n: fiatflw.
kwéww »,
$4.39..

. v n;
35?...

{a}
Em A
.. .a

. . uh . .9. .z .
I: 3.49.?" 1.15...

.3. 1551

s .‘U .1. .
#«flflfi

.2. a
1: g

.1

1 t u
‘ ‘ :5). . 9. ‘
1W hum-mhMJnFIJ
«in. Run}- ,.3 t

.3. a..!...)..,! .I

fit .ukm

' ‘1
3min)?
‘5: 1‘4... r. .
.21 .1. INA}

t
:‘Prnntuywrvnl

 

   
 
 

 

(95/37 . L'BRARY
Michzgan State
University l

 

This is to certify that the
dissertation entitled

AN ECOLOGICAL PREDICTION MODEL OF BULLYING '
BEHAVIORS AMONG SOUTH KOREAN MIDDEL SCHOOL
STUDENTS

presented by
Chang-Hun Lee

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in School of Criminal Justice

/ Majnyrof'essor’s Signature

 

 

—.-._<--<—.—-.-.-

 

MSU is an affirmative—action, equal-opponunity employer

 

 

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

 

DATE DUE DATE DUE DATE DUE
1 2 0 3 Late
.\ K 1! wk)"

‘T

U"

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/DaleDue.indd-pt1

AN ECOLOGICAL PREDICTION MODEL OF BULLYING BEHAVIORS AMONG
SOUTH KOREAN MIDDLE SCHOOL STUDENTS

By

Chang-Hun Lee

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

School of Criminal Justice

2007

ABSTRACT

AN ECOLOGICAL PREDICTION MODEL OF BULLYING BEHAVIORS AMONG
SOUTH KOREAN MIDDLE SCHOOL STUDENTS

By

Chang-Hun Lee

This study is designed to address shortcomings found in bullying
literature. Specifically, this study applies ecological system theory
(Bronfenbrenner, 1979) as a theoretical framework to identify different levels of
factors influencing bullying behavior in schools. These levels include the
microsystem, mesosystem, exosystem, and macrosystem levels. In addition, this
study develops four measures of peer influence relevant to bullying behavior in
school, and improves the definition of bullying and its measures by including
severity, duration, and visibility dimensions of bullying. Using a multiple cluster
sampling method, this study randomly selected 36 classes from 6 different areas
of South Korea. Data were collected from 1238 middle school students in those
selected classes in 2007. In the first phase of the analysis, basic statistical
analyses, hypothesis testing with OLS regression, and sensitivity analysis were
performed. In the second phase, the second-order confirmatory factor analysis
and structural equation modeling were performed. A multiple group analysis for

different types of bullying was attempted. Findings suggest the utility of the

ecological approach, and that this model accounted for a high portion of variance
in bullying. All of the ecological systems as well as individual traits were found to
be statistically significant influences on bullying either directly or indirectly. Based

on the findings, policy implications and future research are discussed.

To my dear wife, Jung-Mi Kim, and my precious son, Eugene Kim Lee

iv

ACKNOWLEDGEMENT

The author wishes to thank the fine faculty of School of Criminal Justice
at Michigan State University for assisting me in the realization of this project.
Especially, the author is grateful to Dr. Merry Morash for her guidance and
support. Without her guidance and assistance, this research could not have been
completed. The author also would like to thank the dissertation committee
members, Dr. Vincent Hoffman, Dr. Ed McGarrell, and Dr. Ken Frank, for their
knowledgeable guidance and advice. The author extends special thanks to
Jihyun Kim, Woonjin Lee, and many others who helped and supported this
author to complete this project from the beginning to the end.

The author is grateful to my parents, Pyoung-Ho Lee and Young—Ran Lee,
and my mother-in-law, Myung-Kyu Lee. Most of all, the author would like to thank
my dear wife, Jung-Mi Kim, and my precious son, Eugene Kim Lee, for their
endless smiles, support, and unconditional love. Love and thanks to them for

their presence in my life.

TABLE OF CONTENTS

 

 

 

LIST OF TABLES xi
LIST OF FIGURES xiii
CHAPTER 1
INTRODUCTION
I. INTRODUCTION 1
II. A BRIEF HISTORIC OVERVIEW OF BULLYING STUDIES ------------ 6

III. SIGNIFICANCE OF BULLYING AND

 

 

 

IMPORTANCE OF THE CURRENT STUDY 8

1. Prevalence and Incidence Rates of Bullying in Schools ----------- 8

2. Outcomes of Bullying 10

3. Importance of the Current Study 11
CHAPTER 2

LITERATURE REVIEW

 

 

 

 

 

 

 

 

I. DEFINITION OF BULLYING 14
1. Physical, Verbal, and Relational Bullying 16
2 Chronic vs. Non-chronic

and Severe/overt vs. Moderate/covert Bullying 18
II. REVIEW OF PREVIOUS STUDIES OF BULLYING 22
1. Thematic Overview of Previous Bullying Studies 22
A. Types of Bullying Behaviors 22

B. Roles-played in Bullying Situations and
their Characteristics 23
C. Correlates of Bullying Behaviors 26

D. The Social Network Approach to Explaining Bullying -------- 30

vi

IV. ECOLOGICAL MODEL OF BULLYING IN SCHOOLS
1.

V.

VI.

SHORTCOMINGS OF RESEARCH ON BULLYING

5.
6.

A SUMMARY OF THE CURRENT STUDY

HYPOTHESES
1.

 

 

Child as a Center of School Ecology

 

Age
Prior Experience of Bullying Victimization
Dominance and Impulsivity

 

 

 

 

Attitude towards Aggression
Fun Seeking Tendency
Microsystem
A. Experiences in the Family
B. Experiences with Teachers
C. Experiences with Peers
Mesosystem
Exosystem
A. Gender
B. Family SES

C. School Climate
Macrosystem
Summary of the Current Ecological Model of School Bullying

mpow>

 

 

 

 

 

 

 

 

 

 

 

 

 

Child as Center of School Ecology

 

 

 

2. Microsystem Level (Interactions among Individuals)
3. Mesosystem Level (Interactions among Microsystems) -----------
4. Exosystem Level (Interactions among Mesosystems) ..............
5. Macrosystem Level
CHAPTER 3
METHODOLOGY
INTRODUCTION

CROSS-SECTIONAL RESEARCH DESIGN

 

 

vii

36

39

42
42
43
44
44
44
45
46
47
48
55
57
58
58
60
62
67

69

72

72
72
73
74
74

76

79

SAMPLING AND DATA COLLECTION PROCEDURES
1. Sampling

 

 

2. Data Collection Procedure and Response Rates

 

 

MEASUREMENT

 

1. Level of Analysis
2. Self-evaluative and Anonymous Measures

 

3. Operationalized Variables

 

 

A. Dependent Variable: Bullying Behaviors
B. Independent Variables

 

ANALYSIS PROCEDURE

 

CHAPTER 4
FINDINGS
DATA PREPARATION

 

FACTOR ANALYSIS, SCALES, AND RELIABILITY TESTS

UNIVARIATE, BIVARIATE, AND MULTIVARIATE ANALYSES

1. Univariate Analysis

 

A. Demographic Characteristics and

 

Academic Achievement of Sample
B. General Characteristics of Sample

 

 

C. Validity of School Climate Measures
2. Bivariate Analysis

 

Individual Traits

 

 

Microsystem Variables
Mesosystem Variables

 

 

Exosystem Variables
Macrosystem Variables

m90w>

 

3. Multivariate Analysis

 

 

A. Individual Traits (H1a — H1f)
B. Microsystem Level (H2a — H2k)

 

 

C. Mesosystem Level (H3a — H3b)

viii

84

84
87

90

90
91
93
93
94

102

106
112

115
115

115
118
121
125
125
128
131
133
135
137
137
139
140

VI.

' MULTIPLE GROUP ANALYSIS

 

D. Exosystem Level (H4a — H4g)
E. Macrosystem Level (H5a — H50)

 

SENSITIVITY ANALYSIS: ITCV

 

STRUCTURAL EQUATION MODELING: BASIC MODEL -----------

 

1. Confirmatory Factor Analysis
2. Structural Equation Modeling

 

A. The Initial Ecological Prediction Model for Bullying (ME)

 

B. Revised Model 1 (Mn)
C. The Full Theoretical Model

 

(reciprocal relationship of climate, MFT)
D. Revised Model 2 (Mg)

 

E. Revised Model 3 (Mr3)

 

STRUCTURAL EQUATION MODELING:

 

CHAPTER 5
DISCUSSION AND CONCLUSION

 

SUMMARY AND DISCUSSION

 

1 . Multivariate Analysis
2. Structural Equation Modeling

 

LIMITATIONS AND RECOMMENDATIONS

 

 

CONCLUSION AND POLICY IMPLICATIONS

APPENDICES

Appendix A. 232 Korean Governmental Districts Categorized into

Appendix B. Survey Questionnaire - Students

Appendix C. Survey Questionnaire - Teachers

Three Levels of Social Disorganization

 

 

 

ix

140
140

143

149

149
152
154
157

160
161
161

164

166

166
169

173

175

178
181

196

BIBLIOGRAPHY

Bibliography 202

 

Table 1

Table 2

Table 3

Table 4
Table 5

Table 6

Table 7
Table 8

Table 9

Table 10
Table 11

Table 12

Table 13

Table 14

Table 15

LIST OF TABLES

Typical Categorization of Bullying 17

 

Expanded Definition of Bullying: Temporal and Visibility

Elements Added to the Traditional Physicality / Severity
Dimension of Bullying Behavior 20

 

Bronfenbrenner’s Ecological System Model and

 

 

Factors Relevant to Bullying in Schools 67
Response Rates for Survey Questionnaire 89
Univariate Normality Tests: Skew and Kurtosis Values --------- 107

Univariate Statistics of the Non-normally Distributed

 

 

Items 120 — 125 108
Group Comparison of Frequencies of Missing Data ------------- 109
Reliability Coefficients for Scales Used 113

Demographic and Academic Achievement

 

 

 

Characteristics of Sample (N = 1238) 116
General Characteristics of Sample (N = 1238) 119
Mean Comparison across Six Schools 122

Correlations between Average Scores of Teachers and

Students Responses to Measures of Academic Achievement,
Moral Authority, and Moral Atmosphere 123

 

Correlations between Individual Traits

 

and Bullying Behaviors 127

Correlations between Microsystem Variables

 

and Bullying Behaviors 130

Correlations between Mesosystem Variables

xi

and Bullying Behaviors 132

 

Table 16 Correlations between Exosystem Variables

 

 

 

 

 

 

and Bullying Behaviors 134
Table 17 Correlations between Macrosystem Variables

and Bullying Behaviors 136
Table 18 Multivariate Regression 140
Table 19 ITCV Scores 145
Table 20 Comparison between ITCV and Impact of Covariates ----------- 147
Table 21 Confirmatory Factor Analysis 150
Table 22 Model Fit Summary for All Models (N = 1238) 154
Table 23 Standardized Path Coefficients and R2 for All Models ----------- 155
Table 24 Results of Modification Indexes from All Models 158

 

xii

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

LIST OF FIGURES

Bronfenbrenner’s Ecological System Model

Applied to the School Setting

 

Multiple Cluster Sampling of Korean
Middle School Students

 

An Ecological Prediction Model of School Bullying

Mean Difference in Academic Achievement

across Six Schools

 

 

Mean Difference in Teachers’ Moral Authority

across Six Schools

 

Mean Difference in Moral Atmosphere

across Six Schools

 

Initial Ecological Prediction Model of Bullying Behaviors

Revised Model 3,

Ecological Prediction Model of Bullying Behaviors

xiii

 

40

87
104

123

124

124
153

163

CHAPTER 1
INTRODUCTION

I. Introduction
II. A Brief Historic Overview of Bullying Studies
III. Significance of Bullying and Importance of the Current Study

I. INTRODUCTION

Bullying has devastating consequences. Some victims take their lives,
and others seek revenge. According to the 2005 Youth Statistics by the Korean
National Statistical Office, the suicide rate among 10,000 youths in Korea
dramatically increased from 3.8 to 8.2 from 2000 to 2005 (KNSO, 2005). In 2000,
suicide was the third leading cause of death of teenagers in Korea after traffic
accidents and cancer (Choi, 2003), but in 2003, suicide became the second
leading cause of death (KNSO, 2005). A large portion (about 54.2%) of those
who took their lives were reported victims of bullying in their schools (Kim, Koh, &
Leventhal, 2005).

In the United States, school shootings have been reported as a
disastrous response to bullying in schools (Leary, Kowalski, Smith, & Phillips,
2003; Verlinden, Hersen, & Thomas, 2000). A total of 220 school shootings
occurred between 1994 and 1999, and the number of victims killed by the
shootings reached 253 (Anderson, Kaufman, Simon, Barrios, Paulozzi, Ryan,
Hammond, Modzeleski, Feucht, & Potter, 2001). It is difficult to compare
prevalence rates of bullying for different countries, because of different definitions

1

of bullying and different reflection time periods used in studies (Farrington, 1993;
Kim, 2004; Smith, Cowie, Olafsson, & Liefooghe, 2002). Although comparisons
are difficult, there is evidence that bullying has became a more common type of
school violence in many countries around the world (for review, see Olweus,
1993, 1995).

One nationwide study to determine the prevalence of bullying in the US.
showed that it affected 29.9% of students in middle schools; 13% of the students
were bullies, 10.6% victims, and 6.3% both bullies and victims (Nansel, Overpeck,
Pilla, Ruan, Simons-Morton, & Scheidt, 2001). South Korea is not an exception to
this pattern. Using a definition of bullying similar to the one used in Olweus’
studies, a study found that, in South Korean middle schools, the prevalence rate
for bullying has increased dramatically between the late 1990s and 2006. In 2006,
12% of youths were bullies, 5.3% of them were victims, and 7.2% of them were
both bullies and victims (Yang, Kim, Kim, Shin, & Yoon, 2006).

These startling figures provoke a number of questions. Why are children
bullying classmates? What causes bullying? Who are the bullies? Is there any
effective intervention for bullying in schools? To answer these questions, many
researchers from different areas of academics, such as psychology,
psychopathology, and sociology, have invested their energy in carrying out
research. The result is a large body of literature that investigates factors related
to bullying behavior in school. Those factors found to be significant include
individual personality and psychological traits (Ando, Asakura, & Simons-Morton,
2005; Bosworh, Espelage, & Simon, 1999; Dill, Vernberg, Fonagy, Twemlow, &
Gamm, 2004; Espelage & Swearer, 2003; Gladstone, Parker, & Malhi, 2006;

2

Haynie, Nansel, Eithel, Crump, Saylor, Yu, & Simons-Morton, 2001; Unnever &
Cornell, 2003), harsh physical discipline, inconsistent parenting, and family
experiences (Batsche & Knoff, 1994; Bowers, Smith, & Binney, 1994; Christie-
Mizell, 2003; Curtner-Smith, Culp, Scheib, Owen, Tllley, Murphy, Parkman, &
Coleman, 2006; Flouri & Buchanan, 2003; Gladstone et al., 2006; Shields &
Cicchetti, 2001), membership in peer groups and friendship networks that
encourage bullying (Bosworh et al., 1999; Boulton & Smith, 1994; Boulton,
Trueman, Chau, Whitehand, &Amatya, 1999; Buhs, Ladd, & Herald, 2006;
Espelage, Bosworh, & Simon, 2000; Gest, Graham-Bermann, & Hartup, 2001;
Haselager, Hartup, Lieshout, & Riksen-Walraven, 1998; Haynie, 2001; Hodges,
Boivin, \fttaro, & Bukowski, 1999; Hodges, Malone, & Perry, 1997; Huttunen,
Salmivalli, & Lagerspetz, 1996; Mouttapa, Valente, Gallaher, Rohrbach, & Unger,
2004; Poulin, Cillessen, Hubbard, Coie, Dodge, & Schwartz, 1997; Salmivalli,
Huttunen, & Lagerspetz, 1997; Salmivalli, Lappalainen, & Lagerspetz, 1998), and
school environments and community characteristics (Farrington, 1993; Hoshino,
2001; Laub & Lauritsen, 1998; Menesini, Melan, & Pignatti, 2000; Naito & Gielen,
2005; Twemlow, Fonagy, & Sacco, 2001).

The outcomes of prior research, however, are somewhat disappointing for
several reasons. First, bullying in school has not been featured in the
criminological literature, especially in the United States (Batsche & Knoff, 1994;
Espelage 8 Swearer, 2003; Farrington, 1993; Spivak & Prothrow-Stith, 2001).
Second, there is lack of a theoretical framework to explain the causal linkages
between bullying and the various facilitating influences on bullying (Espelage &

Swearer, 2003; Morrison, 2002; Sullivan, Cleary, & Sullivan, 2004; Swearer &

3

Doll, 2001). Third, bullying has not been understood within social contexts,
particularly school settings that are the ecological environment where it occurs
(Morrison, 2001; Olweus, 1993; Smith & Brain, 2000). Fourth, several aspects of
bullying are left underinvestigated or uninvestigated by researchers. Examples
are teacher’s involvement in bullying phenomenon, school climate (Espelage &
Swearer, 2003; Farrington, 1993) and quality of friendship networks (e.g., Naylor
& Cowie, 1999; Pakaslahti & Keltikangas-Jarvinen, 2001; Park, 2002). These
shortcomings are discussed in detail in a later section of Chapter 2,
“Shortcomings of Research on Bullying.”

This dissertation is designed to address shortcomings in bullying
literature. Specifically, the proposed research will apply ecological system theory
(Bronfenbrenner, 1979) as a theoretical framework to identify different levels of
factors influencing bullying behavior in schools. These levels include the
microsystem, mesosystem, exosystem, and macrosystem levels. These key
concepts from ecological system theory will be defined and discussed in the
section, Ecological Model of Bullying in Schools, in Chapter 2.

This study develops four measures of peer influence relevant to bullying
phenomenon in schools. These four measures are power dynamics, level of
difference-acceptance, coercion of public self, and pseudo friendship. Also
children’s interactions with teachers and parents are included as an important
microsystem in this ecological model. In addition, as a part of mesosystems,
parental communication with teachers and peers is studied, and school climate,
which has been neglected in prior empirical work, is also included as a part of the

exosystem in the ecological model.

Finally, this study uses an improved categorization of bullying behaviors.
Traditional typologies of bullying focus on physical, verbal, and relational (or
direct vs. indirect) dimension of bullying behaviors. This study adds temporal,
visibility and severity elements to the traditional bullying typology. It will consider
chronic/severe/overt (e.g., $34 or wangtfa), non-chronic/severe/overt or
chronic/moderate/covert (e.g., Bill} or bantta), and non-chronic/moderate/covert

bullying (e.g., III or tta) in Korean schoolsI.

 

' The Korean terms for bullying behavior are provided since they are used in the
Korean scholarly literature.

II. A BRIEF HISTORIC OVERVIEW OF BULLYING STUDIES

The phenomenon of bullying was first investigated by the Norwegian
scholar, Heinemann, in 1973 (for review, see Olweus, 1993; Smith et al., 2002).
Heinemann used the Norwegian term mobbning to refer to sudden group
violence against individuals (Heinemann, 1973). Shortly after, the term bullying
was first used by a prominent bullying scholar, Dan Olweus, in 1978. In his book,
Aggression in the Schools: Bullies and Whipping Boys, he suggested that
bullying was a one-on-one physical attack of a stronger child against a weaker
child (Olweus, 1978). Later, Olweus incorporated verbal aggression and group
action into the previous bullying concept, and suggested that bullying is repeated
exposure to negative actions against an individual by one or more individuals,
and that the negative actions could involve aggressive behaviors as well as
aggressive words (Olweus, 1993, p 9). Since the initial work by Heinemann and
Olweus, Scandinavian research on bullying continued to flourish throughout the
19703 and 19805 in Nonivay, Denmark, Sweden, and Finland, and this tradition
spread to other European countries, such as England, Germany, and Spain (for
review, see Farrington, 1993; Olweus, 1993).

During the late 19803 and early 19905, in some Asian countries,
especially in Japan, researchers started investigating the phenomenon of bullying
in schools (e.g., Matsueda & Heimer, 1987; Naito, 1990; Naito & Gielen, 2005). It
was more recent that scholars in South Korea, Turkey and the United States
began to carry out research to understand the bullying phenomenon. Especially

in South Korea, the phenomenon of bullying in schools has drawn academic

6

attention since the media first used the term Ijime, a Japanese term for bullying,
in the middle of the 19905, and since then the term has become popular in
Korean society (Ahn, 2002; Farrington, 1993; KEDI, 1998; Kepenekci & Cinkir,
2006; Kim, 2004; Park & Son, 1998; Seo & Han, 2004).

III. SIGNIFICANCE OF BULLYING AND
IMPORTANCE OF THE CURRENT STUDY

1. Prevalence and Incidence Rates of Bullying in Schools

Atypical approach to understanding the phenomenon of bullying in
schools is to see how many students are involved as perpetrators or victims of
bullying behaviors during a specified period of time. Even though there are
contradictions between definitions of bullying across different studies, bullying
and victimization rates could provide insightful information about the seriousness
and significance of the bullying phenomenon.

Criminologists commonly focus on prevalence and incidence rates as
reflections of the scope of illegal behavior, including bullying. Prevalence is
typically defined as “the number of different persons committing crimes or the
participation rate” (Farrington, 1986, p 189). In contrast, incidence of offending is
defined as “the rate at which offenders commit crimes or the individual crime
rate" (Farrington, 1986, p 214). Thus, incidence is the ratio of the number of
incidents in a period of time to the number of people being studied.

Studies of the prevalence of bullying found that in some schools everyday
a significant number of students experience bullying. For instance, the US.
Department of Justice reported that one out of every four students experiences
bullying in schools (for review of further analysis, see Liepe-Levinson & Levinson,
2005). Using data collected from 4,263 students in a Maryland school district,
Haynie and her colleagues (2001) found that 7.4% of students reported acting as
bullies, 30.9% of students reported bullying victimization, and more than one half

of the bullies reported that they also have been bullied. Very similar prevalence

8

rates have been found in other US. studies. Based on analysis of nation wide
data from 15,686 students in 6th through 10th grades, Nansel and others (2001)
found that 13% of students were bullies, 10.6% of them were victims, and 6.3%
of students were bullies and victims. More recently, Seals and Young (2003)
found that 24% of 454 public school students were involved with either or both
bullying and victimization (Seals & Young, 2003). Similar rates were found in a
study of a predominantly Latino and black populated, U.S., low socioeconomic,
urban community, where 22% of youths were involved in bullying behaviors, 7%
of the 6th graders were bullies, 9% of them were victims, and 6% of them were
both bullies and victims. (Juvonen, Graham, & Schuster, 2003). These results
indicate that, in terms of prevalence, the US. rates do not seem to vary by
demographic characteristics of the students.

