Content uploaded by Chad A. Rose
Author content
All content in this area was uploaded by Chad A. Rose
Content may be subject to copyright.
© 2010 Chad A. Rose
SOCIAL-ECOLOGICAL FACTORS RELATED TO THE INVOLVEMENT OF
STUDENTS WITH LEARNING DISABILITIES IN THE BULLYING DYNAMIC
BY
CHAD A. ROSE
DISSERTATION
Submitted in partial fulfillment of the requirements
for the degree of Doctor of Philosophy in Special Education
in the Graduate College of the
University of Illinois at Urbana-Champaign, 2010
Champaign, Illinois
Doctoral Committee:
Associate Professor Lisa Monda-Amaya, Chair
Professor Dorothy Espelage
Associate Professor Steven Aragon
Assistant Professor Karrie Shogren
ii
Abstract
Bullying has emerged as one of the most fundamental problems facing school-
aged youth to date. While bullying is generally considered a problem that involves the
entire student body, research suggests that students with learning disabilities are
overrepresented in the bullying dynamic. Additionally, existing literature suggests that
involvement in bullying is based on complex interactions between an individual and
social-ecological factors. Few empirical studies have examined the interplay between
these social-ecological factors and disability status. Therefore, the current study
investigated demographic variables, sense of belonging, and social supports as predictors
for involvement in the bullying dynamic for students with learning disabilities (n = 83)
and students without disabilities (n = 360). While the two groups of students are
characteristically different, results of the current study suggested involvement in bullying
was invariant between students with learning disabilities and students without disabilities.
However, gender, race, grade point average, and participation in extracurricular activities
emerged as significant predictors for involvement in the bullying dynamic. Additionally,
increased peer social support was found to be the most significant predictor of decreased
bullying, victimization, fighting, and anger for both students with learning disabilities and
students without disabilities. Educational implications from the current study suggests
that schools should consider adopting multi-tiered anti-bullying programs that foster
increased social supports and incorporate targeted interventions for at-risk subpopulations
of students.
iii
To my wife and best friend Cathy
iv
Acknowledgements
Sir Isaac Newton once said, “If I have seen farther than others, it is because I was
standing on the shoulders of giants.” With that said, I would like to take a moment to
recognize the ‘giants’ who have allowed me to stand on their shoulders. First, I would
like to thank my Lord and Savior, Jesus Christ. His love, support, forgiveness, and
guidance have allowed me to accomplish feats far beyond my expectations.
Second, I would like to thank my advisor, mentor, and friend, Dr. Lisa Monda-
Amaya. Over the past four years, Dr. Monda-Amaya has provided me with the support
and guidance necessary to navigate through the doctoral program at the University of
Illinois. More importantly, Dr. Monda-Amaya has offered a shoulder to cry on, an ear to
listen, and a hand to push me in the right direction. While I was probably a little more
high maintenance than she anticipated, she has been my biggest advocate and supporter.
Third, I would like to thank Dr. Dorothy Espelage for her passion and work ethic.
Dr. Espelage has served as my mentor in the field of bullying, and has provided me with
countless opportunities to grow as an academician. She introduced me to large-scale
research, late night writing sessions, advanced statistics, and concise email responses.
Most importantly, Dr. Espelage taught me the importance of our research, and the value
of working until the task is complete.
I would also like to thank a series of professors at the University of Illinois for
their continued support throughout my doctoral program. Dr. Steven Aragon has served
as a friend, mentor, and colleague for the past four years. His advice has been
immeasurable, and he believed in me even when I didn’t believe in myself. I would also
like to recognize Dr. Karrie Shogren for her support and guidance throughout the
v
dissertation process. I was truly blessed to have such a brilliant mind guide my statistical
analyses, assist with interpretations, and shape my dissertation. Next, I would like to
thank Dr. James Shriner for providing me with the opportunity to work on his research
grant for four years and present at several national conferences. While I am sure that I
could write a paragraph about everyone in the Department of Special Education, for
brevity sake I would like to thank all of the faculty and staff for their continued support
and guidance throughout my doctoral career.
Fifth, I would like to recognize all of my friends who have unknowingly
embarked on this journey with me. I would like to thank Tony Plotner and Lance Neeper
for their encouragement throughout this process and allowing me to vent during our
brainstorming sessions. I would also like to recognize all of the doctoral students in the
Department of Special Education for their hard work, passion, and dedication to the field.
Additionally, I would like to thank Lee Wehrle and Jim Gusloff for their support and
unconditional friendship.
Sixth, I would like to thank my family for their love, support, and guidance over
the past four years. I would like to recognize my in-laws, Nancy Zimmerman and Jerry
Chalwick, for introducing me to the University of Illinois. I would also like to thank
Grandma and Grandpa Bales, Jason and Karen Bales, and Shirley and Dave Applegate
for their continued support. Most importantly, I would like to thank my mother and
father, Diana and Rollie Bell, for their unconditional love and guidance. My mother is my
rock, and I cannot express my gratitude for the life that she has provided for me.
Additionally, my father, Rollie, is a true father figure, and I only hope that I am half the
father for my children as he has been for me.
vi
Finally, I would like to thank my wife, partner, and best friend Cathy. Although
the stressors associated with doctoral student life were intense, I always knew that I had a
partner that I could lean on. Cathy’s glowing personality and optimistic outlook on life
are contagious, and these attributes allowed me to stay positive during the most stressful
times. She often reminded me why I decided to pursue my doctoral degree, and never let
me lose sight of the bigger picture. She believed in me through the sleepless nights,
exhaustion, triumphs, and failures. Most importantly, she viewed the pursuit of a doctoral
degree as a partnership, not an individual endeavor. Overall, the completion of the
doctoral program is a direct reflection of her unconditional love and encouragement.
Acknowledgements for Sources Used
Portions of Chapters 1 and 2 from “Bullying Among Students with Disabilities: Impact
and Implications” by C. A. Rose, 2010, In D. L. Espelage & S. M. Swearer (Eds.),
Bullying in North American Schools: A Socio-Ecological Perspective on Prevention and
Intervention (2nd ed.). Copyright 2010 Lawrence Erlbaum. Used with Permission.
Portions of Chapters 1 and 2 from “Bullying Perpetration and Victimization in Special
Education: A Review of the Literature” by C. A. Rose, L. E., Monda-Amaya, & D. L.
Espelage, 2010, Remedial and Special Education. Advance online publication.
Doi.10.1177/0741932510361247. Copyright 2010 Hammill Institute on Disabilities.
Used with Permission.
Research was supported by Centers for Disease Control & Prevention
(#1U01/CE001677) to Dorothy Espelage (PI).
vii
Table of Contents
List of Tables .................................................................................................................... ix
List of Figures ................................................................................................................... xi
Chapter 1 Introduction......................................................................................................1
General Statistics on Bully Perpetration and Victimization ..............................2
Involvement of Students With Disabilities ..........................................................4
Middle School Bullying and Sexual Violence:
Measurement Issues and Etiological Models .......................................................8
Fundamental Gaps in the Literature .................................................................12
Statement of the Problem ....................................................................................14
Purpose of the Study and Research Questions ..................................................14
Theoretical Framework .......................................................................................15
Summary ...............................................................................................................20
Chapter 2 Review of the Literature ...............................................................................21
Definition of Terms ..............................................................................................21
Victimization of and Perpetration by Students With Disabilities ...................26
The Influence of Class Placement on Bully Perpetration and
Victimization ........................................................................................................36
Disability Type and Severity ...............................................................................39
Disability Characteristics ....................................................................................42
Social-Ecological Factors .....................................................................................45
Conclusion ............................................................................................................48
Chapter 3 Methods ..........................................................................................................50
Site and Participant Demographics ....................................................................50
Description of Instrument and Measures ..........................................................57
Procedures ............................................................................................................70
Data Analysis ........................................................................................................73
Chapter 4 Results .............................................................................................................80
Sample Demographics .........................................................................................80
Missing Data Procedures .....................................................................................86
Construct Equivalence.........................................................................................94
Associations and Latent Mean Differences......................................................101
Demographic Predictors ....................................................................................106
Sense of Belonging and Social Supports
as Predictors in the Bully Dynamic ..................................................................110
Chapter 5 Discussion .....................................................................................................116
Overall Findings .................................................................................................118
Bullying Construct Measurement ....................................................................119
Direct Comparison of Student Groups ............................................................121
viii
Demographic Predictors ....................................................................................124
Predictive Relationships ....................................................................................129
Limitations of the Study ....................................................................................136
Educational Implications...................................................................................138
Future Research .................................................................................................140
Conclusion ..........................................................................................................142
References .......................................................................................................................143
Appendix A Student Behavior Survey .........................................................................165
Appendix B IRB School Record Review Approval .....................................................186
Appendix C Waiver of Informed Consent ...................................................................187
Appendix D Waiver of Informed Consent for School Record Review .....................189
Appendix E Descriptives for Crosstabs by Gender, Ethnicity, and
Disability .........................................................................................................................191
ix
List of Tables
Table Page
1 Means (Standard Deviations) of Bullying and Fighting Perpetration and
Victimization Among Students With and Without Disabilities Across 1009
Middle School Participants ....................................................................................11
2 Bronfenbrenner’s Social-Ecological Model of Child Development ......................17
3 Definitions of Bullying ...........................................................................................22
4 Victimization and Bullying Rates of Students With Disabilities ............................30
5 Sample Population Demographic Data by School ................................................51
6 Self-Reported Sample Demographics at Wave 4 ...................................................52
7 Demographic Data of Teachers From Sample District .........................................54
8 Sample Population Demographics ........................................................................54
9 Sample Population Disability Data .......................................................................55
10 Description of Measures Used on the University of Illinois
and Wellesley College: Student Behavior Survey ..................................................58
11 Measures Selected From the SBS for the Current Study .......................................65
12 Statistical Analyses and Items for Research Questions .........................................74
13 2Difference Tests for Variables Across Disability Type ......................................81
14 Descriptive Statistics for Students From the Two Groups .....................................82
15 2 Statistics for Cross Tabulations ........................................................................84
16 Missing Data Specifics per Item ............................................................................88
17 Item Parceling Procedure for the Eight Subscales ................................................91
18 Means and Standard Deviations by Subgroup for Individual Parcels ..................93
19 Loadings, Intercepts, and Estimated Latent Variance
From Strong Metric Invariance Model ..................................................................97
x
20 Fit Indices for Multi-Group Confirmatory Factor Analysis ..................................98
21 Mean Scores, Unique Residuals, and Square Multiple Correlations
for Individual Parcels Across Disability Groups ................................................100
22 Correlations Between the Latent Constructs .......................................................103
23 Latent Mean Scores Merged From 2-Group Model ............................................104
24 Fit Indices for Variance, Covariance, and Latent Means CFA
Evaluations ..........................................................................................................105
25 Estimates, Standard Errors, and Significance of Covariates ..............................107
26 Beta Weights and Z-Scores of the Final Structural Model ..................................112
E1 Crosstabs for Students Without Disabilities by Race ..........................................191
E2 Crosstabs for Students With Learning Disabilities by Race ................................193
E3 Crosstabs for Females Without Disabilities by Race ..........................................195
E4 Crosstabs for Males Without Disabilities by Race ..............................................197
E5 Crosstabs for Females With Learning Disabilities by Race ................................199
E6 Crosstabs for Males With Learning Disabilities by Race ....................................201
xi
List of Figures
Figure Page
1 Middle School Bullying and Sexual Violence data collection design .....................9
2 Espelage and Swearer’s (2004) Social-Ecological Framework
for Bullying/Victimization .....................................................................................18
3 Data collection waves and cohort groups used for current study ..........................73
4 Multi-group confirmatory factor analysis ..............................................................78
5 Theoretical structural equation model for predictive and preventative
factors associated with bullying, victimization, fighting, and
anger for two-group model ....................................................................................79
6 Missing data patterns of total items .......................................................................87
7 Strong metric invariance measurement model for multi-group
confirmatory factor analysis ..................................................................................99
8 Structural model with covariates and significant paths .......................................115
1
Chapter 1
Introduction
Bullying has emerged as one of the most fundamental problems facing our nation’s
schools to date (Espelage & Swearer, 2003). Since 1999, state legislators have taken a keen
interest in this issue, and a majority of the nation’s states have enacted legislation that prohibits
bullying and harassment, and have taken measures to report policies, programs, and procedures
to students and parents (Swearer, Espelage, & Napolitano, 2009). In addition to adopting
specific policies regarding bullying, schools are often encouraged to adopt research-supported
programs that focus on reducing perpetration and victimization through teacher awareness,
social skill development, and curricular instruction (Rose, Espelage, & Monda-Amaya, 2009).
While increased state mandates are a critical first step in reducing bullying, many of the
programs and policies neglect to provide targeted approaches for addressing marginalized
student populations.
When the continuum of the bullying dynamic is considered (i.e., bullies, victims, bully-
victims, bystanders), evidence suggests that it involves the overwhelming majority of the
nation’s student population (Espelage, Bosworth, & Simon, 2000). However, conventional
research has investigated this phenomenon in a whole school context by comparing students
based on general demographic descriptors (e.g., school, age, gender, race). Unfortunately, the
statistics and implications from these studies may significantly underestimate the prevalence of
bullying within certain subpopulations of students (Rose, Monda-Amaya, & Espelage, 2010).
For example, the National Center for Educational Statistics documented that 28% of adolescents
reported being victimized within a six month period prior to being surveyed (Dinkes, Cataldi,
Kena, & Baum, 2006), while several studies involving students with disabilities have yielded
2
victimization rates in excess of 50% (see Dawkins, 1996; Doren, Bullis, & Benz, 1996; Little,
2002; O’Moore & Hillery, 1989; Whitney, Smith, & Thompson, 1994). Therefore, consideration
must be given to the bullying dynamic as it relates to students with disabilities.
Understanding this discrepancy involves attending to several variables that may place
students with disabilities at a greater risk for involvement in bullying as both the victims and
perpetrators. While Whitney and colleagues stated (1994), “Often just being different in a
noticeable way can be a risk factor for being a victim” (p. 213), careful consideration must be
given to the factors that contribute to this “difference” for students with disabilities. Broadly
defined, students with disabilities include those who receive special education services for
academic, behavioral, physical, or functional performance. Generally, these students have an
Individualized Education Plan (IEP), but this definition may include students with 504 plans or
those who have been diagnosed via the Diagnostic and Statistical Manual of Mental Disorders
(DSM-IV; American Psychiatric Association, 2000). This chapter will focus on bullying among
students with disabilities as it relates to the discrepancies between students without disabilities,
the severity of the disability, classroom placement and instruction, and disability characteristics
that may place students with disabilities at a greater risk for being bullied.
General Statistics on Bully Perpetration and Victimization
While the national legislation focuses on increasing academic outcomes through the use
of evidence-based practices, behavioral interventions has become more germane to state
legislative bodies. At the present time, one of the most common and pervasive behavior
problems in the school setting is bullying (Espelage & Swearer, 2003), where 40 states have
enacted legislation that prohibits bullying and harassment (Swearer et al., 2009). The
pervasiveness of this phenomenon stems from a general outlook in which bullying and
3
victimization are regarded as social ritual (Brendtro, Ness, & Mitchell, 2001), a typical part of
the adolescent experience or even a student’s rite of passage (Carter & Spencer, 2006; Dawkins,
1996; Thompson, Whitney, & Smith, 1994; Walker, Ramsey, & Gresham, 2004). While
perpetration and victimization are fundamental issues facing our nation’s youth, much of the
recent research in the United States on the bullying phenomenon was preceded by a government
campaign called The Safe School Initiative (Vossekuil, Fein, Reddy, Borum, & Modzeleski,
2002). This collaborative initiative between the United States Secret Service and the Department
of Education examined planning and pre-attack thoughts and behaviors of the 41 perpetrators of
37 U.S. school shootings occurring between 1974 and 2000. While a clear and concrete
perpetrator profile could not be established, common characteristics were determined. Most
importantly, researchers discovered that approximately 71% of the school shooters had been
victimized prior to the incident (Vossekuil et al., 2002).
Following the Safe School Initiative, a national survey was conducted to determine the
prevalence of bullying perpetration in the United States (Nansel et al., 2001). Findings indicated
that approximately 30% of the school-age population experienced bullying either as a
perpetrator, victim, or bully-victim (students who bully others and who are victimized; Swearer
et al., 2009). More recently, the National Center for Educational Statistics documented that 28%
of adolescents reported being victimized within a six month period prior to being surveyed
(Dinkes et al., 2006). Espelage and colleagues (2000) conducted a survey in which they found
that only 19.5% of middle school students had not observed, been a victim of, or participated in
bullying perpetration within the last month of being surveyed. While several reports have
documented a decline in juvenile violence (Brener, Lowry, Barrios, Simon, & Eaton, 2005;
Dinkes et al., 2006; Dinkes, Kemp, Baum, & Snyder, 2009), evidence suggests that bullying
4
victimization and perpetration have remained relatively stable over the past decade (Garrity,
Jens, Porter, & Stoker, 2002; Swearer et al., 2009).
Involvement of Students With Disabilities
Although bullying consists of complex interactions between participants, research
suggests that students with disabilities are more likely to be the perpetrators and victims when
compared to their general education peers (Rose, Espelage, & Monda-Amaya, 2009). While few
empirical studies have examined bullying perpetration and victimization rates among American
schoolchildren with disabilities, international research has indicated that students receiving
special education services are the perpetrators and victims of more bullying occurrences than
their peers without disabilities (Conti-Ramsden & Botting, 2004; Kaukiainen et al., 2002;
Kumpulainen, Räsänen, & Puura, 2001; Marini, Fairbairn, & Zuber, 2001; Nabuzoka & Smith,
1993; O’Moore & Hillery, 1989; Sheard, Clegg, Standen, & Cromby, 2001; Singer, 2005;
Whitney et al., 1994; Yude, Goodman, & McConachie, 1998). Additionally, it has been
documented that students with disabilities may engage in more fighting behaviors (Rose et al.,
2009; Rose, Espelage, Aragon, & Elliott, under review) and students with learning disabilities
may exhibit more aggressive behaviors than students without disabilities (Kuhne & Wiener,
2000). Therefore, it is necessary to investigate the prevalence of bullying perpetration and
victimization among American students with disabilities.
In a recent review of the literature, Rose and colleagues (2010) determined that the
documented national average for adolescent victimization might underestimate the victimization
and perpetration rates of students with disabilities. Of the 32 articles reviewed, several authors
described victimization rates of students with disabilities in excess of 50%. Additionally, 13
articles explored perpetration among adolescents, and reported that on average students with
5
disabilities were twice as likely to be identified as perpetrators when compared to their general
education peers. Based on this review, it can be concluded that students with disabilities are at a
greater risk for involvement in the bullying dynamic when compared to their general education
counterparts (Rose, 2010; Rose et al., 2010).
When considering bullying among students with disabilities, attention must be given to
the spectrum of disabilities and the types of placements in which these students are served. For
example, inclusive and segregated settings may elicit varying rates of victimization and
perpetration based on educational practices, classroom structure, percentage of instructional time
dedicated to special education services, and disability labels (Rose et al., 2010). In their seminal
work, Whitney and colleagues (1994) investigated the victimization rates of 93 students with
disabilities and their demographically matched peers within an inclusive setting. Through
student and teacher interviews, the researchers determined that 55% of students with mild
learning difficulties and 78% of students with moderate learning difficulties experienced
moderate to severe levels of victimization. Conversely, only 25% of their demographically
matched peer group reported being victimized in the same setting. These findings are
corroborated in several studies in which students and teachers consistently nominated students
with disabilities in their classrooms as frequent victims of bullying (Nabuzoka, 2003; Nabuzoka
& Smith, 1993; Sabornie, 1994).
More recently, in an investigation of bullying, fighting, and victimization rates among a
large sample of American middle school (n = 7,331) and high school students (n = 14,315)
enrolled in general and special education programs, Rose and colleagues (2009), attempted to
replicate international findings. Data indicated that students with disabilities engaged in higher
rates of bullying and fighting perpetration, and were victimized more than their general
6
education peers. Additionally, the restrictiveness of educational placement (i.e., inclusion, self-
contained) served as a predictor for fighting and bullying perpetration. For example, as the
restrictiveness of placement increased, students engaged in higher rates of bullying and fighting
behaviors.
Potential factors associated with increased involvement in the bullying dynamic.
Although the involvement of American students with disabilities has a limited empirical base,
several theories have attempted to explain the discrepant rates of victimization and perpetration
between students with and without disabilities. While these hypotheses will be further explained
in Chapter 2, it seems appropriate to provided a brief synopsis within this section. Some studies,
for example, have documented a significant difference between victimization rates among
students with disabilities in inclusive and self-contained settings (Martlew & Hodson, 1991;
Morrison, Furlong, & Smith, 1994; O’Moore & Hillery, 1989; Sabornie, 1994). To explain this
discrepancy, researchers point to inclusive practices as a buffer for victimization because
inclusive settings may enhance social skill acquisition, improve overall social and academic
development (Brown et al., 1989), increase acceptance, reduce negative stereotypes (Martlew &
Hodson, 1991), and increase participation in classroom activities (Sabornie, 1994). Most
importantly, inclusive settings may allow students with disabilities to utilize peer behavior
models to learn, practice, and validate age appropriate social skills among their same aged peer-
group (Baker & Donlley, 2001; Mishna, 2003).
While all students are legally obligated to receive educational services in their “Least
Restrictive Environment,” disability characteristics may predict increased special education
services. This consideration is necessary because several behavioral characteristics of students
with disabilities may increase the likelihood of victimization and perpetration regardless of
7
service placement. For example, students with Emotional and Behavioral Disorders (EBD)
demonstrate the highest levels of perpetration when compared to other sub-groups of students
(Monchy, Pijl, & Zandberg, 2004; Van Cleave & Davis, 2006), while students with more
observable disabilities are victimized at a greater rate when compared to students with less
obvious disabilities (Dawkins, 1996). Additionally, students with learning disabilities are
identified as victims (Kaukiainen et al., 2002; Martlew & Hodson, 1991; Nabuzoka, 2003;
Nabuzoka & Smith, 1993; Sabornie, 1994), bullies (Kaukiainen et al., 2002; Nabuzoka & Smith,
1993), and aggressors (Kuhne & Wiener, 2000) more often than their general education peers.
This subgroup of students, represents approximately 40% of the population of students with
disabilities, and receives varying degrees of special education services (NCES, 2009). The
prevalence of involvement will be discussed in greater detail in Chapter 2.
Since each variable (i.e., disability characteristics, percentage of special education
services) can be considered a predictor, it is difficult to pinpoint which factor is most closely
related to increased involvement in the bullying dynamic. While these factors must be
considered, students with disabilities may be overrepresented within the dynamic for reasons
other than disability label and percentage of special education services. Students may be
victimized more frequently because they are too passive or exhibit timid responses that may
reinforce bullying behaviors (Sabornie, 1994). On the other hand, students with disabilities may
act too aggressively or misinterpret social stimuli because of social information processing
deficits (Burks, Laird, & Dodge, 1999; Crick & Dodge, 1994, 1996; Dodge et al., 2003;
Sabornie, 1994). Additionally, a growing body of literature supports the idea that students with
disabilities develop aggressive characteristics as a method of combating prolonged victimization
(Kumpulainen et al., 2001; O'Moore & Hillery, 1989; Singer, 2005; Van Cleave & Davis, 2006).
