificationa
Wave 1 89.21 14.78 85.73 15.24 .000*
Wave 2 90.24 15.17 85.99 15.69 .000*
Wave 3 91.09 15.48 86.90 15.67 .000*
Passage Comprehensiona
Wave 1 88.80 13.45 86.00 13.60 .001*
Wave 2 88.90 13.84 86.58 14.98 .0129:
Wave 3 89.22 12.35 86.79 12.98 .003*
Oral Reading Fluency
Wave 1 93.1 1 44.92 82.49 43.38 .000*
Wave 2 107.59 45.59 99.88 45.38 .009*
Wave 3 127.31 44.05 121.80 44.50 .054
Reading Grades
Wave 1 2.32 0.80 2.14 0.82 .001*
Wave 2 2.37 0.76 2.12 0.71 0009‘
Wave 3 2.43 0.78 2.25 0.80 .000*

 

T = equal variances not assumed: *significant (p<.05): a = although standard scores are reported here for
ease of interpretation, analyses were completed using W scores

97

Appendix Table 6

Impact of Covariates on Oral Reading Fluency

 

 

 

 

Initial Reading Status Reading Growth

Unstand. S.E. Stand. Unstand. S.E. Stand.

Estimate Estimate Estimate Estimate
Grade 5.902 1.334 250* -2.268 0.708 -.219*
Gender 13.706 5.997 .1 1 1* 0.353 2.495 .009
Ethnicity -3.209 5.894 -.031 -3.632 3.873 -.106
Household Income 0.819 0.651 .088 -0.124 0.418 -.040
Mother’s Education 5.131 3.480 .104 1.654 2.658 .102
Functional Skills -4.600 0.901 -.266* -0.215 0.798 -.037
Reading Goal -4.574 3.282 -.089 -4.130 1.464 -.242*
Extemalizing 0.071 1.393 .003 -0.022 1.069 -.003
Behavior
Student Behavior -3.084 .917 -. 168* 1.763 0.728 290*
Medication Status 4.183 4.345 .045 0.718 3.158 .023
Behavioral -2.282 2.646 -.050 3.801 1.819 .250*
Intervention
Academic -7.358 1.199 -.361* 0.148 0.768 .022
Intervention

 

*significant (p<.05)

98

Appendix Table 7

Impact of Covariates on Letter Word Identification

 

 

 

Initial Reading Status Reading Growth

Unstand. S.E. Stand. Unstand. S.E. Stand.

Estimate Estimate Estimate Estimate
Grade 5.303 0.983 308* -2.536 0.515 -.660*
Gender -1.249 3.826 -.014 1.839 2.396 .092
Ethnicity -2.895 4.352 -.038 -4.186 2.388 -.248
Household Income 1.143 0.508 .168* 0.188 0.277 .124
Mother’s Education 3.181 2.283 .089 0.316 1.374 .039
Functional Skills -3.816 0.660 -.303* 0.020 0.367 .007
Reading Goal -5.222 1.819 -.139* -0.933 1.371 -.11 1
Extemalizing -0.219 0.933 -.013 -0.183 0.681 -.048
Behavior
Student Behavior -1.829 0.772 -.136* 0.628 0.709 .209
Medication Status 0.248 3.51 1 .004 2.474. 2.347 .164
Behavioral -1.747 1.978 -.052 3.755 1.159 .501 *
Intervention
Academic -3.941 0.938 -.265* 0.724 0.528 .218
Intervention

 

*significant p<.05

99

Appendix Table 8

Impact of Covariates on Passage Comprehension

 

Initial Reading Status Reading Growth

 

 

 

Unstand. S.E. Stand. Unstand. S.E. Stand.