There seems to be variation in prevalence rates of bullying in different
socio-cultural environments. Compared to the United States, in Japan, research
has revealed a slightly higher rate of victimization in primary school. With a
nationwide sample, the Japanese Council for Research on Children and
Students’ Problematic Behaviors found that 22% of primary school students (4th
to 6th graders), 13% ofjunior high school students, and 4% of senior high school
students reported being victims of bullying (for review, see Naito & Gielen, 2005).
Korean students seem to enjoy relatively lower rates of victimization than those
in Japan and the United States. A recent study of 1,344 students showed that
12% of 4th graders bullied others, 5.3% of them were victimized, and 7.2% of
them were both bullies and victims (Yang et al., 2006).

Comparison of prevalence rates of bullying across different studies and

9

different countries is somewhat problematic. There is a considerable amount of
discrepancy between prevalence rates of bullying in different countries, and even
between different studies in a country. This discrepancy stems from different
definitions of bullying behaviors (Kim, 2004; Smith et al., 2002), different
reflection time periods considered by research subjects (Farrington, 1993), and
different systems of school semesters (Naito & Gielen, 2005). It also could be
due to actual differences.

Despite these difficulties, however, studies of prevalence of bullying in
various countries have revealed that bullying and being bullied by classmates in
schools are widespread phenomenon. Every school day, roughly between 20 and
30 percent of students are involved with bullying behaviors in many countries,
and about 10 to 15 percent of them were bullies and roughly 10 percent of them
were victims of bullying. These figures on students’ involvement in the bullying
phenomenon indicate the significance of bullying in contemporary school

environments.

2. Outcomes of Bullying
Aftermaths of bullying phenomenon can be categorized into two types:

outcomes in terms of bullies and victims. Bullying produces devastating
outcomes for victims. Victims suffer from physical and psychological health
problems including depression and anxiety in childhood (Fekkes, Pijpers,
Fredriks, Vogels, & Verloove-Vanhorick, 2006; Rigby, 1998) as well as in
adulthood (Gladstone et al., 2006), suicidal ideation (Liepe-Levinson & Levinson,
2005; Park, Schepp, Jang, & Koo, 2006), especially among girls (Kim et al.,

10

2005), and psychosocial maladjustment to school, such as fighting, drinking,
smoking, academic achievement problems, difficulty in making friends, and
alienation (Nansel et al., 2001).

For bullies, researchers have found that bullying behavior is a strong and
significant predictor of delinquency and violent crime during puberty and early
adulthood. Research conducted in Hong Kong found that unaddressed bullying
behavior leads to violent behavior during the school years and adulthood (Wong,
2004). Similarly, Andershed and his colleagues found that in Sweden bullying
behavior in schools predicted a higher likelihood of engagement in street
violence (Andershed, Kerr, & Stattin, 2001). Bullying behavior is also a strong
predictor of antisocial behavior and drop-out from school (Morrison, 2002; Rigby
& Cox, 1996). Other researchers suggested that bullying behavior in schools is
easily aggravated into more severe criminal behavior in and out of school

environments (Kinlock, Battjes, & Gordon, 2004).

3. Importance of the Current Study

As studies on prevalence rates of bullying phenomenon and seriousness
of outcomes of bullying behaviors and victimization show, it is of importance that
researchers adequately understand bullying in schools and provide appropriate
methods to deter it, or at least reduce it to some extent. Since bullying has
devastating consequences, such as suicide or school shootings, adequate
understanding and appropriate prevention methods will greatly benefit children
and society.

In addition, the existing Korean literature on bullying does not provide a

11

tool to allow prediction of bullying behavior, because there is no attempt made to
apply any theoretical model to analysis of empirical data. Most studies have
simply correlated characteristics of society and culture with individual bullying
behaviors and victimization without providing any theoretical rationale for
expecting an association. For example, one study correlated living in urban/rural
areas with bullying behavior and victimization without a theoretical framework
(Yang, 2005). Another study correlated the number of game rooms and karaoke
stores with bullying behavior and victimization, and argued that capitalized
society and contemporary Korean cultural emphasis on seeking pleasure
produce an increase of bullying in school (Shin, 2000). Unfortunately, in the latter
study, there was no discussion of whether game rooms and karaoke stores are
valid indicators of the levels of capitalization and pleasure-seeking in a culture.

Further, in South Korea, most bullying studies have a disjuncture between
the literature review and actual statistical analysis of data. In other words, the
studies focus on previous studies that suggest a holistic explanation of the
bullying phenomenon, but the data collection and analysis do not adequately
measure and consider the range of potential influences. Most of the studies
reviewed suggested that bullying has been influenced by contemporary societal
change from collectivistic to individualistic culture, problematic philosophies of
education and educational systems, or violent and sexual content of the media.
But none of them accurately measured these aspects of the environment (e.g.,
Lee, 2003; Lee & Kim, 2000; Park, 2003).

The lack of adequate empirical research in South Korea makes it difficult
for researchers as well as educational practitioners to prevent, or at least

12

alleviate, bullying problems in Korean schools. In addition, to the extent that
bullying is a precursor to other types of delinquency, adequate understanding of
bullying phenomenon may also result in prevention of other illegal behavior.
Thus, it is of importance that researchers conduct an adequate study to promote
better understanding of the bullying phenomenon, and this effort will greatly
improve the quality of education in schools as well as quality of life among

children.

13

CHAPTER 2
LITERATURE REVIEW

l. Definition of Bullying

II. Review of Previous Studies of Bullying
Ill. Shortcomings of Research on Bullying
IV. Ecological Model of Bullying in Schools
V. A Summary of the Current Study

VI. Hypotheses

I. DEFINITION OF BULLYING

Defining bullying is a difficult task. A consensus on the definition of
bullying has not yet been achieved among researchers. For example, Coloroso
(2003) defined bullying as “a conscious, deliberate hostile activity intended to
terrorize and harm others through the threat of further aggression” (p13).
Farrington (1993) described bullying as “repeated oppression, psychological or
physical, of a less powerful person by a more powerful one” (p 381).

The most widely used definition of bullying was developed by Olweus.
Olweus (1993) wrote “a person is bullied when he or she is exposed, repeatedly
and over time, to negative actions on the part of one or more other persons” (pp
8 — 10). In this definition, there are three elements, negative actions, repetition,
and imbalance in power. The negative actions include intentional infliction of
injury or discomfort on another, which is carried out through physical aggression
or aggressive words, for example threatening, taunting, teasing, and name
calling. These actions should be repeated to constitute bullying, and there should

14

be imbalance of power between inflictors (more powerful) and victims (less
powerful). More recently, it is recognized that there is a continuum between being
a bully and being a victim. Espelage and Swearer (2003) found that some portion
of bullies also simultaneously suffer from victimization by others.

Even though Olweus’ definition is most often used, bullying studies in
different countries use this definition with some degree of dissatisfaction. For
example, Japanese researchers expressed dissatisfaction with Olweus’ definition,
based on lack of sensitivity to socio-cultural differences in the nature of school
bullying. Naito and Gielen (2005) found that Japanese bullying participants and
onlookers perceived bullying more as psychological intimidation rather than
physical violence, and argued that in the Japanese context, bullying is repeated
negative behavior that dehumanizes others either physically or psychologically.
Using a sample of 1245 8 and 14 year-old students from 14 countries, Smith et al.
(2002) also found that countries differed in what is perceived as bullying.
Especially in Japan, China, Thailand, and Norway, bullying is perceived as less
physically aggressive, but more socially manipulative aggressive behaviors
(Smith et al., 2002). In addition, an acute cultural difference in verbal aggression
has also been found in research. Naito and Gielen (2005) argued that it is verbal
aggression in Western countries when a person calls another names openly, but
in Japan, “name whispering” behind the victim’s back would be verbal aggression.

Typical definitions of bullying used in Korean research emphasize the
importance of social isolation as an outcome of bullying behavior. Thus, the terms
used to refer to bullying communicate collective ostracism, collective social
exclusion, or collective harassment (Ahn, 2002; KDE, 1999; Lee, 2000; Shin,

15

2000). The term, Wangtta, has been used to include all types of bullying
behaviors, such as social ostracism and physical and verbal harassment (Park &
Son, 1998; Shim, 2004). Originally the term Wangtta came from the noun, tta,
which means ostracizing others. The prefix, wang, originally means a king, but in
this case, it means big or worst.

A replication study of the research of Smith et al. (2002) was recently
conducted in Korea. This study found that the most relevant and concurrent
Korean term with the English term bullying was ‘Wangtta,’ and that the most
important aspects of Wangtta (in the order of statistics significance) were social
exclusion, verbal bullying, and physical bullying (Shim, 2004).

The current study conceptualizes bullying as having subcategories, such
as Wangtta (see Table 2), and defines bullying as repeated behaviors by one or
more individuals that are intended to impose psychological and physical harms to
and social isolation of one or more victims through physical, verbal and relational
aggression for an extended period of time. This definition goes beyond the
Olweus’ definition and the definition of the Japanese term ljime in two ways, by
considering not only severity, but also visibility and duration of bullying behaviors.
To discuss the expansion of the definition in detail, first, the dimensionality of
bullying is discussed in the next section, and then the severity, visibility and

duration aspects are discussed in a later section.

1. Physical, Verbal, and Relational Bullying
According to Dishion, French and Patterson (1995), there are three
aspects of aggression, dimensionality, continuity, and progressions.

16

Dimensionality refers to variation in visibility (visible, or invisible (or relational)),
physicality (physical vs. non-physical), and time (early vs. late onset). Continuity
recognizes persistence of aggression in early childhood through puberty into
early adulthood, and progressions refers to escalation of noncompliant behavior
to temper tantrums to violence (Dishion, French, & Patterson, 1995). Some
bullying researchers consider only the dimensionality of bullying and categorize
bullying into physical, verbal, and relational bullying (for example, Olweus, 2003).
Other researchers categorize bullying as either direct or indirect (for example,
Bjorkqvist, Lagerspetz, & Kaukiainen, 1992). Direct bullying includes direct
physical and direct verbal violence, and indirect bullying includes relational
bullying (see also, Carney, 2005; Coloroso, 2003). As Table 1 shows, some
definitions are overlapped and used interchangeably. For example, physical
bullying typically refers to visible and direct bullying behaviors that physically
harm others, and verbal bullying typically refers to visible and direct verbal
violence. In contrast, relational bullying is typically characterized as invisible and

indirect harm to others through damaging their interpersonal relationships.

Table 1. Typical Categorization of Bullying

 

 

 

 

 

Physicality
Physical Non-physical (or verbal)
Invisible Indirect/relational
Visibility
Visible Direct physical Direct verbal

 

 

 

 

So far in bullying studies, however, an operational definition of bullying

17

 

has been limited to measures of the dimensionality aspect of bullying behavior
(typically physical, verbal and relational aggression). In this operational definition,
continuity and progression dimensions have been ignored by researchers. This
limitation was partly due to the difficulty of getting longitudinal data to assess the
continuity and progression aspects of aggression. In addition, lack of consensus
on the definition of bullying also contributed to this limitation. Bullying is typically
defined with three factors, negative action, imbalance of power, and repetition
(Olweus, 1991, 1993, 1995). But there is no clear way to establish repetition of
bullying over time. These problems contribute to difficulties with interpreting
findings about differences in the prevalence of bullying in different schools and
countries (Farrington, 1993; Naito & Gielen, 2005; Smith et al., 2002). More
relevant to the present study, they also raise questions about the appropriate

measure of bullying in Korea.

2. Chronic vs. Non-chronic and Severe/overt vs. Moderate/covert Bullying
The present study considers the persistence of bullying behavior rather

than the continuity of bullying, which is usually conceptualized as the onset of
bullying behavior (Dishion et al., 1995). Since bullying is defined as “repeated”
aggression, duration of aggressive behavior should be recognized by the
operational definition. This time dimension of bullying behavior allows
differentiation of chronic bullying from non-chronic situational bullying behavior.
Research found that although bullying may affect a large number of children, only
a small portion of students were chronically and persistently victimizing others

and victimized by others (Boulton & Smith, 1994; for review, see Graham &

18

Juvonen, 1998; Salmivalli et al., 1998; Smith & Brain, 2000).

The Korean Department of Education published a study on bullying and
its prevention policy in 1999. In this report, collective ostracism refers to “inflicting
harms to certain students for more than two weeks by one or more students
through psychological and verbal aggression, money extortion, or imposing
physical harm.” Other Korean literature suggests different types of bullying
victims and bullying behaviors (Ahn, 2002; Coloroso, 2003, p 21; KDE, 1999;
KEDI, 1998; Kim, 2001). For example, Ahn (2002), based on a literature review,
suggested two types of bullying victims; SUI or euntta and flu} or jeontta, and
the term wangtta was used to refer to both types of bullying victimizations. Others
foolish), bantta (those who are bullied by classmates in a class), and jeotta (those
who are bullied by all students of a school) (KEDI, 1998; Kim, 2004; Shim, 2004).
Still others use two categories; tta (those who are bullied at minor/moderate level
of aggression and short term exclusion) or wangtta (those who are bullied at
severe level of aggression for a relatively long period of time) (KDE, 1999).

These different types of victims and bullying behaviors could be
distinguished by different levels of severity, visibility and duration of bullying.
Bullying behaviors are not only indicated by harshness and openness of the
behavior, but also by the extent and duration of social isolation and rejection.
Thus, chronic bullying and non-chronic bullying would differ in visibility,
physicality/severity, and duration. In other words, bullies use different forms of
aggression and social isolation for different types of victims for different lengths of

time with different levels of openness. Table 2 shows different types of victims

l9

and the related bullying behaviors based on the expanded definition of bullying

with temporal, severity and visibility elements added together.

Table 2. Expanded Definition of Bullying: Temporal and Visibility Elements Added

to the Traditional Physicality/Severity Dimension of Bullying Behavior

 

 

 

 

Severe/Overt Moderate/Minor/Covert
Chronic wangtta/jeontta bantta/euntta
Non-Chronic bantta/euntta tta

 

 

 

It is important to distinguish different types of bullying behaviors based on
temporal, severity and visibility elements, because this categorization will
distinguish use of different levels of physical, verbal, and relational aggression
with different levels of openness for different types of victims (e.g., Kwon, 1999).
For instance, the wangtta/jeontta group will be the most severely, openly and
chronically victimized students by other students through all types of physical,
verbal, and relational aggression. Sometimes wangttajeontta status ends when
the wangttafleontta students graduate from schools and move up to higher
grades. In most cases, however, the wangttajeontta status remains through the
entire primary, middle, and high school years (KEDI, 1998). This is why parents
and other school professionals use transferring those students to another school
as a solution for bullying victimization. Compared to this, the tta group may suffer
from more open covert minor or moderate relational aggression or covert minor
verbal aggression in a class for a short period of time, such as one semester or
until the emergence of a new tta victim in the class (KEDI, 1998). Thus, the

20

 

identification of level of severity, visibility and temporal duration across different
types of bullying may provide different approaches in school policy and programs
to address bullying. In addition, this identification of different forms of aggressive
behaviors may help to understand what types of students using what types of

bullying behaviors become more anti-social and deviant adolescents and adults.

21

ll. REVIEW OF PREVIOUS STUDIES OF BULLYING

1. Thematic Overview of Previous Bullying Studies

Previous bullying studies have several focal concerns. In addition to the
prevalence and outcomes of bullying discussed in the previous chapter, there are
four more major topics: types of bullying behaviors, roles-played in bullying
situations and their characteristics, correlates of bullying, and the social network
approach to understanding bullying. In this section, studies within each of the
four additional topic areas are reviewed in detail.

Specifically, the literature on types of bullying behavior and roles played in
bullying situations will be examined so that conclusions can be drawn about how
to define and measure bullying behavior in this dissertation. The correlates of
bullying research will be reviewed to identify factors influencing bullying
behaviors, e.g., individual traits, perceptions of and experiences with parents,
teachers, and school environments. Those identified factors will be used in the
ecological prediction model developed by this study. The literature on the social
network approach to understanding bullying will be used to develop measures of
peer influences. In particular, this part of the review is the basis for advancing a
measure of the pseudo friendship network to indicate the quality of friendship, as
a replacement for previous measures (e.g., mutuality, popularity, centrality, or

similarity) used in social network studies of bullying.

A. Types of Bullying Behaviors

The most common categorization of bullying behaviors was borrowed

22

from a typology of aggressive behaviors. When Olweus first used the term,
bullying, to refer to a one-on-one physical attack of a stronger child against a
weaker child, he only considered physical aggression (Olweus, 1978). Later he
expanded the definition of bullying to capture verbal aggression, group action
and repeated exposure to negative actions against individuals, and then he
categorized the bullying behaviors into physical and verbal aggression (Olweus,
1993). Recent bullying studies have used a three dimensional approach to
bullying behaviors, physical, verbal, and indirect aggression (Bjorkqvist et al.,
1992). Indirect aggression refers to aggressive behaviors intended to damage
victims’ social relationships through social isolation and exclusion. The examples
of this type of aggression include gossiping and spreading rumors, and social
exclusion using third parties (Bjorkqvist et al., 1992). Some researchers use the
term social aggression (Carney, 2005), and others use the term relational

aggression (Coloroso, 2003) to refer to indirect aggression.

B. Roles-played in Bullying Situations and their Characteristics

The typical and traditional categorization of roles played in bullying
situations is based on the distinction between youths who act as bullies, who are
victims, and who bully others but who also are victimized by bullies (Haynie et al.,
2001; Juvonen et al., 2003). Haynie et al. (2001) found that members of the three
groups share some common characteristics, but also are distinct in other ways,
in terms of involvement in problem behaviors, self-control, deviance acceptance,
deviant peer influences, and depressive symptoms. For example, Juvonen et al.
(2003) found that bullies are psychologically the strongest students who have

23

high levels of self—esteem and enjoy high social status among classmates (see
also Mouttapa et al., 2004), and that victims were distressed and socially
marginalized. Olweus (1993; 1995) argued that bullies typically have
aggressiveness and physical strength, and that victims usually suffer from
loneliness, anxiety and physical weakness. According to Kim and his colleagues,
the individual who is both a bully and a victim suffers the most serious suicidal
ideation compared to youths who are either a bully or a victim (Kim et al., 2005).

Research has also recognized the importance of the psychosocial
function of non-participants or indirect participants in bullying. For example,
Coloroso (2003) suggested that bystanders who take no action against bullying
or who support it have a tremendous effect on bullying because they tacitly
allow deviant behaviors of bullies in schools (Coloroso, 2003; see also Harris &
Petrie, 2003).

In some studies, the bystanders were subdivided into different types of
students according to more specific roles they play in bullying situations. For
example, Salmivalli and her colleagues (1997) argued that reinforcement of
bullying behavior through a peer group network is one of the most important
factors supporting bullying, and that the reinforcement could come from groups
who were partly involved with actual bullying behaviors (assistant groups) or from
groups who were not considered to be active participants in bullying behaviors,
but who were emotionally supporting the behaviors (reinforcer groups). Those
who do nothing when they witness bullying are referred to as the outsider group.
Defender groups are those who become actively involve with protection of

victims from bullying behaviors. Six participant roles, thus, are bully, assistant,

24

reinforcer, outsider, defender, and victim (Andreou & Metallidou, 2004; Salmivalli,
Lagerspetz, Bjorkqvist, Osterman, & Kaukiainen, 1996).

Olweus’s (2003) categorization of participant’s roles is also consistent
with this previous categorization, ‘bully,’ ‘followers, supporters but passive bully,’
‘passive supporters but possible bully,’ ‘disengaged onlookers,’ ‘possible
defenders,’ ‘defenders of the victim,’ and ‘victim’ (Olweus, 2003). According to
Olweus (2003), ‘followers and supporters but passive bullies’ are those who will
not act as bullies but who will assist bullies (similar to the assistant group), and
‘passive supporters but possible bullies' are those who might become bullies
(similar to the reinforcer group). ‘Disengaged onlookers’ are similar to the
outsider group and ‘possible defenders’ and ‘defenders’ are similar to the
defender group in the previous categorization. Youth in the defender group enjoy
the highest sociometric status among classmates (Salmivalli et al., 1996).

In Japan, instead of assistant, reinforcer and outsider groups, slightly
different types of participant roles were identified. According to Morita and
Kiyonaga (1994), some students were just amused by the bullying situation
(audience group), and some acted as if they do not know the bullying victims in
class (bystander group) (Morita & Kiyonaga, 1994). Students who enjoy watching
bullying or who ignore bullying behavior are similar to the reinforcer and outsider
groups. However, the significance of identifying bullying as “fun seeking”
behavior among members of the audience group should be noted, and this will
be discussed more fully in the later section.

In Korea, types of victims have been one of the main concerns for studies
on bullying (for review, see Ah, Jeong, & Cha, 2005). Based on analyses of

25

languages and terms used among youths in Korea, studies suggested that there
were two major types of victim; euntta and jeontta (Ahn, 2002; Kim, 2004). The
term tta refers to overall behaviors that ostracize, bully, and harass others
physically and emotionally, and also refers to the victim of those behaviors. As a
combination of a prefix and ‘tta,’ euntta refers to victims who are covertly
ostracized (the term eun refers to ‘covert’), and jeontta refers to those who are
overtly victimized not only by others in a class, but by all students in a school (the
term jeon refers to ‘entire’).

Aforementioned findings indicate the existence of different types of
bullying behaviors based on levels of severity (types of bullying behavior, such as
physical, verbal or relational bullying), involvement (different roles played in
bullying), visibility (covert vs. overt bullying), and duration (chronic vs. non-
chronic bullying victims). These findings were used to replace the traditional
definition of bullying with a new expanded definition of bullying, which is used in
this dissertation. The suggested new definition of bullying reflects three aspects
of bullying behavior; severity, visibility, and duration (see the section, Definition of
Bullying, p 14 — 21). Research could also be done on which youth take which
roles, but this is not the main focus of this dissertation, which instead develops

and tests theory to explain the frequency and nature of youths’ bullying behavior.

C. Correlates of Bullying Behaviors
Studies of correlates of bullying focus on psychological, psychosocial,
psychopathological and sociological characteristics of individuals, families and

schools. Each of these types of correlates will be considered in the following

26

secfions.

1) Psychological, Psychosocial and Psychopathological
Correlates

Studies focusing on individual psychological and psychopathological
differences stress the importance of intra- and inter-individual traits. Those
individual traits found to be significantly associated with bullying behavior include
self-esteem (Batsche & Knoff, 1994; Smith & Brain, 2000), anger and aggression
(Bosworh et al., 1999), ADHD and low self-control (Unnever & Cornell, 2003),
depression and deviance acceptance (Haynie et al., 2001), anxiety and empathy
(Espelage & Swearer, 2003), illness, disability and temperament (Gladstone et al.,
2006), and dominance and impulsivity (Lee, 2000). Shyness and withdrawal
behavior are related to increased victimization by bullying (Dill et al., 2004).