8
Overall, the literature points to poor social skills as the common contributing factor to increased
perpetration and victimization among students with disabilities (Baker & Donelly, 2001; Doren,
Bullis, & Benz, 1996; Kaukaiainen et al., 2002; Kuhne & Wiener, 2000; Llewellyn, 2000;
Miller, Beane, & Kraus, 1998; Woods & Wolke, 2004).
Middle School Bullying and Sexual Violence:
Measurement Issues and Etiological Models
Although evidence suggests that students with disabilities are overrepresented within the
bullying dynamic, the fundamental problem of bullying impacts the entire school community.
Therefore, the current project is in coordination with a larger, concurrent study entitled Middle
School Bullying and Sexual Violence: Measurement Issues and Etiological Models (MSBSV)
funded by the Center for Disease Control and Prevention (#1U01/CE001677) under the direction
of the Principal Investigator, Dr. Dorothy L. Espelage. The impetus for this overarching study is
to assess the interplay between sexual harassment/violence and bully perpetration during early
adolescence. Additionally, this study aims to examine psychometric dimensions of bullying and
sexual violence to explore unique and shared predictive and preventative factors associated with
bullying and sexual violence (Espelage, 2006).
The MSBSV project is a longitudinal study that follows 5 cohort groups of middle
schools students, from 4 different schools, over five waves of data collection. The overall
timeframe for this project is 3 years with each wave of data collection occurring in the fall and
spring semester of each academic year beginning in the spring of 2008 (see Figure 1). This
method of data collection allowed for the longitudinal and cross-sectional analysis of each
cohort group and all subgroups contained within (Espelage, 2006).
9
Figure 1. Middle School Bullying and Sexual Violence data collection design.
To assess the interplay between sexual violence and bully perpetration, the University of
Illinois and Wellesley College: Student Behavior Survey (SBS; Espelage & Stein, 2006) was
developed. This instrument will be explicitly described in Chapter 3, but it is pertinent to provide
a brief synopsis within this section. The SBS is comprised 50 separate scales measuring over 30
different constructs including scales related to bullying, sexual violence/harassment, school
belonging, social supports, and exposure to violence.
While these data from the MSBSV have not yet been evaluated longitudinally, several
notable findings have been established cross-sectionally at Wave 1. In the Spring of 2008,
surveys were administered to 1009 middle schools students ranging from 6th through 8th grade.
Findings at Wave 1 suggested that bullying was strongly associated with homophobic teasing
and perpetration, but was only slightly correlated with sexual harassment perpetration (Espelage,
2010). This finding suggests that bullies and sexual violence perpetrators are not the same
students.
According to Espelage’s (2010) executive summary, when gender was considered cross-
sectionally, separate predictors emerged for bullying and sexual violence. For males, individual,
10
peer, family, and neighborhood factors could explain 50% of the variance for bully perpetration
and 30% of the variance for sexual harassment perpetration. Specific predictors of bullying for
males included higher levels of anger, greater involvement in delinquent behaviors, greater bully
perpetration directed toward siblings, and alcohol and drug use. Specific predictors for sexual
violence perpetration for males included less sense of school connectedness, greater involvement
in delinquent behavior, greater bully perpetration directed toward sibling, and increased
pornography consumption (Espelage, 2010).
For females, individual, peer, family, and neighborhood factors explained 70% of the
variance for bully perpetration and 30% of the variance for sexual harassment perpetration.
Specific predictors of bullying for females included higher levels of anger, greater involvement
in delinquent behavior, greater bullying perpetration directed toward sibling, increased attitudes
that are supportive of aggression, and increased family violence. Specific predictors of sexual
violence perpetration for females included higher levels of anger, greater involvement in
delinquent behavior, greater bullying perpetration directed toward sibling, increased family
violence, greater alcohol and drug use, less parental supervision, increased pornography
consumption, and increased dismissive attitudes toward sexual harassment (Espelage, 2010).
In addition to the overall predictors described above, Rose and colleagues (under review)
examined bully perpetration, fighting, and victimization among students with disabilities cross-
sectionally at Wave 1. Students were asked whether they had a disability, and were placed in a
self-reported dichotomy (i.e., student with disability, student without disability). Of the 1009
students in Wave 1, 18% (n = 182) indicated that they had a disability. The overall purpose of
this cross-sectional study was to determine if students with disabilities were overrepresented
within the bullying dynamic.
11
A multivariate analysis of variance (MANOVA) was conducted with the University of
Illinois Bully, Fighting, and Victimization scales (Espelage & Holt, 2001) as dependent
variables and special education status, gender, and school as the independent variables. An
overall MANOVA effect was found for special education status (Wilks’ λ = .99, p < .01, η2 =
.02), and univariate analyses indicated that the groups differed on victimization and fighting (η2s
= .02, .01), but did not differ significantly on bullying perpetration (η2s = .001). Mean scale
scores for victimization and fighting perpetration by special education interaction indicated that
students with disabilities were victimized more and exhibit higher levels of fighting behaviors
than their general education peers. However, the mean scale scores for bullying perpetration did
not differ significantly across special education status (see Table 1). This preliminary study
served as the foundation for the proposed study with the addition of primary labels and
percentage of time the students receive special education services.
Table 1
Means (Standard Deviations) of Bullying and Fighting Perpetration and Victimization Among
Students With and Without Disabilities Across 1009 Middle School Participants
Students without
Disabilities
Students with
Disabilities
F
Bullying 1.42 (.57) 1.47 (.68) .84
Fighting 1.50 (.70) 1.66 (.84) 6.07
*
Victimization 1.50 (.72) 1.77 (1.03) 15.85
**
*p< .05
** p < .001
12
Fundamental Gaps in the Literature
In examining the existing literature and foundational studies, several fundamental gaps
emerged. First, and most importantly, the body of literature is extremely limited, and often
involves homogenous subgroups of students outside the United States (see Rose et al., 2010).
This issue is critical to our understanding of bullying among students with disabilities because
American schools are comprised of diverse learners, populations, and subcultures that may not
be represented in studies outside the U.S. Therefore, it is essential to replicate international
efforts to ensure the phenomenon transcends international studies.
Second, the notion of disability labels versus percentage of time students receive special
education services has been relatively untested. As noted previously, students are required by
law to receive their educational services within their Least Restrictive Environment. While Least
Restrictive Environment is a legal requirement, it is commonplace for students with disabilities,
especially those with learning disabilities, to receive varying degrees of special education
services (varying percentages of time receiving services).
Students with learning disabilities represent the largest proportion of students with
disabilities (NCES, 2009), and often are overrepresented as bullies, victims, and aggressors in
the bullying dynamic when compared to their general education counterparts (Kaukiainen et al.,
2002, Kuhne & Wiener, 2000; Martlew & Hodson, 1991; Nabuzoka, 2003; Nabuzoka & Smith,
1993; Sabornie, 1994). Theoretically, this overrepresentation could be a function of disability
characteristics, percentage of time receiving special education services, or an interaction
between the characteristics and services. However, the interplay between these variables has not
been empirically validated, and it is necessary to investigate whether disability labels or
13
percentage of time the student receives special education services serves as predictors for
involvement in the bullying dynamic.
Third, aggression and bullying have been used synonymously in the literature, and it is
difficult to confirm that students with learning disabilities actually engage in more bullying than
their peers without disabilities. This issue has become more pressing because bullying is
generally considered a social construct (Espelage & Swearer, 2009; Swearer & Espelage, 2004),
and students with disabilities, including those with learning disabilities, are often characterized
as having below average social skills (Baker & Donelly, 2001; Doren et al., 1996; Kaukaiainen
et al., 2002; Kuhne & Wiener, 2000; Llewellyn, 2000; Miller et al., 1998; Woods & Wolke,
2004). Consequently, it is important to understand the discrepancies between definitions of
aggression and bullying to determine if students with learning disabilities are actually engage in
more bullying behaviors or if they are more accurately identified as aggressive.
Finally, unique predictors and preventative factors associated with bullying among
students with learning disabilities are not present in the current literature base. For example,
existing literature implies that students with learning disabilities are overrepresented within the
bullying dynamic, but fail to provide definitive explanations of this overrepresentation.
Moreover, it is conceivable to believe that some subgroups of students are more at-risk for
involvement than others, but mediators and moderators for bullying and victimization among
these subgroups remain untested. These findings could serve as a tool for increase intervention
efforts for specific subpopulations of students. Therefore, the current study attempts to add to the
literature by addressing these fundamental gaps.
14
Statement of the Problem
Although national mandates have traditionally neglected provisions for behavioral
interventions (see IDEA, 1997; IDEIA, 2004; NCLB, 2001), evidence suggests that bullying
involves the overwhelming majority of the nation’s students (Espelage et al., 2000). As stated
previously, a majority states in the U.S. have enacted legislation that prohibits bullying and
harassment, and have taken measures to report policies, programs, and procedures to students
and parents (Swearer et al., 2009). While increased state mandates regarding bully prevention
are necessary to eliminate perpetration and victimization, empirical evidence suggests that
certain subgroups of students are at-risk for increased involvement in the bullying phenomenon.
At the present time, research on bullying indicates that approximately 28% of American school
children have been victimized during their educational career (Dinkes et al., 2006), and
approximately 13% of American school children exhibit bullying characteristics (Nansel et al.,
2001). Unfortunately, an overwhelming majority of the literature on bullying is reported in a
“whole school context,” and neglects to collect data or report findings for individual sub-groups
of students who may be at increased risk for victimization or perpetration (Rose et al., 2009;
Rose et al., 2010). When sub-group data are collected and reported, it becomes evident that
disability status should be factored in when exploring the bullying dynamic.
Purpose of the Study and Research Questions
The purpose of the proposed study is to examine factors associated with the involvement
of students with learning disabilities in the bullying dynamic. First, this study is designed to
determine if the constructs that are typically used to define bullying are representative of
students with learning disabilities. Second, the influence of the learning disability label on self-
reported involvement will be examined. Third, this study will investigate demographic (i.e.,
15
gender, ethnicity, grade point average, school involvement, percentage of special education
services received) and social (i.e., sense of belonging, social supports) predictors for the
involvement of students with learning disabilities and students without disabilities. The
following five research questions will guide this study.
1. Can the constructs that define the bullying dynamic be measured equivalently across
students with learning disabilities and students without disabilities?
2. To what extent does being identified with a learning disability influence associations and
mean levels of bullying, victimization, fighting, anger, sense of belonging, and social
supports?
3. To what extent do gender, ethnicity, grade point average, participation in extracurricular
activities, and for students with learning disabilities, percentage of time receiving special
education services, predict involvement in the bullying dynamic for students with?
4. To what extent does sense of belonging and social supports predict involvement within
the bullying dynamic for students with learning disabilities and students without
disabilities?
Theoretical Framework
According to Swearer and colleagues (2009), “bullying/victimization does not occur in
isolation and, in fact, results as a complex interaction between the individual and his or her
family, peer group, school community, and societal norms” (pp. 7-8). Additionally, Dempsey,
Fireman, and Wang (2006) reported that these interactions influence the stability or fluidity of
bully and victim roles. Due to the complexity of these interactions, the foundational framework
for this study is based on Bronfenbrenner’s (1977, 1979, 1986) Social-Ecological Model of
Child Development, which was extended by Espelage and Swearer (2004) and Swearer and
colleagues (2006) to include predictive models of bully perpetration and victimization (Swearer
et al., 2009). This model will be used as the initial framework to identify predictors and
preventative factors uniquely associated with students with learning disabilities. While this is not
a novel model for bully perpetration and victimization investigations among school aged
16
students, it has yet to be utilized as an investigative tool for understanding bullying among any
subgroup of students with disabilities.
The Social-Ecological Model was selected to serve as the theoretical framework for this
study for several reasons. First, experts in the field of bullying research believe this model
encapsulates all factors that perpetuate bullying among school aged students (Espelage &
Swearer, 2004; Swearer et al., 2006). This belief is based on the notion that bullying is a series
of complex interactions stemming from environmental variables such as family interactions, peer
group involvement, community norms, and societal influences (Swearer et al., 2009). Second,
the Social-Ecological Model allows for factor comparison among specific subgroups of students
in order to investigate unique contributors to increased bully involvement. For example, by using
this model, we can investigate the magnitude of influence associated with the label of learning
disability in comparison to the aforementioned environmental variables and the outcome
variables of bullying, victimization, fighting, and anger. Finally, this model accounts for
individual and environmental changes that occur over time (i.e., chronosystems), which may
influence the stability and fluidity of contextual roles (i.e., bully, victim, bully-victim).
Bronfenbrenner’s (1977, 1979, 1986) Social-Ecological Model of Child Development is the
basis for the Social-Ecological Framework for Bullying/Victimization, and includes five distinct
domains. First, Microsystems refers to the complex relations between the individual and their
immediate setting. Second, Mesosystems refers to interrelations among an individual’s major
settings at a specific point in time. Third, Exosystems refers to formal and informal social
structures that impinge upon the individual’s immediate setting. Fourth, Macrosystems refers to
institutional patterns of the culture or subculture. Fifth, since the interactions between the
different systems are not static, Bronfenbrenner (1986) introduced Chronosystems , which refers
17
an individual’s developmental changes overtime within the environment that the individual
resides (see Table 2; Brofenbrenner 1977, 1979, 1986).
Table 2
Bronfenbrenner’s Social-Ecological Model of Child Development
System Definition
Microsystem* A microsystem is the complex of relations between the developing
person and environment in an immediate setting containing that
person.
Mesosystem* A mesosystem comprises the interrelations among major settings
containing the developing person at a particular point in his or her
life.
Exosystem* An exosystem is an extension of the mesosystem embracing other
specific social structures, both formal and informal, that do not
themselves contain the developing person but impinge upon or
encompass the immediate settings in which that person is found,
and thereby influence, delimit, or even determine what goes on
there.
Macrosystem* A macrosystem refers to the overarching institutional patterns of
the culture or subculture, such as the economic, social, educational,
legal, and political systems, of which micro-, meso-, and exo-
systems are the concrete manifestations.
Chronosystem** A chronosystem refers to the influence on the individual’s
development of changes over time in the environment in which the
person is living
Note. * refers to excerpts from pp. 514-515 of Bronfenbrenner’s 1977 article titled Toward an
experimental ecology of human development. Full reference located in reference section.
** refers to excerpts from p. 724 of Bronfenbrenner’s 1986 article titled Ecology of the family
as a context for human development: Research perspectives. Full reference located in the
reference section.
Espelage and Swearer’s (2004) extension of this model, the Social-Ecological
Framework for Bullying/Victimization, incorporates the five domains that coincide with
Bronfenbrenner’s model. First, Individual Factors include intrapersonal factors such as
depression, anxiety, and impulsivity. Second, Family Factors include interpersonal relationships
18
between the individual and his/her immediate family members. Third, Peer Group and School
Factors include school climate and interpersonal relationships between the individual and
his/her peers, teachers, and other school personnel. Fourth, Community Factors include
community resources, neighborhood influences, and school-community partnerships. Finally,
Societal Factors include global influences such as media and popular culture (Swearer et al.,
2009). Figure 2 is a graphic depiction of the relationships involved within the Social-Ecological
Framework for Bullying/Victimization.
Figure 2. Espelage and Swearer’s (2004) Social-Ecological Framework for
Bullying/Victimization.
Theoretically, an infinite number of variables can influence the interactions between the
individual and each subsequent factor grouping within the Social-Ecological Framework for
Bullying/Victimization (Espelage & Swearer, 2004). Historically, empirical evidence has
suggested that social behaviors, influences, and supports may set students with learning
disabilities apart from their peers without disabilities (Pearl, Donahue, & Bryan, 1986). These
social factors are embedded within the first three tiers of the Social-Ecological Framework (i.e.,
19
individual, familial, peer and school factors). At the individual level, students with learning
disabilities have historically reported higher levels of loneliness (Pavri & Luftig, 2000) and
lower levels of social competence and sense of belonging (Pearl et al., 1986).
These individual factors are initially influenced by parental and familial perceptions and
interactions. Over the past three decades, researchers have attempted to ascertain the perceptual
differences between parents with children who have learning disabilities and parents who have
children without disabilities. Foundational studies have determined that parents of children with
learning disabilities maintain lower expectations (Boersma & Champman, 1982; Tollison,
Palmer, & Stow, 1987) and perceive their children more negatively than parents of children
without disabilities (Bryan, Pearl, Zimmerman, & Matthews, 1982; Owen, Adams, Forrest,
Stolz, & Fisher, 1971).
Individual factors are also directly influenced by peer and school factors. For decades,
student s with learning disabilities have been described by their classmates as unpopular (Kuhne
& Wiener, 2000; Martlew & Hodson, 1991; Nabuzoka & Smith, 1993; Pavri & Luftig, 2000).
Similarly, students with learning disabilities who report medium to high levels of loneliness, also
report lower levels of worth received from their teachers (Pavri & Monda-Amaya, 2001).
Educators often maintain lower academic and behavioral expectations for students with learning
disabilities (Boersma & Chapman, 1982), and often perceive them to have below average
socializing behaviors (Nowicki, 2003). This feeling of perceived rejection by peers and teachers,
compounded by low expectations, may lead to a decreased sense of belonging.
While each of the aforementioned factors will be explicitly described in Chapter 2, the
components of the theoretical framework that guide this study are based on the complex
interactions between individual, familial, and peer and school influences. Grounded in the
20
foundational literature, students with learning disabilities often struggle with membership and
social supports. Therefore, sense of belonging, family social support, peer social support, and
school support will be investigated to determine if these variables serve as unique predictors for
students with learning disabilities’ involvement within the bullying dynamic.
Summary
An overview of the background of bullying research, significance regarding students
with learning disabilities, purpose, theoretical framework, research questions of the current
study, and a description and preliminary findings of the global study were presented. This study
will investigate students with learning disabilities involvement with the bullying dynamic. More
specifically, this study will examine unique predictors and preventative factors associated with
students identified with a learning disability.
21
Chapter 2
Review of the Literature
The literature currently available in bully perpetration and victimization among students
with disabilities is reviewed in this chapter. More specifically, the following variables related to
involvement will be reviewed: a) definitions and participants b) restrictiveness of placement, c)
disability type, d) severity of disability, e) disability characteristics and f) social-ecological
factors as it relates to the current study. Overall, a global view of involvement in bullying among
students with disabilities as well as a justification for the current study will be discussed within
this chapter.
Definition of Terms
To understand bullying conceptually, it is necessary to explore the components that
directly influence the defining characteristics of perpetration and victimization. Definitions of
bullying vary considerably and as a consequence empirical data around bullying often yield
inconsistent results (Miller et al., 1998). Generally, bullying is defined as “a negative and often
aggressive or manipulative act or series of acts by one or more people, against another person or
group of people usually over a period of time. It is abusive and is based on an imbalance of
power” (Sullivan, Cleary, & Sullivan, 2004, pp. 4-5). The concept is complex, with perpetration
and victimization rarely occurring in isolation of other behaviors. Bullying can only be
understood in relations among individuals, families, peer groups, schools, communities and
cultures (Smith, 2004; Swearer & Espelage, 2004). Table 3 demonstrates inconsistencies in
operational definitions found in the literature.
22
Table 3
Definitions of Bullying
Citation Definition
Dawkins, 1996, p. 603 Bullying is the intentional, unprovoked abuse of power by
one or more children in order to inflict pain or cause
distress to another child on repeated occasions.
Olweus, 1993, p. 9 A student is being bullied or victimized when he or she is
exposed, repeatedly and over time, to negative actions on
the part of one or more other students.
Nansel et al., 2001, p. 2095 A student is being bullied when another student, or a
group of students, say or do nasty and unpleasant things to
him or her. It is also bullying when a student is teased
repeatedly in a way he or she doesn’t like. Any form of
verbal or physical hurtful behavior, such as name-calling,
punching, repeated teasing, kicking, hitting, spreading
malicious rumors, pestering, socially isolating can be
considered bullying if the peer persists with it after it is
apparent that the victim is traumatized by what is being
said or done.
O’Moore & Hillery, 1989, p. 431 Bullying is longstanding violence, mental or physical,
conducted by an individual or a group and directed against
an individual who is not able to defend himself/herself, in
the actual situation.
Although definitions vary across studies, three commonalities emerged (Espelage &
Swearer, 2003; Garrity et al., 2002; Langevin, Bortnick, Hammer, & Wiebe, 1998; Marini et al.,
2001; Miller et al., 1998; Nansel et al., 2001; Walker et al., 2004). First, for an act to be
considered bullying, there must be an imbalance of physical, social, or emotional power between
the victim and the bully. Second, the act of perpetration is systematic with intent to cause
emotional or physical harm to the victim. Third, victimization and/or perpetration are generally
repeated over the course of days, months, or years. In 1995, Dr. Olweus introduced a fourth
concept that should be considered, that of unequal level of effect, in which the victim is left
traumatized while the bully maintains a lack of concern and compassion.
23
Participants in the bullying dynamic. Bullying perpetration and victimization involve
the overwhelming majority of the school population because involvement falls on a participatory
continuum (Espelage & Swearer, 2003). The bullying dynamic includes three possible
participants: (a) the bully, (b) the victim, and (c) the bystander (Marini et al., 2001; Olweus,
1993; Walker et al., 2004). All participants play an integral role by engaging in, experiencing, or
reinforcing the aggressive behavior.
A bully is defined as an individual who perpetrates emotional or physical power over the
victim. Bullies can be classified into three categories: (a) aggressive bully, (b) anxious bully, and
(c) passive bully (Olweus, 1993). An aggressive bully usually displays violent characteristics
and the desire to dominate others. The passive bully is often less violent and aggressive, and
usually plays a supporting role to the aggressive bully. An anxious bully is generally a bully-
victim who has adopted bullying behaviors as a way to combat victimization (Olweus, 1993).
However, it is difficult to characterize the bully because s/he may exhibit either negative
(e.g., low self control, poor academic performance, externalizing behaviors, alcohol abuse) or
desirable (e.g., classroom leader, popular, high spirited, active) personality traits (Kumpulainen
et al., 1998; Marini, Koruna, & Dane, 2006; Miller et al., 1998; Nansel et al., 2001; Perren &
Alsaker, 2006). Perpetration is reinforced by social or peer group dynamics. These dynamics,
may become established at a young age (Perren & Alsaker, 2006) and exacerbate prolonged
bullying perpetration (Espelage, Holt, & Henkel, 2003; Espelage & Swearer, 2003). While data
are available on bullying behavior, it is difficult to profile a bully on demographic, physical, or
social characteristics because of the heterogeneity across students.
Victims of bullying have been classified into two separate subgroups: a) the passive
victim and b) the bully-victim. Passive victims account for 80 to 85 percent of the victimized
24
population (Olweus, 2003). Generally the passive victim does not aggress or act out toward the
bully, and is characterized as being physically weaker, having fewer friends, demonstrating
lower self-esteem, being rejected by peers, being dependent on others, having observable
differences, or possessing weaker social skills (Kumpulainen et al., 1998; Marini et al., 2006;
Nansel et al., 2001; Whitney, Nabuzoka, & Smith, 1992). Research has indicated that passive
victims may have preexisting internalizing behavior problems prior to school enrollment that
could serve as a predictor for victimization (Arseneault et al., 2006).