Estimate Estimate Estimate Estimate
Grade 3.853 0.484 362* -1.671 0.207 -.452*
Gender - l .999 4.792 -.036 0.954 2.230 .049
Ethnicity -5.579 2.539 -.1 19* 0.648 0.841 .040
Household Income 0.378 0.259 .090 -0.201 0.127 -.137
Mother’s Education 3.230 1.557 .146* -0.092 0.696 -.012
Functional Skills -l.562 0.375 -.200* -0. 150 0.174 -.055
Reading Goal -2. 167 1.208 -.093 -0. 166 0.479 -.021
Extemalizing Behavior 0.324 0.606 .031 -().410 0.273 -.1 13
Student Behavior 0.705 0.507 .085 -0.379 0.245 -. 131
Medication Status -0933 2.263 —.022 1.632 0.883 .1 r3
Behavioral Intervention 1.540 1.558 .074 -1.071 0.664 -. 149
Academic Intervention -2.270 0.537 -.247* 0.614 0.204 .192“

 

*significant p<.05

100

Appendix Table 9

Impact of Covariates on Reading Grades

 

 

 

 

Initial Reading Status Reading Growth

Unstand. S.E. Stand. Unstand. S.E. Stand.

Estimate Estimate Estimate Estimate
Grade 0.1 13 0.063 .186 0.000 0.040 .000
Gender -().239 0.225 -.075 0.000 0.124 .000
Ethnicity 0.140 0.271 .053 -0.042 0.144 -.046
Household Income -().()04 0.024 -.016 -0.008 0.016 -.092
Mother's Education 0.065 0.1 1 1 .051 -0.160 0.072 -.364*
Functional Skills 0.1(15 0.043 .235* -0.040 0.023 -.256
Reading Goal 0.177 0.094 .133 -0.044 0.049 -.095
Extemalizing Behavior -(.).()48 0.059 -.081 0.010 0.033 .048
Student Behavior -0.31 1 0.058 -.657* 0.092 0.029 .561*
Medication Status 0.139 0.207 .058 -0.138 0.126 -.167
Behavioral Intervention -0.082 0.1 18 -.069 0.019 0.083 .046
Academic Intervention -0.074 0.047 -.140 0.028 0.029 .156

 

*significant p<.05

101

 

 

Biological
Influences, e.g.,
genencs

 

 

 

 

Primary Dx

NBUF 0131010810311 Criteria/Core Variables:
Substrate Inattention
Hyperactivity
Impulsivity

 

 

 

 

 

 

 

 

 

 

/ l /

 

 

 

 

 

 

Environmental MPH CBT/Behavioral
Demands
\ Skills Training Peripheral
Features
Treatment

 

 

 

Appendix Figure 1

Conceptual Model of ADHD (Rapport, Chung, Shore & Isaacs, 2001)

102

Behavioral
Disinhibition

   
   
   
 

  
  

  

Delayed
Internalization
of Speech

 

 
  
  
   
 

Immature Self-
Regulation of
Affect/Motivation

Reconstitution

Reduced Motor
Control/Fluency/

Syntax

Appendix Figure 2

Disinhibition Model (Barkely, 1997)

103

 
  
 

Pnor
Achievement

  
  
  
 

  

Current
Achievement

  

Engagement

 

 
 

Interpersonal
Skills

    
 

Appendix Figure 3

Model of Academic Enablers (DiPerna, Volpe & Elliott, 2002)

104

Scholastic
Achievement

Classroom
Performance

 

 

   
 

Vigilance

 

 

ADHD _ 1Q

 

Appendix Figure 4

Model of Dual Developmental Pathways (Rapport, Scanlon & Denney, 1999)

105

 
  
     
 
  

  

Prior
Reading
Achievement

  
     

Current
Reading
Achievement

Engagement

Appendix Figure 5

Model of ADHD & Academic Enablers (Volpe et al., 2006)

106

 

 

Demographics Achievement
Wave 1

 
     

Initial
Reading

| Reading

 

 

 

     

Skills

 

 

Student
Variables

   
   

 

  

Reading
Rate of \ Achievement

 

 

 

   
 

 

 

Reading Wave 2
, Growth
Intervention
Variables
Reading
Achievement
Wave 3

 

 

 

Appendix Figure 6

Hypothesized Latent Growth Model of Reading Achievement

107

140 fi

 

120
100

80 + ADHD

60 + No-ADHD

40

l
l
l
l
l
l
1
1
l
l
20 1

Words Read Correctly Per Minute

 

0 1‘
ORF 1 ORF 2 OFIF 3

 

Appendix Figure 7

Means of ORF over time and between groups

108

BIBLIOGRAPHY

109

Bibliography

American Academy of Child and Adolescent Psychiatry (AACAP; 1997), Practice
parameters for the assessment and treatment of children, adolescents, and adults
with attention—deficit/hyperactivity disorder. Journal of the American Academy of
Child & Adolescent Psychiatry, 36(105upp1).