Among these, self-esteem has been found to have inconsistent effects on
bullies and victims. Specifically, studies found contradicting evidence for
relationships between self-esteem and bullying behavior and between self-
esteem and victimization. Batsche and Knoff (1994) found that bullies have
higher or average levels of self-esteem compared to lower levels of self-esteem
among victims or bystanders, but others found that bullies, especially females,
have low levels of self-esteem (Rigby & Cox, 1996) or average levels of self-

esteem (Seals & Young, 2003). 2

 

2 Unlike the contradicting evidence for the relationship between self-esteem and
bullying behavior, a consistent finding across different studies indicates that
victims of bullying behaviors have low levels of self-esteem (Rigby & Slee, 1993;
Smith & Brain, 2000).

27

Psychosocial factors associated with bullying behaviors were also
identified by numerous studies. Those factors are self-efficacy, academic and
social cognition (Andreou & Metallidou, 2004), normative beliefs about bullying
and social skill deficits (Bosworh et al., 1999; Espelage & Swearer, 2003), low
levels of confidence in using nonviolent strategies (Bosworh et al., 1999), and
positive attitudes towards bullying (Bentley & Li, 1995; Boulton, Trueman, &
Flemington, 2002; Shin, 2000). For example, a recent study found that
psychosocial factors, such as attitude in school; self-control of aggressiveness
and impulsiveness; self-assertive efficacy against bullying; and euphemistic

thinking were associated with all types of bullying (Ando et al., 2005).

2) Sociological and Social Psychological Correlates

Factors that have been given considerable attention by researchers are
social psychological variables, such as experiences with family, peers, and
community. The effects of parents and their parenting skills on bullying behaviors
have been tested based on the idea that children internalize their family
environments and that this internalization affects development of a self-concept
and self-control. Studies, for instance, suggested that negative parental
experiences develop negative self-concepts among children, and in turn they
influence bullying behavior (Christie-Mizell, 2003; Curtner—Smith et al., 2006;
Flouri & Buchanan, 2003). Other researchers argued that parental maltreatment
causes emotional dysregulation, i.e., low self-control, and in turn causes more
physical bullying (Shields & Cicchetti, 2001). The types of parenting thought to
affect bullying behavior are authoritarian parenting using physical means of

28

discipline (Gladstone et al., 2006; Lee & Kim, 2000), hostile, rejecting and
inconsistent parenting, little supervision, parenting that leads to poor problem-
solving skills, and encouragement of striking-back against aggressors (Batsche &
Knoff, 1994).

It also should be noted that perceived parenting by children is an
important predictor of children’s behavioral pattern. Using the quota sampling
method to collect data from 20 students who were bullies, 20 victims, and 20 who
were both bullies and victims, Browers and others (1994) found distinct
differences between those groups in terms of family system, family relations, and
parenting styles. Their findings suggested that children’s perspective on family
relations, rather than parents’ perception of family relations, is an important
influence on their behaviors (Bowers et al., 1994).

Since bullying is a phenomenon that occurs within peer relationships,
relationships among classmates have been extensively investigated. For
example, Buhs et al. (2006) found that the best predictor of bullying victimization
is peer exclusion and rejection within peer groups. Also the influence of deviant
peers has been found to be a significant predictor of bullying behavior (Haynie et
al., 2001). Recent research tends to empirically measure the peer relationship
using social network analysis (SNA), and this area will be discussed in detail in
the next section, “The Social Network Approach to Explaining Bullying.”

School and community environments are likely important influences on
individuals who engage in the various behaviors that carry out and support
bullying behaviors. As noted previously, these factors, however, are by far the
most underinvestigated. Japanese scholars emphasized the significance of

29

norms and value systems among students, teachers’ moral authority, and the
existence of collective punishment in classes as influences on bullying behaviors
(Masataka, 1998; Naito & Gielen, 2005; Tai, 2001). In addition, a few scholars
have argued that community characteristics have distal but significant effects on
bullying behavior in schools. Farrington (1993), for example, wrote that bullying is
more common in big inner cities where social disorganization is prominent (see
also Laub & Lauritsen, 1998). These explanations linking school and community

environment to bullying, however, have not been empirically tested.

D. The Social Network Approach to Explaining Bullying

Since an early application of the social network analysis (SNA) approach
to bullying (i.e., Cairns, Cairns, Neckerman, Gest, & Gariepy, 1988), SNA
became a popular method to understand peer dynamics of bullying behavior
among classmates. The core argument is that peer rejection, acceptance, and
affiliation through friendship networks determine the occurrence of and
involvement in bullying behaviors and the prevention of victimization. Gest et al.
(2001) argued that changes of friendship, measured by sociometric status,
centrality, and mutuality, cause changes in level of acceptance and rejection
within the peer group, which in turn cause aggressive behaviors against those
who precipitate the change of friendship. Currently, there are four focuses in the
SNA approach: sociometric status, mutuality and popularity, centrality, similarity

and peer influence.

1) Sociometric Status

Being accepted or rejected by friends is an important element of school
life among students. Typically measured through the peer nomination method,
the sociometric status of an individual among classmates provides a unique idea
about acceptance and rejection in friendship networks. Studies found that
rejected status among classmates provokes aggression (Gest et al., 2001), and
more bullying and victimization (Boulton & Smith, 1994). Highly accepted youth,
that is those with high sociometric status, are typically bullies and bully assistants
(Lee, 2000; Mouttapa et al., 2004). According to one study, classmates perceive
bullies to be high on sociability, leadership, aggressive-disruptive behavior, and
low on sensitive-isolated traits (Collins & Bell, 1996). However, those who enjoy
the highest sociometric status are found to be defenders of victims. According to
Salmivalli et al. (1996, p 12), defenders have the highest sociometric status,
“possibly because 1) they react to bullying, so defending the victim is appreciated
by peers; or 2) a high-status child does not have to be afraid of being victimized,
even if he takes sides with the victim.” Their results, however, suggested that the
lowest status was occupied by both bullies and victims. This contradicting
evidence indicates the necessity of using multidimensional measures of
friendship networks, as is proposed below, and moving away from the simple

measure of sociometric status.

2) Mutuality and Popularity
Reciprocity is identified as an important characteristic of friendship

networks. Reciprocated peer nomination indicates that perceived friendship is not

31

uni-dimensional, but there is mutual interdependency. This reciprocated
relationship is demonstrated in tests of what is called the friendship protection
hypothesis (Boulton et al., 1999; Hodges et al., 1999; Hodges et al., 1997). For
instance, using longitudinal data collected from 1170 youth at time 1 and 1158
youth at time 2 in 5 junior schools in the United Kingdom, Boulton et al. (1999)
found that reciprocated best friendship could provide protection from bullying
victimization (see also Hodges et al., 1997). Furthermore, Hodges et al. (1999)
found that supportive friendship helps youths avoid internalizing behaviors, such
as depression and low self-esteem, which are causes of bullying victimization.

The size of reciprocated friendship networks (popularity) is also
recognized as an important characteristic of peer relationships. Gest et al. (2001)
found that bullies have a great number of friends (see also Mouttapa et al., 2004),
and bullies, assistants, and reinforcers belonged to larger networks than
defenders, outsiders, and victims (Salmivalli et al., 1997).

In another study, popularity and mutuality were compared for bullies and
victims. Huttunen et al. (1996) measured the number of friendships and
reciprocated friendships among 459 students in 6th grade from 18 classes in
Finland, and found that assistants, reinforcers, and bullies had larger friendship
networks than outsiders, defenders, and victims. But their results suggested that
mutuality is irrelevant to the participant roles in bullying situations and that

mutuality is higher among girls than boys.

3) Centrality

ln SNA, centrality stems from “the sociometric concept of the ‘star’ - that

32

person who is the most ‘popular’ in his or her group or who stands at the centre
of attention” (Scott, 2000, p 82). The idea of centrality is relevant to leadership
among peers in bullying studies. Gest et al. (2001), for example, found central
figures in both pro-social and anti-social peer groups. Similarly, Mouttapa et al.
(2004) identified ‘model boys’ who are leaders of pro-social peer groups and
‘tough boys’ who are the center of anti-social peer groups. Based on social
cognitive theory (i.e., learning theory in criminology) and social dominance theory,
they found that bullies have larger friendship networks, higher sociometric status,
and central network positions (Mouttapa et al., 2004). In another study, a high
level of centrality was associated with the high levels of popularity among girls,
but it was associated with aggressive behavior among boys (Xie, Cairns, &
Cairns, 1999). These studies suggest the existence of central figures among

peer groups, and that bullies typically are the central figures among their friends.

4) Similarity and Peer influence

It is common to find that a member of a peer group behaves in a very
similar manner as do other members of the group. This similarity of behaviors
among participants of a network group causes a controversy in SNA. The
selection and attraction hypotheses suggest that those who are more similar will
be more likely to be friends, and similarity factors include gender, race, poverty,
aggression, withdrawn behavior, achievement, and sociometric status
(Kupersmidt, DeRosier, & Petterson, 1995; Xie et al., 1999). Based on 506
students in 4th through 7th grades from 4 schools in inner-city areas, Xie et al.
(1999) found that members of the same peer group were similar on multiple

33

behavioral dimensions, such as aggression, popularity, academic competence,
affiliation, and emotional and behavioral disorder. Furthermore, Haselager et al.
(1998) found that similarity among members is even greater in socially
nonaccepted, anti-social groups. In addition, it is found that this similarity
hypothesis is applicable only to proactive aggression, but not for reactive
aggression (Poulin et al., 1997).

In contrast to the selection and attraction hypothesis, the influence model
is supported in tests of the homophily hypothesis, which is that a member’s
behavior is influenced by other members of the peer group (Espelage, Holt, &
Henkel, 2003; Salmivalli et al., 1997). Espelage et al. (2003) suggested that peer
group membership influences bullying behavior, and found supportive evidence
from analysis of longitudinal data from 422 students. Espelage et al. (2000) also
found that involvement with delinquent peers mediates the associations of social
bonds to family and delinquency (see also Elliott, Huizinga, &Ageton, 1985).
Furthermore, peer influence was found to be greater among female students.
Salmivalli et al. (1998) found that friends’ behaviors were a better predictor of
how the female students behave in bullying situations than their own former
behavior in earlier grades.

These competing hypotheses (selection/attraction and influence) were
cross-tested in a research project with a cross-lagged panel model using the
National Youth Survey data (Matsueda & Anderson, 1998). The researchers
compared learning and interactional theories of delinquency with control theory,
and found that "the effect of delinquency on peer associations is larger than that

of peer associations on delinquency” (Matsueda & Anderson, 1998, p 269).

34

However, it is important to note that their analysis also found that there is
reciprocal relationship between delinquent peer association and delinquent
behavior.

Above mentioned findings from the SNA studies in bullying suggest the
necessity of developing a different approach to measuring peer relationship.
Particularly, the contradicting findings indicate that sociometric status is not a
reliable measure of peer relationship. In addition, the measures used in the SNA
studies are typically based on size of friendship networks or reciprocated
friendship using peer nomination. Those measures for the mutuality, popularity,
and centrality are simply counts of the frequency of particular persons’ names
nominated, thus those measures do not adequately address the quality of
friendship. Finally, since the measures for testing the similarity and influence of
peers typically require longitudinal data, this dissertation instead develops a
measure of the quality of friendship networks (i.e., pseudo friendship network).
This measure, as well as three other measures for peer influence, are discussed

in detail in the section, Experiences with Peers.

35

III. SHORTCOMINGS OF RESEARCH ON BULLYING

The outcomes of aforementioned research efforts invested in
understanding bullying phenomenon are somewhat disappointing for several
reasons. First, bullying in school has not been a popular topic among
criminologists, especially in the United States (Batsche & Knoff, 1994; Espelage
& Swearer, 2003; Farrington, 1993; Spivak & Prothrow-Stith, 2001). According to
this researcher’s effort to search the literature in the field of criminal justice and
criminology, a very limited number of studies have been related to bullying, and
most of these studies focus on school violence, rather than bullying as a major
dependent variable. Since bullying is generally considered as a subtype of anti-
social behavior (Farrington, 1993), understanding bullying and the escalation of
bullying into anti-social behavior will help criminologists to better understand the
etiology of illegal behaviors among youths as well as adults in later life. Thus,
more in-depth and comprehensive study on bullying behaviors is necessary for a
better understanding of overall anti-social behaviors among school children.

Second, bullying has been understood as independent incidents within a
context of different aspects of school life. Even though researchers acknowledge
complex social fabrics influencing the incidence of bullying in schools, they
typically focus on the individual or incident level in studies of bullying (Flores,
2005; Morrison, 2001; Rodkin & Hodges, 2003; Sexton-Radek, 2005; Swearer &
Doll, 2001). Bullying, however, is notjust an individual act against victims, but it is
a manifestation of group acts against victims (Morrison, 2001; Olweus, 1993;

Smith & Brain, 2000). Morrison (2001) argued that bullying and victimization in

36

schools is a systemic problem and is a cumulative and multifaceted phenomenon.
According to her, understanding and pathologizing the problem should not be
based on blaming one factor, one person, or one place (Morrison, 2001).
However, most prior research has been limited to a focus on a single area of the
researchers’ interests, such as individual psychological traits, peer networks,
family influences, or school environments.

Third, and more strikingly, no effort has previously been made to
construct a theoretical framework to accommodate causal linkages between
bullying and various facilitating factors for bullying (Espelage & Swearer, 2003;
Morrison, 2002; Sullivan et al., 2004; Swearer & Doll, 2001). There is a body of
literature investigating factors associated with bullying behavior (see
aforementioned correlates in the previous section), but the absence of a
theoretical framework made it difficult to develop a multi-level explanation of
bullying in schools (Sullivan et al., 2004).

Fourth, several aspects of bullying are left underinvestigated or
uninvestigated by researchers. For example, a number of researchers suggested
the importance of possible effects of school climate and teacher’s attitude
towards bullying on bullying behavior, but there has not been any empirical
research to exclusively and empirically investigate such influences (Espelage &
Swearer, 2003; Farrington, 1993). Japanese scholars emphasized the
importance of studying norms and value systems among students, teachers’
moral authority, referring to how well teachers’ morality is respected by students,
and collective punishment (Masataka, 1998; Naito & Gielen, 2005; Tai, 2001). But

those factors are neither comprehensively nor empirically investigated so far.

37

Finally, a recent trend in studies of bullying focuses on the application of
SNA to peer relationship and social status of students involved with bullying. SNA
researchers have attempted to measure peer influence through personal
networks. However, those attempts failed to recognize differences in the quality
of friendship networks for different peer groups. It is problematic that these three
dimensions (sociometric status, centrality, and mutuality) of friendship networks
have a limited association with each other. For example, some bullies belong to
popular cliques (i.e., popularity or mutuality) even though they are generally
disliked (i.e., sociometric status) by other classmates (Salmivalli et al., 1996).
This indicates that a concept of quality of friendship networks may be useful for
developing a clearer understanding of friendship dynamics as they affect bullying
in schools.

Recently, the negative quality of social capital in networks has received
considerable attention from social network analysts (see Portes, 2000). However,
the SNA measures used in bullying studies to investigate friendship quality (i.e.,
mutuality, popularity, centrality, or similarity) could not reflect the negative quality
of social capital in friendship networks in bullying situation. Thus, it is necessary
to investigate the negative quality of friendship networks, which may increase
bullying or decrease protection from bullying among friends. The negative quality
of friendship networks refers to characteristics like helplessness in friendship
networks or using friendship networks to avoid retaliation against bullying. The
concept used in this study to signify the negative quality of a friendship network is
pseudo friendship network, and this is further discussed in detail in the section,

“Experiences with Peers.”

38

IV. ECOLOGICAL MODEL OF BULLYING IN SCHOOLS

This study borrows Bronfenbrenner’s (1979) ecological system model as
a theoretical framework to understand multilevel etiological factors affecting
bullying behavior within the social contexts of school environments. With the
ecological system approach to bullying, school environment is viewed as “a set of
nested structures” (Bronfenbrenner, 1979, p 3), which include home, school, and
neighborhood settings (Flores, 2005). Each layer of the set of structures has an
independent effect on a child’s development of bullying behavior, but also it is
interconnected and interacts with each other layer to generate a contextual effect
over the development of bullying behavior. In the ecological model, the social
contexts are not merely limited to immediate social settings, such as family or
school, but they are extended to larger surroundings, such as community
(Bronfenbrenner, 1979). Figure 1 visualizes the ecological system model applied

to the school setting.

39

Figure 1. Bronfenbrenner’s Ecological System Model Applied to the School

Setting

 
   
 
      
 
 
  

Macrosystem

Culture Exosystem

   

Microsystem
Family

 
    
       
  
 

 

  

School

 

 

(Parents)

 

Meso- M880-

system ® system

Classroom / W Classroom

(Teachers) (Peers)

 

 

   
   
 
   
 

 

 

 

 

 

Mass Media

   

  

Political
System

The importance of the contextual effects of the school environment as
well as the effects of individual differences on aggressive and bullying behaviors
has been previously recognized by many scholars. For example, Fagan and
Wilkinson (1998) criticized studies that focus only on individual traits as causes of
violence for overlooking other factors, such as school environments and

community characteristics (Fagan & Wilkinson, 1998; Laub & Lauritsen, 1998).

40

Cairns and Cairns (1991) argued that bullying phenomenon should be
understood as a continuum of children’s developmental process, and that the
developmental approach to bullying requires an ecological perspective, social
network analysis, and a multilevel explanatory model, which measures different
levels of causal factors. Since a developmental model of social behavior
presumes that social behavior patterns are determined by multiple factors
including ontogenetic and social contexts, it is a necessary step to use an
ecological approach to explaining bullying for further development in this area of
study (see also, Sexton-Radek, 2005).

Despite recognition of the problems with existing theory to explain
bullying, there has been no empirical research guided by an ecological model.
Theory on bullying in schools has neglected the importance of the interactions of
people across different systems, for example parents and teachers, or parents
and a youth’s peers. As a result, there is much empirical evidence of the
correlates of bullying, but there is no theory to make sense of the different sorts
of predictors.

Because it views bullying as a developmental outcome for children,
Bronfenbrenner’s ecological system model could provide a framework for
explaining how the different systems influence children to behave as bullies
(Flores, 2005). Dishion, French, and Patterson (1995) also argued that the
ecological model does not ignore the importance of the context in which children
interact with each other and in which social relationships are embedded. Based
on vigorous acknowledgement of the importance of both individual differences

and social contexts, Dishion and his colleagues argued that using one theory to

41

explain antisocial behavior is reductionistic, and suggested that an integrative
model that considers individual, school, and other influences is essential to
understand individuals in context.

Since bullying is an “ecological phenomenon that emerges from social,
physical, institutional and community contexts as well as the individual
characteristics of youth of both bully and victim” (Swearer & Doll, 2001, p 7), it is
of importance that each component of the ecological structures of school
environment as well as social interactions between the components are
thoroughly investigated. Bronfenbrenner’s (1979) ecological system model
includes 4 layers of systems relevant to the school environment, microsystem,
mesosystem, exosystem, and macrosystem. In this section, the components of

each system and their relevance to bullying are discussed in detail.

1. Child as a Center of School Ecology
According to Bronfenbrenner (1979), children are perceived as a
developmental entity that plays an important role in interacting with surroundings.
The interactions and interconnections are characterized by types of ecological
settings as well as individual traits (Bronfenbrenner, 1979, 1989). Some
individual traits relevant to explaining bullying include age, prior experience of
bullying victimization, dominance, impulsivity, attitude towards aggression, and

fun-seeking tendency.

A. Age
Studies have documented the age-graded nature of the bullying

42

phenomenon. Most of the studies found that prevalence of bullying peaks during
middle school between 6th grade and 8th grade (Espelage et al., 2003; Harris &
Petrie, 2003; Lee, 1999; Nansel et al., 2001; Pellegrini & Long, 2002; Seals &
Young, 2003). Within the range, researchers found that prevalence of bullying
escalates from 6th to 8th grade (Espelage et al., 2003; Seals & Young, 2003), and
then it gradually decreases during high school (Pellegrini & Long, 2002). Despite
the general pattern, a small portion of bullies and victims continues to experience
aggression at the end of middle school and in high school. Thus, there is some
stability and continuity of bullying behaviors and victimization even when youth
move into the upper grades with new classmates (Boulton & Smith, 1994; Kim,

2001; Salmivalli et al., 1998).

B. Prior Experience of Bullying Victimization

Previous studies have recognized that prior experience of bullying
victimization could result in bullying behaviors to others. Two explanations are
possible for the association between prior bullying victimization and bullying
behavior. First, exposure to violence and experience of bullying could lead to
bullying behaviors to other students through a social learning process (Ah et al.,
2005; Baldry, 2003; Lee, 1999). Second, fear of victimization could precipitate
student engagement with powerful, deviant peer groups and involvement in
bullying behaviors to avoid bullying victimization (Farrington, 1993; Kim, 2001;
Park, 2002). Both explanations suggest that prior experience of bullying

victimization influence bullying behavior to other students in school.

43

C. Dominance and Impulsivity

Other personal psychological factors found to be important influences on
bullying behavior are dominance and impulsivity. Particularly, one study in Korea
found that personal psychological characteristics, i.e., impulsivity and dominance,
are significantly different between bullies and victims (Lee, 2000). Based on
comparison of 54 bullies and 41 victims, Lee (2000) found that bullies scored
significantly higher than victims on impulsivity (average scores on impulsivity
scale: 69.09 vs. 50.95) and dominance (average scores on dominance scale:
67.44 vs. 50.59). In Western countries, researchers also found that impulsivity is
significantly and positively associated with levels of bullying behaviors (e.g.,

Bosworth, Espelage, & Simon, 1999).

D. Attitude towards Aggression

Many studies found that one of the most important psychosocial factors
that significantly predicts high levels of bullying behaviors is attitude towards
aggression (Bentley & Li, 1995; Bosworth et al., 1999; Boulton et al., 2002;
Espelage & Swearer, 2003; Shin, 2000). Bentley and Li (1995) found that
elementary school bullies (grades between 4 and 6) were more likely to have
positive beliefs toward aggression than youth who did not bully. Bosworth et al.
(1999) also found that beliefs supportive of violence significantly increased the

levels of bullying behaviors.

E. Fun Seeking Tendency
Social information processing theory has been used to explain bullying as

44

a fun seeking behavior in Japan. Even though Naito and Gielen (2005) did not
specifically mention the theory, they argued that some students misinterpret
bullying behaviors as normal and acceptable forms of joking and kidding. More
specifically, they argued that school norms are interpreted ambiguously and
situation-specifically. Thus, students tend to disregard some norms in some
situations, and go into a “play mode” when they engage in bullying as pleasure
seeking (Naito & Gielen, 2005, p 182). They said that, such misinterpretation of
normal school norms helps students to more easily engage in bullying behaviors,
and the engagement makes their boring school lives less dull and more fun in the
short run. A similar finding has been reported by Rigby (2004), who wrote that
bullying was a response to peer pressure, and that powerful peer groups were
motivated to bully less powerful others by grievance, or imaged grievance, a
prejudice, and desire to have fun. In Korea, Shin (2000) also found that fun
seeking is an important determinant of bullying behavior especially in urban

areas (see also Seo & Han, 2004).