Conversely, the bully-victim develops bullying characteristics as a result of exposure to
victimization. This group of victims is often described as having internalizing and externalizing
behavior problems, being reactively aggressive, maintaining poor interpersonal relationships, or
displaying a negative demeanor (Kumpulainen et al., 1998; Marini et al., 2006; Nansel et al.,
2001). These findings imply that students may be predisposed to or develop social roles at a
young age, and early behavior problems may serve as a predictor for future victimization
(Schwartz, McFadyen-Ketchum, Dodge, Pettit, & Bates, 1999) and possible identification for
special education placement. Overall, victims of bullying may possess or develop character traits
that have long-term consequences and adversely impact their social, emotional, or academic
development.
In addition to the bully and victim, bystander participation and support networks that
reinforce perpetration should be examined (Smith, 2004). A bystander is not directly involved in
the act of bullying but can reinforce the bully (observer) or support the victim (defender; Marini
et al., 2006). Bystanders may include followers (who actively engage in bullying after the initial
onset), supporters (reinforcing the bully but not actively engaging), passive supporters (support
the bully but do not take an open stand), disengaged onlookers (watch but do not support either
25
party), possible defenders (dislike the bully but do not intervene), and defenders (help the victim
when they feel it is appropriate) (Olweus, 2003; Salmivalli, Karhunen, & Lagerspetz, 1996).
Types of bullying. Bullying involves proactive or reactive aggression through direct or
indirect means (Doll & Swearer, 2006; Espelage & Swearer, 2003; Walker et al., 2004). The US
Department of Education identified four distinct categories of bullying perpetration: (a) physical,
(b) verbal, (c) indirect (i.e., relational, emotional, social), and (d) sexual (Walker et al., 2004).
Researchers suggest aggression is more direct during the early stages of educational
development, becoming more indirect with age (Björkqvist, 2001; Björkqvist, Österman, &
Kaukiainen 1992; Monks, Smith, & Swettenham, 2005). Björkqvist et al. (1992) noted that
physical, verbal, and indirect aggression followed distinct developmental phases. Younger
students without well-developed verbal or social skills resort to physical aggression. As verbal
skills develop, they transition to less physical forms of aggression. Finally, as social skills
develop and students learn to analyze and manipulate situations in their favor, they used more
indirect means of aggression. Although developmental stages of aggression differ, Björkqvist et
al. (1992) note that physical, verbal, and indirect aggression can be observed throughout each
stage.
Physical bullying can range from intentional shoving to aggressive fighting, and may
include damage to personal property. Verbal bullying can consist of intimidation, abusive
language, mimicking, and racist remarks. It often begins with teasing but can transition into
threats of violence. Relational (indirect) bullying is purposeful manipulation and damage to the
victim’s peer relationships (Crick & Grotpeter, 1995), and occurs when the bully tells lies,
spreads rumors, ignores, or intentionally isolates a victim in order to destroy or damage the
victim’s reputation (Doll & Swearer, 2006; Hill, 2003; Marini et al., 2001; Sullivan et al., 2004;
26
Walker, Colvin, & Ramsey, 1995). One more recent and common form of relational bullying is
cyber-bullying. Sexual bullying includes sexually explicit language and/or sexually abusive
actions, and is more accurately described as sexual harassment (American Association of
University Women Educational Foundation, 1993, 2001). Although indirect bullying and sexual
harassment are the focus of increased study, the majority of extant special education literature
addresses verbal and physical aggression.
Because bullying can be defined so broadly (physical, verbal, indirect, and sexual), it is
important to understand contexts in which behaviors are not characterized as such. Three types
of aggression typically are not interpreted as bullying: instrumental, retaliatory, and jostling.
Instrumental aggression occurs when someone takes a stand to defend their property, reputation,
or the well being of a peer. Retaliatory aggression, generally interpreted as a “typical” physical
altercation, which is impulsive and displayed in the “heat of the moment.” Finally, jostling
(rough and tumble play) is perceived as an enjoyable and mutually reinforcing interaction (Doll
& Swearer, 2006). Most importantly when two students of similar strength or social standing
fight or quarrel, their behaviors are not generally regarded as bullying (Nansel et al., 2001;
Olweus, 1993). Although assessment of intent is desirable, the examples above do not
demonstrate an imbalance of power, repetition of occurrence, intent to cause harm, or unequal
levels of effect.
Victimization of and Perpetration by Students With Disabilities
The aforementioned defining characteristics allow for the comparison of bullying
involvement across specific subgroups of student. Interestingly, the belief in a social hierarchy in
our system of education, “in which bullying and victimization are generally considered a social
ritual, a typical part of adolescent experience, or even a student’s rite of passage” (Rose et al.,
27
2010, p. 1) may prove to be more detrimental for students with disabilities. While evidence
suggests that special education status (being identified with a disability) does not directly predict
victimization among primary aged students (Woods & Wolke, 2004), pre-school aged victims
may be characterized as having preexisting internalizing problems (Arseneault et al., 2006).
These internalizing problems may be exacerbated by the early development of group dynamics
in which students migrate into social clusters based on social, physical, or environmental
similarities (Perren & Alsaker, 2006). The development of these early social clusters may
exclude students with disabilities, because evidence suggests that students with disabilities tend
to be regarded as unpopular and have fewer close friendships than students without disabilities
(see Baker & Donelly, 2001; Davis, Howell, & Cooke, 2002; Martlew & Hodson, 1991;
Nabuzoka & Smith, 1993; Whitney et al., 1994), thereby placing them at a greater risk for
victimization.
Although special education status may not serve as a predictor for victimization at the
primary level, as students’ progress through their educational careers, the discrepancy between
students with and without disabilities becomes increasingly more evident. Contextually, special
education status may not be a direct predictor during the early stages of education because
cognitively, students may not be able to identify the differences, the disability may not be
noticeable, or the disability may yet to have been identified (Langevin et al., 1998; Monks et al.,
2005). Presumably, once these differences have been established within a social context,
disability status emerges as a potential predictor for involvement within the bullying dynamic
across all groups with disabilities. This broad assumption is grounded in the majority of the
extant literature that identifies adolescents with disabilities as being victimized significantly
more often than their general education peers (Rose et al., 2010).
28
It is important to note that when general and special education are viewed as a dichotomy
(i.e., presence or absence of a disability), research suggests that students with disabilities are
victimized significantly more than students without disabilities. For example, typical estimates
suggest that approximately 20 to 30 percent of the student population have experienced bullying
either through victimization or perpetration (Rose et al., 2010). Conversely, several reports
suggest that students with disabilities are victimized at least twice as much as their general
education peers (Kaukiainen et al., 2002; Monchy et al., 2004; Nabuzoka & Smith, 1993). More
specifically, by making the dichotomous distinction between general and special education, Rose
and colleagues (under review) found that in a large-scale sample of middle school students (n =
1009), students with disabilities reported significantly higher rates of victimization as compared
to their general education peers (see Table 4).
Additionally, significant differences between students with and without disabilities are
not necessarily specific to victimization. A growing number of research reports have
investigated the bullying behaviors of students with disabilities. While approximately 13% of the
American school population exhibit bullying characteristics (Nansel et al., 2001), several
research reports suggest that students with disabilities are identified as bullies twice as often as
students without (Dawkins, 1996; Kumpulainen et al., 2001; Rose et al., 2009; Woods & Wolke,
2004). However, escalated victimization rates among students with disabilities may lead to
increased bullying rates, because victimized students may develop aggressive characteristics to
combat prolonged victimization (Kumpulainen et al., 2001; O’Moore & Hillery, 1989; Van
Cleave & Davis, 2006; see Table 4).
Unfortunately, bullying and overt aggression may be interpreted synonymously even
though the terms are distinctly different, which may negate findings that suggest students with
29
disabilities engage in more bullying behaviors than their peers. For example, Rose and
colleagues (under review) found that students with and without disabilities reported similar rates
of bullying behaviors, but students with disabilities reported significantly higher rates of fighting
behaviors. Interestingly, students without disabilities who reported being victimized also
reported higher levels of bullying behaviors, while students with disabilities who reported being
victimized reported higher levels of fighting behaviors. These findings suggest that victimization
may lead to more aggressive behaviors in students with disabilities, but not necessarily more
bullying behaviors (Rose et al., under review).
The distinction between students with and without disabilities, in reality, is more
complex than a simple dichotomous approach. While the term “disability” is used to refer to a
large subgroup of students, in actuality, the term refers holistically to 13 separate disability
categories that maintain different eligibility criteria and service needs (Smith, 2007). However,
eligibility criteria may differ from state-to-state, and each disability category maintains a range
of severity. This range of severity leads to a range of supports and instructional placements for
students with disabilities. Therefore, it becomes necessary to breakdown discrepancies in bully
involvement for students with and without disabilities in terms of class placement (i.e., inclusive
classrooms, segregated settings), the severity and overt nature of the disability, and the specific
disability characteristics.
30
Table 4
Victimization and Bullying Rates of Students With Disabilities
Author Disability Type Victimization Rates Bullying Rates
Baker & Donelly, 2001
Fragile X Syndrome 100% (n = 4) Not Measured
Bramston, Fogarty, &
Cummins, 1999
Intellectual
Disabilities
37% - Victimized
47% - Teased
25%, 30% - Control
Not Measured
Conti-Ramsden & Botting,
2004
Language Impairment 36% - At-Risk for Victimization 17% - Bullied Others
Davis et al., 2002
Stuttering 37.5% - Victimized
10.6% - Control
Not Measured
Dawkins, 1996
Observable and
Unobservable
Disabilities
Observable Disabilities:
50% - At Least Once
30% - On a Regular Basis
9% - Severe
Unobservable Disabilities:
21% - At Least Once
14% - On a Regular Basis
10% - Severe
Not Measured
Doren et al., 1996
Adult Transition 54% - Victimized Not Measured
Fuijki, Brinton, Isaacson, &
Summers, 2001
Language Impairment >1% - Victimized Not Measured
(continued)
31
Table 4 (continued)
Author Disability Type Victimization Rates Bullying Rates
Kaukiainen et al., 2002
Learning Disabilities 10.7% - Victimized
6.3% - Control
21.4% - Bullied Others
6.3% - Control
Knox & Conti-Ramsden, 2003
Language Impairment 36.2% - Total Special Education
Victims
14.9% - Mainstream Setting
21.3% - Pull-Out Setting
12% - Control
Not Measured
Kuhne & Wiener, 2000
Learning Disabilities
Not Measured 83% of the Aggressive Group
(LD)
Kumpulainen et al., 2001
Psychiatric Disorders 24.8% - Victimized 5.7% - Total Population
Langevin et al., 1998
Stuttering 59% - Victimized (For Stuttering)
56% - On a Regular Basis
69% - Victimized (For Something
Other than Stuttering)
50% - On a Regular Basis
Not Measured
Little, 2002
Asperger’s Syndrome 94% - Victimized During the Past
Year
Not Measured
Llewellyn, 2000
Physical Disabilities 67% - Severely Victimized Not Measured
Marini et al., 2001
Developmental
Disabilities
28% - Victimized 13% - Bullied Others
(continued)
32
Table 4 (continued)
Author Disability Type Victimization Rates Bullying Rates
Martlew & Hodson, 1991
Learning Disabilities Students with disabilities reported
significantly more teasing/bullying
than their mainstream peers (this was
especially true for older students).
Not Measured
Monchy et al., 2004
Behavior Problems 50% - Victimized
50% - Rejected
19% - Control
Not Measured
Morrison et al., 1994
Various Disabilities Special day class students
experienced higher rates of verbal
assaults and bullying when compared
to the other subgroups.
Not Measured
Nabuzoka, 2003
Learning Disabilities Teachers and peers nominated
students with learning disabilities as
being victims of bullying significantly
more than students without
disabilities.
Not Measured
Nabuzoka & Smith, 1993
Learning Disabilities 25% - Rejected
9% - Control
66.7% - Female with Learning
Disabilities Rejected
7.4% - Control
No Significant Difference for Males
20% - Females with LD
1.5% - Females without LD
No Significant Difference for
Males
Norwich & Kelly, 2004
Various Disabilities 84% - Victimized Not Measured
(continued)
33
Table 4 (continued)
Author Disability Type Victimization Rates Bullying Rates
O’Moore & Hillery, 1989
Various Disabilities 67.9% - Remedial Class Victimized
17.5% - Frequently
77.2% - Special Class Victimized
14.3% - Frequently
62.1% - Control
6.1% - Control (Frequently)
43.1% - Remedial Class
68.6% - Special Class
42% - Control
Rose et al., 2009 Disabilities Not
Specified
18.5% - Inclusive Settings
21.7% - Self Contained Settings
12.0% - Control
15.6% - Inclusive Settings
20.9% - Self Contained
Settings
10.2% - Control
Rose et al., under review Disabilities Not
Specified
19.6% - Students with Disabilities
10.4% - Control
13.7% - Students with
Disabilities
10.1% - Control
Sabornie, 1994
Learning Disabilities Students with learning disabilities
were 3.5 times more likely to be
victimized.
Not Measured
Sheard et al., 2001
Intellectual
Disabilities
21% - Victimized
10% - Residential Placements
11% - Placement within the Home
27% - Bullied Others (Total)
19% - Residential Placement
8% - Home Placement
Singer, 2005
Dyslexia 85% - Teased
25% - Frequently
28% - Reported Teasing
Others
(continued)
34
Table 4 (continued)
Author Disability Type Victimization Rates Bullying Rates
Sweeting & West, 2001
Various Disabilities 39% - Language Impairments
(Weekly)
30% - Reading Difficulties (Weekly)
15% - Control
Not Measured
Unnever & Cornell, 2003
ADHD 34% - Victimized (2-3 times per
Month)
22% - Control
12% - Bullied Others
8% - Control
Van Cleave & Davis, 2006
Special Health Care
Needs
42.9% - Victimized
22% - Control
31.8% - Bullied Others
51.1% - Students with EBD
21.1% - Control
Whitney et al., 1994
Various Disabilities 67% - Victimized (SE Total)
55% - Mild Learning Difficulties
78% - Moderate Learning Disabilities
50% - Physical Disabilities
100% - Hearing Impairments
29% - Visual Impairments
25% - Control
33% - Bullied Others (Total
Special Education)
27% - Mild Learning
Difficulties
29% - Moderate Learning
Disabilities
33% - Physical Disabilities
50% - Hearing Impairments
29% - Visual Impairments
17% - Control
Woods & Wolke, 2004
Disabilities Not
Specified
No Significant Difference Not Measured
(continued)
35
Table 4 (continued)
Author Disability Type Victimization Rates Bullying Rates
Yude et al., 1998
Hemiplegia 43% - Victimized
13% - Control
6% - Bullied Others
11% - Started Fights
17% - Control (Bullied)
13% - Control (Started
Fights)
Note. Control = nondisabled peer group.
This table was adapted from Rose, Monda-Amaya, & Espelage (2010).
36
The Influence of Class Placement on Bully Perpetration and Victimization
One of the central issues currently facing students with disabilities is access to the general
curriculum. The 1997 amendments of the Individuals with Disabilities Education Act (IDEA,
1997) required that all students with disabilities have access to the general curriculum. More
specifically, the Individualized Education Plan (IEP) must include statements regarding how
disability affects participation in the general curriculum, annual measurable goals geared toward
increasing the participation in the general curriculum, and program modifications (e.g., services,
adaptations, supports) necessary to achieve these goals (Agran, Alper, & Wehmeyer, 2002).
More recently, the revisions of IDEA, now referred to as the Individuals with Disabilities
Education Improvement Act (IDEIA, 2004), placed a strong emphasis on improving the
educational outcomes for students with disabilities through the use of evidence-based practices.
These provisions allow school districts to use up to 15% of their federal budget for early
intervening services, which include extra academic and behavioral supports in general education
classrooms (Yell, Shriner, & Katsiyannis, 2006). However, all of the provisions to IDEA or
IDEIA to date have allowed for the continuum of services for students with disabilities (e.g.,
inclusion, self-contained classrooms, and segregated schools) as long as the placement is
justified by the student’s least restrictive environment (Smith, 2007).
The continuum of services available for students with disabilities may be necessary for
some students to be successful either functionally or academically. These additional services,
however, provide a fundamental difference between students with and without disabilities,
because they often include alternative classroom placements, academic accommodations, or
increased personnel support. In relation to bully perpetration and victimization, it is necessary to
explore how differences in class placement potentially serve as a predictor for increased
37
perpetration and victimization. Traditionally, class placement is broadly defined in terms of
inclusive or segregated settings. Inclusion represents a philosophy of education geared toward
the provision of services in the general education classroom with the purpose of providing a
meaningful, challenging, and appropriate curriculum for everyone (Salend, 2008). In contrast, in
segregated settings (e.g. pullout programs) academic instruction and/or behavioral supports are
provided outside the general education classroom (Smith, 2007). While these two approaches are
distinctly different, students with disabilities receive multiple variations of each defined by their
least restrictive environment and percentage of time they spend receiving special education
services. Based on the ambiguity of the definitions and the general assumption that all students
with disabilities require some level of academic or behavioral supports, this chapter will consider
inclusive services where the student receives a majority of their core academic instruction in a
general education classroom.
In general, students and teachers consistently rank students with disabilities as frequent
victims of bullying (Nabuzoka, 2003; Nabuzoka & Smith, 1993; Sabornie, 1994). When
consideration is given to class placement, rates of victimization often vary between students in
inclusive settings and students in more restrictive placements. This variation could be attributed
to educational practices, classroom structure, percentage of educational supports, or the severity
of the disability (Rose et al., 2010). For example, Whitney and colleagues (1994) investigated the
victimization rates of 93 students with disabilities in an inclusive setting and their
demographically matched peers and determined that the students with disabilities were
victimized significantly more than their general education classmates. Similarly, O’Moore and
Hillery (1989) explored the victimization rates of students with disabilities in inclusive and
restrictive settings and compared them to their general education peers. The researchers reported
38
that students in self-contained settings were victimized significantly more than their peers with
disabilities in inclusive settings and their general education counterparts. These findings are
supported by current literature that has documented that students who receive increased service
are victimized by their peers twice as often than any other sub-group of students (Martlew &
Hodson, 1991; Morrison et al., 1994; Sabornie, 1994).
Similar to victimization, class placement also could serve as a predictor of bullying
perpetration. Although current research is limited regarding bullying among students with
disabilities in inclusive and restrictive settings, foundational research suggests that perpetration
follows the same pattern as victimization (O’Moore & Hillery, 1989; Rose et al., 2010). For
example, in a large scale middle school sample, Rose and colleagues (2009) determined that
students with disabilities in a more restrictive environment engaged in more bullying and
fighting behaviors than students with disabilities in inclusive settings and their general education
peers. Whitney, Nabuzoka, and Smith (1993) also suggested that students with disabilities who
were victimized in inclusive environments tended to exhibit bullying behaviors when moved to a
more restrictive environment. Unfortunately, as previously stated, bullying and aggressive
behaviors could be interpreted synonymously, and this distinction will be discussed further in the
disabilities characteristics section.
Although current research suggests that students with disabilities are victims and
perpetrators more often than their general education peers, inclusive practices could serve as a
preventative factor for the victimization of and perpetration by students with disabilities. The
preventative characteristics of inclusive settings could be attributed to positive behavior
modeling, acquisition of social skills, increased social and academic development (Brown et al.,
1989), increased acceptance, reduction in negative stereotypes (Martlew & Hodson, 1991), and
39
increased participation in classroom activities (Sabornie, 1994). However, it should be noted that
not all extant literature has documented the discrepancy between victimization rates among
students in inclusive and restrictive settings (Reiter & Lapidot-Lefler, 2007; Rose et al., 2009),
indicating that inclusion does not always maintain these preventative characteristics. For
example, if students are not fully integrated into peer groups, inclusion may maintain or
exacerbate victimization and perpetration (Martlew & Hodson, 1991). This lack of integration
could hinder the development of a protective peer base (Morrison et al., 1994; Whitney et al.,
1994) and limit students’ opportunities to learn, practice, and validate social skills (Mishna,
2003). Thus, ineffective inclusive practices could be detrimental for students with disabilities in
regards to involvement in bullying as perpetrators and victims.
Disability Type and Severity
Given the Least Restrictive Environment mandate for students with disabilities, the
discrepancy between perpetration and victimization among students in inclusive or restrictive
settings could partially be explained by the disability type and severity. For example, current
educational trends and national mandates are placing a strong emphasis on Response to
Intervention (RtI; Batsche et al., 2006) and Positive Behavior Supports (PBS; Bambara & Kern,
2005; Ross & Horner, 2009), defined by a multi-tiered framework for providing academic
interventions and behavioral accommodations for all students. Based on this framework, as a
student’s academic or behavioral needs increase the level of support also increases. Therefore,
once a student’s needs exceed pre-set criterion, their supports and classroom placement become
more individualized in order to provide the most appropriate curriculum and accommodations.
Often, the restrictiveness of this placement, which is generally based on the severity of the
40
student’s disability, causes the student to be removed from the general education classroom for
an extended period of time.
Based on the aforementioned framework, with the general assumption that students have
been placed in their Least Restrictive Environment, an argument can be made that the
discrepancy in victimization and perpetration rates among students in inclusive and self-
contained settings may more likely be due to the disability label as opposed to the percentage of
time students receive special education services. Therefore, attention must be paid to the overall
severity and overt nature of the disability. For example, Dawkins (1996) investigated the
difference between victimization rates of students with observable and unobservable disabilities.
The researchers documented that 50% of the students with observable disabilities reported being
victimized at least once during the current term, with 30% victimized on a regular basis.
Conversely, 21% of students with unobservable disabilities reported being victimized at least
once during the current term, and 14% on a regular basis. Therefore, students with unobservable
disabilities (e.g., learning disabilities) reported victimization rates similar to the United States
average, where students with observable disabilities reported significantly higher rates of
victimization.
While empirical research supports the Dawkins’ study, it is important to note that
visibility of disabilities also fall upon a continuum. For example, Whitney and colleagues (1994)
noted that students with mild to moderate learning difficulties were two to three times more
likely to be victimized, where students with physical disabilities and hearing impairments were
two to four times more likely to be victimized than their general education peers. Similarly,
students with language impairments (Davis et al., 2002; Knox & Conti-Ramsden, 2003;
Sweeting & West, 2001) and psychiatric disorders (Unnever & Cornell, 2002; Van Cleave &
41
Davis, 2006) reported being victimized 20% more, and students with emotional/behavioral
disorder (EBD; Monchy et al., 2004; Van Cleave & Davis, 2006) reported being victimized 30%
more than students without disabilities. Additionally, recent reports suggested that students with
Asperger’s syndrome or autistic traits were victimized as much, if not more than any other sub-
group of students (Bejerot & Mörtberg, 2009; Little, 2002). Interestingly, all of the
aforementioned disability labels account for a significant proportion of the students who receive
increased special education services.