American Psychiatric Association (APA). (2000). Diagnostic and Statistical Manual of
Mental Disorders (4"' Ed.) Text Revision. Washington DC: author.

Asparouhov, T. (2005). Sampling Weights in Latent Variable Modeling. Structural
Equation Modeling, 12, 41 1—434.

Baker, S. K., Smolkowski, K., Katz, R., Fien, H., Seeley, J .R., Kame’enui, E.J., & Beck,
C. T. (2008). Reading fluency as a predictor of reading proficiency in low-
performing, high-poverty schools. School Psychology Review, 37, 18-37.

Barbaresi, W. J ., Katusic, S. K., Colligan, R. C., Weaver, A. L. & Jacobson, S. J.
(2007).Long-term school outcomes for children with Attention-
Deficit/Hyperactivity Disorder: A population-based perspective. Journal of
Developmental & Behavioral Pediatrics, 28, 265-273.

Barkley, R. A. (1997). ADHD and the Nature of Self-Control. New York: Guilford Press.

Barkley, R. A. (1998). Attention deficit hyperactivity disorder: A handbook for diagnosis
and treatment (2rd ed. ). New York: Guilford.

Barkley, R. A. (2003). Attention-Deficit/Hyperactivity Disorder. In E. J. Mash & R. A.
Barkley Child Psychopathology (2"‘1 Edition). New York: Guilford Press.

Brown, R. T. (2005). Recent advances in pharmacotherapies for the externalizing
disorders. School Psychology Quarterly, 20, 118-134.

Bussing, R., Zima, B. T., Perwien, A. R., Belin, T. R., & Widawski, M. (1998). Children
in special education programs: Attention deficit hyperactivity disorder, use of
services, and unmet needs. American Journal of Public Health, 88, 880-886.

Deno, S., Fuchs, L., Marston, D., & Shin, J. (2001 ). Using curriculumObased
measurement to establish growth patterns for students with Ieaming disabilities.
School Psychology Review, 30, 507-524.

Deno, S. L., Mirkin, P. K., & Chang, B. (1982). Identifying valid measures of reading.
Exceptional Children, 49, 36-45.

110

Department of Education. (2006). Assistance to States for the Education of Children with
Disabilities and Preschool Grants for Children with Disabilities; Final Rule.
Accessed March 11, 2007 at http://idea.ed.gov/down]oad/finalregulations.html.

DeShazo Barry, T., Lymon, R.D., & Grofer Klinger, L. (2002). Academic
underachievement and attention-deficity/hyperactivity disorder: The negative
impact of symptom severity on school performance. Journal of School
Psychology, 40, 259-283.

DiPerna, J. C. & Elliott, S. N. (2000). Academic Competence Evaluation Scales (ACES).
San Antonio, Texas: Pearson Education, Inc.

DiPerna, J. C. & Elliott, S. N. (2002). Promoting academic enablers to improve student
achievement: An introduction to the miniseries. School Psychology Review, 31,
293-297.

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.

DuPaul, G. J. & Eckert, T. L. (1997). School-based interventions for children with
attention deficit/hyperactivity disorder: A meta-analysis. School Psychology
Review, 26, 5-27.

DuPaul, G. J., Volpe, R. J ., Jitendra, A. K., Lute, G., Lorah, K. S., & Gruber, R. (2004).
Elementary school students with AD/HD: predictors of academic achievement.
Journal of School Psychology, 42, 285-301.

DuPaul, G. J ., & Stoner, G. (2003). ADHD in the Schools: Assessment and intervention
strategies. New York: The Guilford Press.

DuPaul, G. J. & Weyandt, L. L. (2006). School-based intervention for children with
attention deficit hyperactivity disorder: Effects on academic, social, and
behavioural functioning. International Journal of Disability, Development and
Education, 53, 161-176.