2. Microsystem
Microsystem is defined as “a pattern of activities, roles, and interpersonal
relations experienced by the developing person in a given setting with particular
physical and material characteristics” (Bronfenbrenner, 1979, p 22). According to
Bronfenbrenner, the most critical notion in the microsystem is experience in face-
to-face interaction through activity, role, and interpersonal relations at home,
school, and community. The experience is constituted not only with “objective

properties” of the environment, but also with “properties [that] are perceived by

45

the persons in that environment” through interaction (p 22). Thus, this
phenomenological view emphasizes the importance of meanings of experience
within interpersonal interactions, which are “most powerful in shaping the course
of psychological growth” (p 22).

The microsystems not only contain the objective characteristics of places
or social institutions, but also include other persons, such as parents, peers, or
teachers. Griffore and Phenice (2001) argued that family, as an example of a
microsystem, is not simply defined by the physical setting, and that “the essence
of the family as a microsystem is the interaction that occurs there” (Griffore &
Phenice, 2001, p 70). Likewise, experiences (i.e., interactions between children
and peers and between children and teachers) in school classrooms have been
recognized as important microsystems in the school setting (Bronfenbrenner,
1979; Griffore & Phenice, 2001). Thus, three interactions a child may have with
the immediate microsystems in family and school settings are experiences with

parents, experiences with teachers, and experiences with peers.

A. Experiences in the Family

In studies of bullying, experiences in the family and perception of the
experiences have been identified as important factors in the prediction of bullying
behavior. For example, for a sample of Italian youths, Baldry (2003) found that
bullying and victimization were predicted by exposure to parental violence and
child abuse, especially for female children (Baldry, 2003). In addition, as
mentioned previously, authoritarian parenting using physical discipline, hostile
and rejecting parenting, lack of supervision, and a permissive attitude toward

46

bullying significantly affect the development of children’s poor self-image and
emotional dysregulation, and then these psychosocial factors influence children’s
bullying behaviors in school (Batsche & Knoff, 1994; Christie-Mizell, 2003;
Curtner-Smith et al., 2006; Espelage et al., 2000; Espelage & Swearer, 2003;
Shields & Cicchetti, 2001). In particular, children’s perception of family and
interactions with parents have been empirically found to be significant predictors
of bullying behaviors (Bowers et al., 1994; Stevens, Bourdeaudhuij, & Oost,
2002)

In addition, parents’ emphasis on the importance of education was
recognized as having important impacts on the interaction between parents and
children in Korea and Japan. Research found that parents who demand
academic success tend to discourage their children from involvement in any
types of school activities other than studying in school (Kim, 2001; Kim & Lee,
2000; Lee & Kim, 2000; Moon, 2002; Naito & Gielen, 2005; Shin, 2000). This
parenting practice could generate stress among students (Agnew, 1992, 2001;
Lee & Kim, 2000; Naito & Gielen, 2005), but also it may promote a positive

school climate through parental involvement in school.

B. Experiences with Teachers

Relatively little research has been conducted to investigate the effect of
teacher’s attitude towards bullying and interaction between students and
teachers in regard to bullying behavior in schools (Espelage & Swearer, 2003;
Rodkin & Hodges, 2003). Recently, a study suggested that teachers perceived

relational bullying as the least serious type of bullying, and responded to that with

47

the least severe actions (Bauman 8. Rio, 2006). In Japan, teachers’ lack of moral
authority has been found to be associated with an increase in bullying
phenomenon in the classroom (Naito, 1990). According to Naito (1990), teacher’s
moral authority is how much students respect teachers for their morality. Naito
(1990) argued that once students think teachers lack moral authority, then
teachers do not have influence over what students should and should not do.

In South Korea, studies found that teachers’ ineffective intervention in
bullying situations as well as teachers’ attitude towards bullying were significantly
associated with levels of bullying in schools (Park, 2003; Yang, 2005). Based on
a literature review, Park (2003) argued that teacher’s dysfunctional supervision
and instruction in relation to bullying behavior could be possible causes of
bullying in schools. In addition, Yang (2005) suggested that students’ perceptions
of teacher’s positive attitude towards bullying might increase bullying in the
classroom. However, there is a lack of empirical study on perceptions of
teacher’s attitude toward bullying and student-teacher interaction around the

issue of bullying.

C. Experiences with Peers

Since students spend most of their day with friends and since they are in
the middle of a developmental process in which collective peer judgment
provides a guideline for their behaviors (Kim, 2004), peer influence can be
enormous. Studies of peer influence, so far, have focused on various topics, such
as relationships with peer groups, social roles and status among classmates, and

perceived sociability, leadership, and sensitivity to others. More specifically,

48

research found that social exclusion from peer groups (Bollmer, Milich, Harris, &
Maras, 2005; Buhs et al., 2006; Dill et al., 2004; Natvig, Albrektsen, &
Qvarnstrom, 2001), friendship with deviant peers (Haselager et al., 1998; Poulin
et al., 1997; Xie et al., 1999), peer dissimilarity (Lease, McFall, & Viken, 2003),
and seeking a higher sociometric status that are typically occupied by bullies
(Collins & Bell, 1996; Mouttapa et al., 2004) increase the likelihood of bullying
others. However, these studies were limited in that each focuses on a single
aspect of peer relationship, or because they ignore the possibility that friendship
networks can have a negative quality.

Based on the results of the previous studies on peer relationship in
bullying situations, this dissertation considers four factors relevant to peer
influence in school environments. They are the power-dynamic (PD), the level of

difference-acceptance, coercion of public self, and pseudo friendship networks.

1) Power-Dynamic (PD)

Some youth view bullying as a way to gain higher social status among
classmates. Especially when there is power inequality between two or more peer
groups, a more powerful group will (both individually and collectively) bully the
less powerful groups in an effort to obtain dominance (Seo & Han, 2004). The
tendency for seeking power has been captured by the concept, power dynamic.
Twemlow and his colleagues (2001) suggested that the power dynamic refers to
“a conscious or unconscious pattern in which an individual or group controls the
thoughts and actions of others” (p 377). Long and Pelegrini (2003; see also

Pellegrini 8 Long, 2002) found that middle school youth used bullying to achieve

49

dominance over other classmates, and once dominance was established,

bullying tended to increase, resulting in even more dominance. Consistent with
this finding, based on an extensive literature review, Rigby (2004) concluded that
among social groups with different levels of power, such as males and females,
certain groups sought domination to enjoy a powerful status. According to Rigby,
since boys were trying to assert hegemonic masculinity, and therefore dominance,

they were more likely to bully other youth (see also, Young & Sweeting, 2004).3

2) Level of Difference-Acceptance

Level of difference-acceptance refers to how much students and their
peer groups tolerate dissimilarity with other classmates, especially victims and
themselves. A low level of difference-acceptance among students seems to be
significantly associated with bullying others.

Consistent with the notion that bullying involves tolerance, Coloroso
(2003, p 20) argued that “bullying is not about anger, [but] it’s about contempt — a
powerful feeling of dislike toward somebody considered to be worthless, inferior,
or undeserving of respect.” Further, she argued that contempt consists of three
psychological advantages, a sense of entitlement [power], an intolerance toward
differences, and a liberty to exclude (Coloroso, 2003, pp 20 - 21). According to
her, intolerance is a major cause of bullying based on racial and gender

differences (see also Young & Sweeting, 2004).

 

3 Due to the fact that boys’ tendency for hegemonic masculinity, gender
arrangements that are part of the exosystem (discussed in the Section,
Exosystem), and which are a constant in the present research, may influence the
power dynamic.

50

There is empirical evidence that students with similar characteristics are
more likely to be friends (Huttunen et al., 1996; Kupersmidt et al., 1995), and this
tendency is especially strong for delinquent peer groups (Haselager et al., 1998).
In contrast, based on analysis of data collected from 478 students from 26
classrooms in 4th through 6th grades, Lease et al. (2003) found that perceived
dissimilarity was positively related to views that a student was odd, which in turn
was related to social exclusion and bullying.

In Japan, there is some evidence that interpersonal comparison is a
major reason for bullying (Naito & Gielen, 2005). In Korea, Ahn (2002) suggested
that one of the most important motives to bullying others is that victims would not,
or could not, follow patterns and trends of behaviors, clothing, or even hair styles
common in peer groups. Even though contemporary youths in Korea seem to
emphasize individuality, they tend not to accept others who do not share
common interests or characteristics (Shin, 2000). This may occur in Korea
because there is an enormous investment of energy, concern, and money into
rearing children (Moon, 2002), which can lead to unusually high levels of self-
esteem among students, and a related low tolerance for peers who are “different”

(Kim & Lee, 2000).

3) Coercion of Public Self
A discrepancy between self-image and perceptions of how peers see you
(or ‘internal image’ and ‘public self’ in Cairns and Cairns (1991, p 256)) seems to
be another important reason for bullying. This means that youth who act

consistently with their own self images, but differently than how other youth

51

expect them to act, are likely to be bullied. It also means that youth would act as
bullies because they feel that their victims are behaving incongruently with
expectations, or their “public selves.”

The notion of bullying to coerce peers to act in accord with expectations
is particularly applicable to a collectivistic society, like South Korea. Korean
studies of bullying found that about 75% of bullies said that they bullied others
because their victims “pretended” to be beautiful, strong, or smart (KEDI, 1998).
The KEDI study concluded that those who could not live according to the norms
of a collective culture tended to be victimized by others due to their pretense of
being beautiful, strong, or smart, and that to avoid victimization, students needed
to behave according to “public images” in the minds of their peers (see also, Kim
& Park, 1997; see also, Lee & Kaok, 2000). Similarly, other studies found that
most of the students who bullied others said that victims deserved bullying
because they did not behave according to peer culture (Lee, 2003), and that
especially in urban areas, they bullied to conect victims’ wrong doing (Shin,
2000). Another study found that students collectively agreed on images of other
students, and they coerced them to behave according to those images (Kim,
2004). This kind of coercion was more common in collectivistic peer groups or
schools (Lee, 2004). In Korea, consequently, collective judgment of oneself by
others functions as a guideline for behaviors in schools, and this coercion of
public self seems to be more reinforced and practiced in a collectivistic peer

group or school.

52

4) Pseudo Friendship Network

Social network theory focuses on the tie between individuals, specifically
on individual communication (a tie) between people (actors) in an organization.
This perspective emphasizes quantification of interaction (i.e., communication),
selection of actors, and influence of actors over other actors by measuring the
transaction of social capital through personal networks. In this sense, it is a more
direct and methodologically explicit approach to studying relationships and their
influence over individual behavior than traditional approaches to the study of
social relationships.

Recently, a concept of negative social capital has received academic
attention. This concept considers the quality of social capital and suggests that
social capital is not only positive in that it provides help and resources to actors
within a network, but also some social capital does have negative consequences,
such as social exclusion of others who do not belong to the network or
helplessness in the social networks (Portes, 1998).

Some types of peer support networks can produce a negative outcome
for other individuals and groups. It was found that without proper adult
supervision, the presence of a peer support system does not bring a decrease in
bullying (Naylor & Cowie, 1999). Some male peer supporters reported the
experience of ridicule and sabotage of their supporter role from other classmates
(Cowie, 1998). Helplessness in friendship networks was found among aggressive
peer groups (Pakaslahti & Keltikangas-Jarvinen, 2001), and help provided to
bullying victims is withdrawn due to fear of retaliation and victimization (Slee,

1994)

53

Typical measures used in social network analysis have not reflected the
quality of friendship networks, since they focused on measuring reciprocated
friendship. For example, the emphasis has been on sociometric status, mutuality,
popularity, or centrality as indicated by either peer nomination or reciprocated
nomination (Huttunen et al., 1996; Mouttapa et al., 2004; Salmivalli et al., 1997;
Salmivalli et al., 1996). Typical questions used in the peer nomination are ‘who is
your best friend?,’ or ‘who is a bully in your classroom?’ Once a student
nominates a friend, and the friend nominates the student, then researchers
assume the friendship between them brings a positive outcome.

The measure ignores the possibility of negative outcomes of a
relationship. For some students, fear of retaliation and victimization and
helplessness in friendship networks prevents development of a genuine
friendship network which contains positive social capital. For instance, a frequent
method that students use to avoid bullying victimization in Korean schools is to
demonstrate friendship with youth who have a high sociometric status and
dominate the class (Park, 2002). This kind of friendship may serve as a
protection mechanism by showing membership in a powerful peer group, and
thereby borrowing power from the peer group to avoid victimization (see also,
Kim, 2001).

However, in addition to this type of protection, a pseudo friendship
network can facilitate bullying behaviors, because youth behave (i.e., bullying
others) according to their friends’ expectations to maintain their friends’
willingness to protect them. Afocus on pseudo friendship networks is unique in

criminological theory, which has instead focused on how deviant peers learn from

54

each other (Akers, 1996; Akers, 2000; Sutherland & Cressey, 1970; Vold,
Bernard, & Snipes, 1998; Wolfgang & Ferracuti, 1981). The concept of a pseudo
friendship network focuses attention on short-term superficial friendships that
provide protection from bullying victimization. With this kind of friendships,
students may not share favorable definitions or norms toward bullying with the
leader of groups that does have these values, but they may still be influenced to
bully other youth.

In sum, since having a number of friendships helps students to avoid
bullying victimization (Boulton et al., 1999; Rodkin & Hodges, 2003), and
because of fear of victimization (Farrington, 1993; Park, 2002), students may
want to make or maintain a superficial pseudo friendship with peer groups that
dominate in classes. They may even get involved with bullying others. Thus,
pseudo friendship networks with peers who support bullying behaviors will

increase involvement in bullying behavior.

3. Mesosystem
Mesosystem refers to “the interrelations among two or more settings in

which the developing person actively participates (such as, for a child, the
relations among home, school, and neighborhood peer group; for an adult,
among family, work, and social life)” (Bronfenbrenner, 1979, p 25). As a system of
microsystems, the mesosystem comprises a net of social interconnections
among participants, such as students, teachers, and parents, in schools. The
forms of interconnections are “[participation in both or more settings, e.g., family

and school], intermediate links in a social network, formal and informal

55

communications. . .the extent and nature of knowledge and attitudes existing in
one setting about the other” (p 25).

Once a child enters a new environment (e.g., school), which is called a
setting transition, the child becomes a primary link between family and school, so
as to constitute a mesosystem (Bronfenbrenner, 1979). According to
Bronfenbrenner (1979, p 210), a primary solitary link refers to a person who
“participates in more than one setting of a mesosystem” with no other persons
involved in the setting. For instance, according to him, when a child “goes to
school on the first day with no one else from her home” (p 210), she becomes the
primary solitary link between family and the school. Supplementary dual links, in
addition to the primary solitary link, develop when other persons are involved in
the two settings, e.g., “when Mary’s mother attends a PTA meeting, her teacher
pays a visit to the home, or Mary brings home a classmate to play”
(Bronfenbrenner, 1979, p 210).

In terms of school environments, thus, the family relationship remains a
microsystem until children go to school on the first day. Once children go to
school, there is a mesosystem developed by the links among the children,
parents, peers, and teachers, if they participate in each other’s setting, for
instance parents talk to teachers. In contrast, within a school setting, experiences
and communication with teachers and a child’s peers remain microsystems,
since there is no other microsystem involved with it. Parents’ communication with
teachers or with children’s friends in school will develop mesosystems for
bridging between the family system and the school system. This distinction of

experiences with parents (a microsystem) from parental participation within

56

schools (a mesosystem), including communications with teachers and with peers,
could reveal distinct preventive effects of parental involvement with schools on
bullying.

There is research evidence that low parental involvement, an indicator of
a lack of interconnections of microsystems, significantly contributes to bullying
behavior (Flouri & Buchanan, 2003). Scholars also have emphasized that
parental interconnections with peers prevent bullying behavior (Harris & Petrie,
2003; Sullivan et al., 2004). Thus, exchanges of knowledge and information
about attitudes towards bullying through communications between parents and
teachers, and parents and students should have an influence over bullying

phenomenon in schools.

4. Exosystem
Exosystem refers to “one or more settings that do not involve the

developing person as an active participant, but in which events occur that affect,
or are affected by, what happens in the setting containing the developing person”
(Bronfenbrenner, 1979, p 25). The exosystem provides a place where the
microsystem and mesosystem interplay, and examples of an exosystem for
students include gender of children, parents’ socioeconomic status, especially
jobs, and school climate including school policies and rules established by school
boards. Gender of children is considered in this study as an exosystem, because
of the gendered nature of bullying phenomenon in schools. School climate is
generally defined as the total environmental quality within a school (Anderson,

1982). In this section, each part of school’s exosystem is discussed in detail.

57

A. Gender

Researchers have agreed on the gendered nature of bullying (Simpson 8
Cohen, 2004). Most of the bullying studies found that boys are more likely to
bully others (Baldry, 2003; Espelage 8 Swearer, 2003; Farrington, 1993; Nansel
et al., 2001; Natvig et al., 2001; Olweus, 1993, 1995; Rigby, 2005; Rivers 8
Smith, 1994; Seals 8 Young, 2003; Wolke, Woods, Stanford, 8 Schulz, 2001).
Typical roles boys play in bullying situations are bully, reinforcer, and assistant,
while girls are more likely to be defender and outsider (Salmivalli et al., 1996). In
terms of types of bullying behaviors, studies found that boys are more likely to
commit direct physical and verbal bullying against other boys and girls, but that
girls are more likely to use indirect, relational bullying using third-parties to isolate
victims (e.g., Rivers 8 Smith, 1994). More specifically, studies found that girls
tend to be involved more in cyber bullying through emails and instant messages
(e.g., Keith 8 Martin, 2005). Based on prior work, it is expected that there will be
gender differences in prevalence of involvement in bullying behavior and types of

bullying behaviors.

B. Family SES

Batsche and Knoff (1994) argued that family SES is one part of the
exosystem because children have no say about their social class, and social
class has distal effects on children’s behaviors. One way that children are
influenced by the family SES is through distress stemming from economic strain
(Batsche 8 Knoff, 1994). Empirical studies, however, found contradicting

evidence of the effect of parental SES on bullying behavior. For example, Wolke

58

et al. (2001) collected data from 2377 students aged between 6 and 8 in England
and 1538 students aged 8 years in Germany, and found a significant, but weak,
negative association between SES and bullying behavior. In Korea, Shin (2000)
collected data from 578 subjects from both rural and urban areas, and found that
those who have low SES and mothers whose educational level is low are more
likely to bully others. In contrast, based on an extensive literature review, Rigby
(2004) argued that parental SES was an irrelevant factor to school bullying in
different countries, such as France, Spain, and Portugal (for review, see Rigby,
2004; Yang, 2005).

An absence of association between parental SES and bullying behavior is
possibly due to a U-shaped distribution of the relationship of bullying and parental
SES, which cancels out the effects of both extremes. In terms of bullies, Korean
research found two distinct groups of students; the first group consists of
students who are from the high or middle level of SES and the other group
consists of those from the lower level of SES (Park 8 Son, 1998; Yang, 2005).
The first group of bullying students has high academic achievement and self-
esteem and shows leadership, and acts like ‘opinion leaders’ in class (e.g., Kim,
2001; Lee, 1999), and their higher parental educational level is associated with a
higher level of physical bullying (Ah et al., 2005). The second group of students,
by contrast, has low academic achievement and self-esteem, but shows
leadership among delinquent peers (e.g., Lee, 2000; e.g., Lee, 1999), and their
low parental educational and SES levels were associated with verbal bullying (Ah
et al., 2005). These groups are consistent with a categorization of the ‘model boy’

and ‘tough boy’ in Western research (Rodkin 8 Farmer, 2000). In light of these

59

findings, therefore, it will be important to examine the shape of the relationship of

social class to bullying.

C. School Climate

School climate is typically defined as the total environmental quality within
a school (Anderson, 1982). The total environment includes ecology, milieu, social
system, and culture (Tagiuri, 1968). Many studies found that school climate is an
important influence on students’ behaviors. For example, studies found that
improving general school climate could reduce delinquency and misconduct
among non-delinquent youths (Gottfredson, 1986, 2001; Gottfredson 8
Gottfredson, 1985). Specifically, Gottfredson (1986) found that environmental
interventions, which aimed to improve school climate, reduced delinquency and
misconduct slightly but statistically significantly. The environmental interventions
included improving the general image of the school and cohesion between
teachers and students, and generating peer pressure to resolve problems in a
socially acceptable way (Gottfredson, 1986). However, according to Gottfredson
(1986), targeted direct interventions (e.g., services aimed to increase
experiences of academic success, self-concept, and bonds to the social order)
for high-risk students did not reduce overall delinquent behavior and misconduct.

A relatively small number of empirical studies have been carried out to
test the influence of school climate on bullying behaviors. Menesini et al. (2000)
selected 17 bullies, 18 victims, and 35 control group students, whose ages were
between 8 and 11, and conducted an experiment to test the effect of competitive

and cooperative school settings imposed by teachers on children’s behaviors in

60

bullying situations. They found that bullies showed less assertive and aggressive
behaviors in a cooperative setting than in a competitive setting (Menesini et al.,
2000). Espelage and Swearer (2003) argued that school climate could actually
decrease bullying in schools, and that the school climate could be measured by
school policy, rules, and programs, high academic standards, and high parental
involvement in school boards.

In Japan, school norms and value systems, collective justice, and the
moral atmosphere have been studied as factors contributing to school climate
(Masataka, 1998; Naito 8 Gielen, 2005; Tai, 2001). Naito and Gielen (2005)
argued, based on the Japanese literature on bullying, that bullying tends to more
frequently occur in classes where students perceived a poor moral atmosphere.
Another researcher also argued that in classes where bullying is common,
students are not controlled by positive formal school norms accepted by school
authorities (e.g., teachers’ moral authority), and they behave without concern for
others and conform to deviant peer group norms (Hoshino, 2001). In such
classes, collective justice becomes distorted into collective punishment
administered by deviant peer groups on bullying victims (Naito, 1990; Tai, 2001).
These formal and informal norms and values in schools could be revealed
through investigating school policy, rules and programs; perceived effectiveness
of these school policy, rules and programs; collective justice; level of academic
standards; and level of parental involvement in school boards that make
decisions about school policies, rules and programs (Espelage 8 Swearer, 2003;
Rigby, 1996; Swearer 8 Doll, 2001).

It should be also noted that the school climate could be influenced by

61

microsystem and mesosystem factors. According to Bronfenbrenner (1979), each
layer of the set of systems is interconnected and linked with each other layer to
generate a contextual effect over individual behavior. As a place where the micro-
and meso-systems interplay, the school climate is expected to be connected with
the micro-and mesosystem factors. Especially, since children and parents bring
their experiences to school, and since they interact with teachers and peers in
school, the school climate could be affected by them.