More germane to the current study, the involvement of students with learning disabilities
has been investigated in only a small number of studies. Interestingly, when students with
learning disabilities are the primary focus of investigation, perpetration and victimization rates
mirror the rate of involvement of students from other disability groups (Martlew & Hodson,
1991; Nabuzoka, 2003; Nabuzoka & Smith, 1993; Sabornie, 1994). For example, Sabornie
(1994) reported that students identified with learning disabilities are 3.5 times more likely to be
victims of bullying as compared to their general education peers. Additionally, teachers and
classroom peer nominate students with learning disabilities as victims significantly more than
students without disabilities. Perpetration follows a much similar pattern, in which students with
learning disabilities represent a higher proportion of bullies or aggressor as compared to their
peers without disabilities (Kaukiainen et al., 2002; Kuhne & Wiener, 2000; Nabuzoka & Smith,
1993). Based on these studies, students with learning disabilities are as involved, if not moreso,
in the bullying dynamic, when compared to their same aged peers.
While evidence suggests that the observable nature and severity of a disability predicts
escalated victimization, bully perpetration follows a much different pattern. Presumably, the
social nature of bullying, which is reinforced by peers and peer groups, dictates the difference
42
between victimization and perpetration among students with disabilities (Rose et al., 2009). For
example, students with high incidence disabilities (e.g, learning disabilities, EBD) engage in
bullying behaviors twice as often as the United States average (Kaukiainen et al., 2002; Whitney
et al., 1994). Additionally, students with EBD demonstrate the highest level of bully perpetration
when compared to any other subgroup of students (Monchy et al., 2004; Van Cleave & Davis,
2006). However, students with low incidence disabilities (e.g., severe cognitive disabilities)
report much lower rates of perpetration when compared to students with high incidence
disabilities and students without disabilities (Sheard et al., 2001). This discrepancy may be
attributed to minimal interaction opportunities with chronically aged peer groups, social skills
development, and cognitive understanding of bully perpetration. While these factors could be
limited for all students with disabilities, students with high incidence disabilities have a higher
likelihood of being included within the typical school structure (Giangreco, Hurley, & Suter,
2009).
Disability Characteristics
Although educational setting and severity of the disability may serve as predictors for
victimization and perpetration, it is necessary to explore the disability characteristics that may
place students with disabilities at a greater risk for involvement in bullying. Reiter and Lapidot-
Lefler (2007) found that “being a victim was correlated with emotional problems and
interpersonal problems” (p. 179). More importantly, the concept of bullying is complex, based
on the social interplay between perpetration and victimization, and can only be understood in
relations among individuals, families, peer groups, schools, communities, and cultures (Espelage
& Swearer, 2009; Swearer & Espelage, 2004). However, students with disabilities frequently
struggle with these social relationships because they often lack age appropriate social skills (see
43
Baker & Donelly, 2001; Doren et al., 1996; Kaukiainen et al., 2002; Llewellyn, 2000; Woods &
Wolke, 2004).
Based on the general difficulties with social skills combined with the social nature of
bullying, several hypotheses have been developed to explain the escalated rates of victimization
among students with disabilities. According to Sabornie (1994), victims of bullying may be too
passive, exhibit timid responses, misread nonverbal communication, or misinterpret non-
threatening cues. This passivity may reinforce the bullying and misinterpretation may incite
aggressive responses from peers. Additionally, students with disabilities may be at greater risk
for victimization because they lack the appropriate socializing behaviors that help them avoid
being victimized (Nabuzoka, 2003). This lack of socializing behaviors may also lead to the
victim’s inability to develop close friendships, rejection from classroom peers, and the increased
perception that they are dependent on adult assistance (Baker & Donelly, 2001; Llewellyn, 2000;
Martlew & Hodson, 1991; Morrison et al., 1994; Nabuzoka & Smith, 1993). Conversely,
research suggests that when students with disabilities possess age-appropriate social skills with a
positive self-concept, exhibit academic independence, maintain quality relationships, and
participate in school and classroom activities, they are less likely to be targets of bullying (Flynt
& Morton, 2004; Kumpulainen et al., 1998; Martlew & Hodson, 1991; Mishna, 2003; Whitney et
al., 1994).
With respect to perpetration, Rose and colleagues (2010) argue, “bullying perpetration by
students with disabilities is often a learned behavior, possibly a reaction to prolonged
victimization, or an overall lack of social skills” (p. 36). While a lack of social skills may cause
students with disabilities to have greater difficultly with assertion and self-control (Mayer &
Leone, 2007), they may also misread social communication (Whitney et al., 1994), misinterpret
44
social stimuli, or act too aggressively toward the wrong peers (Sabornie, 1994). Additionally,
lack of social skills may also lead students with disabilities to misinterpret rough and tumble play
as a physical attack and thus respond inappropriately with aggressive behavior (Nabuzoka &
Smith, 1999). Although perpetration may be a learned behavior, below average social skills may
also indicate that students with disabilities who engage in bully perpetration could have social
information processing deficits (Crick & Dodge, 1994, 1996; Dodge et al., 2003).
If bully perpetration is a reaction to prolonged periods of victimization, a distinction must
be made between overt aggression (e.g., fighting) and actual bullying behaviors. This distinction
must be made because bullying is a social construct, and as stated above, many students with
disabilities who are involved in bullying display a general lack of age appropriate social skills.
For example, Rose and colleagues (under review) determined that students with disabilities who
are victimized tend to fight, while students without disabilities who are victimized tend to bully.
The work of Björkqvist (2001) and Björkqvist and colleagues (1992) suggests that students
maintain distinct developmental patterns, and many of these patterns hinge on development of
social skills. More specifically, they theorize that aggression is more direct during the early
stages of development, becoming more indirect with age (i.e., physical, verbal, indirect). For
students without disabilities, these developmental patterns are achieved at an age-appropriate
rate, allowing them to process social information and effectively engage in social behaviors.
Therefore, students without disabilities maintain the social skills necessary to engage in more
indirect forms of bullying (Rose et al., 2010). However, students with disabilities often have
delayed social skills (Baker & Donelly, 2001; Doren et al., 1996; Kaukiainen et al., 2002;
Llewellyn, 2000; Woods & Wolke, 2004), placing them in the earlier stages of Björkqvist and
colleagues’ (1992, 2001) developmental trajectory. Therefore, the behaviors displayed by
45
students with disabilities in response to victimization may be more appropriately defined as overt
aggression as opposed to bullying.
Social-Ecological Factors
Since bullying is a social construct based on interactions between the individual and
outside influences, exploration of these factors is necessary to understanding the bullying
dynamic. At the present time, attention in the literature has been given to gender, ethnicity, sense
of belonging, and social supports as they relate to increased perpetration. Overall, these factors
have been found to be predictors or buffers to bullying involvement.
The influence of gender. The relationship between gender and bullying has been the
topic of much debate in recent literature. For years, the “gender dichotomy” has influenced
perceptions of aggression and bullying (Espelage, Mebane, & Swearer, 2004; Swearer, 2008),
where males were considered the disproportionate population of bullies and victims (Nansel,
2001; Olweus, 1993; Ostrov & Keating, 2004; Seals & Young, 2003). To explore the
disproportionality between males and females, several researchers have made a distinction in
type of bullying. More specifically, aggression has been viewed as direct and indirect
(relational), where females have often been identified as more relationally aggressive (Crick,
1996, Crick and Grotpeter, 1995, Ostrov & Keating, 2004). While this distinction and finding is
commonplace in recent social science research, Card, Stucky, Sawalani, & Little (2008) partially
debunk these findings by reporting that males do engage in more direct aggression, but there is
no categorical difference in the rates of relational aggression between males and females.
However, when consideration is given to bullies, victims, and bystanders, bullying involves the
overwhelming majority of school-aged youth (Espelage et al., 2000), which includes both males
and females.
46
The influence of race. Racial consideration has also been the focus of much attention in
the bullying literature. For example, in a nationally representative survey by Nansel and
colleagues (2001), the researchers found that African Americans were less likely to be victimized
than European Americans. Similarly, Wang, Iannotti, and Nansel (2009) reported that African
American adolescents were more likely to be perpetrators, but less likely to be victimized when
compared to other racial groups. These discrepancies may be associated with academic
achievement (Booker, 2006; Wong, Eccles, & Sameroff, 2003), strained peer relationships (Holt
& Espelage, 2007) and increased racial discrimination (Prelow, Danoff-Burg, Swenson, &
Pulgiano, 2004; Seaton, 2009).
Sense of belonging. Sense of belonging is directly related to how one situates him or
herself within a peer group, school, or among friends. An increased sense of belonging has been
documented as a buffer for increased victimization, perpetration, and aggression (Nipedal,
Nesdale, & Killen, 2010; Poteat & Espelage, 2005). For example, schools that have a climate
that is more positive and inclusionary, where developmental needs and academic achievement
are valued equally often have students who maintain a higher sense of belonging (Johnson, 2009;
Nipedal et al., 2010). Similarly, students who have more friends and engage in fewer aggressive
behaviors tend to have higher levels of belonging (Nipedal et al., 2009). Therefore, school
climate and peer relations can increase a sense of belong while decreasing aggressive behaviors.
Family social supports. While students can experience a wide variety of social supports,
family, school, and peer supports have been documented as buffers for involvement in the
bullying dynamic. Families are the primary source of socialization in primary aged youth
(Swearer et al., 2009). Therefore, parental attachment has been documented as a strong predictor
of bullying and victimization (McFadyen-Ketchum, Bates, Dodge, & Petit, 1996; Troy & Sroufe,
47
1987). For example, students who are rated as bullies and victims often report less social
supports from their families (Demeray & Malecki, 2003). Holt, Kantor, and Finkelhor (2009)
extend this finding by reporting that the environments in which bullies live are characterized by
increased maltreatment, less supervision, and higher levels of neighborhood violence. On the
other hand, victims often experience higher levels of family cohesion, have less authoritative
parents, and live in families with low levels of negotiation (Swearer et al., 2009). Therefore,
family characteristics may serve as unique predictors for perpetration and victimization.
School support. Teacher and school support consistently has been documented as buffers
against perpetration and victimization (Birchmeier, 2009; Kochenderfer-Ladd & Pelletier, 2008;
Wang, 2009). As previously stated, schools and classrooms that value academic achievement and
developmental needs equally often have lower levels of victimization and perpetration (Johnson,
2009; Newman, Murray, & Lussier, 2001). Additionally, schools that have comprehensive bully
prevention programs tend to have fewer incidences of bullying (Ttofi, Farrington, & Baldry,
2008). However, the effectiveness of these programs is directly related to teacher awareness of
bullying dynamics as well as teacher knowledge and self-perceived competency in managing
bullying situations (Merrell, Gueldner, Ross, & Isava, 2008). Teacher awareness is also
associated with increased academic achievement, higher levels of support, and lower levels of
bullying (Ma, Phelps, Lerner, & Lerner, 2009) These findings imply that teacher connectedness
and support, coupled with an effective bully prevention program could serve as a vehicle for
decreased victimization and perpetration.
Peer social support. Peer social support has also been documented as a significant
predictor of bullying and victimization. For example, individuals who develop and maintain
quality peer relationships are less likely to be victimized (Hodges, Bovin, Vitraro, Bukowski,
48
1999; Salmivalli, 2010). However, peer groups tend to develop based on the fact that students
share similar behavioral characteristics (Espelage & Swearer, 2004). It also has been noted that
students associate with peers who exhibit similar levels of aggression (Cairns, Cairns,
Neckerman, Gest, & Gariépy, 1988; Espelage et al., 2003). Bullies tend to associate with one
another to maintain status within the dominant peer group (Witvliet, van Lier, Cuijpers, & Koot,
2009), and victims tend to be members of smaller ‘rejected’ peer groups with low social status
(Bagwell, Coie, Terry, & Lochman, 2000). Therefore, bullies tend to report higher levels of
support as compared to their victimized peers (Demaray & Malecki, 2003). On the other hand,
students who are not involved in bullying tend to report higher levels of support when compared
to bullies and victims (Demaray, Malecki, Davidson, Hodgson, & Rebus, 2005). Therefore, the
buffering effect of peer social support is related to number of quality relationships, not the
number of total relationships.
While several social ecological factors impact, or act as predictors or buffers against
involvement in the bullying dynamic, increased research attention has been placed on gender,
ethnicity, sense of belonging, and social supports. These factors are included in the first three
tiers of the social-ecological framework, and seem to have a direct relationship with perpetration
and victimization. However, there is limited information regarding the interplay between these
variables, bully involvement, and students with disabilities. Therefore, these relationships could
provide a better understanding of the overrepresentation of students with disabilities in the
bullying dynamic.
Conclusion
This chapter examined bully perpetration and victimization as it relates to students with
disabilities. While disability is a broad term used to describe 13 subcategories of students defined
49
under IDEA, it becomes evident that both bullying and disabilities fall on a continuum.
Therefore, the interplay between disability label and participation in bullying becomes
exponentially more complex and must be examined longitudinally. Although complexity is an
issue, the social nature of bullying and the lack of social skills among students with disabilities
who are perpetrators or victims remain central to preventing bullying among this population.
Current research in the field of bullying among students with disabilities is limited.
Evidence suggests that these students are victims and perpetrators of bullying more often than
their general education counterparts. However, several questions arise when exploring the
bullying phenomena among this population of students. Most importantly, do the predictive and
preventative factors of involvement in bullying differ for students with and without disabilities?
Evidence suggests that class placement or percentage of special education services received,
disability labels, and disability characteristics play an integral role in predicting victimization
and perpetration. Unfortunately, the extent to which the relationship among these factors’ ability
to predict victimization and perpetration remains untested for all subgroups of students with
disabilities, including students with learning disabilities.
Given the aforementioned gaps in the literature, this study is designed to investigate
predictive and preventative factors uniquely associated with students with learning disabilities. In
doing so, participatory difference between students with and without learning disabilities will be
examined. This investigation will be followed by an inspection of the interplay between the label
of learning disability and percentage of time students receive special education services to
determine if one factor is more predictive of involvement than the other. Finally, social
ecological factors will be explored to see if they are more predictive of involvement for students
with learning disabilities when compared to their peers without disabilities.
50
Chapter 3
Methods
This study was designed to investigate the involvement of students with learning
disabilities in the bullying dynamic. By using bullying, victimization, fighting, and anger as
outcome variables, the influences of the learning disability label, percentage of time spent
receiving special education services, sense of belonging, and social supports were identified. In
this chapter, research questions, school and participant demographics, instrument and subscales,
procedures, and statistical analyses will be presented. As previously stated, five research
questions will guide this study:
1. Can the constructs that define the bullying dynamic be measured equivalently across
students with learning disabilities and students without disabilities?
2. To what extent does being identified with a learning disability influence associations and
mean levels of bullying, victimization, fighting, anger, sense of belonging, and social
supports?
3. To what extent do gender, ethnicity, grade point average, participation in extracurricular
activities, and for students with learning disabilities, percentage of time receiving special
education services, predict involvement in the bullying dynamic for students with?
4. To what extent does sense of belonging and social supports predict involvement within
the bullying dynamic for students with learning disabilities and students without
disabilities?
Site and Participant Demographics
Site demographics. Data have been collected from 6th, 7th, and 8th grade students from
four middle schools in one school district from a diverse Midwestern city over a three year
period. Based on the demographic profile of the school district (Illinois State Board of Education
[ISBE], 2009) the student population is 13,825 students with 29.8% Caucasian, 61.2% African
American, 5.9% Hispanic, 2.5% Asian/Pacific Islander, and 0.6% Multiracial. Additionally, 70%
of the student population are identified as Low-Income Rate, defined by the school district as
51
students from families receiving public aid, living in institutions for neglected or delinquent
children, supported in foster homes with public funds, or are eligible to receive free or reduced-
price lunches. Students with disabilities comprise approximately 24% of the student population
in the sample district (ISBE, 2009). Overall, this district represents an extremely diverse student
population (demographic data by school are presented in Table 5).
Table 5
Sample Population Demographic Data by School
Demographics School 1 School 2 School 3 School 4
Total Population 262 370 111 363
Gender
Male
Female
45%
55%
52%
48%
43%
57%
57%
43%
Ethnicity
White
Black
Hispanic
Asian
Multiracial
29.8%
67.6%
2.3%
0.4%
0.0%
65.4%
24.9%
3.8%
5.4%
0.5%
15.4%
81.8%
2.5%
0.1%
0.1%
8.3%
86.8%
4.7%
0.3%
0.0%
Low SES (%) 66.1% 28.3% 89.3% 96.1%
Students with Disabilities (%) 27.6% 19.5% 23.3% 37.2%
Average Class Size
5th
6th
7th
8th
15.8
13.0
16.3
15.8
22.8
15.7
16.0
23.3
--
--
13.2
20.7
--
14.0
14.4
16.8
Based on the demographics presented in Table 5, schools 1, 3, and 4 report higher
proportions of African American students and students from low socioeconomic backgrounds
when compared to school 2. Due to the student variation, additional self-reported demographic
data were examined to provide a better interpretation of the overall population from each school.
52
These data include “Mother’s Education,” “Father’s Education,” “Grade Point Average,” and
“Participation in Extracurricular Activities.” Based on these data, respondents from Schools 1
and 2 reported the highest levels of education for their parent, respondents from school 3
reported the highest overall grade point averages, and respondents from schools 3 and 4 reported
the highest levels of extracurricular participation. These data are reported in Table 6.
Table 6
Self-Reported Sample Demographics at Wave 4
Demographics School 1
(n = 115)
School 2
(n = 142)
School 3
(n = 89)
School 4
(n = 158)
Mother’s Education
Less than High School 8.7% 7.0% 6.4% 12.3%
High School Diploma or GED 27.0% 25.4% 43.6% 32.7%
Some College 22.6% 14.8% 21.3% 21.0%
College Degree 25.2% 31.0% 14.9% 21.6%
Some Graduate School 0.9% 8.5% 3.2% 4.9%
Graduate Degree
3.5% 5.6% 5.3% 1.9%
Father’s Education
Less than High School 11.3% 4.9% 16.0% 11.7%
High School Diploma or GED 23.5% 29.6% 39.4% 43.8%
Some College 24.3% 13.4% 17.0% 13.6%
College Degree 23.5% 24.6% 12.8% 14.2%
Some Graduate School 2.6% 4.9% 2.1% 8.0%
Graduate Degree 2.6% 13.4% 4.3% --
Grade Point Average
Mostly A’s 8.7% 16.2% 20.2% 12.3%
Mostly A’s & B’s 41.7% 43.0% 57.4% 36.4%
Mostly B’s 3.5% 6.3% 5.3% 4.3%
Mostly B’s & C’s 20.9% 8.5% 8.5% 22.8%
Mostly C’s 3.5% 7.0% 2.1% 7.4%
Mostly C’s & D’s 2.6% 5.6% -- 6.8%
Mostly D’s & F’s 3.5% 1.4% -- 1.9%
Not Sure
9.6% 7.7% 4.3% 5.6%
Participation in Activities
Yes 52.2% 50.0% 59.6% 58.0%
No 42.6% 47.2% 26.6% 39.5%
53
Due to the potential variation among the student samples as described in Tables 5 and 6,
Intraclass correlation coefficients (ICC) were calculated for the outcome variables (bullying,
victimization, fighting, anger) to assess the magnitude of variation among the sample that can be
attributed to the school level variable (Scheier, Griffin, Doyle, & Gilbert, 2002). This analysis
was necessary to determine whether respondent outcome data should be nested within their
school to account for demographic variation among the students. Using the following formula,
measures of ICC magnitude can range from -1.00 to +1.00,
ICC = (MSb-MSw)/MStotal
where MSb represents the mean square between schools and MSw represents the mean square
within schools (Koth, Bradshaw, & Leaf, 2009). Typically, ICCs lower than .10 are considered
to have low variation between the schools (Hedges & Hedberg, 2007), and the nesting would not
need to be accounted for. For the current sample, ICCs were not significant (Bullying = .002,
Victimization = .000, Anger = .005, Fighting = .041), indicating that the school level variable
does not need to be accounted for in the multilevel model.
Teacher demographics. While the demographic student profile for this district is
represents a diverse population, the demographic profile for the teachers in the district is much
more homogeneous. According to the Illinois State Board of Education (2009), 91% of the
teachers in this district are Caucasian, 6.4% black, 1.7 Hispanic, and 0.8% Asian/Pacific
Islander. Additionally, 48.7% of the teachers hold a Bachelor’s Degree, 51.3% hold a Master’s
Degree or above, and 0.6% are working on an Emergency Permit or Provisional Credentials.
Since school level teacher demographic data are unavailable on the ISBE (2009) website, Table
7 represents the demographic profile for the teachers throughout the sample district.
54
Table 7
Demographic Data of Teachers From Sample District
Gender Race Experience
Male
Female White Black Hispanic Asian Years B.A. M.A. Emrg H.Q.
17.8 82.2 91.0 6.4 1.7 0.8 13.7 48.7 51.3 0.6 0.1
Note. All data represent percentages except years, which represents average number of years of
experience. B.A. = % of teachers with Bachelor’s Degrees, M.A. = % of teachers with Master’s
degrees or above, Emrg = % of teachers with emergency or provisional credentials, and H.Q. =
% of classes not taught by Highly Qualified Teachers
Participant demographics. Since this study is a portion of a larger, overarching study,
the collection of special education data was added as an amendment in 2009. Therefore, special
education data from the four schools were collected for students in seventh and eighth grade
during the 2009-2010 academic year. According to school enrollment, the overall sample of
seventh and eighth graders equals 648 students, with a total response rate of 79.2% (n = 513).
Additionally, the total enrollment of students with disabilities in seventh and eighth grade is 212,
with a total response rate of 72% (n = 153). Demographic data for the study sample are presented
in Table 8.
Table 8
Sample Population Demographics
Demographics School 1 School 2 School 3 School 4
Total Population in 7t
h
and 8t
h
Grade 115 142 94 162
Gender (missing)
Male
Female
(2)
50
63
(1)
78
63
(0)
39
55
(4)
82
76
(continued)
55
Table 8 (continued)
Demographics School 1 School 2 School 3 School 4
Ethnicity (missing)
White
Black
Hispanic
Asian
Other/Multiracial
(7)
41
48
4
2
13
(2)
77
38
6
4
15
(5)
5
74
4
0
6
(5)
12
117
10
0
18
Students with Disabilities 24 44 26 59
Although the sample is comprised of 153 students with disabilities, it is important to
distinguish between the various disability labels and percentage of services received. Overall, the
sample population included students from 9 disability categories and 5 models of service
delivery. Due to the limited number of respondents, and the wide range of disability categories,
students with learning disabilities emerged as the only subgroup of students with disabilities
large enough to examine. While this distinction was not expected for the current study, it is
necessary to avoid the major limitation of grouping all students with disabilities together. Table 9
represents the special education data collected for the study sample.