Fomess, S. R. & Kavale, K. A. (2001). ADHD and a return to the medical model of
special education. Education and Treatment of Children, 24(3), 224-247.

Fergusson, D. M. & Horwood, L. J. (1995). Early disruptive behavior, IQ, and later
school achievement and delinquent behavior. Journal of Abnormal Child
Psychology, 23, 183-199.

Fergusson, D.M., Horwood, L. J ., & Linksey, MT (1993). The effects of conduct

disorder and attention deficit in middle childhood on offending and scholastic
ability at age 13. Journal of Child Psychology and Psychiatry, 34, 899-916.

111

Francis, David J ., et a1. (1996). Developmental Lag versus Deficit Models of Reading
Disability: A Longitudinal, Individual Growth Curves Analysis. Journal of
Educational Psychology, 88, 1, 3-17.

Ghelani, K., Sidhu, R., Jain, U., & Tannock, R. (2004). Reading comprehension and
reading related abilities in adolescents with reading disabilities and attention-
deficit/hyperactivity disorder. Dyslexia, 10, 364—384.

Hasbrouck, J ., & Tindal, G. (2006). Oral reading fluency norms: A valuable assessment
tool for reading teachers. The Reading Teacher, 59, 636-646.

Hoagwood, K., & Johnson, J. (2003). School psychology: A public health framework I
From evidence—based practices to evidence-based policies. Journal of School
Psychology, 41 (1), 3-21.

Hu, L. & Benter, P. (1999). Cutoff criteria for fit indices in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-
55.

Individuals with Disabilities Education Improvement Act of 2004 (IDEIA). Public Law
108-446. Accessed December 1, 2006 at http://frwebgate.access.gp
o.gov/cgi—bin/getdoc.cgi?dbname=108_cong_pub1ic_1aws&docid=f:publ
446.108

Jensen, P. S., Hinshaw, S. P., Swanson, J. M., Greenhill, L. L., Conners, C. K., Arnold, L.
E., et a1. (2001). Findings from the NIMH multimodal treatment study of ADHD
(MTA): Implications and applications for primary care providers. Journal of
Developmental & Behavioral Pediatrics, 22, 60-73.

Joreskog, K. & Sbrbom, D. (1996-2001). LISREL 8: User's Reference Guide.
Lincolnwood, IL: Scientific Software International, Inc.

Joreskog, K. & sorbom, D. (2006). LISREL.8.80 Statistical Software. Lincolnwood, IL:
Scientific Software International, Inc.

Juel, C. (1988). Learning to read and write: A longitudinal study of 54 children from first
through fourth grades. Journal of Educational Psychology, 80, 437-447.

Kavale, K. A. & Fomess, S. R. (1999). Effectiveness of special education. In C. R.
Reynolds & T. B. Gutkin (Eds), The handbook of school psychology (3rd ed., pp.
984-1024). New York: John Wiley.

Keith, T. Z. (1999). Structural equation modeling in school psychology. In C. R. Reynolds

& T. B. Gutkin (Eds). The Handbook of School Psychology (3rd Ed. ). New York:
John Wiley & Sons.

112

 

Kline, R. B. (2005). Principles and practices of structural equation modeling, Second
Edition. New York: The Guilford Press.

LeFever, G. B., Villers, M., Morrow, A., & Vaughn, E. (2002). Parental perceptions of
adverse educational outcomes among children diagnosed and treated for ADHD:
A call for improved school/provider collabOration. Psychology in the Schools, 39,

63-72.

Mannuza, S., Gittelman-Klein, R., Bessler, A., Malloy, P. & LaPadula, M. (1993). Adult
outcome of hyperactive boys: Educational achievement, occupational rank, and
psychiatric status. Archives of General Psychiatry, 50, 565-576.

Marston, D. B. (1989). A curriculum-based measurement approach to assessing academic
performance: What is it and why do it. In M.R. Shinn (Ed.) Curriculum-Based
Measurement: Assessing Special Children (pp.18-78).. New York: Guilford.

Marston & Deno (1986). Standard Reading Passages.