Specifically, as noted earlier, parents’ emphasis on the importance of
education may promote a positive school climate, for example, through parental
involvement in school boards. As a part of the microsystem, the parents’
emphasis on education may alter academic standards, which is a part of school
climate. In addition, perceived moral atmosphere is heavily related to teachers’
moral authority and collective punishment (Hoshino, 2001; Naito, 1990; Tai,
2001). Also there is a strong association between the power dynamics among
peers and school climate (Twemlow et al., 2001). These findings indicate that the
school climate could be influenced by the microsystem. Finally, school policies,
rules, and regulations are decided by school boards which could be heavily
influenced by parents and teachers, who discuss and share ideas about bullying
in schools. Consequently, this study hypothesizes that the school climate will be

related to micro- and mesosystem factors.

5. Macrosystem
Macrosystem refers to “consistencies, in the form and content of lower-

order systems (micro-, meso- and exo-) that exist, or could exist, at the level of

62

the subculture or the culture as a whole, along with any belief systems or
ideology underlying such consistencies” (p 26). According to Bronfenbrenner
(1979), belief systems and lifestyles differ for socioeconomic, ethnic, religious,
and other subcultural groups as well as different countries. Such intrasocietal
contrasts, according to him, could reveal functions and effects of each distinct
macrosystem.

Children are reared within social interaction, which is regulated by norms
and values espoused by a culture. Childrearing differs between different cultures,
and children’s behaviors are regulated by parents’ and their culture’s norms and
values. By citing an African phrase, “it does take a village to rear a child,"
Espelage and Sweaer (2003) argued that community has its distal but significant
effect over individual children’s behavior in schools. For example, one study
found that children’s pretend play greatly differs for different countries, and this
discrepancy stems from culture-specific parental teaching norms and values in
terms of personal interaction (Farver 8 Shin, 1997). Thus, understanding cultural
difference may be critical in understanding bullying in schools.

A recent comparative study on bullying found that the likelihood of
bullying others depends on societal characteristics, such as individualism and
collectivism (Nesdale 8 Naito, 2005) and social disorganization (Farrington,
1993; Laub 8 Lauritsen, 1998). Specifically, Nesdale and Natio (2005) collected
data from 158 Japanese and 157 Australian students on behavioral intentions,
and found that compared to their Australian counterparts, Japanese participants
have a greater likelihood of bullying others and a lower likelihood of helping

victims. They concluded that this discrepancy stems partly from different

63

orientations of individualism and collectivism.

Hofstede (1991) explained,

“Individualism pertains to societies in which the ties between individuals
are loose: everyone is expected to look after himself or herself and his or
her immediate family. Collectivism as its opposite pertains to societies in
which people from birth onwards are integrated into strong, cohesive
ingroups, which throughout people’s lifetime continue to protect them in
exchange for unquestioning loyalty.” (Hofstede, 1991, p 51)

These unidemensional definitions of individualism and collectivism have been
criticized and expanded by Triandis (Kim, Triandis, Kagitcibasi, Choi, 8 Yoon,

1994). Triandis (1995) defined individualism and collectivism as follows;

Collectivism may be initially defined as a social pattern consisting of
closely linked individuals who see themselves as parts of one or more
collectives (family, co-workers, tribe, nation); are primarily motivated by
the norms of, and duties imposed by, those collectives; are willing to give
priority to the goals of these collectives over their own personal goals;
and emphasize their connectedness to members of these collectives. A
preliminary definition of individualism is a social pattern that consists of
loosely linked individuals who view themselves as independent of
collectives; are primarily motivated by their own preferences, needs,
rights, and the contracts they have established with others; give priority to
their personal goals over the goals of others; and emphasize rational
analyses of the advantages and disadvantages to associating with
others.” (Triandis, 1995, p 2)

Parents in a collectivistic society teach their children to control the display of

emotion, to value group harmony, and to minimize conflicts with others (Farver 8

64

Shin, 1997; Kim, 1991), and a collectivistic society emphasizes relatedness,
interdependence or “we” (Maday 8 Szalay, 1976). Furthermore, collectivistic
Korean parents discourage children from expressing their own opinions (Kim 8
Choi, 1994; Lee 8 Lee, 1987), and teach them to talk “around” an issue and pay
a great deal of attention to the other’s sensitivity to understand the point of the
conversation (Chu, 1978; Farver 8 Shin, 1997).

Within this kind of collectivistic society, school bullying can be understood
as a group-based problem. When there is a person who violates group goals or
group-oriented norms, or who deviates from group similarities, he or she is more
likely to be bullied by others. In Korea, a study found that students who violate
‘tacit rules’ of peer groups were more likely to be bullied by the groups (Park 8
Son, 1998). Also researchers found that collectivistic orientation within a peer
group induced more lenient and more tolerant reactions towards such rule
violators, but harsher and less tolerant reactions, such as physical attacks,
towards students who do not belong to the peer group (Kim 8 Lee, 2000).
Another empirical study found that students who perceived themselves as more
individualistic were more likely than those who perceived themselves as
collectivistic to bully others as well as to be bullied by others (Lee, 2004).

In Japan, Naito and Gielen (2005) argued that Ijime could be well
explained from a collectivistic point of view. They argued that Ijime has several
collectivistic characteristics, such as viewing bullying as “enmeshment” among
girls, and secrecy surrounding Ijime (p 184), meaning that bullies collectively and
secretly ostracize victims so that only members of the bullies group know what is
happening. Tel (2001) also found that in Japan, bullying occurs more frequently

65

“when classes have some forms of group competition and collective punishment,
in this case anybody seen as interfering with collective goals is in danger of being
bullied” (Tai, 2001).

This collective judgment and punishment, however, depends on the level
of social disorganization. Naito and Gielen (2005) argued that bullying as a form
of collective punishment does not necessarily occur more often in a collectivistic
society than in an individualistic society, but that “poorly integrated collective
society oriented toward collective punishment and relentless group competition
will have more bullying (Naito 8 Gielen, 2005, p 184). Farrington (1993) also
argued, based on his literature review, that bullying occurred more frequently in
big inner-city and socially disorganized areas (see also Laub 8 Lauritsen, 1998).

An empirical study in Korea, which was not based on social
disorganization theory, found that in urban areas, the prevalence of bullying is
significantly greater than in rural areas, and suggested social isolation as a major
source of bullying behavior (Shin, 2000). Even though some have argued that a
higher prevalence of bullying in individualistic communities is due to a higher
reporting tendency among individualistic people (Seals 8 Young, 2003), and
others argued that urbanization negatively affects bullying prevalence (Park 8
Son, 1998), it seems obvious that community characteristics have influence over
students’ behavior in schools. Especially, a comparison of rural areas with low
levels of social disorganization and urban areas with high levels of social
disorganization might reveal acute differences in the effect of community

characteristics on bullying behavior.

66

6. Summary of the Current Ecological Model of School Bullying

Table 3 summarizes the Bronfenbrenner’s (1979) ecological system

model and factors relevant to bullying in schools.

Table 3. Bronfenbrenner’s Ecological System Model and Factors Relevant to

Bullying in Schools

 

Ecological system
model

Factors

Variables

 

Child as a Center
of School Ecology

Individual traits

 

Age

Prior experience of bullying
victimization

Dominance

Impulsivity

Attitude towards aggression
Fun-seeking

 

 

Microsystem

Family interactions

Perceptions of parents’ attitude
toward bullying

Perceptions and experiences
of authoritarian parenting
Child’s experience and witness
of domestic abuse

Importance of education

 

Teacher interactions

 

Perception of teachers’ attitude
toward bullying

Perception of effectiveness of
teacher’s intervention
Perception of teacher’s moral
authority

 

 

Peer interactions

Power dynamic

Level of difference-acceptance
Coercion of public self

Pseudo friendship

 

67

 

Table 3 continued

 

Mesosystem

Parental communication 10

Parents’ communication about
knowledge and information
regarding bullying with
teachers

Parents’ communication about
knowledge and information
regarding bullying with peers

 

Exosystem

Gender

 

[0

Gender

 

Family SES

Financial strain

 

School Climate

 

Academic standards and
general image of school
Competitive vs. cooperative
school climate

Cohesion between students
and teachers

Perception of moral
atmosphere

Parental involvement in school
boards

Policy, rules, programs

 

Macrosystem

 

 

Community
characteristics

Individualism vs. Collectivism
Social disorganization

 

68

 

V. A SUMMARY OF THE CURRENT STUDY

This study recognizes the importance of a holistic and integrative
approach to bullying in schools. Since bullying is not isolated behaviors
(Farrington, 1993, pp 406-408), bullying should be viewed as a systematic abuse
of power within school ecology by an individual or a group of students against
one or more individuals. This study borrows Bronfenbrenner’s (1979) ecological
system model as a theoretical framework to understand multilevel etiological
factors affecting bullying behavior within the social contexts of school
environments.

The current study attempts to expand the definition of bullying and types
of bullying behaviors based on findings from Korean studies. Temporal, severity
and visibility elements are added to measure different types of bullying behaviors.
This study also recognizes that in different countries, there may be different
reasons for bullying. This dissertation, thus, suggests three concepts, fun seeking
tendency, level of difference-acceptance and coercion of public self, as important
culture-specific reasons for bullying behaviors in South Korea“.

In addition, this study examines the effect of perceived teacher’s attitude
toward bullying and social interaction between youths and teachers. These two
potential influences are uninvestigated in prior work. This study also will consider

school climate as an influence on bullying behaviors. Quality of interaction

 

4 Other influences should be considered for other countries. For example,
according to a study conducted in Turkey, the most important and common
reason to bully other students was to pretend to be strong (Kepenekci 8 Cinkir,
2006)

69

between friends is also measured in this study. Unlike previous SNA research on
bullying, this study recognizes the importance of quality of friendship measured
not by mutuality, popularity, centrality, or similarity, but by genuineness of the
helping network (i.e., whether it is or is not a pseudo friendship network). The
proposition is that, to avoid bullying victimization, some students develop a
pseudo friendship network with a certain peer group, and to maintain the pseudo
friendships, they bully others.

In sum, based on the review of literature and the ecological prediction
model developed, this study uses a measure of bullying behaviors, which takes
into account visibility, physicality/severity, duration and involvement. It is
expected that the following factors increase the levels of bullying behavior in
school environments:

- high scores in prior victimization through bullying, dominance,

impulsivity, and fun-seeking tendency

- negative perceptions of and experiences with family

- negative perceptions of teachers

- negative peer interactions

0 high score in power dominance tendency
0 low score in levels of difference-acceptance
0 high score in coercion of public self

0 high score in pseudo friendship network

- low levels of parental communications with teachers and with peers

- being male

- lower and upper levels of family SES

7O

- negative perceptions of school climate
- more collectivistic communities / peer groups

- high level of community social disorganization

71

VI. HYPOTHESES

1. Child as a Center of School Ecology

Individual traits

Hypothesis 1a:

Hypothesis 1b:

Hypothesis 1c:

Hypothesis 1d:

Hypothesis 1e:

Hypothesis 1f:

Age and grade of students are negatively related to bullying
behaviors.

Prior experience of bullying victimization is positively related
to bullying behaviors.

A strong tendency for dominance is positively related to
bullying behaviors.

Impulsivity is positively related to bullying behaviors.

Positive attitude towards aggression is positively related to
bullying behaviors.

Fun-seeking tendency is positively related to bullying
behaviors.

2. Microsystem Level (Interactions among Individuals)

Experiences in the Family

Hypothesis 2a:

Hypothesis 2b:

Hypothesis 2c:

Hypothesis 2d:

Perceived parents’ positive attitude towards bullying is
positively related to bullying behaviors.

Perceived and experienced authoritarian parenting is
positively related to bullying behaviors.

Experience and witness of parental domestic abuse is
positively related to bullying behaviors.

The perceived parents’ view of the importance of academic

72

achievement is negatively related to bullying behavior.

Experiences with Teachers

Hypothesis 2e:

Hypothesis 2f:

Hypothesis 2g:

Perceived teacher’s positive attitude toward bullying is
positively related to bullying behaviors.

Perceived effectiveness of teacher’s intervention in bullying
situations is negatively related to bullying behaviors.

Perceived lack of teacher’s moral authority is positively
related to bullying behaviors.

Experiences with Peers

Hypothesis 2h:

Hypothesis 2i:

Hypothesis 2j:

Hypothesis 2k:

Power dominance tendency is positively related to bullying
behaviors.

Level of difference-acceptance is negatively related to bullying
behaviors.

Coercion of public self is positively related to bullying
behaviors.

Pseudo friendship is positively related to bullying behaviors.

3. Mesosystem Level (Interactions among Microsystems)

Interactions between Family, Peer, and Teacher

Hypothesis 3a:

Hypothesis 3b:

Perceived level of parent-teacher communication about
attitude and information on bullying is negatively related to
bullying behaviors.

Perceived level of parent-peer communication about attitude

and information on bullying is negatively related to bullying
behaviors.

73

4. Exosystem Level (Interactions among Mesosystems)

Interactions between SES, Gender, School Climate and Individuals

Hypothesis 4a:

Hypothesis 4b:

Hypothesis 4c:

Hypothesis 4d:

Hypothesis 4e:

Hypothesis 4f:

Hypothesis 4g:

Hypothesis 4h:

Hypothesis 4i:

Being male is positively related to bullying behaviors.

Very high and very low family SES are positively related to
bullying behaviors.

High academic standard in school is negatively related to
bullying behaviors.

General positive image of school is negatively related to
bullying behaviors.

Competitive school climate is positively related to bullying
behaviors.

Cohesive relationship between teachers and students is
negatively related to bullying behaviors.

Positively perceived moral atmosphere is negatively related to
bullying behaviors.

Perceived level of parental involvement in school boards is
negatively related to bullying behaviors.

Perceived effectiveness of school policies, rules, and
programs targeting prevention and punishment of bullying
behaviors in school is negatively related to bullying behaviors.

5. Macrosystem Level

Interaction between Community and Individuals

Hypothesis 5a:

Perceived level of collectivism is positively related to bullying
behaviors.

74

Hypothesis 5b: Perceived social disorganization is positively related to
bullying behaviors.

Hypothesis 5c: More collectivistic peer groups are more likely to bully others
than less collectivistic peer groups.

75

CHAPTER 3
METHDOLOGY

I. Introduction

ll. Cross-Sectional Research Design

lll. Sampling and Data Collection Procedures
IV. Measurement

V. Analysis Procedure

I. INTRODUCTION

Traditional bullying research has been criticized for its lack of a theoretical
framework and the lack of a multi-level model to view children’s environment as a
multi-layered ecological setting. As previously mentioned, the definition of
bullying also has been controversial, especially when different countries and
cultures are considered. In addition, the lack of empirical study of school climate
and of the influence of a community’s characteristics on bullying behavior in
school, especially in Korean studies, has been criticized. This dissertation is
designed to address these shortcomings in the bullying literature.

Specifically, this study develops a multiple level prediction model based
on ecological system theory. Since the ecological system theory views school
settings as complex and multi-layered social contexts in which children and
others interact and interconnect with each other, this theoretical framework
allows investigation of individual as well as contextual effects. This study also

expands the traditional definition of bullying by considering severity, visibility and

76

duration of behavior all together. For peer influence, instead of relying on social
network measures, such as reciprocated friendship, this study develops four
measures of peer group interaction: the power dynamic, level of difference—
acceptance, coercion of public self, and pseudo friendship networks. This is an
improvement because these four measures of peer group interaction build on the
literature on bullying and allow investigation of different aspects of friendship
relationships as well as quality of friendship networks.

In terms of research design, this dissertation uses cross-sectional data
that will be collected through a multistage cluster sampling method. Even though
some scholars have argued that a developmental, ecological approach to
bullying requires longitudinal data (e.g., Cairns 8 Cairns, 1991; Sexton-Radek,
2005), this study is based on cross sectional data due to the time and resource
constraints. Scholars have argued that cross-sectional design is inferior to
longitudinal design in that the causal direction between factors cannot be
conclusive with cross—sectional data (Cairns 8 Cairns, 1991; King, 2001;
MacCallum 8 Austin, 2000; Singleton 8 Straits, 1999). Even with structural
equation modeling, making causal inferences based on a cross-sectional data is
somewhat problematic due to difficulty in establishing the time sequence of
causes and effects (Gollob 8 Reichardt, 1987, 1991; MacCallum 8Austin, 2000).
This limitation and several reasons for use of cross-sectional data will be
discussed further in the next section, Cross-Sectional Research Design. Also, it
should be noted that the dissertation research takes a first step in providing
evidence of causation, by focusing on hypothesized associations. It is recognized

that next steps would be to design longitudinal research that can establish time

77

order.

Structural equation modeling (SEM) will be used to test and develop the
theoretically driven model that has the best fit to the data. SEM is a series of
statistical methods that allows the researcher to find a best fitting model having
both observed and latent variables (Kline, 1998). In addition, since structural
equation modeling is mainly based on covariance, which represents the strength
of association between variables, SEM explains patterns of correlation among a
set of variables as well as their covariance within a model (Kline, 1998). In other
words, SEM allows the researcher to investigate the nature of covariance (i.e.,
direct and indirect effects) among the variables, and SEM also allows the
researcher to find a model that best fits the data by specifying paths of influences.

Since the ecological prediction model of bullying behaviors developed in
this study requires multivariate analyses among those systems, as well as
investigation of the nature of covariance among those systems, structural
equation modeling is considered to be the most appropriate statistical method.
With SEM used for this ecological model, it is possible to investigate the
influence of each system on bullying behaviors.

In this chapter, specific methodological procedures for research design,
measurement, and sampling will be presented. First the use of a cross-sectional
design will be discussed. Next will be a discussion of the multistage cluster
sampling procedure, as well as specific steps taken for sampling and data
collection. In the measurement section, the level of analysis as well as detailed
information about measures used in this study will be presented. Finally in the

last section of this chapter, analysis procedures will be discussed in detail.

78

II. CROSS-SECTIONAL RESEARCH DESIGN

For this dissertation, a cross-sectional design will be used for several
reasons. First, since there are time and resource constraints, this study will be
based on cross-sectional data. Scholars recognize that cross sectional designs
are economical in terms of time and resource compared to longitudinal design
(e.g., King, 2001). Since data are collected at one time, there is no need to
maintain contacts with subjects, and therefore costs and problems with subject
attrition are reduced.

Second, some ecological variables, e.g., perceived community
characteristics and school climate, are more easily and efficiently compared with
cross-sectional data rather than longitudinal data (Singleton 8 Straits, 1999).
Some scholars have argued that a longitudinal design would be consistent with
the developmental approach to bullying phenomenon in schools (e.g., Farrington,
1993). However, a cross-sectional design has some advantages when the focus
is on contextual and other effects on dependent variables, especially when there
are not many changes over time in the context (Duncan 8 Kalton, 1987;
Singleton 8 Straits, 1999, pp 247 - 248). Since community characteristics and
school climates are not rapidly changing, variation in context is best achieved by
sampling different contexts rather than the same context over time.

Third, studies have found that the anonymous questionnaire is a more
valid and reliable medium for investigating bullying behaviors (see Section 2.
Self-evaluative and Anonymous Measures). Since longitudinal research design
usually requires identifiers to link data collected at different times, it requires

79

researchers to collect any personal information. Collection of any personal
information in longitudinal research on bullying, thus, may possibly hinder valid
and reliable data collection. Consequently, it is possible that a cross-sectional
research design is most suitable for the study of bullying behavior.

In addition, even though cross-sectional data has limitations for drawing
conclusions about causal direction or reciprocal relationships between factors
(Cairns 8 Cairns, 1991; Elliott 8 Harwin, 1974), establishing correlations between
variables is an important first step in drawing any inference about causation
(Singleton 8 Straits, 1999). Furthermore, especially when writing about structural
equation modeling (Gollob 8 Reichardt, 1987, 1991; MacCallum 8Austin, 2000),
some scholars have argued that to some extent investigation of causal direction
and reciprocal relationship is possible when one of the following two assumptions
is met, the instantaneous time interval assumption, and the assumption of
unchanged causal variables (MacCallum 8Austin, 2000). The instantaneous time
interval assumption, which cannot be met in the present study, and which is
inconsistent with the developmental perspective, means that the time interval in
which the causal influence operates is instantaneous. This justifies “concurrent
measurement of variables in a cross-sectional design” (MacCallum 8 Austin,
2000, p 214). The assumption of unchanged causal variables, which is consistent
with the theoretical framework for this research, is that causal variables do not
change between the time that causal effects occur and the time that the causal
variables are measured (MacCallum 8 Austin, 2000). When one of the two
assumptions is met, then, researchers can draw causal inferences using

structural equation modeling and cross-sectional data, although the inference will

80

not be as strong as it would be from a longitudinal research design (Gollob 8
Reichardt, 1987, 1991).

In addition to this justification, this study attempts to test for the possible
impact of omitted variable(s) which may have strong influence on bullying
behavior. With cross-sectional non-experimental data, inferences about causality
are weak, in part because of the possible existence of one or more variables
which may be omitted but important in the relationship between the dependent
variable and independent variables (Frank, 2000; Rosenbaum, 1986; Sobel,
1995, 1996). In other words, to make causal inference stronger, or even feasible,
research using cross-sectional non-experimental data should identify possible
confounding variable(s) that may alter statistical inference based on regression
coefficients. In this dissertation, there is a possibility that a confounding variable
could alter the statistical significance of a correlation between bullying behavior
(a dependent variable) and other independent variables. The confounding
variable suspected but which will not be studied is prior propensity for bullying
behavior.

There are three reasons for omitting variables measuring the prior
propensity for bullying behavior in this study. First, bullying within Korean schools
has not been recognized as illegal behaviors. Not many school authorities have
punished bullies unless there were severe criminal charges against bullies and
physical damage to victims. Thus, there are no official records of punishment of
bullying behavior, such as suspension records, in Korean schools. Second, even
if there were records of punishments for bullying, in Korea family SES has a

significant influence on punishment of deviant youths. Parental involvement,

81

especially when families have high standing in the society, would greatly reduce
the number of formal reactions to bullying behaviors. Finally, it is likely that
youths may not accurately recall their prior bullying behaviors. In addition, it is
also possible that youths may distinguish their prior bullying behaviors from
current bullying behaviors since situations and/or victims have changed.

To investigate the potential influence of the confounding variable, i.e.,
prior propensity for bullying behavior, this study employs a statistical method (i.e.,
ITCV: Impact Threshold for a Confounding Variable, Frank, 2000), which is an
innovation in sensitivity analysis, compared to the sensitivity analysis originally
suggested by Rosenbaum (1986). According to Frank (2000), there are basically
two ways to make causal inference stronger based on regression coefficients,
using experimental data with random assignment or including confounding
variable(s) in a regression model. However, in some situations, such as this study,
it is not possible to include such variables. Frank’s ITCV addresses this issue, by
indexing “the impact of confounding variable on the inference of a regression
coefficient” (p 147). In this case, the confounding variable is a hypothetical
unobserved variable, and the index is a function of “the hypothetical correlations
between the confound and outcome, and between the confound and independent
variable of interest” (p 150). Unlike the original sensitivity analysis, the ITCV
allows researchers to focus on causal inference based on testing the credibility of
the statistical significance test (Frank, 2000). Using this method, this study
examines how large the correlations between the confound and the dependent /
independent variables should be to alter causal inferences that this study will

draw based on results of the statistical significance tests using cross-sectional

82

non-experimental data. In other words, this method will provide a range of

correlations between the prior propensity for bullying behavior with other

variables, within which causal inferences that this study will draw are treated as

credible.