Table 9
Sample Population Disability Data
School/Total Population 20% or
Less
21-60% 61%+ Resource or
Instructional
Total
School 1
Cognitive Disability -- -- 3 -- 3
Other Health
Impairment
1 -- -- -- 1
Learning Disability 11 4 3 -- 18
Speech/Language
Impairment
2 -- -- -- 2
(continued)
56
Table 9 (continued)
School/Total Population 20% or
Less
21-60% 61%+ Resource or
Instructional
Total
School 2
Autism -- -- 1 -- 1
Cognitive Disability -- -- 1 -- 1
Emotional/Behavioral
Disorder
4 -- -- -- 4
Hearing Impairment 2 -- -- -- 2
Other Health
Impairment
4 1 4 -- 9
Orthopedic
Impairment
1 -- 2 -- 3
Learning Disability 8 7 6 -- 21
Speech/Language
Impairment
3 -- -- -- 3
School 3
Cognitive Disability 1 2 -- -- 3
Emotional/Behavioral
Disorder
-- -- 1 -- 1
Other Health
Impairment
4 -- -- -- 4
Learning Disability 5 7 1 1 14
Speech/Language
Impairment
4 -- -- -- 4
School 4
Autism 1 -- -- -- 1
Cognitive Disability -- -- 3 -- 3
Emotional/Behavioral
Disorder
2 2 5 -- 9
Other Health
Impairment
1 -- 2 -- 3
Learning Disability 6 6 17 1 30
Speech/Language
Impairment
12 -- -- -- 12
Traumatic Brain
Injury
-- -- -- 1 1
(continued)
57
Table 9 (continued)
School/Total Population 20% or
Less
21-60% 61%+ Resource or
Instructional
Total
Total
Autism 1 -- 1 -- 2
Cognitive Disability 1 2 7 -- 10
Emotional/Behavioral
Disorder
6 2 6 -- 14
Hearing Impairment 2 -- -- -- 2
Other Health
Impairment
10 1 6 -- 17
Orthopedic
Impairment
1 -- 2 -- 3
Learning Disability 30 24 27 2 83
Speech/Language
Impairment
21 -- -- -- 21
Traumatic Brain
Injury
-- -- -- 1 1
Description of Instrument and Measures
Global instrument. The University of Illinois and Wellesley College: Student Behavior
Survey (SBS; Espelage & Stein, 2006; Appendix A) was developed to assess frequency, types,
and trajectories of bullying perpetration, victimization, witnessing, sexual violence, sexual
harassment, sexual coercion, and homophobia of middle school students. This survey was
developed in combination with a grant funded by the Center for Disease Control and Prevention
(#1U01/CE001677) entitled Middle School Bullying and Sexual Violence: Measurement Issues
and Etiological Models (MSBSV) as described in Chapter 1. Overall, the survey consists of 14
demographic items and 332 individual items, representing 50 separate subscales across 30
different constructs, ranging from school sense of belonging (Espelage & Holt, 2001) to
homophobic teasing (Poteat & Espelage, 2005). It should be noted that each iteration of the SBS
was subjected to a factor analytic procedure resulting in survey modification for each Wave.
However, the modifications did not compromise the integrity of the instrument, and the factor
58
analytic procedure actually allowed for better estimates of the constructs by more accurately
measuring items associated with the construct and eliminating the statistical noise (Little,
Lindenberger, & Nesselroade, 1999; Tynes, Rose, Giang & Williams, under review). Each
subscale and construct is briefly described in Table 10.
Table 10
Description of Measures Used on the University of Illinois and Wellesley College: Student
Behavior Survey
Scale (Author) # of
Items
Description Construct
University of Illinois
Aggression Scale: Sibling
Bullying (Espelage &
Stein, 2006)
5 Measures sibling aggression
either as a victim or
perpetration. Items were
adapted from the University of
Illinois Aggression Scales.
Aggression
University of Illinois
Anger Scale: 4 items
(Espelage & Stein, 2006)
4 Measures anger in the last 30
days.
Anger
Hostility-SCL-90
(Derogatis, Rickels, &
Rock, 1976)
6 Measures symptoms of
underlying hostility, reflecting
qualities such as aggression,
irritability, rage and
resentment.
Anger,
Depression/Anxiety,
Rule-breaking
Impulsivity-Teen Conflict
Survey (Bosworth &
Espelage, 1995)
4 Measures the frequency of
impulsive behaviors (e.g., lack
of self-control, difficulty
sitting still, trouble finishing
things).
Anger,
Depression/Anxiety,
Rule-breaking
Modified Depression Scale
(Orpinas, 1993)
9 Measures feelings of
depression over the 30 days
prior to being surveyed.
Anger,
Depression/Anxiety,
Rule-breaking
(continued)
59
Table 10 (continued)
Scale (Author) # of
Items
Description Construct
Attitudes Toward Bullying
Scale (Espelage, Mebane,
& Adams, 2004)
3 Measures adolescents’
attitudes towards bullying of
others.
Bullying
Peer Nominations 5 Measures the nominations of
peers who are tease, start
fights, get teased, make sexual
comments, and receive sexual
comments often
Bullying
Relational Aggression &
Victimization Scales:
Perpetration (Crick, 1994)
5 Measures self-reported
relational aggression.
Bullying
University of Illinois Bully
Scale: 9 items (Espelage &
Holt, 2001)
9 Measures bullying behavior
(including teasing, group
exclusion, rumor spreading,
name-calling), over the 30
days prior to being surveyed.
Bullying
Computer Aggression
Perpetration Scale (Ybarra,
Espelage, & Mitchell,
2007)
4 Measures computer aggression
perpetration by the respondent
Cyber Aggression
Sexual Text and Computer
Perpetration Scale (Ybarra,
Espelage, & Mitchell,
2007)
5 Measures sexual internet and
text perpetration by the
respondent
Cyber Sexual
Perpetration
Sexual text and computer
victimization (Ybarra,
Espelage, & Mitchell,
2007)
5 Measures sexual text and
computer victimization
experienced by the respondent
Cyber Sexual
Victimization
Computer Aggression
Victimization (Ybarra,
Espelage, & Mitchell,
2007)
3 Measures computer
victimization experienced by
respondent
Cyber Victimization
(continued)
60
Table 10 (continued)
Scale (Author) # of
Items
Description Construct
Friend’s Delinquent
Behavior-Denver Youth
Survey (Institute of
Behavioral Sciences, 2004)
8 Measures respondent’s
knowledge of their friends’
involvement in vandalism,
violence, and drug use during
the past year.
Delinquency
Self-Reported
Delinquency-Problem
Behavior Frequency Scale
(Multisite Violence
Prevention Project, 2004)
8 Measures the frequency of
delinquency behaviors such as
suspension, stealing,
shoplifting, and cheating.
Delinquency
Teen Conflict Survey
(Bosworth & Espelage,
1995)
5 Measures ability to listen,
care, and trust others.
Empathy
Family Conflict and
Hostility – Rochester
Youth Development Study
(Thornberry, Krohn,
Lizotte, Smith, & Tobin,
2003)
3 Measures the extent to which
the parent reports a climate of
hostility and conflict within
the family.
Family Conflict
Femininity Scale (Tolman
& Porche, 2000)
10 Measures respondents’ views
on traditional femininity.
Femininity
University of Illinois Fight
Scale: 4 items (Espelage &
Holt, 2001)
4 Measures, physical aggression
over the 30 days prior to being
surveyed.
Fighting
Motivation for Aggression
Scale: Fighting (Espelage,
2008)
4 Measures adolescent
motivation to fight others.
Fighting for
Impression
Management
Friendship Nominations
(Ennett & Bauman, 1994;
Espelage, Holt, & Henkel,
2003)
1 Measures the friendship
network of the school.
Friendship
Nominations
(continued)
61
Table 10 (continued)
Scale (Author) # of
Items
Description Construct
Homophobic Content
Agent Target Scale
(HCAT) (Poteat &
Espelage, 2005)
5 Measures perpetration
involving homophobic verbal
content.
Homophobic
Perpetration
Homophobic Content
Agent Target Scale
(HCAT): Victimization
Scale (Poteat & Espelage,
2005)
5 Measures victimization
involving homophobic verbal
content.
Homophobic
Victimization
Adolescent Masculinity
Ideology in Relationships
Scale: 12 items (Chu,
Porche, & Tolman, 2005)
12 Measures respondents’ views
on traditional and non-
traditional norms of masculine
behaviors.
Masculinity
Non-traditional
Masculinity Scale (Chu,
Porche, & Tolman, 2005)
5 Measures respondents’ views
on non-traditional norms of
masculine behaviors
Masculinity
Traditional Masculinity
Scale (Chu, Porche, &
Tolman, 2005)
7 Measures respondents’ views
on traditional norms of
masculine behaviors
Masculinity
Pornography Exposure
(Espelage & Stein, 2006)
2 Measures adolescents’
exposure to pornography (i.e.,
internet, print, film)
Pornography
Exposure
School Sense of Belonging
(Espelage & Holt, 2001)
5 Measures sense of belonging
in school and class
Sense of Belonging
Weinberger Adjustment
Inventory (Weinberger &
Schwartz, 1990)
4 Measures an individual’s
perception of his or her value.
Has items from Weinberger’s
Distress Scale.
Self-Esteem
Caring Behavior Scale
(Crick, 1994; Espelage,
Mebane, & Adams, 2004)
4 Measures the frequency of
caring behaviors.
Self-Perceptions
(continued)
62
Table 10 (continued)
Scale (Author) # of
Items
Description Construct
Dominance in Peer Groups
Scale (Espelage & Stein,
2006)
7 Measures of ability to
influence others.
Self-Perceptions
American Association of
University Women Sexual
Harassment Survey
(AAUW, 2001)
30 Questionnaire lists 15 types of
harassment that the
participants might have
experienced either as a victim
or perpetrator in and/or out of
school. These items were
asked 4 different times (i.e.,
perpetrator in school,
perpetrator out of school,
victim in school, victim out of
school).
Sexual Violence
NIJ Survey of Attitudes
and Behaviors Related to
Sexual Harassment:
Dismissal of Sexual
Harassment (Stein &
Taylor, 2006)
10 Measures perceptions,
attitudes, and beliefs regarding
sexual harassment.
Sexual Violence
NIJ Survey of Attitudes
and Behaviors Related to
Sexual Harassment:
Knowledge Scale (Stein &
Taylor, 2006)
10 Measures respondent’s
knowledge of laws, rules, and
regulations regarding sexual
harassment.
Sexual Violence
NIJ Survey of Attitudes
and Behaviors Related to
Sexual Harassment: Sexual
Harassment Prevention
Scale (Stein & Taylor,
2006)
2 Measures perceptions,
attitudes, and beliefs regarding
sexual harassment.
Sexual Violence
(continued)
63
Table 10 (continued)
Scale (Author) # of
Items
Description Construct
NIJ Survey of Attitudes
and Behaviors Related to
Sexual Harassment: Sexual
Harassment Skills and
Intentions to Intervene
Scale (Stein & Taylor,
2006)
10 Measures perceptions,
attitudes, and beliefs regarding
sexual harassment.
Sexual Violence
Who Did This To You?
(Espelage & Stein, 2006)
30 Measures the individual or
individuals who the
respondent victimized or who
victimized them according to
the AAUW scale.
Sexual Violence
Parental Supervision-
Seattle Social
Development Project
(Arthur, Hawkins, Pollard,
Catalano, & Baglioni,
2002)
8 Measures students’
perceptions of what rules their
parents have established and
how closely their parents
monitor those rules.
Social Support
Vaux Social Support
Record (Vaux, 1988)
9 Measures satisfaction with
perceived emotional advice
and guidance, and practical
social support.
Social Support
Vaux Social Support
Family Scale (Vaux, 1988)
3 Measures satisfaction with
perceived emotional advice
and guidance, and practical
social support.
Social Support
Vaux Social Support
Friends Scale (Vaux,
1988)
3 Measures satisfaction with
perceived emotional advice
and guidance, and practical
social support.
Social Support
(continued)
64
Table 10 (continued)
Scale (Author) # of
Items
Description Construct
Vaux Social Support
School Scale (Vaux, 1988)
3 Measures satisfaction with
perceived emotional advice
and guidance, and practical
social support.
Social Support
Drug & Alcohol Use-
Problem Behavior
Frequency Scale (Multisite
Violence Prevention
Project, 2004)
8 Measures the frequency of
drug and alcohol use in the
past month.
Substance Abuse
Motivation for Aggression
Scale: Teasing: 4 items
(Espelage, 2008)
4 Measures adolescent
motivation to tease others.
Teasing for
Impression
Management
Relational Aggression &
Victimization Scales
(Crick, 1994)
5 Measures self-reported
relational aggression.
Victimization
Student Health and Safety
Survey: Past Victimization
(CDC, 2004)
3 Measures the victimization
experienced in the home
environment (i.e., parents
physical abuse, injuries,
sexual abuse).
Victimization
University of Illinois
Victimization Scale: 4
items (Espelage & Holt,
2001)
4 Measures victimization in the
last 30 days.
Victimization
Children’s Exposure to
Community Violence
(Modified) (Richters &
Martinez, 1990)
5 Measures frequency of
exposure (through sight and
sound) to violence in one’s
home and neighborhood.
Violence Exposure
Willingness to Intervene
Scale (Espelage, Mebane,
& Adams, 2004)
5 Measures adolescents’
willingness to intervene when
others are being victimized.
Willingness to
Intervene
Note. References to Espelage & Stein, 2006 indicates scales developed for the MSBSV project.
Some descriptions of measures taken from Dahlberg et al., 2006.
65
Although the MSBSV project utilized the entire SBS instrument, this study focused on
constructs associated with bully perpetration and victimization, and how these constructs related
to students with learning disabilities. Therefore, the 346 item SBS instrument was reduced to 5
demographic items (i.e., gender, grade, ethnicity, self-reported grade point average, and
extracurricular participation), disability data (described in procedures section of Chapter 3), and
8 scales comprised of 34 items (see Table 11). These scales were purposely selected due to their
potential impact on bullying among students with learning disabilities. More specifically, the
dependent measures and outcome variables (i.e., bully, victim, fighting, anger) were selected
because they theoretically provided a holistic representation of bullying involvement and
aggressive behaviors. The manifest variables and latent constructs (i.e., sense of belonging,
school, peer, family social supports) were selected because it was hypothesized that these
variables might partially predict the overrepresentation of students with learning disabilities in
the bullying dynamic.
Table 11
Measures Selected From the SBS for the Current Study
Scale # of
Items
Scale Description Chronbach
Alpha
Bullying: Dependent Variable and Outcome Measure
University of Illinois
Bully Scale
9 Measures bullying behavior
(including teasing, group exclusion,
rumor spreading, name-calling),
over the 30 days prior to being
surveyed.
.88
Victimization: Dependent Variable and Outcome Measure
University of Illinois
Victimization Scale
4 Measures victimization in the last
30 days.
.79
(continued)
66
Table 11 (continued)
Scale # of
Items
Scale Description Chronbach
Alpha
Fighting: Dependent Variable and Outcome Measure
University of Illinois
Fight Scale
4 Measures, physical aggression over
the 30 days prior to being surveyed.
.70
Anger: Dependent Variable and Outcome Measures
University of Illinois
Anger Scale
4 Measures anger in the last 30 days. .81
Latent Constructs
School Sense of
Belonging
4 Measures sense of belonging in
school and class
.62
Vaux Social Support
School Scale
3 Measures satisfaction with
perceived emotional advice and
guidance, and practical social
support from School.
.79
Vaux Social Support
Family Scale
3 Measures satisfaction with
perceived emotional advice and
guidance, and practical social
support from Family.
.82
Vaux Social Support
Friends Scale
3 Measures satisfaction with
perceived emotional advice and
guidance, and practical social
support from friends.
.87
Note. The Chronbach alpha coefficients reported were found at Wave 1 for the entire sample
population of the MSBSV project.
Independent measures. The first section of the SBS is comprised of 14 demographic
items (see Appendix A). To provide an overall picture of the participants, demographic data for
this study were collected related to gender, grade, ethnicity, self-reported grade point average,
and participation in extracurricular activities.
In addition to the aforementioned independent measures, specific disability data were
collected from school officials. These data included primary labels of the respondents, and
percentage of time each student with a disability received special education services. These data
67
were used to provide descriptive statistics for comparing students with learning disabilities with
respondents without disabilities, and provided the foundation for each of the 5 research
questions.
Dependent measures and outcome variables. Eight scales (34 items) from the MSBSV
were used for analysis within this study. The University of Illinois Bully Scale (UIBS; Espelage
& Holt, 2001), University of Illinois Victimization Scale (UIVS; Espelage & Holt, 2001),
University of Illinois Fight Scale (UIFS; Espelage & Holt, 2001), and University of Illinois
Anger Scale (UIAS; Espelage & Stein, 2006) were used in the analysis to respond to the to five
research questions. For research questions 1 and 2 these scales were used as outcome measures
in a multi-group confirmatory factor analysis. For research questions 3 and 4, these measures are
used as outcome measures in a multi-group structural equation model, all of which will be
described in more detail in the data analysis section of Chapter 3.
University of Illinois Bully Scale (Espelage & Holt, 2001). The nine-item University of
Illinois Bully Scale (UIBS; Espelage & Holt, 2001) was used to assess bullying behavior, which
includes teasing, social exclusion, name-calling, and rumor spreading. This scale was developed
based on interviews with middle school students, a review of the extant bullying measures
literature, and extensive factor analytic procedures (Espelage et al., 2000; Espelage et al., 2003).
Students were to indicate how often in the past 30 days they have engaged in each behavior (e.g.,
“I teased other students.” and “I upset other students for the fun of it.”). Response options
included “Never”, “1 or 2 times”, “3 or 4 times”, “5 or 6 times”, and “7 or more times.” These
response options assess bullying persistence where higher scores indicated more self-reported
bullying behaviors. Espelage and Holt (2001) found a Cronbach’s alpha coefficient of .87 and
the UIBS was found to be moderately correlated (r = .65) with the Youth Self-Report Aggression
68
Scale (Achenbach, 1991), suggesting convergent validity. Concurrent validity of this scale was
established with significant correlations with peer nominations of aggression. This scale
converged with peer nomination data (Espelage et al., 2003). However, this scale was not
significantly correlated with the Illinois Victimization Scale (r = .12), and thus provided
evidence of discriminant validity (Espelage et al., 2003).
University of Illinois Victimization Scale. (Espelage & Holt, 2001). Victimization from
peers was assessed using the 4-item University of Illinois Victimization Scale (UIVS; Espelage
& Holt, 2001). Students were asked how often the following things have happened to them in the
past 30 days: “Other students called me names”; ”Other students made fun of me”; “Other
students picked on me”; and “I got hit and pushed by other students”. Response options included
“Never”, “1 or 2 times”, “3 or 4 times”, “5 or 6 times”, and “7 or more times,” where higher
scores indicate more self-reported victimization. Initial examination of this factor accounted for
6% of the variance, and factor loadings ranged from .55 through .92 for the four items (Espelage
& Holt, 2001), which fall well above the minimum criteria of .35 (Worthington & Whittaker,
2006). This factor has continuously documented acceptable alpha coefficients ranging from .85
(Espelage & Holt, 2001) to .93 (Espelage et al., 2003).
University of Illinois Fight Scale (Espelage & Holt, 2001). Fighting was assessed using
the 4-item University of Illinois Fighting Scale (UIFS; Espelage & Holt, 2001). This scale
assesses physical fighting behavior (e.g., “I got in a physical fight” and “I fought students I could
easily beat”), where higher scores indicated more self-reported fighting behavior. Response
options included “Never”, “1 or 2 times”, “3 or 4 times”, “5 or 6 times”, and “7 or more times.”
Factor loadings in the development sample for the UIFS ranged from .50 through .82, which is
well above the minimum criteria of .35 (Worthington & Whittaker, 2006), and accounted for
69
12% of the variance with a Cronbach alpha coefficient of .83 (Espelage & Holt, 2001). These
findings have been replicated with Chronbach alpha coefficients ranging .81 (Poteat & Espelage,
2005) to .88 (Espelage et al., 2003). The UIFS also maintains a low correlation with the UIVS (r
= .21) indicating discriminate validity, and was moderately correlated with the UIBS (r = .58),
providing evidence of convergent validity.
University of Illinois Anger Scale (Espelage & Stein, 2006). Self reported anger was
assessed using the 4-item University of Illinois Anger Scale (UIAS; Espelage & Stein, 2006),
which was developed specifically for the MSBSV project. Students were asked how often the
following things happened to them in the past 30 days: “I got in a physical fight because I was
angry”; “I lost my temper for no reason”; “I was mean to someone when I was angry”; and “I
was angry all day”. Response options included “Never”, “1 or 2 times”, “3 or 4 times”, “5 or 6
times”, and “7 or more times,” with higher scores indicating more self-reported anger. Factor
loadings in a 4-factor solution (Bully, Fight, Victimization, Anger) were .42, .68, .63, and .67
respectively, which accounted for 7.46% of the variance (Espelage & Stein, 2006).
Manifest variables and latent constructs. As stated above, all respondents completed
the 50 separate scales comprised of 346 items on the SBS. However, to answer the 5 research
questions, 4 separate subscales (13 items) were used as Manifest Variables to create four Latent
Constructs. According to MacCallum and Austin (2000), latent constructs “are hypothetical
constructs that cannot be directly measured…[but are] typically represented by multiple manifest
variables that serve as indicators of the construct” (p. 202). For this study, confirmatory factor
analysis and a structural equation model are used to assess predictive and preventative factors
(i.e., Sense of Belonging, Social Support) associated with the overrepresentation of students with
learning disabilities within the bullying dynamic.
70
School Sense of Belonging (Espelage & Holt, 2001). Perceived belonging at school was
assessed with 4 of the 20 items from the Psychological Sense of School Members Scale
(Goodenow, 1993). Students were asked how much they agree with the following four
statements: 1) “I feel proud of belonging to X middle school,” 2) “I am treated with as much
respect as other students,” 3) “The teacher here respect me,” and 4) “There is at least on teacher
or other adult in this school I can talk to if I have a problem.” Response options included
“Strongly Disagree,” “Disagree,” “Agree,” or “Strongly Agree.” Chronbach alpha levels for this
scale have ranged from .68 (Poteat & Espelage, 2005) to .75 (Poteat & Espelage, 2007).
Vaux Social Support Record (Vaux, 1988). The VSSR is a 9-item questionnaire that is
an adaptation of Vaux and colleagues’ (1986) Social Support Appraisal’s (SSA) 23-item scale
that was designed to assess the degree to which a person feels cared for, respected, and involved
(Vaux et al., 1986). The VSSR is comprised of three subscales of three items each, that measure
the support available from family, peers, and school. Scores range from 0 to 6 on each subscale,
and 0 to 18 on the total scale, with higher scores indicating greater perceived support. A sample
item is "I have friends I can talk to, who care about my feelings and what happens to me." The
SSA total scale and family and peer subscales showed good internal consistency across samples
(Vaux, 1988).
Procedures
Institutional review board approval. The initial MSBSV project received approval
from the University of Illinois at Urbana-Champaign’s Institutional Review Board (IRB) in the
fall of 2007. Upon receiving IRB approval, four schools were secured for survey administration
for the duration of the study. In addition to the overall IRB approval, the University IRB
approved an amendment for a record review of students with disabilities in the Spring of 2010
71
(Appendix B). This approval allowed for the collection of information on primary disability
labels and percentage of time students with disabilities received special education services. This
amendment was developed in coordination with the participating schools and district
administrators.