Muthe’n, L. K. & Muthe’n, B. (2006). MPlus Version 4.21. Available
www.statmode1.com

National Association of State Directors of Special Education. (2006). Response to
Intervention: Policy Considerations and Implementation. Alexandria, VA:
National Association of State Directors of Special Education, Inc.

National Institutes of Health. (2000). Consensus Development Conference Statement:
Diagnosis and treatment of attention-deficit/hyperactivity disorder (ADHD).
Journal of the American Academy of Child and Adolescent Psychiatry, 39, 182-
193.

National Reading Panel (2000). Teaching children to read: An evidence-based
assessment of the scientific research literature on reading and its implications for
reading instruction [on-line]. Available:
http://www.nichd.nih.gov/publications/nrp/smal1book.htm.

No Child Left Behind Act (2002; NCLB). Public Law 107-110. Accessed December 1,
2006 at http://www.ed.gov/policy/e1sec/1eg/esea02/ 107-1 10.pdf

Ollendick, T.H., & King, NJ. (2004). Empirically supported treatments for children and
adolescents: Advances toward evidenced-based practice. In P.M. Barrett and
TH. Ollendick (Eds), Handbook of Interventions that Work with Children and
Adolescents: Prevention and Treatment. NY: John Wiley & Sons, Ltd.

Pallant, J. (2007). SPSS Survival Manual (Third Ed. ). Berkshire, England: Open
University Press.

113

Pastor, P. N. & Reuben, C. A. (2002). Attention deficit disorder and learning disability:
United States, 1997-98. In National Center for Health Statistics: Vital Health
Statistics (DHHS Publication No. PHS 2002-1534). Hyattsville, MD: Department
of Health and Human Services.

Pelham, W. E. & Gnagy, E. M. (1999). Psychosocial and combined treatments for
ADHD. Mental Retardation and Developmental Disabilities Research Reviews,
5, 225-236.

Pelham, W. E., Gnagy, E. M., Greiner, A. R., Hoza, B., Hinshaw, S. P., Swanson, J. M.,
et a1. (2000). Behavior versus behavioral and pharmacological treatment in
ADHD children attending a summer treatment program. Journal of Abnormal

Child Psychology, 28, 507-525.

Phelps, L., Brown, R. T., & Power, T. J. (2001). Extemalizing Disorders. In L Phelps,
RT Brown & TJ Power (Eds.) Pediatric Psychopharmacology: Combining
Medical and Psychosocial Interventions. Washington DC: APA.

President's Commission on Excellence in Special Education (2002). A new era:
Revitalizing special education for children and their families. Available
http://www.ed.gov/inits/commissionsboards/whspecialeducation/

Rapport, M. D., Chung, K., Shore, G. & Isaacs, P. (2001). A conceptual model of child
psychopathology: Implications for understanding attention deficit hyperactivity
disorder and treatment efficacy. Journal of Clinical Child Psychology, 30, 48-58.

Rapport, M. D., Scanlon, S. W., & Denney, C. B. (1999). Attention-
Deficity/Hyperactivity Disorder and scholastic achievement: A model of dual
developmental pathways. Journal of Child Psychology & Psychiatry, 40, 1 169-
83.

Raykov, T. (2007). Lecture 9: Analysis of non-normal continuous variables with SEM.
CEP 991 b: Advanced Structural Equation Modeling. Michigan State University.

Raykov, T. & Marcoulides, GA. (2006). A First Course in Structural Equation Modeling
(2"d Ed. ). Mahwah, NJ: Lawrence Erlbaum Associates.

Redden, S. C., Fomess, S. R., Ramey, C. T., Ramey, S. L., Brezausek, C. M. & Kavale,
K. A. (2003). Head start children with a putative diagnosis of ADHD: A four-year
follow-up of special education placement. Education & Treatment of Children,

26, 208-223.

Reid, R., Maag, J . W., Vasa, S. F., & Wright, G. (1994). Who are the children with
ADHD: A school based survey. Journal of Special Education, 28, 117-137.

114

Rowland, A. S., Lesesne, C. A., & Abramowitz, A. J. (2002). The epidemiology of
attention-deficit/hyperactivity disorder (ADHD): A public health view. Mental
Retardation and Developmental Disabilities Research Reviews, 8, 162-170.