83

III. SAMPLING AND DATA COLLECTION PROCEDURES

1. Sampling

This dissertation used a multistage cluster sampling method. Since it was
impractical to collect the complete list of all middle school students in South
Korea, this project broke down the population into several clusters. According to
sampling theory, sampling error can be reduced by two factors in the sample
design: 1) a large sample produces a smaller sampling error than a small sample
does, and 2) a homogeneous population produces samples with smaller
sampling error than a heterogeneous population does. The stratification is based
on this second element in the sampling theory. The ultimate function of
stratification is to organize the population into homogeneous subsets by
maintaining heterogeneity between subsets and to select proportionate samples
from each stratum.

To ensure variation in effects of community characteristics on bullying
behaviors, this study used levels of social disorganization, measured by mobility
and population density (for more information, see Bursik, 1988; Moriatry, 1999;
Sampson 8 Groves, 1989; Shaw 8 McKay, 1969), as the first stratification of the
population. This stratification involves forming lists of high, medium, and low
disorganization areas, and proportionately sampling communities from these lists
(see Singleton 8 Straits, 1999). Since Korean school districts are based on
governmental administrative districts (the total number of the districts = 232), the
annual governmental report on population mobility and density was used to

stratify the districts. The population mobility per 10,000 is calculated by,

84

Mobility : Input + Output

 

x 10,000
Total

where the input is the total number of emigrant population, the output is the total
number of immigrant population, and the total is the total number in the
population of each district during 2004. The population density is simply

calculated by,

Total

Density 2 ~———-——,—-
Area(Km‘ )

where the total is the total number in the population, and the area is the total
square kilometers of each district. These two proxy measures were then
combined to create a scale for the levels of social disorganization. Based on the
score, the 232 Korean governmental districts were divided into three categories,
low, medium, and high levels of social disorganization areas. The result of this
process is presented in Appendix A. 232 Korean Governmental Districts
Categorized into Three Levels of Social Disorganization.

Based on the sampling frame, this study randomly selected two districts
from each stratum, i.e., high, medium, and low levels of social disorganization
areas. This study, then, performed random selection of one middle school
(secondary sampling units) from each area. This random sampling process
generated six middle schools (two middle schools from each high, medium, and

low social disorganization district). Typically, Korean middle schools contain three

85

grades (from 7th to 9th grade). Within each school, two classes (third sampling
units) were selected from each grade of the schools. Studies found that bullying
rates dramatically decrease in the third grade of the middle school period due to
emphasis on academic achievement to enter better high schools (KDE, 1999;
Moon, 2002). Hoping to capture this variation, this study included all three grades
of middle school to see the effect of emphasis on academic achievement on
bullying behavior. This multistage cluster sampling process produced the total
number of 36 classes. Finally, a self-report survey was administered to students
who were in the classes selected. Since typical classes contain about 40
students, the total expected sample size was 1440. Figure 2 shows the

multistage cluster sampling process used in this study.

86

Figure 2. Multiple Cluster Sampling of Korean Middle School Students

  

232 Districts

    
   

   

 

‘ Medium SD Areas I Low SD Areas

 

High SD Areas

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Middle Middle Middle Middle Middle Middle
School School School School School School
A B C D E F
2°29 2°29. 90291020. 9°20. 1°29
S" 3 :T :1" :1" :1" :7" :1" 3‘ 3" 3" 3" =" 3" :1" 7 :1" 2"
0G 00 0‘0 00 0’0 GO DC (IO 00 00 GD 00 0C) 01') 0Q 00 0‘1) 00
H H H H H H H H H H H H H H H H H H
9) 93 90 D) ID DJ 9) a: 9) ID $3) I)? 1:0 CD in 9) 52-) D)
D. D. D. D. D. C). C)- D. D- C)- D. C). O. D. O. C). D. D.
(D (D (D (D (D (D (D (D (D (D (D (D (D (D (D (D (D (D
[Q N N N N N IQ Ix.) N N N N N to [Q N Ix.) N
222222 222222 222222
9) 1)) SD A) 9) 9) I'D $2) 133 ID N N D) D) CD 9) ID I'D
U' U) {D m U) U) U) U) U) U) U) U) U) U) m U) U) U)
U) U) U) U) U) U) U) U) U) U) U) U) U) m U) U) U) U)
(D (D (D (D (D (D (D (D (D (D (D (D (D (D (D (D (D (D
U) U) U) U) U) U) U) U) U) U) U) U) U) U) U) U) U: U)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2. Data Collection Procedure and Response Rates
Before administration of the survey, this researcher conducted a field test
of the Korean version of the survey questionnaire. For the field test, this
researcher recruited five Korean students who were enrolled in middle schools in
the United States. (It is common for Korean students to study in the United States

for one or two years during middle school.) Each individual was asked to read

87

about 30 items without responding to the questions. Once they finished reading
them, in individual interviews, they were asked questions, such as “Is there any
question you don’t understand?”, “Is there any word that you don’t know the
meaning?”, “Is there any suggestion to improve this survey?”, or “What do you
think these questions are asking?” After individual interviews with five students,
the youths were asked to discuss what this researcher wanted to know. During
the interview sessions, students were encouraged to talk about whatever they felt
and thought of this survey and their ideas to improve this survey. Based on field
test results, necessary revisions were made.

To administer the self-report survey developed for this study, the
researcher recruited data collectors to actually administer the survey. The data
collectors were recruited from appropriately educated individuals (e.g., teachers,
social scientists) known personally to the researcher, and were further educated
and trained about subjects’ right to privacy and appropriate procedures to handle
the survey questionnaires. Specifically, the data collectors were trained in a
protocol to recruit youth and teachers to participate in the study, and obtain
parental consent in writing. To obtain the consent forms, the data collectors first
sent out the parental consent forms to the subjects’ schools, and explained the
right to privacy, anonymity, and volunteerism to students. During the second visits
to the schools (between February 12 and 17, 2007), the data collectors collected
the consent forms. Once the data collectors fulfilled the consent process
adequately, then the data collectors handed out the survey questionnaire to
students whose parents agreed to voluntary participation of their children in the

survey. Children also could refuse to participate in the research at this point.

88

Since the survey was anonymous, there was no chance to identify subjects once
they handed in the questionnaire to the data collectors.

Table 4 shows the response rates for each of six different schools. Overall,
the average response rate was 92%. Interestingly, the largest and the smallest

schools had lower response rates than other schools.

Table 4. Response Rates for Survey Questionnaire

 

 

 

School Total number of students Number of survey returned %
A 219 205 93
B 298 234 78
C 237 234 98
D 199 198 99
E 159 132 83
F 238 235 98
Total 1350 1238 92%

 

89

IV. MEASUREMENT

This dissertation uses self-evaluative anonymous measures of bullying
behaviors. Consistent with ecological systems theory and research in the
tradition of that theory, the measurements used in this study are carefully
developed to reflect individual level perceptions or experiences on specific
factors. In this section, the individual level, self-evaluative and anonymous
measures used in this study are discussed in detail. In addition, a threat to
measuring context through perceptions of exo- and macro-system factors, and a

technique used to minimize the threat to validity are discussed.

1. Level of Analysis

According to Bronfenbrenner (1979), the effects of different levels of
factors on individual behaviors can be measured at the individual level, because
the effects are perceived not as properties of the factors at different levels, but as
individual psychological outcomes of interactions between the person and the
factors at different levels. In other words, individual developmental changes are
not measured by correlations between static characteristics of an individual and
environments, but they are studied in terms of changes in “content - what is
perceived, desired, feared, thought about, or acquired as knowledge, and how
the nature of this psychological material changes as a function of a person’s
exposure to and interaction with the environment” (Bronfenbrenner, 1979, p 9).
Consequently, it is assumed that individual level measures of experiences and

perception can be used to measure the multiple factors at different levels.

90

Despite the tradition of measuring multilevel influences with individual
reports in the research that has built on Bronfenbrenner’s ideas, the
appropriateness of use of individual level measures should receive serious
consideration. It can be argued that once school climate and community
characteristics are measured by individual level perceptions, then such social
contexts may change quickly, because perception as a function of individual
behaviors and attitudes could change fairly quickly. For example, how
consistently do children perceive that their school is competitive or is well
equipped with anti-bullying policies? How consistently do children perceive that
their neighborhood is collectivistic? Do their perceptions remain unchanged
during a month, a semester, or a year? If the perception does change fairly
quickly, then the individual measures of perception of the social contexts may not
be reliable measures for the exo— and macro-system levels. In addition, this
further weakens the strength of causal inference to be drawn from this cross-
sectional research.

To deal with the threat to the validity of individual level measures of
organizational level factors, this study provides choices that encourage
respondents to consider stability of the context they are rating. Specifically, youth
are asked to rate how often parents and teachers have interacted during the last

school year, not just how strongly they agree that such interaction has occurred.

2. Self-evaluative and Anonymous Measures
Researchers in criminology have validated self-evaluative anonymous
measures of delinquency by comparing those measures with official delinquency

91

records. Since the earliest prototype of self-reported delinquency scales was
developed by Short and Nye (Short 8 Nye, 1957), the self-report measure of
delinquency has greatly evolved to become a reliable and valid method (Huizinga
8 Elliott, 1986; Thornberry 8 Krohn, 2000). More relevant to bullying, previous
bullying studies found that self-report anonymous measures are reliable and valid.
Even though there is contradicting evidence in a study that suggests that peer
and teacher reports yield consistent identification of bullies and victims (Cornell 8
Brockenbrough, 2004), the majority of bullying studies found that a self-
evaluative and anonymous measure of bullying is a reliable method (Ahmad 8
Smith, 1990; Chan, Myron, 8 Crawshaw, 2005; Farrington, 1993; Xie et al., 1999).
For example, based on review of six previous studies on bullying, Farrington
(1993) argued that the self-report survey is more reliable than the interview
method. Ahmad and Smith (1990) also suggested that the anonymous
questionnaire is more valid and reliable because denial of bullying behavior could
be the biggest source of invalidity of measurement through the interview method
with identification. In another study, Xie et al. (1999) found that teacher
evaluation is far different from self-evaluation in bullying studies, because
teachers often were not aware of bullies and victims. More recently, one study of
562 students in Toronto, Canada, suggested that respondents reported the same
amount of bullying and victimization regardless of whether they put their names
on the surveys (Chan et al., 2005).

In addition to these previous findings, this study uses the self-evaluative
method to identify bullying behaviors because it is possible that teacher
evaluation in Korea may not be reliable due to the Asian culture. Naito and

92

Gielen (2005) found that in Japan, teacher reports of bullying are unreliable,
because teachers tend to under-report bullying incidents to protect their
reputation and jobs. They suggested that the Asian culture’s emphasis on the
importance of dignity and reputation might prevent teachers from reporting school
violence (Naito 8 Gielen, 2005). The actual anonymous self-report survey

questionnaire used for this dissertation is presented in Appendix B.

3. Operationalized Variables

A. Dependent Variable: Bullying Behaviors

There are at least 9 instruments developed to measure bullying behaviors
in schools (for review, see Crothers 8 Levinson, 2004). Among them, this
dissertation uses an improved Korean version of the Olweus’ bully/victim
questionnaire. This survey questionnaire is commonly used in Korean studies
and its validity and reliability have been tested in a couple of studies (e.g., Lee,
2004, reliability =.92). In addition to the questions about frequency of bullying
behaviors, self-reports will also be obtained to categorize youth according to the
type of bullying they are involved in (i.e., severe chronic, moderate chronic /
severe non-chronic, and moderate/minor non-chronic bullying). The source of

scale items for the different dimensions of bullying are:

1) Visibility (Dishion et al., 1995)

I Covert bullying

I Overt bullying

93

2) Physicality/Severity (Carney, 2005; Coloroso, 2003; Dishion et

al., 1995; Olweus, 1995, 2003)

I Relational bullying: spreading rumors, leaving out of things
on purpose, ignoring

l Verbal bullying: threatening, taunting, teasing, name
calling, harassing sexually or talking about physical
defects

I Physical bullying: kicking, hitting, cutting with knife,

breaking bones

3) Duration (Boulton 8 Smith, 1994; KEDI, 1998; Kim, 2004;
Salmivalli et al., 1998; Shim, 2004; Smith 8 Brain, 2000)
I Week, month, semester, year
I Middle school, high school

All items are presented in Appendix B.

B. Independent Variables

The independent variables in this study are divided into the five
categories: 1) individual traits, 2) microsystem, 3) mesosystem, 4) exosystem,
and 5) macrosystem variables. Whenever possible, existing items shown to be
valid in prior research, and items previously used in Korea, were used. Of course,
in some cases, these were no measures that had been used in research in Korea
or in any country. Consistent with Bronfenbrenner’s and ecological theorists’

approach, all variables are measured at individual level.

94

1) Individual Traits

This dissertation measures individual traits, such as age, prior experience
of bullying victimization, tendencies for dominance and impulsivity, attitude
towards aggression, and fun-seeking tendency (Ah et al., 2005; Baldry, 2003;
Bentley 8 Li, 1995; Bosworth et al., 1999; Boulton 8 Smith, 1994; Espelage 8
Swearer, 2003; Farrington, 1993; Kim, 2001; Lee, 2000; Lee, 1999; Naito 8
Gielen, 2005; Park, 2002; Rigby, 2004; Seals 8 Young, 2003; Seo 8 Han, 2004;
Shin, 2000). The respondents’ age is simply measured in an open-ended
question. The prior experience of bullying victimization is measured by an item:
Have you experienced the above things [15 items of bullying behaviors] in an
earlier grade? To measure tendency for dominance, a modified version of the
Standardized Personality Evaluation Test (Lee, Byun, 8 Jin, 1969) is used. This
version includes four items: 1) It is difficult to refuse when my friends ask me to
do a favor, 2) I usually try to avoid confrontation even if I am right, 3) I usually try
to observe rules and regulations, and 4) I hate to lose in an argument. To
measure impulsivity, a modified version of the four-item scale in the Standardized
Personality Evaluation Test (Lee et al., 1969) is used. The four items are 1) It is
difficult for me to sit still during the class, 2) I start things but have a hard time
finishing them, 3) I do things without thinking or planning, and 4) I need to use a
lot of self-control to keep out of trouble. To measure attitude towards aggression,
a modified version of the Beliefs Supportive of Violence Scale from the Houston
Community Project Scale (Dahlberg, Toal, 8 Behrens, 1998) is used. This
modified version was developed by Bosworth et al. (1999), and the reliability

was .71. The present study will use four of the six original items: 1) If I walked

95

away from a fight, I’d be a coward, 2) It’s okay to hit someone who hits you first,
3) If a kid teases me, I usually cannot get him/her to stop unless I hit him/her, and
4) If I refuse to fight, my friends will think I’m afraid. Fun-seeking tendency is
measured by the agree-disagree items: 1) Did you do these things [bullying
behaviors] to another person? For fun? and 2) My friends enjoy it, because they
think I am having fun with them, 3) It is fun to watch Wangtta students go through
these things [bullying behaviors], and 4) There is nothing wrong with these things
[bullying behaviors] because they are just for fun. These items were developed
specifically for the present research, since no similar measures have been used

in other studies.

2) Microsystem Variables

Family interactions are indicated by perceptions of parents’ attitude
towards bullying, perceptions and experiences of authoritarian parenting, child’s
experiences and witness of domestic abuse, and perceived parental views on the
importance of education (Baldry, 2003; Batsche 8 Knoff, 1994; Christie-Mizell,
2003; Curtner-Smith et al., 2006; Espelage et al., 2000; Gladstone et al., 2006;
Kim, 2001; Kim 8 Lee, 2000; Lee 8 Kim, 2000; Moon, 2002; Naito 8 Gielen,
2005; Shields 8 Cicchetti, 2001; Shin, 2000). Parents’ attitude towards bullying in
schools is measured by six agree-disagree items: 1) I think my parents do not
like me teasing other students, 2) I think my parents do not like me hitting other
students, 3) I think my parents do not like me making fun of other students, 4) I
think my parents do not care about me teasing other students, 5) I think my

parents do not care about me hitting other students, and 6) I think my parents do

96

not care about me making fun of other students. The two measures of
authoritarian parenting were developed based on the results of previous studies
(Baldry, 2003; Batsche 8 Knoff, 1994; Flouri 8 Buchanan, 2003; Shields 8
Cicchetti, 2001). The two items are: 1) My parents use physical discipline for
punishments, and 2) I think my parents want to have control over almost every
aspect of my life.

Teacher interactions are measured by youths’ perception of teachers’
attitude towards bullying, perception of effectiveness of teachers’ intervention in
bullying situation, and perception on teachers’ moral authority (Espelage 8
Swearer, 2003; Naito, 1990; Park, 2003; Rodkin 8 Hodges, 2003; Yang, 2005).
The six items for perception of teachers’ attitude towards bullying are: 1) I think
my teacher does not like me teasing other students, 2) I think my teacher does
not like me hitting other students, 3) I think my teacher does not like me making
fun of other students, 4) I think my teacher does not care about kids teasing
classmates in school, 5) I think my teacher does not care about kids hitting
classmates in school, and 6) I think my teacher does not care about kids leaving
out classmates in school. Three items for the perception of teachers’ moral
authority are developed based on the results of the study conducted by Naito
(1990). The three items are: 1) I don’t care about what my teacher says about
morality, 2) I think my classmates do not care about what my teacher says about
how classmates treat each other, and 3) I think my teacher’s moral standard is
old-fashioned, so that it isn’t acceptable in my generation. These items were
developed specifically for the present research, since no similar measures have

been used in other studies.

97

Peer interactions are measured by four aspects of peer relationship
interactions: 1) power dynamic, 2) level of difference-acceptance, 3) coercion of
public self, and 4) pseudo friendship (Cairns 8 Cairns, 1991; Coloroso, 2003;
Graham 8 Juvonen, 1998; KEDI, 1998; Kim 8 Lee, 2000; Kim, 2004; Lee, 2004;
Long 8 Pellegrini, 2003; Naito 8 Gielen, 2005; Park, 2002; Park 8 Son, 1998;
Pellegrini 8 Long, 2002; Rigby, 2004, 2005; Sec 8 Han, 2004; Shin, 2000;
Twemlow et al., 2001; Young 8 Sweeting, 2004). Each of these four aspects of
peer interactions is measured by a four-item scale carefully developed based on
the results of the previous studies. In addition, since Korean school districts have
a very high proportion of students who stay together as classmates from
elementary through high school, peer groups are usually not different inside and

outside of school. Thus, the focus will be on school-based peer groups.

3) Mesosystem Variables

Mesosystem variables measure interactions between two microsystems.
Two measures are 1) perceived frequency of parent-teacher communication
regarding knowledge and information about bullying in school, and 2) perceived
frequency of parent-peer communication regarding knowledge and information
about bullying in school (Flouri 8 Buchanan, 2003; Natvig et al., 2001). The items
for parents’ communication with teachers are: 1) how often in this school year did
your parents and teachers talk to each other to share information about students
teasing others?, 2) how often in this school year did your parents and teachers
talk to each other to share information about students hitting others?, and 3) how

often in this school year did your parents and teachers talk to each other to share

98

information about students making fun of others? Three similar questions were
developed to measure the parents’ communication with peers: 1) how often in
this school year did your parents and friends talk to each other to share
information about students teasing others?, 2) how often in this school year did
your parents and friends talk to each other to share information about students
hitting others?, and 3) how often in this school year did your parents and friends
talk to each other to share information about students making fun of others?
Responses for these perception measures are intended to encourage students to
report not on just a recent communication, but rather on a consistent pattern of

communication over time. Thus the answers are “this never happened this school

9’ fl ” It

year, this happened once this school year, this happened a few times this

school year, this happened more than a few times this school year,” and “this

happened very often this school year.”

4) Exosystem Variables

Exosystem variables include gender, family SES, and school climate. The
family SES is measured by 1) perceived levels of family SES, 2) parents’ jobs,
and 3) monthly parental income (Batsche 8 Knoff, 1994; Kim, 2001; Lee, 2000;
Lee, 1999; Park 8 Son, 1998; Rigby, 2004; Rodkin 8 Farmer, 2000; Shin, 2000;
Wolke et al., 2001; Yang, 2005). The school climate variable consists of five
subcategories, perceived school’s academic standards, perceived competitive vs.
cooperative school climate, perception of moral atmosphere, perceived parental
involvement in school boards, and perceived effectiveness of school’s policies,

rules, and programs for anti-bullying phenomenon. Each of these subcategories

99

is measured by two agree-disagree questions (three questions for perception of
effectiveness of anti-bullying policies, rules and programs) developed based on
the results of the previous studies (Espelage 8 Swearer, 2003; Hoshino, 2001;

Masataka, 1998; Naito 8 Gielen, 2005; Swearer 8 Doll, 2001; Tai, 2001).

5) Macrosystem Variables

Macrosystem variables include levels of individualism or collectivism, and
levels of social disorganization. The levels of individualism and collectivism are
measured by the Korean version of the INDCOL (Individualism and Collectivism)
measure developed by Kim (1996). The original INDCOL measure was
developed by Singelis, Triandis, Bhawuk, and Gelfand (1995). The Korean
version of the INDCOL included some culture-specific measures and excluded
culturally irrelevant questions to Korean context (Kim, 1996). According to the
original developers of the measure, a vertical dimension emphasizes hierarchical
order, and a horizontal dimension emphasizes equality among members
(Singelis, Triandis, Bhawuk, 8 Gelfand, 1995). Thus, vertical individualists (VIND)
are competitive and emphasize victory over others, horizontal individualists
(HIND) emphasize independence and autonomy. By contrast, vertical collectivists
(VCOL) value the group and its goals over family and individuals, and horizontal
collectivists (HCOL) emphasize cooperation and affection among members. Each
measure contains 8 questions, but this dissertation uses 4 questions for each
category based on reliability scores presented by Kim (1996).

To measure youths’ perception of levels of social disorganization, this

study develops three items based on previous research results (Sampson 8

100

Groves, 1989; Sampson, Raudenbush, 8 Earls, 1997): 1) my neighbors do not
care what my friends do in this area, 2) teenagers in my neighborhood are out of

control, and 3) it is difficult for kids to make friends in my neighborhood.

101

V. ANALYSIS PROCEDURE

The analysis of the data in this study consisted of two phases. In the first
phase, data preparation, the first-order factor analysis, basic statistical analyses,
hypothesis testing, and sensitivity analysis were carried out. In the second phase,
the second-order confirmatory factor analysis, SEM, and multiple group analysis
were carried out.