Survey administration. Because this study primarily involved secondary data analysis,
the research procedures described have occurred over a 3-year period. In early Spring 2008, the
primary investigator attended parent-teacher conferences and staff meetings and announced the
study in school newsletters, district newsletters, and emails from school principals. Letters
describing purpose and procedures for the study were sent to parents through mail from the
school principals along with parental consent forms for his/her child’s participation in data
collection (see Appendix C). Parents were asked to return the form only if they did not want their
child to participate in the study. In addition, to ensure that participants understood their rights
and risks, signed student assent forms were obtained at each data collection time point. After the
assent script was read out loud to those students whose parents had passively consented to their
participation, students were asked to indicate their consent by signing the first page of the
survey. Students were told that their participation was strictly voluntary and they could stop
responding at any point during the survey and skip questions they did not want to answer.
Students also were told that their answers would remain confidential unless they indicated that
they had intentions of harming themselves or that someone else was harming them. They were
also told that their names would be converted to numbers and all identifying information would
be removed from their survey answers prior to data entry.
The self-report surveys were administered in classrooms of 20 to 25 students during
designated periods over two consecutive days for Wave 1 and one period for each following
72
Waves. The periods and class selections were predetermined by school officials, and usually
occurred during one block of the academic day (i.e., one classroom period). Survey
administration lasted approximately 40 minutes each day. At each data collection, trained
graduate and undergraduate students read the survey items out loud to participants, monitored
participants’ progress, and ensured data integrity by answering questions and noting when
participants appeared to be responding randomly to survey items. The importance of privacy was
emphasized during survey administration and students were given a blank sheet of paper to cover
their answers as they worked. Due to factor analytic procedures as described earlier in Chapter 3,
the SBS was consolidated to fit into one day of data collection for Wave 2 through 5. However,
other than the reduction in survey administration days, the study procedures remained consistent
for Wave 2, Wave 3, Wave 4, and Wave 5.
Participant names were converted to unique ID numbers within three hours of survey
administration and removed from the survey and shredded. Participant names and ID numbers
are stored in an Excel spreadsheet accessible only to the primary investigator. The dataset
provided for the purpose of the current study only contains ID numbers.
Special education data. While the current study is primarily reliant on secondary data
analysis, special education data were collected in the Spring of 2010 for Cohort groups 1 and 4.
Consent forms were mailed by the school district to parents of all registered students with
disabilities. Parents were provided with phone numbers, addresses, and fax numbers to return the
form if they do not wish for their son/daughter to participate in the project (Appendix D). School
officials (e.g., special education department chair) from each school filled out a spreadsheet to
document primary disability labels and percent of time receiving special education services for
each student with a disability. In order to maintain anonymity, school officials were instructed to
73
remove the student names from the spreadsheet, leaving only the participant number, prior to
providing the researchers with the data.
Since special education data were not collected until the Spring of 2010, survey data and
special education data were only available for Cohort groups 1 and 4, and were analyzed cross-
sectionally at Wave 4 (see Figure 3). Once the special education data were collected, they were
merged with the survey data and represented as independent variables.
Figure 3. Data collection waves and cohort groups used for current study.
Data Analysis
Several procedures were used to analyze the data from the selected scales in the SBS.
Once individual surveys were returned, data were entered into Statistical Package for the Social
Sciences 18.0 (PSAW). Additionally, all disability data were merged with the survey dataset to
represent special education status, primary labels, and percentage of time receiving special
education services. This section will detail preliminary statistical analyses and primary analyses
for addressing the five research questions.
Preliminary statistical analyses. Since data were examined cross-sectionally, a data
imputation procedure was used to account for missing data. It is hypothesized that this method
was necessary to maintain subjects from Cohorts 1 and 4 within Wave 4, and to account for the
limited number of respondents with learning disabilities. Using SAS 9.2 (SAS Institute Inc.,
2008), the SPSS data set with the aforementioned variables was converted to a SAS database to
74
execute the imputation procedure. Multiple EM imputation using the SAS PROC MI procedure
was used to account for the missing data from Wave 4 (Fonagy et al., 2009; Rubin, 1996).
Overall, this imputation method is appropriate for the sample and maintained the integrity of the
original data set.
Primary data analysis. Following data imputation, several analytic procedures were
conducted to examine the five proposed research questions (see Table 12). First, it should be
noted, that data were examined at the scale (or factor) level, so item level data were parceled to
create the scaled constructs (Little, Cunningham, Shahar, & Widaman, 2002). First, to gain a
global understanding of the sample population and involvement within the bullying dynamic,
descriptive statistics and crosstabs were calculated. To address research questions 1and 2, a
multi-group confirmatory factor analysis procedures was used. Questions 4 and 5 utilized a
multi-group Structural Equation Model (SEM).
Table 12
Statistical Analyses and Items for Research Questions
Research Question Measures Analysis
1: Measurement
Invariance
- Learning Disability (Group)
- UIBS (Factor)
- UIVS (Factor)
- UIFS (Factor)
- UIAS (Factor)
- Sense of Belonging (Factor)
- Social Support (3 Scales; Factor)
- Multi-Group CFA
(continued)
75
Table 12 (continued)
Research Question Measures Analysis
2a: Associations - Learning Disability (Group)
- UIBS (Factor)
- UIVS (Factor)
- UIFS (Factor)
- UIAS (Factor)
- Sense of Belonging (Factor)
- Social Support (3 Scales; Factor)
- Multi-Group CFA
2b. Mean Differences - Learning Disability (Group)
- UIBS (Factor)
- UIVS (Factor)
- UIFS (Factor)
- UIAS (Factor)
- Sense of Belonging (Factor)
- Social Support (3 Scales; Factor)
- Multi-Group CFA
4. Demographic
Predictors
- Learning Disability (Group)
- Percentage of Time (PV)
- Ethnicity (PV)
- Gender (PV)
- Grade Point Average (PV)
- School Involvement (PV)
- UIBS (Outcome)
- UIVS (Outcome)
- UIFS (Outcome)
- UIAS (Outcome)
- Multi-Group SEM with
Demographic Predictors
5. Social Support and
Sense of Belonging
Predictors
- Learning Disability (Group)
- UIBS (Outcome)
- UIVS (Outcome)
- UIFS (Outcome)
- UIAS (Outcome)
- Sense of Belonging (PV)
- Social Support (3 Groups; PV)
- Structural Equation Model
with Latent Constructs as
Predictors
Note. IV = Independent Variable, DV = Dependent Variable, Group = Grouping Variable, Factor
= Outcome Variable in Factor Analytic Procedure, PV = Predictor Variable.
Research question 1. To ensure that the constructs investigated are measured
equivalently for both students with learning disabilities and students without disabilities, a multi-
group confirmatory factor analysis procedure was conducted across the four outcome variables
76
(i.e., bullying, victimization, fighting, anger) and the four latent constructs (i.e., sense of
belonging, social supports; see Figure 4). While each of the four factors have undergone
extensive confirmatory factor analyses, it is necessary to confirm that each factor holds for
students with learning disabilities to ensure the validity of the constructs and allow for a distinct
comparison between the two groups. To examine the 8-factor model, a maximum-likelihood
estimation multi-group confirmatory factor analysis was conducted in LISERL 8.80 (Jöreskog &
Sörbom, 2007).
This multi-group confirmatory factor analysis, allowed for the assessment of differences
in latent space, and included three distinct steps in order to test levels of measurement,
equivalence, and invariance. First, a configural invariance process was employed to examine the
pattern of fixed and free parameters across the outcome variables. Second, weak factor
invariance was tested to explore the relative factor loadings. Third, strong factorial invariance
was examined to investigate the relative indicator means. Overall, this process was necessary to
ensure the constructs were measured equivalently across the two groups.
To examine model fit throughout each step, several statistics will be reported. First, the
chi-square statistic divided by the degrees of freedom will be examined to assess the overall
model fit. While chi-square is overly sensitive to sample size, it is usually the null-hypothesis
significance test (Cheung & Rensvold, 2002), and a chi-square/df ratio below 3 is often
considered an acceptable fit (Kline, 1998). Additionally, several relative fit indices will be
examined, as they may be more appropriate in predicting model fit because they are less reliant
on sample size (Hu & Bentler, 1999; Immekus & Maller, 2009). For this study, the Tucker-
Lewis Index (TLI), Incremental Fit Index (IFI), Comparative Fit Index (CFI), and Root Mean
Square Error of Approximation (RMSEA) will be used. Where TLI, IFI, and CFI scores greater
77
than .95 are considered an acceptable fit (Pinterits, Poteat, & Spanierman, 2009; Schermelleh-
Engel, Moosbrugger, & Müller, 2003), and RMSEA scores of above .1 are considered a poor fit,
between .08 and .1 a mediocre fit, between .05 and .08 an acceptable fit, .01 and .05 a close fit,
and .00 an exact fit (Hu & Bentler, 1999). Finally, Cohen’s d will be reported as a measure of
effect size for latent mean differences (Cohen, 1988).
Research question 2. Once factoral invariance has been established for the two groups,
associations (question 2a) and latent mean (question 2b) differences were examined.
Specifically, for question 2, variances and covariances were investigated to determine if the
associations and patterns of covariances/correlations varied significantly across the two groups
for the eight latent constructs. Following the examination of associations, latent mean
relationships were evaluated to determine if the two groups maintained variability across the
latent constructs (Shogren et al., 2007).
Research questions 3 and 4. To assess predictors and preventative factors uniquely
associated with students with learning disabilities, a structural equation model (SEM) was
executed for the two- group model (i.e., learning disabilities, general education) using LISERL
8.80 (Jöreskog & Sörbom, 2007). Using McDonald and Ho’s (2002) recommendations, the SEM
was justified using the latent variables (i.e., sense of belonging, social support), and demographic
predictors (i.e., percentage of special education time, gender, ethnicity, grade point average,
extracurricular participation). To assess model fit, the same procedures as described in the CFA
were used. Therefore, chi-square, TLI, IFI, CFI, and RMSEA are reported. Additionally, per
McDonald and Ho (2002), plausible alternative models and explanations are presented. The
theoretical model for the two-group model is presented in figure 5.
78
Figure 4. Multi-group confirmatory factor analysis.
79
Figure 5. Theoretical structural equation model for predictive and preventative factors associated
with bullying, victimization, fighting, and anger for two-group model.
80
Chapter 4
Results
The purpose of this study was to examine the differences in bully involvement of middle
school students with learning disabilities and their peers without disabilities. Data were
examined to assess whether factors related to the bullying dynamic could be measured
equivalently across the two subgroups of students. Once equivalence was established,
comparative analyses were conducted to determine differences in associations, latent means, and
predictors related to the bullying dynamic. The following section will explicitly detail analyses
conducted to address the four research questions. Specifically, this chapter will detail the sample
demographics, imputation process, confirmatory factor analysis equivalence model, tests of
associations related to bullying constructs, demographic predictors, and social supports
predictors.
Sample Demographics
As reported in Chapter 3, Intraclass Correlation Coefficients (ICC) were not significant
for school level outcome variables (i.e., bullying, victimization, fighting, anger), indicating that
school level nesting variables are unnecessary for the current analyses. Therefore, data were
analyzed at the respondent level for students with learning disabilities and students without
disabilities. Self-reported demographic information was collected for gender, age, grade, race,
grade point average (GPA), and participation in extracurricular activities. While this study is a
cross-sectional investigation of differences between students with learning disabilities and
students without disabilities, exploration of Waves 1 through 3 was necessary to recover some
unreported demographic information due to respondents’ failure to report. Since it is reasonable
to assume gender and race are static variables, and age and grade will increase as a function of
81
time, missing data for these variables were inserted as necessary. GPA and participation in
extracurricular activities are not fixed variables, so missing data were imputed using a multiple
imputation procedure, which is described in the following section.
To investigate sampling differences between students with learning disabilities and
students without disabilities, 2 statistics were calculated. Overall, there was not a significant
difference between the groups for school, age, grade, race, GPA, and extracurricular
participation, indicating the samples are proportionally similar (see Table 13). However, the 2
statistic revealed significant differences for gender (2 (1)= 5.18, p < .05). Descriptive statistics
(see Table 14) revealed that gender was relatively proportional for students without disabilities
(53.6% female, 46.4% male), but males were overrepresented in the group of students with
learning disabilities (39.8% female, 60.2% male).
Table 13
2Difference Tests for Variables Across Disability Type
Demographic
Variables
2 df p
School
1.87 3 .600
Gender
5.18 1 .023*
Age
5.35 4 .254
Grade
0.28 1 .599
Race
1.63 5 .897
GPA
7.94 7 .326
Extracurricular
0.05 1 .824
* represents significant at the .05 level
** represents significant at the .001 level
82
Table 14
Descriptive Statistics for Students From the Two Groups
No Disability (%) Learning Disability (%)
Demographics N = 360 N = 83
School
School 1 91 (25.3) 18 (21.7)
School 2 98 (27.2) 21 (25.3)
School 3 68 (18.9) 14 (16.9)
School 4 103 (28.6) 30 (36.1)
Gender
Female 193 (53.6) 33 (39.8)
Male 167 (46.4) 50 (60.2)
Age
11 3 (0.8) --
12 118 (32.8) 21 (25.3)
13 159 (44.2) 35 (42.2)
14 74 (20.6) 24 (28.9)
15 6 (1.7) 3 (3.6)
Grade
7 172 (47.8) 37 (44.6)
8 188 (52.2) 46 (55.4)
Race
American Indian or Alaska Native 8 (2.2) 2 (2.4)
African American 193 (53.6) 48 (57.8)
Asian 4 (1.1) 2 (2.4)
Hispanic 21 (5.8) 4 (4.8)
White 97 (26.9) 20 (24.1)
Other 37 (10.3) 7 (8.4)
Grade Point Average Missing: 11 (3.1) Missing: 2 (2.4)
Mostly A’s 53 (14.7) 10 (12.0)
Mostly A’s & B’s 161 (44.7) 27 (32.5)
Mostly B’s 15 (4.2) 6 (7.2)
Mostly B’s & C’s 57 (15.8) 16 (19.3)
Mostly C’s 21 (5.8) 8 (9.6)
Mostly C’s & D’s 16 (4.4) 4 (4.8)
Mostly D’s & F’s 5 (1.4) 2 (2.4)
Not Sure 21 (5.8) 8 (9.6)
(continued)
83
Table 14 (continued)
No Disability (%) Learning Disability (%)
Demographics N = 360 N = 83
Participation in Extracurricular
Activities
Missing: 18 (5.0) Missing: 3 (3.6)
No 142 (39.4) 33 (39.8)
Yes 200 (55.6) 47 (56.6)
Percentage of Services Missing: 2 (2.4)
No Services Received 360 (100) --
20% or Less -- 30 (36.1)
21 – 60% -- 24 (28.9)
61% or More -- 27 (32.5)
Descriptive statistics by gender and disability. To further investigate sampling
differences, separate cross tabulations with corresponding 2 statistics were calculated for gender
by disability, race by disability, females by gender and disability, and males by gender and
disability. Specific demographic cross tab data are reported in Appendix E, and separate 2
statistics are reported in Table 15. The 2 statistics for gender by disability uncovered
nonsignificant results for school, age, grade, race, participation in extracurricular activities, and
percentage of special education services. The 2 for grade point average of students without
disabilities (2 (7)= 25.10, p < .001) revealed that females report higher GPAs (67.4% with at
least A’s and B’s) than males (50.3 with at least A’s and B’s).
Descriptive statistics by race and disability. Similar to gender, 2 statistics for race by
disability uncovered nonsignificant results for gender, age, grade, GPA, participation in
extracurricular activities, and percentage of special education services. However, 2 for school of
students without disabilities (2 (15)= 115.17, p < .001) revealed that school 1 and 2 represented
91.8% of the white students without disabilities, and schools 3 and 4 represented 67.4% of the
African American students without disabilities.
84
Descriptive Statistics by race, gender, and disability. This variation is also evident in
the 2 statistics for Females by race and disability (2 (15)= 62.10, p < .001), males by race and
disability (2 (15)= 63.91, p < .001) for students without disabilities, and males by race and
disability (2 (15)= 32.29, p < .01) for students with disabilities. More specifically, school 4
represented 57.6% of the African American males with learning disabilities. However, school
level analyses were not conducted because Intraclass Correlations Coefficients revealed a
nonsignificant difference among the schools. Finally, 2 for Males and Females by race and
disability label revealed nonsignificant results for age, grade, GPA, participation in
extracurricular activities, and percentage of special education services. Overall, the 2 statistics
reported in Table 15 reveal that the current sample is relatively similar across the selected
demographic items.
Table 15
2 Statistics for Cross Tabulations
Demographic 2 df P
2 by Gender and no Disability
School 3.28 3 .350
Age 9.39 4 .052
Grade 3.02 1 .082
Race 3.09 5 .686
GPA 25.10 7 .001**
Extracurricular .71 1 .413
2 by gender and learning disability
School 3.77 3 .288
Age .98 3 .805
Grade .10 1 .748
Race 4.81 5 .439
GPA 7.13 7 .435
Extracurricular .01 1 .622
Percentage of
Services
1.19 2 .516
(continued)
85
Table 15 (continued)
Demographic 2 df P
2 by race and no disability
School 115.17 15 .000**
Gender 3.09 5 .686
Age 25.26 20 .192
Grade 3.03 5 .695
GPA 34.91 35 .473
Extracurricular 9.90 5 .078
2 by race and learning disability
School 24.38 15 .059
Gender 4.81 5 .439
Age 9.47 15 .851
Grade 3.00 5 .700
GPA 47.67 35 .075
Extracurricular 5.67 5 .339
Percentage of
Services
8.631 10 .567
2 by race, female, and no disability
School 62.10 15 .000**
Age 18.61 15 .232
Grade 8.22 5 .145
GPA 25.08 35 .892
Extracurricular 5.41 5 .368
2 by race, male, and no disability
School 63.91 15 .000**
Age 19.04 20 .519
Grade 2.23 5 .816
GPA 38.07 35 .331
Extracurricular 9.91 5 .078
(continued)
86
Table 15 (continued)
Demographic 2 df P
2 by race, female, and learning disability
School 12.32 15 .654
Age 9.62 15 .843
Grade 3.00 5 .700
GPA 28.43 35 .776
Extracurricular 3.92 4 .561
Percentage of
Services
8.70 10 .561
2 by race, male, and learning disability
School 32.29 15 .006*
Age 15.74 15 .399
Grade 5.43 5 .365
GPA 48.28 35 .067
Extracurricular 8.84 5 .116
Percentage of
Services
14.78 10 .140
** p < .001
* p < .01
Missing Data Procedures
Data imputation. To address the issue of missing data within the current sample, a
missing data pattern analysis was conducted in SPSS 18.0 (PSAW, 2009) and a multiple
imputation procedure was executed using the PROC MI function in SAS 9.2 (SAS, 2008). Since
missingness can bias a sample (Davey, Savla, & Luo, 2005; Rubin, 1976), it was necessary to
account for the missing values to best represent students with learning disabilities and students
without disabilities. While Little’s MCAR (Little, 1988) test was insignificant (2 (1136)= 1150.03,
p = .379), the sample may not necessary be characterized as Missing Completely at Random
because the missingness on any given variable may be related to the observed data (Enders &
Peugh, 2004; Luengo, García, & Herrera, 2010). Therefore, the data for the current sample will
be approached as Missing at Random (MAR), because the missingness on any given variable is
87
not related to itself, but it may be related to another measured variable (Enders & Peugh, 2004;
Luengo et al., 2010). However, MAR is an assumption and according to Schafer and Graham
(2002), “there is no way to test whether MAR holds in a data set” (p 152) because data cannot be
directly obtained from nonresponders. Collins, Shafer, and Kam (2001) demonstrated that the
MAR assumption has minimal impact on estimates and standard errors.
Overall, 36 (81.8%) out of the 44 measured variables included some missing data, with
71 (16.0%) out of the 443 respondents having some level of missingness. However, the
missingness of the total sample was extremely low, with only 1.7% (332) missing from the
measured items by total number of respondents (see Figure 6). Overall, missingness per item
ranged from 0 to 4.7%. A missingness breakdown of the items is reported in Table 16.
Figure 6. Missing data patterns of total items.
88
Table 16
Missing Data Specifics per Item
Item Total Missing Percent Missing
Participation in Extracurricular Activities 21 4.74
I am treated with as much respect as other students
are.
14 3.16
What is your overall grade average this year? 13 2.93
There is at least one teacher or other adult in this
school I can talk to if I have a problem.
12 2.71
I feel proud of belonging to ___ Middle School 12 2.71
I have friends who help me with practical problems… 11 2.48
I have friends I can talk to, who give good suggestions
and advice about my problems.
11 2.48
I lost my temper for no reason. 10 2.26
I threatened to hurt or hit another student. 10 2.26
I fought other students I could easily beat. 10 2.26
I was angry all day. 9 2.03
I encouraged people to fight. 9 2.03
Other students called me “gay.” 9 2.03
I spread rumors about other students. 9 2.03
I got in a physical fight. 9 2.03
In a group I teased other students. 9 2.03
There are people in my family who help me with
practical problems…
9 2.03
At school, there are adults who help me with practical
problems…
9 2.03
(continued)
89
Table 16 (continued)
Item Total Missing Percent Missing
I have friends I can talk to… 9 2.03
As school, there are adults I can talk to… 9 2.03
The teachers here respect me. 9 2.03
I got hit and pushed by other students. 8 1.81
I was mean to someone when I was angry. 8 1.81
I teased other students. 8 1.81
I hit back when someone hit me first. 8 1.81
I started (instigated) arguments or conflicts. 8 1.81
Other students picked on me. 8 1.81
There are people in my family I can talk to, who give
good suggestions and advice about my problems.
8 1.81
As school, there are adults I can talk to, who give
good suggestions and advice about my problems.
8 1.81
There are people in my family I can talk to… 8 1.81
I called other students “gay.” 7 1.58
Other students called me names. 7 1.58
I got in a physical fight because I was angry. 7 1.58
I helped harass other students. 7 1.58
I upset other students for the fun of it. 7 1.58
Percentage of Special Education Services 2 0.05
Total 332 1.70
90
Although Luengo and colleagues (2010) suggest that missing data between 1 and 5% are
generally manageable, a multiple imputation procedure was employed to preserve the integrity of
each group of respondents and create a parsimonious dataset. Using Kärnä and colleagues (in
press) as a model, data were imputed with the SAS PROC MI function, using the MCMC
algorithm. In total, 100 imputations were conducted separately for students with learning
disabilities and students without disabilities. Next, the average imputed value for each missing
data point was calculated, which according to Kärnä and colleagues (in press) “represents the
best population estimate of the value need to reproduce the population parameters” (p. 55).
Overall, one parsimonious data set was created, which best represents the sample population.