Ryan, J. B., Reid, R. Epstein, M. H, Ellis, C., &Evans, H. (2005). Pharmacological
intervention research for academic outcomes for students with ADHD. Behavioral
Disorder, 30, 135- 154.

Sattler, J. M. (2001). Assessment of Children: Cognitive applications (4” Ed. ). San
Diego: Jerome M. Sattler, Publisher, Inc.

Schnoes, C., Reid, R., Wagner, M., & Marder, C. (2006). ADHD among students
receiving special education services: A national survey. Council for Exceptional
Children, 72, 483-496.

Shaywitz, S. E., Escobar, M. D., Shaywitz, B. A., Fletcher, J .M., & Makuch, R. (1992).
Distribution and temporal stability of dyslexia in an epidemiological sample of
414 children followed longitudinally. New England Journal of Medicine, 326,
145-150.

Shinn, M. R. (2002). In A. Thomas & J. Grimes (Eds) Best practices in school
psychology IV (pp. 671-697). Bethesda, MD: The National Association of School
Psychologists.

Smith, B. H., Barkley, R. A., & Shapiro, C. .1. (2006). Attention- -Deficit/Hyperactivity
Disorder. In E. J. Mash & R. A. Barkley Treatment of Childhood Disorders ( 3"!
ed. ). New York: Guilford.

Special Education Elementary Longitudinal Study, Wave 1 [CD-ROM database, with
accompanying documentation] (Produced Under Contract No. ED-OO-CO-0017).
(2003). Available from U. S. Department of Education, Office of Special
Education Programs.

SRI International. (1999). Special Education Elementary Longitudinal Study (SEELS):
Revised timeline and data collection, sample, and analysis plans. Accessed
December I, 2006 at http://www.seels.net/designdocs/ design_contentshtm

Stanovich, KB (1986). Matthew effects in reading: Some consequences of individual
differences in the acquisition of literacy. Reading Research Quarterly, 21, 360-
407,

Tabachnick, B. G. & Fidell, L. S. (2001). Using Multiviariate Statistics (4rd Ed.) New
York: HarperCollins Publishers, Inc..

U. S. Department of Education. (2006). Assistance to States for the Education of
Children With Disabilities and Preschool Grants for Children With Disabilities;

115

 

Final Rule. Federal Register, 71, 46540-46845. Accessed
http://www.ed.gov/1egislation/FedRegister/finrule/2006-3/081406a.pdf.

US. Department of Education. (2005). 26th Annual Report to Congress on the
Implementation of the Individuals with Disabilities Education Act, 2004 Vol. 1.
Washington DC.

Volpe, R. J., DuPaul, G. J., DiPerna, J. C. Jitendra, A. K., Lutz, J. G., Tresco, K., & Vile
Junod, R. (2006). Attention Deficit Hyperactivity Disorder and scholastic
achievement: A model of mediation via academic enablers. School Psychology
Review, 35, 47-61.

Wagner, M., Newman, L., Cameto, R., Levine, P. and Gonazlez, P. (2006). The
Academic Achievement and Functional Performance of Youth With Disabilities:
A Report From the National Longitudinal Transition Study-2 (NLTS2). Available
http://ies.ed.gov/ncser/pdf/20063000.pdf

Wang, M. C., Haertel, G. D., & Walberg, H]. (1993). Toward a knowledge base for
school learning. Review of Educational Research, 63, 249-294.

Wilcutt, E. G., Betjemann, R. S., Wadsworth, S. J ., Samuelsson, S., Corley, R., DeFries,
J. C. et a1. (2007). Preschool twin study of the relation between attention-
deficit/hyperactivity disorder and prereading skills. Reading & Writing, 20, 103—
125.

Woodcock, R. W., McGrew, K. S., & Mather, N. (2001). Woodcock Johnson 111. Itasca,
IL: Riverside Illinois. .

116

Ill 1111

11

S
E
R
A
R
m1
L
Y
n
S
R
.1:

HI

STATE

11

ll

MICHIGAN
1“

ll