Specifically, in the first phase, frequency distributions were examined for
all variables to check accuracy of data entry. In addition, by examining the
descriptive statistics, outliers were identified. To perform SEM, careful treatment
of missing data is required. Thus, this study carefully dealt with missing
information by using the multiple imputation method. More specific procedures
which have been followed are presented in the next section. In the next step,
factor analysis and reliability tests were performed to develop scales for the
variables identified in the literature review section. This first-order factor analysis
was carried out to prepare for testing the hypotheses presented in Chapter 2.
The multiple regression analysis was used to test for the connection of each
independent variable to the dependent variables. Descriptive statistics were
obtained, and bivariate and multivariate analyses were performed.
Multicollinearity tests (i.e., tolerance and variance inflation factor) were also
performed, when suspected intercorrelations among some variables were
observed. Once multivariate regression analysis was completed, the ITCV, a type
of sensitivity analysis, was performed to estimate impacts of the omitted variable,

e.g., prior bullying propensity, on significant relationships found in the regression.

102

In the second phase, the second-order confirmatory factor analysis was
performed for the dependent variables and some independent variables (i.e., the
scales created through the factor analyses and the reliability tests in the data
preparation stage) to make appropriate decisions about which items to include as
indicators of latent variables. With path modeling, this study suggested a basic
hybrid model (a combination of the confirmatory factor analysis and the path
model), and respecified the basic model based on the results of a series of SEM
analyses. Finally, multiple group analysis was attempted to examine any
difference between subgroups which were divided by severity, duration, and
visibility of bullying behaviors.

As a simplified basic model, the following theoretically derived model was

tested to establish the fit to the sample data:

103

Figure 3. An Ecological Prediction Model of School Bullying

,//—\\,
l Macrosystem l\

\J \\\.\\

(Family SES

/_..

/

\\ I/ ,
\ (I,
ll
/ /
I \ / \ ’1‘
// 'I
f l
/ I

(Ind ivid u al Traits l/\ / /
\ / [I
/

\
,r'
\ \, \ \ /
\ \ \\ / // “/V/

\ \\ ,\ \ X. /
l \\ \‘\ “\‘\ \\
1 \‘x, . \\ \‘
\ ( ‘g/Famlly

\ - -
\ ‘\ Interactions "
l l \V/

\\

1/ Teacher
\lnteractions

\ //

l

\ / h “R“ -’l l

1 \\\ {/1/

y/ Peer \yf/

‘~ Interactions
v//

104

  
    

School
Climate

 
   
   
 

The .05 alpha level was the criterion for the statistical significance test in
this analysis. For testing the model fit, the following standardized fit indexes were
used: xz/df, the Jereskog-Stjrbom Goodness of Fit Index (GFI), and the Adjusted
Goodness of Fit Index (AGFI). Other fit indexes, such as the Bentler Comparative
Fit Index (CFI), and Root Mean Square Error of Approximation (RMSEA), were
reported. For the multiple imputation for missing data, LISREL 8.80 was used.
For the descriptive statistics, correlation, and the multicollinearity tests, the
Statistical Package for the Social Science (SPSS) 13.0 was used, and for the

structural equation model, the AMOS 5.0 program was used.

105

CHAPTER 4
FINDINGS

I. Data Preparation

ll. Factor Analysis, Scales, and Reliability Tests

llI. Univariate, Bivariate and Multivariate Analyses

IV. Sensitivity Analysis: ITCV

V. Structural Equation Modeling: Basic Model

VI. Structural Equation Modeling: Multiple Group Analysis

I. DATA PREPARATION

Structural equation modeling (SEM) requires careful preparation and
screening of a raw data set to avoid violations of assumptions and failure in the
iteration process, which is often referred to as a “crash" (Kline, 1998, p 67). In
this section, the procedures taken to prepare the raw data for a model-fitting
analysis will be described. The procedures included treating outliers, checking
normality in distributions, checking patterns of incomplete data, and later
imputing the missing observations for SEM analysis.

First, the results of simple univariate frequency tests of all the variables
identified no extreme values. Second, since SEM requires more conservative
adherence to the normal distribution assumption than other statistical analyses,
such as ANOVA (Kline, 1998), this study examined normality of frequency
distributions by running normal probability plots of all variables except the
dependent variables measuring frequency, severity, visibility, and duration of

bullying behaviors. Table 5 shows the results of the univariate normality tests.

106

Table 5. Univariate Normality Tests: Skew and Kurtosis Values

 

Items Skew Kurtosis

 

120. How often in this school year did your parents 42 18.8
and teachers talk to each other to share '
information about students teasing others?

121. How often in this school year did your parents 45 229
and teachers talk to each other to share
information about students hitting others?

122. How often in this school year did your parents 53 30.4
and teachers talk to each other to share
information about students making fun of others?

123. How often in this school year did your parents 44 19.9
and friends talk to each other to share information
about students teasing others?

124. How often in this school year did your parents 4.6 22.6
and friends talk to each other to share information
about students hitting others?

125. How often in this school year did your parents 4_7 245
and friends talk to each other to share information
about students making fun of others?

 

All other items -3.0 ~ 3.0 -10.0 ~ 10.0

 

The values of the univariate skew index indicate the direction and
magnitude of skew. It is difficult to provide clear guidelines about how much non-
normality is problematic, but an acceptable range for skewness is -3.0 ~ +3.0,
and an acceptable range for kurtosis is -10.0 ~ +10.0 (Chou 8 Bentler, 1995; Hu
8 Bentler, 1995; West, Finch, 8 Curran, 1995). As shown in the Table 5, the
values of skew and kurtosis for all independent variables fell in the acceptable
ranges respectively, meaning that those variables have normal distributions.
However, six items measuring parental communication with teachers and peers

107

were extremely positively skewed. Almost all subjects answered that their parents

have never met teachers or their friends during the 2006 school year (see Table

6). These high percentages were not unexpected within Korean schools.

Unfortunately, it is unlikely that any transformation will normalize those variables.

Table 6. Univariate Statistics of the Non-normally Distributed Items 120 — 125

 

 

Item Number Mean St.D. Percentage of “Never”
120 1.13 .50 92,0
121 1.13 .48 91,4
122 1.09 .43 93,9
123 1.14 .56 91,3
124 1.14 .55 91.7
125 1.12 .51 92,5

 

Finally, this study tried to identify any pattern of incomplete data. Most of

the variables have less than 16 missing cases. Six variables were identified as

having more than 19 missing cases. To see whether there is any systematic

pattern of missing data, this study examined the frequencies of missing data by

grouping those six variables (see Table 7).

108

Table 7. Group Comparison of Frequencies of Missing Data

 

 

 

 

 

 

Items 031 Q32 051 0148 0149 0156
(# of missing data) (20) (22) (24) (22) (19) (19)
School (Region)
School A 6 (2.9)* 6 (2.9) 4 (2.0) 6 (2.9) 6 (2.9) 5 (2.4)
School B 8 (3.4) 9 (3.8) 4 (1.7) 3 (1.3) 5 (2.1) 3 (1.3)
School C 1 (0.4) 1 (0.4) 2 (0.9) 2 (0.9) 1 (1.4) 1 (0.4)
School D 4 (2.0) 4 (2.0) 5 (2.5) 3 (1.5) 2 (1.0) 2 (1.0)
School E 0 (0.0) 0 (0.0) 5 (3.8) 6 (4.5) 4 (3.0) 5 (3.8)
School F 1 (0.4) 2 (0.9) 4 (1.7) 2 (0.9) 1 (0.4) 3 (1.3)
Sex
Male 8 (1.1) 10 (1.4) 14 (2.0) 11 (1.5) 8 (1.1) 9 (1.4)
Female 12 (2.3) 12 (2.3) 10 (1.9) 11 (2.1) 11 (2.1) 9 (1.7)
Grade
7th grade 6 (1.5) 6 (1 .5) 6 (1 .5) 8 (2.0) 6 (1 .5) 6 (1 .5)
8th grade 5 (1 .0) 6 (1.2) 7 (1.4) 6 (1.2) 6 (1.2) 5 (1 .0)
9th grade 9 (2.6) 10 (2.9) 11 (3.2) 8 (2.3) 7 (2.1) 8 (2.3)
Family SES
Missing 0 (0.0) O (0.0) 2 (9.5) 0 (0.0) 0 (0.0) 0 (0.0)
Upper class 2 (1.9) 2 (1.9) 5 (4.7) 4 (3.8) 3 (2.8) 1 (0.9)
Upper-middle 2 (0.6) 3 (0.9) 2 (0.6) 5 (1 .5) 4 (1.2) 4 (1.2)
Middle class 16 (2.6) 17 (2.7) 13 (2.1) 12 (1.9) 10 (1.6) 13 (2.1)
Lower-middle 0 (0.0) 0 (0.0) 2 (1.4) 1 (0.7) 2 (1.4) 1 (0.7)
Lower class 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)
Academic achievement
Missing 1 (4.3) 1 (4.3) 3 (13.0) 1 (4.3) 1 (4.3) 1 (4.3)
Upper 10% 3 (1.5) 3 (1.5) 3 (1.5) 5 (2.6) 3 (1.5) 1 (0.5)
Upper 20% 3 (1.7) 4 (2.3) 0 (0.0) 3 (1.7) 3 (1.7) 2 (1.1)
Upper 30% 0 (0.0) 0 (0.0) 0 (0.0) 2 (1.7) 2 (1 .7) O (0.0)
Middle 8 (2.1) 9 (2.3) 9 (2.3) 5 (1.3) 4 (1.0) 8 (2.1)
Lower 30% 3 (1 .6) 3 (1.6) 3 (1 .6) 4 (2.1) 3 (1 .6) 4 (2.1)
Lower 20% 0 (0.0) 0 (0.0) 5 (5.4) 2 (2.2) 2 (2.2) 3 (3.2)
Lower 10% 2 (3.3) 2 (3.3) 1 (1.7) 0 (0.0) 1 (1.7) 0 (0.0)

 

* Note: Figure in the parenthesis is percentage of the missing data.

109

The numbers in percentages are the proportion of missing data for each of the
subgroups being compared. For example, in the first column, for 031, 2.9% of
responses were missing for School A, 3.4% of responses were missing for
School 8, and so on.

Items 31 and 32 asked about visibility of bullying victimization (see
Appendix B). Item 51 measured duration of bullying behavior, and items 148 and
149 measured subjects’ perception of having an independent personality. Item
156 measured subjects’ perception of collectivistic characteristics. As shown in
the table, the missing data of those six variables have been categorized into
different subgroups of each of the demographic variables, regions (schools), sex,
grade, family SES, and academic achievement. The results suggest that there is
no systematic pattern of missing data. In other words, the missing observations
occurred completely at random (missing completely at random, MCAR) (Kline,
1998; Rubin, 1976). Since there is no systematic loss of data, dealing with
missing data can be carried out, with a certain level of confidence, by either
deletion (listwise or pairwise) or imputation. For model-fitting analysis (SEM), this
study used the multiple imputation method for dealing with missing data to retain
as many cases as possible. According to Rubin (1987), multiple imputation
replaces missing values with a set of possible values, and the multiple imputation
contains uncertainty of selecting the right value to impute (Rubin, 1987, see also;
Schafer, 1997). Many scholars suggested that the best ad hoc method of
imputation is the multiple imputation method (Canchola, Neilands, 8 Catania,
2002; SAS, 2002; Schafer, 1997; Toit 8 Mels, 2005). This study used LISREL 8.8

to conduct the multiple imputation, and used the EM (Expected Maximization)

110

algorithm rather than MCMC (Markov chain Monte Carlo) algorithm, since the
data set showed monotone missing data patterns. The monotone missing data
pattern is the case in which a certain variable is missing, and then a set of
subsequent variables are missing. The results from the multiple imputation
showed that the convergence was achieved after 6 iterations, and the percentage
of the missing values was 0.76, meaning that only .76% of the total data were

missing.

111

II. FACTOR ANLAYSIS, SCALES, AND RELIABILITY TESTS

Factor analysis and reliability tests were used to develop scales. First, a
factor analysis was done for the items hypothesized to be part of each scale to
determine whether one or more constructs was being indicated. All factor
analyses were done with Varimax rotation. All variables presented in Table 8 had
all items load on one factor, except two variables, perceptions of parents’ attitude
towards bullying and perceptions of teachers’ attitude towards bullying. These
two variables loaded in two factors: parents (or teachers) “don’t like bullying” and
parents (or teachers) “care about bullying.” Thus two separate scales were
developed for perceived parent and teacher attitude towards bullying.

For all scales, reliability coefficients were calculated to determine whether
items formed a reliable scale. When there was a possibility of increasing a
reliability coefficient, items have been selected and/or deselected based on
changes to the reliability when items were dropped from the scale. Table 8 shows
the final results of factor analysis followed by reliability tests for all scales
developed for this study. The alpha levels for most of the scales were relatively
high and fell in an acceptable range. However, several scales have weak, but
acceptable, alpha scores. Those scores of the scales are highlighted in the Table

below.

112

Table 8. Reliability Coefficients for Scales Used

 

 

 

Scales # of Items Alpha
Dependent Variables
Relational bullying 4 .84
Verbal bullying 6 .81
Physical bullying 5 .78
Duration 3 .61
Visibility 5 .67
Involvement 3 .50
Independent Variables
Individual traits
Academic achievement 2 .92
General positive behaviors 3 .76
Prior experiences of bullying victimization
Relational bullying 4 .76
Verbal bullying 6 .80
Physical bullying 5 .81
Dominance 3 .45
Impulsivity 3 .55
Attitude toward aggression 4 .61
F un-seeking tendency 4 .80
Microsystem level
Family interaction
Perceptions of parents’ attitude towards 3 .90
bullying: Parents don’t like bullying
Perceptions of parents’ attitude towards 3 .93
bullying: Parents care about bullying
Authoritarian parenting 2 .41
Experience/witness of domestic abuse 4 .68
Parents’ view of the importance of education 2 .76
Teacher interaction
Perceptions of teachers’ attitude towards 3 .95
bullying: Teachers don’t like bullying
Perceptions of teachers’ attitude towards 3 .92

bullying: Teachers care about bullying

 

113

Table 8 continued

 

Effectiveness of teachers’ intervention
Teachers’ moral authority
Peer interaction
Power dynamic
Level of difference-acceptance
Coercion of public self
Pseudo friendship
Mesosystem level
Parents’ communication with teachers
Parents” communication with peers
Exosystem level
Family SES
Academic standard and general image of school
Competitive vs. cooperative school climate
Student — teacher cohesion
Moral atmosphere
Parents’ involvement with school boards
Effectiveness of policies, rules, and programs
Macrosystem level
Individualism
Vertical individualism
Horizontal individualism
Collectivism
Vertical collectivism
Horizontal collectivism
Levels of collectivism of peer group
Social disorganization

(0N

OJOO-h-h

0000

NNNOONNU'I

NN

NODNN

.58
.68

.79
.66
.71
.72

.87
.90

.71
.76
.71
.78
.42
.56
.79

.80
.70

.72
.76
.61
.43

 

114

III. UNIVARIATE, BIVARIATE AND MULTIVARIATE ANALYSES

1. Univariate Analysis

A. Demographic Characteristics and Academic Achievement of Sample

Table 9 summarizes the demographic and achievement characteristics of
the students who participated in this study. There are more male subjects (57.8%
of the total subjects) than female subjects (42.2% of the total). Among the
randomly selected six schools, there is one non-coed male school (School F).
Slightly more female subjects participated in the survey (in 4 schools). However,
the sex ratio is similar to that for the Korean population age 10 — 19. According to
the census data reported by the Korean National Statistical Office in 2005, the
male vs. female ratio was 112.8:100 among teenagers (KNSO, 2005). The
sample in this study consists of roughly 15% more male students than female
students, indicating that the sex ratio is similar to that of Korean population. Age
of most of the subjects is that of the typical Korean students for the grade level,
but there are 62 students who are 1 year younger than the regular 7th grade
students. There is one student whose age is 17.

The majority of the students’ parents have an educational level higher
than a high school diploma. Overall, subjects’ fathers had higher levels of
education than subjects’ mothers. For both parents, however, 4 year university
graduate is the second largest category following the high school graduate
category. This result is consistent with the demographic characteristics of the
general population (i.e., parents who have middle school students, KNSO, 2005).

About 30 percent of students reported that their parents’ monthly income

115

is greater than 3,000,000 won, which is the average monthly household income

of South Korea in 2006 (Korean National Statistical Office, 2007). The monthly

income of the majority of students’ parents is less than the average monthly

income level of South Korea. In terms of academic achievement, about the same

portion (approximately 30 %) of students reported that their academic

achievement level is either upper 30%, middle, or lower 30% of their class.

Table 9. Demographic and Academic Achievement Characteristics of Sample (N

 

 

= 1238)
n %
Sex (Total)
Male 716 57.8
Female 522 42.2
School
School A Male 98 47.8
Female 107 52.2
School B Male 112 47.9
Female 122 52.1
School C Male 101 43.2
Female 133 56.8
School D Male 105 53.0
Female 93 47.0
School E Male 65 49.2
Female 67 50.8
School F Male 235 100
Female 0 0
Age
13 62 5.0
14 409 33.0
15 465 37.6
16 301 24.3

 

116

Table 9 continued

 

17
Grade
7th Grade
8th Grade
9‘“ Grade
Family SES
Father’s educational level
No education
Elementary school graduate
Middle school graduate
High school graduate
2 year college graduate
4 year university graduate
Graduate school or more
Mother’s educational level
No education
Elementary school graduate
Middle school graduate
High school graduate
2 year college graduate
4 year university graduate
Graduate school or more
Parents’ monthly income
500,000 won or lower
500,000 — 1,000,000 won
1,000,000 — 1,500,000 won
1,500,000 — 2,000,000 won
2,000,000 — 2,500,000 won
2,500,000 — 3,000,000 won
More than 3,000,000 won
Academic achievement
Upper 10% of the class
Upper 20% of the class
Upper 30% of the class

407
490
341

44
62
537
55
362
137

35
79
692
32
286
66

28

69
138
195
194
193
352

196
174
115

32.9
39.6
27.5

0.2
3.7
5.2
44.8
4.6
30.2
11.4

0.3
2.9
6.6
58.0
2.7
24.0
5.5

2.4

5.9
11.8
16.7
16.6
16.5
30.1

15.8
14.1
9.3

 

117

Table 9 continued

 

Middle of the class 384 31.0
Lower 30% of the class 193 15.6
Lower 20% of the class 93 7.5
Lower 10% of the class 60 4.8

 

B. General Characteristics of Sample

The general characteristics of subjects are presented in Table 10. The
table includes the mean and the standard deviation scores of the scales
developed based on the results of the first-order factor analysis and the reliability
tests. As mentioned in the previous section, Data Preparation, the variables seem
to be normally distributed around means, except the variables measuring
parental communication with teachers and peers. In addition, overall, the
standard deviations for all variables seem to be low and acceptable. The mean
for relational bullying behaviors (4 — 20) is 6.14, the mean for verbal bulling (6 —
30) is 9.89, and the mean for physical bullying (5 — 25) is 6.53. Many responded
that they have positive behaviors (11.82 out of range between 3 and 15). The
mean for impulsivity (3 — 15) is 9.30. Also, on average students said that they
perceived that their parents and teachers would have a negative attitude towards
bullying behaviors and they would be concerned with bullying behaviors (12.89
and 12.65 respectively out of ranges between 3 and 15). The mean for the
teachers’ moral authority (3 — 15) is 10.02. In terms of peer interactions, the
mean for coercive conformity (3 — 15) is 10.89 and the mean for pseudo
friendship (3 — 15) is 9.04. The mean for levels of collectivism of peer group (3 -

15) is 9.22.

118

Table 10. General Characteristics of Sample (N=1238)

 

 

 

Scales Mean St.D.
Dependent Variables
Relational bullying (4 — 20, more bullying) 6.14 3.10
Verbal bullying (6 — 30, more bullying) 9.90 4.82
Physical bullying (5 — 25, more bullying) 6.53 3.02
Duration (3 — 10, longer) 4.16 2.06
Visibility (5 - 25, more visible) 14.02 1.99
Involvement (3 — 15, more involved) 10.31 2.47
Independent Variables
Individual traits
Academic achievement (2 — 15, higher) 9.99 3.44
General positive behaviors (3 — 15, positive) 11.82 2.38
Prior experiences of bullying victimization
Relational bullying (4 — 20, more bullying) 5.80 2.60
Verbal bullying (6 — 30, more bullying) 9.51 4.60
Physical bullying (5 — 25, more bullying) 6.82 3.42
Dominance (2 — 10, more dominance) 7.82 2.16
Impulsivity (3 — 15, more impulsive) 9.30 2.68
Attitude toward aggression (4 - 20, more aggressive) 12.45 3.26
Fun-seeking tendency (4 - 20, higher tendency) 7.80 3.50
Microsystem level
Family interaction
Perceptions of parents’ attitude towards bullying: 12.89 2.78
Parents don’t like bullying (3 - 15, positive)
Perceptions of parents’ attitude towards bullying: 12.46 3.00
Parents care about bullying (3 - 15, positive)
Authoritarian parenting (2 - 10, more 6.29 2.11
authoritarian parenting)
Experience/witness of domestic abuse (4 — 20, 15.99 3.64
more experiences/witness)
Parents’ view of the importance of education (2 — 6.12 2.49

10, less important)
Teacher interaction

 

119

Table 10 continued

 

Perceptions of teachers’ attitude towards bullying:

Teachers don’t like bullying (3 — 15, positive)

Perceptions of teachers’ attitude towards bullying:

Teachers care about bullying (3 - 15, positive)

Effectiveness of teachers’ intervention (2 — 10,

less effective)

Teachers’ moral authority (3 - 15, less authority)
Peer interaction

Power dynamic (4 - 20, less power-seeking)

Level of difference-acceptance (4 — 20, more

tolerance)

Coercive conformity (3 — 15, less coercive)

Pseudo friendship (3 — 15, less pseudo)

Mesosystem level

Parents’ communication with teachers (3 — 15, more)
Parents’ communication with peers (3 - 15, more)

Exosystem level

Family SES (5 - 34, higher SES)

Academic standard and general image of school (2 —
10, higher standard, better image)

Competitive vs. cooperative school climate (2 — 10,
more cooperative)

Student — teacher cohesion (3 — 15, more cohesive)
Moral atmosphere (2 — 10, more positive)

Parents’ involvement with school boards (2 — 10,
more involvement)

Effectiveness of policies, rules, and programs (2 —
10, more positive perception)

Macrosystem level

Individualism
Vertical individualism (2 — 10, less individualistic)
Horizontal individualism (2 — 10, less
individualistic)

12.65

12.00

5.81

10.02

8.26
9.46

10.89
9.04

3.35
3.40

23.04
5.36

5.98
8.94
5.92

6.20

5.37

6.66
4.73

2.89

3.11

2.07

2.61

3.62
3.14

2.69
2.87

1.25
1.48

4.64
2.16

2.04
2.97
1.69

1.66

1.94

1.85
1.75

 

120

Table 10 continued

 

Collectivism
Vertical collectivism (2 - 10, more collectivistic) 6.01 1.59
Horizontal collectivism (2 — 10, more collectivistic) 6.61 1.81
Levels of collectivism of peer group (3 — 15, more 9.22 2.34
collectivistic)
Social disorganization (2 — 10, more socially 5.94 1.74

disorganized)

 

C. Validity of School Climate Measures

In this section, the individual level measures (i.e., measures of individual
students’ perceptions) of school climate are compared with the teachers’
perceptions of school climate, teacher’s moral authority, and moral atmosphere.
Those measures are also examined across six schools. Although this study
provided choices that were worded to encourage students to consider stability of
the context they were rating, the analysis in this section will provide useful insight
into the validity of the students’ perceptions. To do this, the items (q31, q40, q37
in teacher survey) similar to the items in student survey (q128, q101, q135) were

selected for the comparison. Table 11 shows the results of the comparison.