Factor parceling. Following the multiple imputation process, an item-to-construct
balancing procedure was conducted to create parcels for bullying, victimization, fighting, and
sense of belonging (Little et al., 2002). Parcels, which are aggregate-level indicators, were
created to establish a just-identified measurement model because the focus of this study hinges
on constructs, not item-level indicators. Additionally, “a just-identified construct has only one
unique solution that optimally captures the relation among the items, no matter what other
constructs are considered or included in a model” (Little et al., 2002, p. 162). Since three
individual indicators theoretically define the anger and sense of belonging scales, the parceling
procedure was unnecessary because they are already just-identified. Therefore, separate single-
construct models were created for bullying, victimization, fighting, and sense of belonging using
maximum likelihood estimation. Once the models were created, the three highest loadings were
used to anchor the construct, and the next highest loadings were added to the anchors in inverse
order (Little et al., 2002). Due to the item total for victimization, fighting, and sense of
belonging, the two items with the highest loadings represented two separate parcels and the two
91
items with the lowest loadings were combined and averaged to create the third parcel. Table 17
contains the final constructed parcels.
Table 17
Item Parceling Procedure for the Eight Subscales
Parcel
#
Items Factor
Loadings
Bullying
1 Ques23A: I upset other students for the fun of it.
Ques23O: I encouraged people to fight.
Ques23H: I helped harass other students.
.876
2 Ques23B: In a group I teased other students.
Ques23G: I started (instigated) arguments or conflicts.
Ques23K: I threatened to hurt or hit another student.
.832
3 Ques23P: I teased other students.
Ques23F: I spread rumors about other students.
.773
Victimization
1 Ques23N: Other students called me names. .901
2 Ques23D: Other students picked on me. .798
3 Ques23T: I got hit and pushed by other students.
Ques23J: Other students called me “gay.”
.713
Fighting
1 Ques23E: I got in a physical fight .846
2 Ques23L: I got in a physical fight because I was angry. .702
3 Ques23I: I hit back when someone hit me first.
Ques23C: I fought other students I could easily beat.
.612
Anger
1 Ques23S: I was angry all day. .729
2 Ques23R: I was mean to someone when I was angry. .709
3 Ques23M: I lost my temper for no reason.
.683
(continued)
92
Table 17 (continued)
Parcel # Item Factor
Loading
Sense of Belonging
1 Ques14B: I am treated with as much respect as other students are. .737
2 Ques14C: The teachers here respect me. .705
3 Ques14A: I feel proud of belonging to __ Middle School.
Ques14D: There is at least one teacher or other adult in this school I
can talk to if I have a problem.
.605
Support: School
1 Ques17D: At school, there are adults I can talk to, who give good
suggestions and advice about my problems.
.926
2 Ques17A: At school, there are adult I can talk to, who care about my
feelings and what happens to me.
.764
3 Ques17G: At school, there are adults who help me with practical
problems…
.655
Social Support: Family
1 Ques17E: There are people in my family I can talk to, who give me
good suggestions and advice about my problems.
.837
2 Ques17H: There are people in my family who help me with practical
problems…
.810
3 Ques17B: There are people in my family I can talk to, who care
about my feeling and what happens to me.
.720
Social Support: Peers
1 I have friends I can talk to, who give good suggestions and advice
about my problems.
.811
2 I have friends I can talk to, who care about my feelings and what
happens to me.
.801
3 I have friends who help me with practical problems…
.760
Once the parcels were established for the eight separate constructs, data were aggregated
across the 100 imputations to represent the best population estimate for the imputed data (Kärnä
et al., in press) using the Aggregate function in SPSS 18.0 (PSAW, 2009). This aggregation
provided a single “super matrix” based on the mean scores of the 100 imputations for each
respondent on the eight constructs. This aggregated dataset was used for all consequent data
93
analytic procedures, and mean scores and standard deviations of each parcel for students without
disabilities and students with learning disabilities are reported in Table 18.
Table 18
Means and Standard Deviations by Subgroup for Individual Parcels
Students without Disabilities Students with Learning
Disabilities
Parcel Mean sd Mean sd
Bully
1 1.35 .57 1.31 .45
2 1.44 .59 1.39 .51
3 1.31 .56 1.26 .48
Victimization
1 1.58 1.07 1.53 .98
2 1.62 1.09 1.48 .97
3 1.36 .68 1.38 .67
Fighting
1 1.52 .99 1.47 .84
2 1.35 .82 1.45 .93
3 1.64 .81 1.94 .92
Anger
1 1.56 1.02 1.48 1.01
2 1.52 .92 1.42 .66
3 1.37 .83 1.31 .60
Sense of Belonging
1 2.87 .78 2.94 .70
2 3.06 .76 3.04 .76
3 3.07 .66 3.12 .65
Support: School
1 2.11 .59 2.27 .66
2 2.15 .63 2.22 .70
3 2.23 .63 2.18 .64
(continued)
94
Table 18 (continued)
Students without Disabilities Students with Learning
Disabilities
Parcel Mean sd Mean sd
Social Support: Family
1 2.56 .59 2.51 .65
2 2.54 .59 2.38 .65
3 2.71 .48 2.59 .62
Social Support: Peers
1 2.23 .61 2.21 .67
2 2.29 .64 2.24 .71
3 2.25 .60 2.25 .62
Construct Equivalence
Can the constructs that define the bullying dynamic be measured equivalently
across students with learning disabilities and students without disabilities? This research
question addressed measurement invariance on the University of Illinois Aggression Scales
(Espelage & Holt, 2001), Sense of Belonging Scale (Espelage & Holt, 2001) and Social Support
Record (Vaux, 1988). To evaluate this question, a multi-group confirmatory factor analysis
procedure was utilized. This stepwise process enables the measurement equivalences of
constructs and allows for direct comparisons among groups (Little, 1997; Shogren et. al., 2007).
To establish measurement invariance and discern that students with learning disabilities and
students without disabilities are interpreting the constructs equivalently, strong (e.g., intercept)
invariance must be established (Little, 1997). This process includes three distinct steps: (a) test
the model fit based on manifest indicators, (b) equate factor loadings across groups and evaluate
model fit, and (c) equate intercepts across groups and evaluate model fit.
Overall, eight latent constructs were used in the present measurement model to test
measurement equivalence between students with learning disabilities and students without
95
disabilities. These constructs include: (a) bullying, (b) victimization, (c) fighting, (d) anger, (e)
sense of belonging, (f) school support, (g) family social support, and (h) peer social support. As
stated in the data imputation section of this chapter, three parcels or item level indicators were
used to create each construct to maintain a just-identified model.
Using Shogren and colleagues’ (2007) CFA procedure as a model, an indicator loading
procedure (i.e., effects coding) was used by constraining the sum of the indicator’s loadings to
the total number of indicators (e.g. LY(1,1) = 3 – LY(2,1) – LY(3,1)). While traditional
techniques are generally used for CFA procedures, the effects coding method allows for the
estimation of a construct’s latent variance in a non-arbitrary metric (Little, Slegers, & Card,
2006; Shogren et al., 2007). To maintain consistency, intercepts were estimated using a similar
procedure (CO TY(1) = 0 – TY(2) – TY(3)).
As a preliminary step, separate models were fit for each group of students to determine if
they if the initial parameters were tenable for each group of students. The freely estimated model
for students without disabilities demonstrated acceptable model fit (2(224) = 558.06, p <.001,
RMSEA = .061, NNFI = .95, CFI = .096), and the freely estimated model for students with
learning disabilities demonstrated acceptable model fit on RMSEA and mediocre fit for NNFI
and CFI (2(224) = 384.28, p <.001, RMSEA = .071, NNFI = .86, CFI = .88). Since both freely
estimated models fell within the acceptable range on at least one of the fit indices, it was
appropriate to move forward with the CFA.
Using the effects coding method described earlier, a multiple group confirmatory analysis
was conducted with two groups. Since parcels were used to estimate the constructs, a just-
identified model was established for both groups with 24 parcels estimating 8 separate
constructs. During the configural invariance step, all parameters (e.g. loadings, intercepts) are
96
freely estimated. The configural model demonstrated an acceptable fit based on the relative fit
indices (2(448)=942.344, p < .001, RMSEA = .063, NNFI = .92, CFI = .94). The initial step of
the CFA indicates that with the same measurement model compared across the two groups, the
overall fit was acceptable.
Following the configural invariance step, factor loadings are equated to determine if they
are invariant between the two groups of students. The loading invariance test revealed that the
model remained within the acceptable range for the appropriate fit indices (2(464) = 942.344, p <
.001, RMSEA = .063, NNFI = .92, CFI = .94). The RMSEA model test was conducted to
establish if the constraints are tenable, where the RMSEA value of the constrained model is
examined to determine if it falls within the 90% confidence interval of the freely estimated
model (Little, 1997). Additionally, changes in the Comparative Fit Index (CFI) of less than .01
indicate that the constraints are tenable (Cheung & Rensvold, 2002). Based on the RMSEA
model test and evaluation of the CFI, it was concluded that the loadings are invariant between
students with learning disabilities and students without disabilities.
After establishing loading invariance, constraints were placed on the intercepts to
determine if they were invariant between groups. As with the previous two models (i.e. freely
estimated, loading invariant) the strong metric invariance model maintained acceptable model fit
(2(480) = 1007.933, RMSEA = .063, NNFI = .92, CFI = .94). Additionally, the strong metric
model met the criteria of the RMSEA model test and CFI evaluation, indicating that no
significant changes were documented in model fit as constraints increased. Based on these
statistics; bullying, victimization, fighting, anger, sense of belonging, and social supports (i.e.,
Teacher, Family, Peer) are being equivalently assessed for students with learning disabilities and
97
students without disabilities. Table 19 contains loadings, intercepts, and estimated latent
variances from the strong metric invariance model.
Table 19
Loadings, Intercepts, and Estimated Latent Variance From Strong Metric Invariance Model
Indicator - Loading
Estimates (SE)
- Intercept
Estimates (SE)
- Standardized
Loadingsa
Bully
Parcel 1 .98 (.03) .02 (.04) .82
Parcel 2 1.11 (.03) -.08 (.04) .89
Parcel 3 .91 (.03) .07 (.04) .77
Victimization
Parcel 1 1.22 (.04) -.27 (.06) .89
Parcel 2 1.12 (.04) -.10 (.06) .80
Parcel 3 .66 (.03) .37 (.05) .74
Fighting
Parcel 1 1.16 (.05) -.26 (.08) .77
Parcel 2 .97 (.05) -.10 (.07) .73
Parcel 3 .87 (.05) .35 (.07) .66
Anger
Parcel 1 1.12 (.05) -.09 (.08) .70
Parcel 2 1.07 (.05) -.07 (.07) .77
Parcel 3 .81 (.05) .16 (.07) .65
Belonging
Parcel 1 1.04 (.06) -.25 (.17) .68
Parcel 2 1.04 (.06) -.07 (.17) .68
Parcel 3 .92 (.05) .32 (.16) .69
Support: School
Parcel 1 1.09 (.03) -.23 (.08) .88
Parcel 2 1.04 (.04) -.09 (.08) .79
Parcel 3 .88 (.04) .31 (.09) .68
aCommon Metric Completely Standardized Solution (continued)
98
Table 19 (continued)
Indicator - Loading
Estimates (SE)
- Intercept
Estimates (SE)
- Standardized
Loadingsa
Social Support: Family
Parcel 1 1.11 (.04) -.32 (.10) .84
Parcel 2 1.10 (.04) -.32 (.10) .82
Parcel 3 .80 (.04) .64 (.09) .70
Social Support: Peers
Parcel 1 1.03 (.03) -.11 (.08) .82
Parcel 2 1.03 (.04) -.03 (.08) .79
Parcel 3 .94 (.04) .14 (.08) .77
aCommon Metric Completely Standardized Solution
The overall model fit statistics for the three step, multi-group confirmatory factor analysis
are presented in Table 20, with Figure 7 representing the strong metric invariance model. In
addition to establishing strong metric invariance between the groups and across the eight latent
constructs; latent means, unique residuals, and squared multiple correlations were calculated
when constraining the loadings and intercepts. These statistics are reported in Table 21. Given
the results of the multi-group CFA, measurement invariance has been established, and research
question one confirmed.
Table 20
Fit Indices for Multi-Group Confirmatory Factor Analysis
Model 2 Df P RMSEA RMSEA
90% CI
NNFI CFI Constraint
Tenable
Configural
Invariance
942.344 448 <.001 .063 .056 - .070 .92 .94 --
Loading
Invariance
979.358 464 <.001 .063 .057 - .070 .92 .94 Yes
Intercept
Invariance
1007.933 480 <.001 .063 .056 - .069 .92 .93 Yes
99
Figure 7. Strong metric invariance measurement model for multi-group confirmatory factor analysis.
100
Table 21
Mean Scores, Unique Residuals, and Squared Multiple Correlations for Individual Parcels
Across Disability Groups
Students without Disabilities Students with Learning Disabilities
Indicator Mean
Scores
- Residual
(SE)
R2Mean
Scores
- Residual
(SE)
R2
Bully
Parcel 1 1.35 .12 (.01) .65 1.31 .05 (.01) .77
Parcel 2 1.44 .07 (.01) .80 1.39 .06 (.02) .78
Parcel 3 1.31 .12 (.01) .61 1.26 .12 (.02) .52
Victimization
Parcel 1 1.58 .26 (.04) .76 1.53 .15 (.06) .85
Parcel 2 1.62 .40 (.05) .65 1.48 .44 (.09) .61
Parcel 3 1.36 .23 (.02) .53 1.38 .16 (.03) .59
Fighting
Parcel 1 1.52 .39 (.05) .59 1.47 .37 (.09) .56
Parcel 2 1.36 .31 (.03) .56 1.45 .40 (.08) .45
Parcel 3 1.64 .34 (.03) .49 1.94 .63 (.11) .30
Anger
Parcel 1 1.56 .51 (.05) .53 1.48 .66 (.12) .27
Parcel 2 1.52 .32 (.04) .61 1.42 .28 (.06) .45
Parcel 3 1.37 .41 (.04) .42 1.31 .21 (.04) .39
Belonging
Parcel 1 2.87 .32 (.03) .46 2.94 .30 (.06) .48
Parcel 2 3.06 .32 (.03) .45 3.04 .27 (.06) .50
Parcel 3 3.07 .24 (.03) .47 3.12 .20 (.04) .51
Support: School
Parcel 1 2.11 .08 (.01) .77 2.27 .09 (.03) .80
Parcel 2 2.15 .14 (.02) .63 2.22 .19 (.04) .63
Parcel 3 2.23 .22 (.02) .44 2.18 .18 (.03) .56
(continued)
101
Table 21 (continued)
Students without Disabilities Students with Learning Disabilities
Indicator Mean
Scores
- Residual
(SE)
R2 Mean
Scores
- Residual
(SE)
R2
Social Support: Family
Parcel 1 2.56 .10 (.01) .71 2.51 .14 (.03) .69
Parcel 2 2.54 .12 (.01) .66 2.38 .13 (.03) .70
Parcel 3 2.71 .12 (.01) .50 2.59 .18 (.03) .47
Social Support: Peers
Parcel 1 2.23 .11 (.01) .71 2.21 .20 (.04) .58
Parcel 2 2.30 .14 (.02) .64 2.24 .24 (.05) .53
Parcel 3 2.25 .15 (.02) .59 2.25 .13 (.03) .63
Associations and Latent Mean Differences
To what extent does being identified with a learning disability influence associations
and mean levels of bullying, victimization, fighting, anger, sense of belonging, and social
supports? The second research question was to evaluate whether there are differences in
associations (e.g., variance, covariance) and latent means between students with learning
disabilities and students without disabilities across the eight latent constructs. Since factorial
invariance was established in the first research question, an extension of the three-step, multi-
group confirmatory factor analysis can be conducted. Once again, this procedure is a three-step
process, where variances are evaluated first, followed by an evaluation of covariances, and
concluded by an examination of latent means. The overall purpose of this procedure is to discern
whether the groups differ significantly on each construct.
The homogeneity of variance test was conducted to determine if there was variability
between the groups on the eight latent constructs. Model fit for the homogeneity of variance test
was in acceptable range, and fell within the parameters of the previous model (2(488) = 1021.635,
RMSEA = .062, NNFI = .92, CFI = .93). To establish whether this model is significantly
102
different from the strong metric invariance model, the 2 difference test was conducted. For the
current model (2(8) = 13.703, p > .05) no significant difference was found. Based on these
results, it can be concluded that the two groups of students do not significantly differ the
variability of the constructs.
Since the variance constraints are tenable, the next sequential step is to evaluate if the
variance/covariance matrix is significantly different between students with learning disabilities
and students without disabilities. Once the constraints were placed on both the variances and
covariances, model fit maintained acceptable fit and fell within the parameters of the other
models (2(516) = 1050.717, RMSEA = .06, NNFI = .93, CFI = .93). Once again, the 2 difference
test was employed (2(36) = 42.784, p > .05), and it was determined that the models are not
significantly different. Therefore, the associations among the latent constructs between the two
groups of students are not significantly different. Since these associations are invariant, it is
reasonable to equate the correlations among the latent constructs for students with learning
disabilities and students without disabilities. The aggregate correlations between the
constructions are reported in Table 22.
103
Table 22
Correlations Between the Latent Constructs
Construct Bully Victim Fight Anger Belonging TeachSS FamilySS PeerSS
Bully 1.00
Victim .42 1.00
Fight .47 .17 1.00
Anger .63 .38 .61 1.00
Belonging -.18 -.08 -.15 -.20 1.00
TeachSS -.11 .04 .04 -.08 .60 1.00
FamilySS -.13 -.08 -.11 -.05 .33 .47 1.00
PeerSS -.22 -.15 -.18 -.17 .34 .53 .56 1.00
Note. TeachSS = School Support Scale, FamilySS = Family Social Support Scale, PeerSS = Peer
Social Support Scale.
Once invariance was established for the variances and covariances, latent mean
invariance was evaluated. Similar to the variance and covariance procedure, means were equated
across the two groups to discern if these constraints significantly impeded the model. Evaluation
of the latent means invariance model revealed the model had acceptable fit (2(488) = 1019.969,
RMSEA = .06, NNFI = .93, CFI = .93), and it fell within the limits of the previous models (see
Table 24). When compared to the strong metric invariance model, using the 2 difference test, it
was determined that there is not a significant difference between the models (2(8) = 12.036, p >
.05), thereby confirming that the groups are not significantly different on any of the constrained
parameters. Therefore, latent mean scores do not differ between students with learning
disabilities and students without disabilities on any of the eight latent constructs. More
importantly, when latent mean scores do not differ across groups of respondents; it is acceptable
104
to merge the samples and evaluate as a one group model (Little, 1997; Shogren et al., 2007).
Based on the results of the subsequent analyses, and preferred practices of structural equation
modeling, all of the following analyses will be evaluated based on a one-group model. Due to the
convergence of the groups, aggregate latent mean scores were calculated and reported in Table
23.
To directly address research question 2, it is important to dissect the previous analytic
procedure. Based on the results of the CFA in research question 1, it was determined that the
constructs were measured equivalently between students with learning disabilities and students
without disabilities. Invariance on this procedure allows for the direct investigation of variances,
covariances, and latent mean scores between the groups. Interestingly, the current sample of
students with learning disabilities and students without disabilities were invariant across all
constraints. Therefore, since the associations and latent means are not significantly different, it is
reasonable to assume that proceeding with a two-group model will provide two equivalent
models with statistically similar parameter estimates. Additionally, common practice in structural
equation modeling is to merge the two samples since they are essentially equivalent (Little,
1997). Consequently, the evaluation of the structural equation model will be conducted as a one-
group model.
Table 23
Latent Mean Scores Merged From 2-Group Model
Bully Victim Fight Anger Belonging TeachSS FamilySS PeerSS
1.35 1.50 1.52 1.46 3.01 2.17 2.58 2.26
Note. TeachSS = School Support Scale, FamilySS = Family Social Support Scale, PeerSS = Peer
Social Support Scale.
105
Table 24
Fit Indices for Variance, Covariance, and Latent Means CFA Evaluations
Model 2 Df p
2 p RMSEA RMSEA 90% CI NNFI CFI Constraint
Tenable
Intercept
Invariance
1007.933 480 <.001 -- -- .063 .056 - .069 .92 .93 --
Homogeneity of
Variances
1021.635 488 <.001 13.703(8) <.05 .062 .056 - .068 .92 .93 Yes
Homogeneity of
Variances and
Covariances
1050.717 516 <.001 42.784(36) <.10 .061 .054 - .067 .93 .93 Yes
Latent Mean
Invariance
1019.969 488 <.001 12.036(8) <.10 .062 .056 - .069 .93 .93 Yes
106
Demographic Predictors
To what extent do gender, race, GPA, participation in extracurricular activities, and
percentage of special education services predict involvement in the bullying dynamic?
Research question three assessed demographic predictors within a structural framework. The
demographic predictors used for the subsequent analyses were gender, race, grade, grade point
average, participation in extracurricular activities, and percentage of special education services.
All demographic items functioned as covariates within the model, and were entered by using a
dummy coding procedure. For dichotomous items, 0 was entered for the first group and 1 was
entered for the second. The following was used for the dichotomous items: (a) Female = 0 and
Male = 1, (b) 7th Grade = 0 and 8th Grade =1, (c) Do Not Participate in Extracurricular Activities
= 0 and Do Participate in Extracurricular Activities = 1. For Race, two dummy codes were
created to represent African American, Caucasian, and Other. Therefore, the first variable was
Not African American = 0 and African American = 1, and the second variable was Not
Caucasian = 0 and Caucasian = 1. A similar procedure was used for grade point average and
percentage of special education services. Once all of the dummy codes were created, the raw
data were read into Lisrel 8.80 (Jöreskog & Sörbom, 2007) to evaluate model fit and significance
of the covariates. The initial structural model for the full dataset with all covariates freely
estimated demonstrated close model fit (2(416) = 818.98, RMSEA = .044, NNFI = .94, CFI = .96)
To evaluate the significance of each covariate, the path between each individual covariate
and each construct was freely estimated. Therefore, the 88 paths were freely estimated (i.e., 8
paths from each of the 11 covariates) to assess significance. The covariate estimate () and their
z-scores were evaluated for significance, where a z-score above 1.96 represents a significant
predictor. To extend the significance evaluation of each covariate, as step-wise deletion process
107
was used (Shogren et al., 2007). First, all values and their relative z-scores were calculated.
Once calculated, the covariate with the lowest z-score was constrained, and the model was
reevaluated. This procedure was conducted until only significant predictors remained. Overall,
71 iterations of this procedure was conducted, leaving only 17 significant predictors.
The freely estimated structural equation model with significant covariates included
demonstrated close model fit (2(375) = 754.87, RMSEA = .045, NNFI = .95, CFI = .98).
However, not all covariates demonstrated significant paths within the model. Nonsignificant
predictors included 61% or more special education services, grade, and below Mostly A’s and
B’s. Therefore, all of the nonsignificant covariates were removed, as well as all nonsignificant
covariate paths. Table 25 includes all of the significant covariate items, constructs, and
significance statistics.
Table 25
Estimates, Standard Errors, and Significance of Covariates
Demographic Group Construct
- Gamma (SE) z-score
Male Peer Social Support -.36 (.09) -4.08*
Male Victim -.24 (.10) -2.49*
Male Anger -.36 (.10) -3.61*
African American Belonging -.28 (.10) -2.80*
African American Peer Social Support -.34 (.09) -3.83*
African American Victim -.41 (.11) -3.75*
African American Anger -.31 (.11) -2.41*
Caucasian Fight -.40 (.11) -3.37*
(continued)
108
Table 25 (continued)
Demographic Group Construct
- Gamma (SE) z-score
Mostly A’s or B’s Bully -.20 (.10) -2.07*
Mostly A’s or B’s Fight -.54 (.12) -4.65*
Mostly A’s or B’s Anger -.33 (.11) -2.96*
Less than 20%
Services
Peer Social Support .36 (.17) 2.15*
Between 21 and 60%
Services
School Support .53 (.18) 2.90*
Extracurricular Peer Social Support .23 (.09) 2.58*
Extracurricular Bully .47 (.10) 4.62*
Extracurricular Fight .43 (.12) 3.58*
Extracurricular Anger .48 (.12) 3.84*
* Represents significance at the .05 level.
By investigating the significance of the covariates, direct influences on the constructs can
be determined. For brevity purposes, significant covariates will be discussed by specific item
covariates. The Male item exerted a significant influence on Peer Social Support ( = -.36, p <
.05), Victimization ( = -.24, p < .05) and Anger ( = -.36, p < .05; see Table 25) Based on the
negative values, the significance of these items are representative of females. Therefore, females
tend to have higher levels of peer social support when compared to males. More interestingly,
females tend to report higher levels of victimization and anger as measured by the University of
Illinois Anger Scale (Espelage & Holt, 2001; Espelage & Stein, 2006). This association will be
discussed in detail in Chapter 5.
109
In addition to gender, the race item for African Americans produced four significantly
negative covariates, indicating that individuals who identify themselves as African American
directly influence the latent construct. Specifically, significant covariates included sense of
belonging ( = -.28, p < .05), Peer Social Support ( = -.34, p < .05), Victimization ( = -.41, p <
.05), and Anger ( = -.31, p < .05; see Table 25). Individuals who are not African American
within this sample tended to have a significantly higher level of sense of belonging and social
peer support. Additionally, these individuals tend to report higher levels of victimization and
anger when compared to African American students within the sample. The item for Caucasians
students also produced one significant negative covariate for fighting ( = -.40, p < .05; see Table
25), indicating that individuals who are not Caucasian tended to engage in more fighting
behaviors.
When class level covariates were examined, a number of significant paths emerged. First,
students who identified themselves as earning Mostly A’s or Mostly A’s and B’s tended to have
lower levels of bullying ( = -.20, p < .05), fighting ( = -.54, p < .05) and anger ( = -.33, p <
.05; see Table 25) when compared to students who receive B’s or lower. This finding is quite
interesting given the breadth of literature documenting decreases in problem behavior can
influence increased academic achievement. This association will be discussed in greater detail in
Chapter 5.
Similar to class level covariates, percentage of special education service time documented
two significant paths. First, students who receive less than 20% special education services tended
to report higher levels of peer social support ( = .36, p < .05) when compared to students in
general education or students with disabilities who receive more services. Additionally, students
who receive 21 – 60% special education service time tended to report higher levels of teacher or
110
school personnel support ( = .53, p < .05; see Table 25). While this finding is extremely
interesting and relevant, it may be explained as a function of special education services, where as
services increase, access to teachers also increases.
Finally, participation in extracurricular activities exerted a significant influence on four
latent constructs. Not surprisingly, students who are involved in extracurricular activities tended
to have higher levels of social support from their peers ( = .23, p < .05). Conversely, students
who are involved in extracurricular activities also tended to report higher levels of bullying
perpetration ( = .47, p < .05), fighting ( = .43, p < .05), and anger ( = .48, p < .05; see Table
25). Therefore, individuals who are involved in extracurricular activities appear to engage in
more aggressive behaviors than their peers who are not involved.
Sense of Belonging and Social Supports as Predictors in the Bully Dynamic
To what extent does sense of belonging and social supports predict involvement
within the bullying dynamic? The final research question was posed to investigate social
predictors of bullying, victimization, fighting, and anger within a structural equation model
framework. Based on the findings from the CFA in research questions one and two, all eight
constructs were measured equivalently across all of the respondents in the sample. Initially, this
model was going to be estimated as a two-group model comparing students with learning
disabilities and students without disabilities. However, the latent invariance procedures
determined that the two groups were not significantly different, and should be evaluated as a
single sample. Therefore, within this sample, students with learning disabilities and students
without disabilities would maintain equivalent models, so they will be assessed as one group.
To investigate sense of belonging, social support from family, support from teachers or
school personnel, and social supports from peers as predictors of bullying, victimization,
111
fighting, and anger, a freely estimated structural equation model was constructed to assess initial
model fit. Based on the estimates of this model, close to acceptable fit was achieved (2(224) =
547.15, RMSEA = .057, NNFI = .95, CFI = .96). Once this freely estimated model was fit, the
covariates as described earlier were included (2(375) = 754.87, RMSEA = .045, NNFI = .95, CFI
= .98) while maintaining close model fit.
Although model fit of the freely estimated model was exceptional, it was necessary to
address the regression paths that were not significant to eliminate the ‘noise’ in the model. This
process is necessary, because insignificant regression paths can bias the predictors, and make
result interpretation difficult. Similar to the nonsignificant reduction process of covariates in
question 3, a step-wise reduction process was used to eliminate nonsignificant predictor (i.e.,
regression) paths. Although this process technically identical to the covariate reduction process,
the results are conceptually different. More specifically, the removal of nonsignificant covariates
eliminates static items that do not influence the constructs, where the removal of nonsignificant
regression paths eliminates nonsignificant predictors of the latent constructions from the model.
Initially, six regression paths emerged as insignificant (i.e., z-scores below 1.96) and
removal began in sequential order. Specifically, Family Social Support did not predict fighting
( = -.00, z = -.03, p > .05), Social Support did not predict victimization ( = -.00, z = -.06, p >
.05), Family Social Support did not predict Bullying ( = .05, z = .72, p > .05), Teacher or
School Personnel Support did not predict Bullying ( = .12, z = 1.35, p > .05), Teacher or
School Personnel Support did not predict Anger ( = .13, z = 1.46, p > .05), and Sense of
Belonging did not predict Bullying ( = -.10, z = -1.59, p > .05), resulting in the removal of
these nonsignificant paths. Once the initial nonsignificant paths were removed, additional
nonsignificant paths emerged. Specifically, Sense of Belonging did not predict victimization ( =
112
-.10, z = -1.27, p > .05), Sense of Belonging did not predict anger ( = -.10, z = -1.67, p > .05),
and Family Social Support did not predict anger ( = .12, z = 1.85, p > .05), so these paths were
also removed from further analyses. After the removal of all nonsignificant paths, it was
determined that the entire Family Support Scale was removed from further analyses. The final
model resulted in a close fitting model (2(384) = 769.51, RMSEA = .046, NNFI = .95, CFI = .96),
which represents the dataset and predictors as parsimoniously as possible. The final model and
predictor items are reported in Table 26.
Table 26
Beta Weights and Z-Scores of the Final Structural Model
Construct Beta (SE) z-score Standardized
beta
Path to Bullying
Belonging -- -- --
School Support - -- --
Family Social Support -- -- --
Peer Social Support -.24 (.06) -4.26* -.23
Path to Victimization
Belonging -- -- --
School Support .18 (.07) 2.77* .17
Family Social Support -- -- --
Peer Social Support -.30 (.07) -4.30* -.30
Path to Fighting
Belonging -.16 (.08) -1.99* -.15
School Support .31 (.09) 3.58* .28
Family Social Support -- -- --
Peer Social Support -.29 (.07) -3.82* -.26
Path to Anger
Belonging -- -- --
School Support -- -- --
Family Social Support -- -- --
Peer Social Support -.24 (.06) -3.81* -.23
* Represents significance at the .05 level.
113
Predictors of bully perpetration. Following the removal of all nonsignificant predictors,
a single predictor of bullying remained in the model. Peer social support was a negative predictor
of bullying ( = -.25, z = -4.26, p > .05), indicating that lower levels of social support from
peers predicted higher levels of self-reported bullying. Interestingly, School Sense of Belonging,
Teacher or School Personnel Support, or Family Support did not emerge as significant predictors
of bullying.
Predictors of victimization. Predictors of victimization included support from peers and
Teachers or School Personnel. Specifically, individuals who reported lower levels of peer
support ( = -.30, z = -4.30, p > .05) reported higher levels of victimization. Conversely,
individuals who reported higher levels of support from teachers or school personnel ( = .18, z =
2.77, p > .05) also reported higher levels of victimization. Therefore, increased adult support
could have unwanted effects on victimization, where increased peer support predicts lower levels
of reported victimization.
Predictors of fighting. Fighting emerged as the construct that had the most significant
predictors. Overall, three predictor paths emerged as significant in the current model. First, social
support from peers predicted lower levels of fighting behaviors ( = -.26, z = -3.82, p > .05),
indicating that peer support could directly impact levels of fighting behaviors. Second, lower
levels of belonging predicted increase fighting behaviors ( = -.16, z = -1.99 p > .05), indicating
that the more a student feels s/he belongs among their peer group, the less likely they are to
engage in fighting behaviors. Finally, increased support from adults and school personnel
predicted increased levels of fighting behaviors ( = .31 z = 3.58, p > .05).
114
Predictors of anger. Similar to bullying, victimization, and fighting, peer support served
as a predictor of anger. Once again, Support from Peers was negatively associated ( = -.24, z =
-3.81, p > .05) with anger. Therefore, individuals who reported lower levels of peer support also
reported higher levels of anger. Figure 8 represents the final predictive structural model with
covariates included and significant paths included.
Overall findings from the structural model indicate that sense of belonging, peer social
supports, and teacher or school personnel supports serve as predictors for at least one of the four
latent constructs. Sense of belonging served as a negative predictor of fighting, where students
who reported lower levels of belonging engaged in more fighting behaviors. Second, support
from Adults or school personnel served as a positive predictor of victimization and fighting,
where higher levels of adult support resulted in higher levels of victimization and fighting.
Finally, support from peers served as a negative predictor for all four of the latent construct.
Specifically, increased peer supports predicted decreased levels of bullying, victimization,
fighting, and anger.
115
Figure 8. Structural model with covariates and significant paths.
116
Chapter 5
Discussion
Involvement in the bullying dynamic is based on complex interactions between
individuals, their family, peer group, school, and societal norms (Swearer et al., 2009). To assess
the complexity of these interactions, researchers have started investigating potential predictors
associated with involvement. While existing research has provided a foundation for
understanding factors associated with the bullying dynamic, the devastating outcomes related to
involvement still plague our nation’s youth. Current national statistics suggest that between 20
and 30% of school aged children are involved as bullies, victims, or bully-victims (Dinkes et al.,
2006; Dinkes et al., 2006; Nansel et al., 2001). Recently, bullying has become a national
epidemic that has prompted increased public concern due to the tragic and fatal outcomes for a
number of youth.
Considering its complexity, it is conceivable to believe that involvement in the bullying
dynamic is not equitably distributed across all subgroups of students. Therefore, characteristics
associated with specific subgroups may predict increased involvement as bullies or victims.
Specifically, recent research suggests that when consideration is given to disability status,
individuals with disabilities may be twice as likely to be bullies or victims (see Rose et al.,
2010). While this distinction was necessary to provide an exploratory glimpse of the differences
between students with and without disabilities, fundamental characteristic differences exist
between students identified with a disability (Smith, 2007). Based on the categorical differences,
it is difficult to make the assertion that the entire subgroup of students with disabilities is
overrepresented within the bullying dynamic.
117
To address this fundamental concern, the present study focused on exclusively comparing
the involvement of students with learning disabilities to students without disabilities. Making this
comparison is especially germane to disability studies because students with learning disabilities
represent the largest special education subgroup. To assess participatory differences and
predictors of involvement, four research questions were developed: (a) can constructs used to
assess involvement in the bullying dynamic be measured equivalently?; (b) are their quantifiable
difference between the groups on these constructs?; (c) to what extent do demographic variables
predict involvement?; and (d) to what extent does sense of belonging and social supports predict
involvement?
In interpreting the findings of the study, a number of factors must be considered. First, it
was critical to collect accurate disability information from the school district to evaluate the
differences between the two subgroups. This consideration is important because it eliminated
two major limitations present in the existing literature. Initially, accurate data eliminated the
necessity to aggregate all students with disabilities into one group, thereby allowing the
unconstrained comparison between students with learning disabilities and students without
disabilities. Additionally, these data eliminated bias associated with inaccurate self-reporting of
personal disability.
Second, consideration should be given to the overall sample population as described in
Chapter 4. Several 2 statistics were calculated to discern if demographic variables for students
with learning disabilities differed significantly from students without disabilities (e.g., gender by
disability). While a few demographic categories demonstrated significant differences, the overall
sample was relatively equivalent across the two subgroups of students. Therefore, comparisons
can be made more accurately because the two samples are proportionally equivalent.
118
Finally, structural equation modeling (SEM) was used to evaluate differences between
the two groups. This approach is appropriate for investigating differences between students with
learning disabilities and students without disabilities because the SEM procedure yields accurate
estimates of unbiased population parameters. Therefore, by using the SEM procedure,
similarities and differences among latent constructs can be evaluated while controlling for
measurement error (Kline, 1998). Thus, more accurate interpretations of the group differences
can be made to discern if the groups are statistically different.
Overall Findings
Four findings emerged from the current study. First, measurement invariance was
evaluated for bullying, victimization, fighting, anger, sense of belonging, and social supports
between the two groups of students. Based on the three-step multi-group confirmatory factor
analytic process, measurement invariance was established indicating that the same constructs
were being assessed between the groups. This invariance allowed for accurate comparisons
between latent means and associations.
A second, yet unexpected, finding was revealed due to the analytic procedure. Following
the CFA, comparisons across variances, covariances, and latent means were conducted. Results
for this level of analysis revealed that students with learning disabilities and students without
disabilities did not differ significantly on bullying, victimization, fighting anger, sense of
belonging, school support, family social support, or peer social support. Therefore, the two
groups were statistically equivalent, and were merged to assess predictors associated with the
entire population of students. While the two groups were merged in the current study, it should
be noted that a majority of extant literature suggests the two groups are characteristically
different. Therefore, implications for each group will be discussed in the following section.
119
Third, several demographic variables emerged as significant predictors of the eight latent
constructs. This finding is noteworthy because it indicates that consideration should be given to
gender, ethnicity, participation in extracurricular activities, and grade point average when
assessing involvement within the bully dynamic for students with learning disabilities and
students without disabilities. Additionally, this finding supports the assertion that bullying
involvement is not equitably distributed across all subgroups of school-age youth (e.g., males,
females, students who are involved in extracurricular activities), and provides empirical backing
for the argument regarding individualized supports and multi-tiered bully prevention programs.
Finally, results indicated that social supports offered from peers were the only common
predictor of the four latent constructs (bullying, victimization, fighting, and anger). As expected,
increased levels of peer supports decreased levels of bullying, victimization, fighting, and anger.
Contrary to expectations: school sense of belonging only predicted fighting, teacher or school
personnel social support only predicted victimization and fighting, and family social support
failed to predict any of the four latent constructs. Surprisingly, when school support was
evaluated as a predictor for this sample, increased levels of support predicted increased levels of
victimization and fighting.
Bullying Construct Measurement
Although bullying has become a mainstay in social science literature, few studies have
examined the differences between students with and without disabilities (see, Rose, 2010). Due
to the dearth of literature in this area, construct measurement confirmation has remained
untested. However, this level of analysis is necessary to ensure that the items or scales are
measuring the desired construct (Little, 1997). A factor analytic process is common practice for
psychometric development, but once reliability and validity have been consistently established,
120
the scale measures can be consistently utilized with similar samples (Dahlberg et al., 2005).
However, the overarching concern in the current study was whether the psychometrics
demonstrated construct bias because they were being used to measure responses from a new, and
unique group of students (Kline, 1998). Therefore, the first research question investigated
factorial invariance between students with learning disabilities and students without disabilities.
Results of the CFA process indicated that the variables were measuring the desired
constructs of bullying, victimization, fighting, anger, sense of belonging, and social supports
between students with learning disabilities and students without disabilities. Additionally, the
modeling process maintained acceptable model fit throughout the three-step procedure, and it
was determined that the psychometrics used for assessment in the study were tenable for both
groups of students. These findings are consistent with previous factor analytic procedures for
each of the eight latent constructs.
While direct multi-group CFA procedures have not been used to assess the differences
between involvement in bullying for students with disabilities, factorial invariance was expected
in the current study because results from previous factor analytic procedures substantiate the use
of these scales for individual subgroups of students. For example, the University of Illinois
Aggression Scales (Espelage & Holt, 2001) have been used to reliably measure the bullying
constructs related to social supports (Holt & Espelage, 2007), dating violence and sexual
harassment (Holt & Espelage, 2005), associated risk factors (Holt, Finkelhor, & Kantor, 2007),
peer supports (Espelage et al., 2003) and homophobic teasing (Poteat, 2008; Poteat & Espelage,
2005; Poteat & Rivers, 2010). Additionally, the Sense of Belonging (Goodenow, 1993) and
Social Supports Scales (Vaux, 1988) have demonstrated acceptable internal consistency across
various subgroups of respondents (Poteat & Espelage, 2007; Vera et al., 2008).
121
Based on the historical consistencies of the selected scales, it was not surprising to
establish factorial invariance for students with learning disabilities and students without
disabilities. Invariance was expected for the current sample because each scale (other than
Anger) has been subjected to multiple factor analytic procedures and has been administered to
various diverse populations of students, while maintaining acceptable internal consistency.
Additionally, it is reasonable to conclude that each sample described above included some
students with disabilities. Therefore, internal consistency estimates for each sample included
representation of students with disabilities. However, the statistical rigor of the multi-group
confirmatory factor analysis allowed for a direct comparison between students with learning
disabilities and students without disabilities to ensure that the desired construct was measured
equivalently between the two groups of students.
Direct Comparison of Student Groups
At the present time, an emerging literature base investigating the involvement of students
with disabilities in the bullying dynamic is being constructed. Unfortunately, current
investigations of the involvement of students with disabilities has over-generalized the groupings
of students by constructing broad subsamples regardless of disability labels or characteristics.
For example, Rose and colleagues (2009) investigated self-reported bullying, victimization, and
fighting among a large-scale sample of students with and without disabilities. A trichotomy was
created representing students without disabilities, students who received minimal special
education services, and students who received more restrictive services. Although findings
supported previous research that suggests students with disabilities were overrepresented as
bullies and victims, it was difficult to make generalizations from this study because group
construction was based on arbitrary disability categories.
122
To address this fundamental problem, the current study investigated differences between
students without disabilities and students with learning disabilities. An extension of the initial
confirmatory factor analysis was conducted to directly compare differences in variances,
covariances, and latent mean scores. As expected, when the variances and covariances were
sequentially constrained, the groups did not differ in their direct associations to the latent
constructs. Contrary to the initial hypothesis and extant literature, when the latent means were
constrained, model fit did not change significantly. This level of invariance allowed for the two
groups to be converged because they were statistically similar across all of the sequentially
constrained parameters (Little, 1997).
While the aforementioned result was unexpected, it does represent a significant finding
within the special education bullying literature. Specifically, when students with learning
disabilities were directly compared to students without disabilities it was determined that
students with learning disabilities did not report higher or lower levels of bullying, victimization,
fighting, or anger than students without disabilities. Additionally, since the CFA was an eight
factor model, it was also determined that students with learning disabilities did not report higher
or lower levels of belonging or social supports. Although a majority of the extant literature
conflicts with the current findings, a small number of studies substantiate the results (Wallace,
Anderson, Bartholomay, & Hupp, 2002; White & Loeber, 2008).
Interpretation of these results requires an initial interpretation of the sample population.
First, as stated previously, very few 2 difference tests emerged as significant, indicating that the
sample was relatively proportionate. Secondly, students with learning disabilities represent the
largest subgroup of students with disabilities (Smith, 2007), and questions arise as to how these
students are identified (McKenzie, 2009). This criticism hinges on the ambiguity related to the
123
operational definition of learning disability and the variability of constraints (e.g., discrepancy
model, RTI) used to identify students as learning disabled (McKenzie, 2009). Based on the
variability of the identification process, the label of learning disability can fall on a continuum
and depend on the local (e.g., school, district) identification process and policies. Therefore, the
nonsignificant difference between the groups in the current sample could be based on group
overlap, where the learning disability group contained individuals who have been inaccurately
identified with a disability and the general education group contained individuals who could
benefit from special education services, but have not been formally identified with a disability.
While group overlap is conceivable for any categorical structure among highly diverse
populations of students, some empirical evidence suggests that students with and without
learning disabilities are not fundamentally different. For example, White and Loeber’s (2008)
longitudinal investigation of bullying and disability status as predictors of serious delinquency
corroborated the current findings. The researchers reported that when disability status was
inserted into their longitudinal analysis, a nonsignificant change was documented and the groups
were determined to be characteristically equivalent in their bullying behaviors and experiences.
Additionally, students with learning disabilities have a higher likelihood of receiving
their educational services in a general education environment. These inclusive practices could
serve as a buffer against increased bully dynamic involvement due to positive peer behavior
modeling, acquisition of social skills, increased social and academic development (Brown et al.,
1989), increased acceptance, reduction in negative stereotypes (Martlew & Hodson, 1991), and
increased participation in classroom activities (Sabornie, 1994). Similarly, Wallace and
colleagues (2002) reported no identifiable differences between students with and without
124
disabilities on engagement and behaviors. In the Wallace et al., (2002) study, the majority of
observed students with disabilities were identified with a learning disability.
Although the results of research question two documented an unexpected outcome,
previous research on students with learning disabilities provides a reasonable interpretation.
However, if the current literature on bullying among students with disabilities is valid, then
further analyses are warranted to determine the overrepresented subgroup of students.
Additionally, the majority of literature suggests that these two groups are characteristically
different, and these differences should be examined to develop a more holistic understanding of
unique predictors across these differences. Overall, two questions emerged: a) Does this study, as
well as existing research, compare similar populations of students? and b) Does another
subgroup of students with disabilities represent the highest proportion of students involved in the
bullying dynamic?
Demographic Predictors
The Social-Ecological Framework for Bullying/Victimization (Espelage & Swearer,
2004) was used as the interpretative model for the current investigation. Based on this
framework, factors associated with self, family, school, and peer group directly influence
involvement within the bullying dynamic (Espelage & Swearer, 2004). To assess individual
social ecological predictors associated with increased participation in the bullying dynamic;
gender, grade, ethnicity, GPA, participation in extracurricular activities, and percentage of
special education services were evaluated as covariates within a structural equation modeling
framework. As expected, the grade level variable did not significantly influence any of the latent
constructs. However, gender, grade, ethnicity, GPA, participation in extracurricular activities,
and percentage of special education services resulted in at least one significant path.
125
Gender. Gender emerged as a significant indicator of peer social support, victimization,
and anger, where females tended to report higher levels on all three constructs. The significant
path to peer social support is plausible because