121

Table 11. Mean Comparison across Six Schools

 

 

 

 

Total School

B C D E F

Academic Mean:3.31 2.98 3.08 3.55 3.22 3.24 3.71
.. Achievement St.D:1.22

é Moral authority Mean:2.60 2.32 2.66 2.75 2.63 2.51 2.68
% St.D:1.08

Moral atmosphere Mean:2.53 2.62 2.60 2.12 2.61 2.72 2.59
St.D:1.06

Academic Mean:3.26 3.00 3.20 3.5 3.67 2.40 3.67
_ Achievement St.D:1.08

g Moral authority Mean:3.29 3.50 3.20 3.17 2.67 4.00 3.33
§ St.D:.84

Moral atmosphere Mean:2.82 2.17 3.40 3.50 2.17 3.00 2.83
St.D:.87

 

Also, the followings three figures display the differences in the average
scores of students and teachers’ responses for three questions, academic
achievement, teacher’s moral authority, and school moral atmosphere. Although
there are slight differences in the mean scores for each item (shown in the table
and figures), all of the mean scores for each school from both groups fall within
less than one standard deviation away from the total means. Mean difference
tests between the two groups found no statistically significant difference across
six schools. In addition, correlations between average scores of teachers’ and
students’ responses to above three variables in all schools and in three
classrooms showed that they were highly correlated (see Table 12). These
results indicate that to some extent the individual level measures for the students’

perceptions toward school climate provide adequate results.

122

Table 12. Correlations between Average Scores of Teachers and Students

Responses to Measures of Academic Achievement, Moral Authority, and Moral

 

 

Atmosphere.

All schools Class 1 Class 2 Class 3
Academic achievement .50 .51 .53 .50
Moral authority .48 .50 .51 .49
Moral atmosphere .53 .51 .55 .52

 

Figure 4. Mean Difference in Academic Achievement across Six Schools

 

 

 

 

-_-I— teacher

 

 

 

 

 

Academic Achievement

 

 

 

123

Figure 5. Mean Difference in Teacher’s Moral Authority across Six Schools

 

 

4.5
4
3.5
3 ,2 -2
2.5 ‘—0— student
2 l——Il+ teacher
1.5
1
0.5
0

Q

School
A
School
B
School
C
School
School
E
School
F

Moral authority I

 

 

Figure 6. Mean Difference in Moral Atmosphere across Six Schools

 

4
3.5
3
2': + student
1_5 _7—Il— teacher
1
0.5
0

 

s s 2 a 6 s
_c< .ctIlcO :0 Em 5U.
o o o o o o
(I) 10) U) U) U) U)

Moral atmosphere ‘

 

 

124

 

 

2. Bivariate Analysis

Bivariate correlations were computed between all independent variables
used to predict each dependent variable to see whether there is any serious
intercorrelation among the variables. The results are presented in Table 13
through 17. The correlation results indicated that there were no high
intercorrelations among independent variables, suggesting no multicollinearity
problem in the data. In the presentation of findings, VIF and tolerance statistics
are examined for moderate correlations.

The dependent variables are significantly correlated with each other.
Relational bullying behavior is significantly correlated with verbal bullying
behavior (r =.70, p < .001), and verbal bullying behavior is significantly correlated
with physical bullying behavior (r :68, p < .001). In Table 13, the three types of
previous bullying victimizations are moderately and significantly correlated with

bullying behaviors.

A. Individual Traits

Interestingly, as shown in Table 13, age (13 — 17 year-old) has a
significant positive relationship with three different types of bullying behaviors.
The literature review revealed that bullying behaviors increased from 6th to 8th
grade and then decrease as youth get older. However, there seems to be a
different pattern in South Korea. There was no significant change throughout the
middle school years, but it is suspected that frequency of bullying behaviors
would decrease during the high school years. However, this expectation cannot

be examined with the data collected fro this dissertation. The relationship of

125

bullying to age will be further discussed in the Limitations and Recommendations
secfion.

Previous experiences of bullying victimization were strongly positively
correlated with all types of bullying behaviors. Especially, verbal bullying
victimization was strongly correlated with verbal bullying behaviors (r = .50, p
< .001), and physical bullying victimization was also strongly correlated with
physical bullying behaviors (r= .51, p < .001).

Dominance was significantly positively correlated with all three types of
bullying behaviors (r= .12, r= .08, r= .07 respectively). As the literature
suggested, impulsivity, attitude towards aggression, and fun-seeking tendency

were significantly positively correlated with all types of bullying behaviors.

126

8. Vet... .8. vet. .8. ve-

 

 

......2. :8. .19. ES. 8. :8. 58. i8. .58. Beams-ea“. :>
few. .8. :8. .8. 8. so. 18. ER. 8. 825m 88.05 o e>
03 8.- 3.8. $9. 53. 8. ES. 58. 3.8. £2235. ¢>
03 8.- 8.- 8.- .8.- :8. :8. the. 88558 m>
03 i8. is... :8. EB. E8. £8. 555 .8522 s>
03 .38. :8. i8. 58. $8. 555 seems e>
8.3 8. i8. 58. 18. E85 .8822 m>
03 .8. 2:2. .8. 8e. v>
84 i8. 18. 83:8 622:5 m>
8.? .12. 83:8 88> ~>
9: 823m .8828. 5
m> m> S e> m> 4> m> N> F>

 

@0323 93:5 new 9.9..- _m:u_>_cc_ c6653 mcozflotoo .9 £an

127

B. Microsystem Variable

Overall, youths’ experiences and perceptions of their experiences have
significant relationships with their bullying behaviors (see Table 14). Specifically,
youth’s perceptions of authoritarian parenting and their experiences / witness of
domestic abuse have significant positive relationships with bullying. The more
negative the perceptions and experiences of parenting and abuse, the more
frequent are bullying behaviors. Interestingly, parent’s emphasis on the
importance of education is significantly positively correlated with all three types of
bullying behaviors, meaning that the more parents emphasize education, the
more youths bully others in all three different ways. In addition, youths’
perception of their parents’ attitude towards bullying behaviors was not
significantly correlated with their bullying behaviors. Only the parents’ concern
about bullying behaviors was significantly negatively correlated with verbal and
physical bullying behaviors.

Teachers’ attitude towards bullying behaviors also did not have significant
relationships with youths’ bullying behaviors. Only teachers’ concern about
bullying was negatively correlated with verbal and physical bullying behaviors.
Youth’s perception of the effectiveness of teacher’s intervention has significant
negative correlations with relational and verbal bullying behaviors. However,
teacher’s moral authority has negative relationships with all types of bullying
behaviors.

Finally, peer variables seem to have some important associations with
bullying. Specifically, the power dynamic among peers has positive significant

correlations with all types of bullying behaviors (r= .21 for relational and physical

128

bullying, and r = .23 for verbal bullying). Level of difference-acceptance also has
significant but weak positive relationships with relational, verbal and physical
bullying behaviors (r= .06, r= .05, r= .07 respectively). Coercion of public self
has significant relationships with relational (r = .14, p < .001) and verbal bullying
behaviors (r = .09, p < .01), but not with physical bullying behaviors. As predicted
based on the literature review, the pseudo friendship network has significant
positive relationships with relational, verbal and physical bullying behaviors (r

= .13, r= .14, and r= .08).

It is worth noting that the two measures of perceived parents’ attitudes
toward bullying behaviors are moderately and significantly correlated with the
corresponding measures of perceived teachers’ attitude toward bullying
behaviors (r= .51 for don’t like bullying, and r= .44 for care about bullying).
However, neither correlation indicates any severe multicollinearity issue
(tolerance = .74, VIF = 1.35 for don’t like bullying, and tolerance = .75, VIF = 1.34
for care about bullying). Similarly, Table 14 shows that the parents’
communication with teachers is highly and significantly correlated with the
parents’ communication with peers (r=.67, p < .001). Although the level of
correlation is high, there is no multicollinearity issue in this level (tolerance = .56,
VIF = 1.79). Table 15 and 16 also show that there is no single variable that is
highly intercorrelated with other variables, suggesting that there is limited risk of

having multicollinearity problem in this data set.

129

@539: 8:90.3va .SEBES o>_o._ooovmm> .Aoocmfiooom-oococotfi u.o _o>o_vwm> .3526 Loaoarms
.3553 .908 mLocomooom> .EozcoZEE mzocomg .6 mmmcozfimtovo; A98 ”335m mzocomosm F> .35.: rcon
”museum Moose—moon; .Eozmosoo ho mocmtancog .Aowsnm ozmmEoEm F> .Bczcmema cmtmEocvavg .828 ”museum

—

25.8va .22: can “eases seesaw; .3533 .8623? 8523.... .8832, .3528 .mee_a_sm:> 838g

:8. vet... .8. vs... .8. ve-

 

ooe stow. stow. stow. story :32... .mo.. mo. «tow. riot- «1:.- For oo. two. 3.3.. time.
00... ..oo. 5.- 11:.- too.- .Ioo. ism F. :oo. oo. Lo.- 3... _. _.. 15m w. vo. too. .13..
oo._. simm. ism _. .- senor Sim .- .- sio r .- I... .- _.. rim _- .- .mor tin .- .- 1.2.- tho. .mo. «oo.
oo..- ..tom .- 25m F .. team... rim _. .. rim r. 2:. 5.. tin _. .. sssvwr assomr :5 FN. 31mm. :3 _.N.
oo.? issmm. isom. too. stv _. .. Ito _.. too. sssNN. too. :30 _. .. stow.- ..to r .-

006 :L _.. toor senor stem _.. too. too. too.- mo.- .19. .i 1:. _. r .-
oo._. :imv. too... {to F. tho. rivv. stow. too... smor vor
oo._. «oo. vo. mo..- tiwo. 2:5. or oo. _.o.
oo... simm.- isvmr 3:3.- No.. 23:. .13. too.
oo._. tax-v. «1mm. to... time.- ..imwr time.-
ooe rim _.. vow foor rim F .i {oo.-

oo._. tswv. «floor toor vor
oo.? vor No.. No.-

ooe isoo. atom.

oo... atom.
oo._.

VN>
mN>
NN>
FN>
ow>
or>
o F>
t>
o_.>
m F>
3>
m F>
NF>

m>

N>

_.>

 

VN> mN> NN> FN> oN> or> ow> t> or> mr> vw> mr> N_.> m> N> P>

 

meo_>mcmm 9.2.5 was mo_nm:m> E65322). comics 20:38.00 .3 flow...

130

C. Mesosystem Variables

Unexpectedly, as shown in Table 15, youth’s perception of parental
communication with teachers and peers has weak but positive relationships with
all types of bullying behaviors. Parents’ communication with teachers regarding
bullying behaviors was correlated with relational (r = .15), verbal (r = .18), and
physical bullying (r = .21), suggesting that youth, who perceived their parents
talking about bullying to their teachers frequently also bullied other students more
often. Also, youth, who perceived their parents talking to their friends about
bullying behaviors, said they bullied others more often (r = .16, r = .21, r = .22 for
relational, verbal and physical bullying).

It is worth noting that there is a strong significant relationship between
parents’ communication with teachers and parents’ communication with peers (r
= .76). Although the level of correlation is high, there is no multicollinearity issue

(tolerance = .55, VIF = 1.81).

131

88. vat. .8. ve: .8. ve-

 

 

84 1.8. 5.8. i8. .59. 585 55, 58853558 .5855 8>
83 i8. 32. $8. 5289 55, 88053558 .5852 8>
83 :8. .58. 852.8 .8552 m>
8.5 .12. 852.8 .855 ~>
8.? 82.5 825.82 F>
8> 8> m> ~> S

 

w§>mcom o:_>__:m ncm woBmtm> 868803.). comics 228.250 .3 min...

132

D. Exosystem Variables

The correlations reveal that male students commit more bullying behavior
(relational, verbal, and physical) (see Table 16). The expectation was that there
would be a U shape distribution between bullying and family SES. However,
scatterplots and mean-difference tests for the relationship between family SES
and all types of bullying behaviors revealed that there was no U shape
distribution, and the correlation was not significant.

Overall, school climate has a strong correlation with bullying behaviors.
Specifically, higher academic standard and general positive image of the school
have significant negative correlations with all types of bullying behaviors.
Especially, a cooperative school climate has a negative relationship with bullying
behaviors ( = -.12, r= -.17, r = -.11 for relational, verbal, and physical bullying).
Moral atmosphere also has negative correlations with relational (r = -.09) and
verbal bullying behaviors (r= -.08). Parental involvement with school boards has
a negative correlation with all three types of bullying behaviors (r = -.05, r = -.05, r
= -.06 for relational, verbal, and physical bullying). Finally, perception of
effectiveness of school policies, rules, and anti-bullying programs has negative

relationships with relational (r = -.08) and verbal bullying behaviors (r= -.05).

133

88. vet... .8. var. .8. ve-

 

 

..8.- :8.- 88 88m 8>
.8.- .8.- 5858225 .5885 8>
.22.- ...8? 8880 ~m>
:8.- ..8.- 88.8852 88.2 8>
it- :8? 8858 825858 88>
:8 a .- :8.- 5888 25882 8N>
8. .8. mum 558“. m~>
.58.- ...8.- 8m R>
i8. i8. 828 8858 m>
83 :85. 858 88> ~>
8.? 823m 8888 e>

N> F>

 

meo_>mcmm_ o:_>__:m new moEmtm> Eofiw>moxm coozcon wcozflotoo or 8382.

134

E. Macrosystem Variables

The statistics in Table 17 suggest a significant relationship between
community characteristics and bullying behaviors. A higher level of horizontal
collectivism has negative correlations with all three types of bullying behaviors (r
= -.10 for relational, r= -.07 for verbal, and r= -.05 for physical bullying).
However, there was no significant correlation between vertical collectivism and
bullying behaviors. As expected, higher levels of social disorganization are
positively correlated with bullying behaviors (r = .07, r = .09, r = .08 for relational,
verbal, and physical bullying), and the more collectivistic peer groups more often

bully other students in two ways, verbally and physically (both r = .07).

135

88. vet... .8. Va... .8. ve-

 

 

84 8. .18.. 8- :8. :8. :8. 52858888 .88 88>
83 i8. .12. :8. t8. 8. 5228.80 888 8>
83 $8. ..8.- :8- .12.- 5m_>=8__eo 8588: 8>
8.? 8. 8.- 8.- 5858.80 88> 8>
8.5 :8. 38. 823m 8855 m>
8.? 52. 828 88> m>
83 853m. 8288 E
8> 8> 8> 8> m> N> e>

 

m.o_>mcom oc_>__:m new moEmtm> 89982085. con-:38 8:28.250 .2. 658-.-

6

3. Multivariate Analysis

To examine the hypothesized relationship between the dependent
variables and the independent variables, and to see which variables have
stronger effects than others, multivariate regression was carried out, and the
results are presented in Table 18. First, a multivariate regression analysis was
carried out with a composite measure of three types of bullying behaviors and a
composite measure of three types of prior bullying victimizations. The result
suggested that the composite measure of prior bullying victimization (Beta=.468)
was the most significant variable in the prediction of overall bullying behavior.
Other important predictors were fun-seeking tendency (Beta=.252), parents-peer
communication (Beta=.126), attitude towards aggression (Beta=.125), dominance
(Beta=.096), difference-acceptance (Beta=-.065), coercive conformity
(Beta=.060), and cooperative climate (Beta=-.056). The adjusted R-square value
(.429) suggested that about 43% of overall bullying behaviors were explained by
the multivariate model. Subsequently, each type of prior bullying victimization and
all other variables were regressed to each type of bullying behaviors. The results

are discussed in the following subsections.

A. Individual Traits (H1a — H1f)

Although there were significant positive correlations between age and
bullying behaviors, there was no significant relationship between them in the
multivariate regression. The literature review revealed that bullying behaviors
increased from 6th to 8th grade and then decrease as youth get older. However,

the correlations and regression results indicates that it seems different in South

137

Korea. Unfortunately, it is impossible to compare the frequency of bullying in
middle school years with that of elementary or high school years due to the
sampling frame.

Previous experience of bullying victimization was the strongest positively
related predictor for the different types of bullying behaviors. Especially, each
type of bullying victimization was the strongest predictor for the same type of
bullying behaviors. Relational bullying victimization had the strongest association,
after controlling for other independent variables, with relational bullying behavior
(Beta=.256). Verbal bullying victimization was the strongest predictor of verbal
bullying behavior (Beta=.313). Physical bullying victimization also was the
strongest predictor of physical bullying (Beta=.452). Also physical bullying
victimization has significant relationships with both relational and verbal bullying
(Beta=. 1 86/.174 respectively).

The multivariate regression results also showed that dominance was a
significant predictor of all three types of bullying behaviors (Beta=.118/.075/.069
for relational, verbal, and physical bullying). Although the literature suggested
that impulsivity, attitude towards aggression, and fun-seeking tendency would be
significantly and positively related to bullying behaviors, the results showed that
there are different relationships. Impulsivity was not a significant predictor of any
types of bullying behaviors. Attitude toward aggression and fun-seeking tendency,
however, were significant predictors of all three types of bullying behaviors.
Especially, fun-seeking tendency was a significant and strong predictor of all
three types of bullying behaviors (Beta=.220/.226/.190 for relational, verbal, and
physical bullying).

138

Overall, relational bullying had the strongest association with relational
bullying victimization. Fun-seeking tendency had the second strongest
association with relational bullying. For verbal bullying, verbal bullying
victimization had the strongest effect, and fun-seeking tendency was the second
strongest predictor of verbal bullying. Physical bullying victimization had the

strongest association with physical bullying.

B. Microsystem Level (H2a — H2k)

In Table 14, the correlations suggested that youths’ family, peer, and
school experiences are related to their bullying behaviors. However, the
multivariate regression analysis revealed that indicators of the microsystems did
not significantly predict all types of bullying behaviors (see Table 18). Parents’
attitude toward bullying behaviors was not a significant predictor of any types of
bullying. Teachers’ attitude toward bullying behaviors only predicted physical
bullying (Beta=.059). Unlike previous research findings, this study showed null
relationships of experience of authoritarian parenting, domestic abuse,
importance of education, teacher intervention, and pseudo friendship with each
type of bullying behaviors. However, teachers’ moral authority was a significant
predictor of verbal bullying (Beta=—.058). The peer variables, dynamic, level of
difference-acceptance and coercive conformity, were significant predictors of
relational bullying. Level of difference-acceptance and coercion of public self

were also significant predictors of verbal bullying (Beta=-.067l.049 respectively).

139

C. Mesosystem Level (H3a — H3b)

As noted previously, youth’s perception of parental communication with
peers has strong positive relationships with all types of bullying behaviors, and
this variable was significant in the multiple regression analyses
(Beta=.089/.122/.131 for relational, verbal, and physical bullying). Youth’s
perception of parental communication with teachers was not significantly

predictive of any type of bullying behaviors.

D. Exosystem Level (H4a - H4g)

As expected, sex was a significant predictor of all types of bullying
behaviors (Beta=.068/-.066/-.058 respectively at p < .05). Many variables
measuring different aspects of school climate were found to be insignificant
predictors of all types of bullying behaviors, but a cooperative climate was a

significant negative predictor of verbal bullying (Beta=-.075).

E. Macrosystem Level (H5a — H50)
Unexpectedly, unlike the previous arguments on the effects of community
and cultural level variables on bullying behavior, the results showed all variables

at the macrosystem level were insignificant direct predictors of bullying.

Table 18. Multivariate Regression

 

Dependent Variables

 

, , , Composite , ,
Hypothesrzed Relationship , Relational Verbal Physucal
of bullying

 

Beta St.E Beta St.E Beta St.E Beta St.E

 

140

Table 18 continued

 

Individual traits

 

 

 

H1a Age .010 .252 .009 .090 .010 .127 -.006 .083
Composite of victimization .468 .025
H1 b Prior bullying victimization
Relational .256 .038 .019 .053 .050 .035
Verbal .028 .026 .313 .037 -.024 .024
Physical .186 .031 .174 .044 .452 .029
H1c Dominance .096 .101 .118 .036 .075 .051 .069 .033
H1d Impulsivity .028 .084 .013 .030 .025 .043 .030 .028
H1e Attitude on aggression .125 .074 .077 .027 .125 .038 .117 .025
H1f Fun-seeking tendency .252 .073 .220 .026 .226 .037 .190 .024
Microsystem level
H2a Parents’ attitude
Don’t like bullying -.004 .095 -.027 .034 .014 .048 -.011 .032
Care about bullying .027 .087 .049 .031 .026 .044 .001 .029
H2b Authoritarian parenting .040 .118 .028 .042 .034 .060 .044 .039
H20 Domestic abuse .019 .069 .025 .025 .011 .035 .030 .023
H2d Importance of education .001 .093 .003 .033 .003 .047 -.011 .031
H2e Teachers’ attitude
Don’t like bullying .046 .091 .029 .033 .043 .046 .059 .030
Care about bullying .041 .083 .046 .030 .035 .042 .015 .028
H2f Teacher intervention -.011 .113 -.041 .041 -.001 .057 -.003 .038
HZg Teacher’s moral authority -.026 .098 -.036 .035 -.058 .050 .022 .033
H2b Power dynamic .040 .072 .076 .026 .017 .036 .032 .024
H2i Difference-acceptance -.065 .074 -.O61 .026 -.067 .037 -.039 .024
H2j Coercive conformity .060 .084 .088 .030 .049 .042 .022 .028
H2k Pseudo friendship -.032 .080 -.020 .029 -.022 .041 -.019 .027
Mesosystem level
H3a P—T communication -.009 .228 -.008 .082 -.030 .116 -.014 .076
H3b P-P communication .126 .191 .089 .069 .122 .097 .131 .063
Exosystem level
H4a Sex (male=1, female=2) -.032 .476 .068 .174 -.066 .245 -.058 .162
H4b Family SES .033 .046 .031 .017 .016 .023 .038 .015
H4c Academic standard -.016 .108 .005 .039 -.028 .055 -.003 .036

 

141

Table 18 continued

 

H4d Cooperative climate -.056 .109 -.028 .039 -.075 .055 -.044 .036
H4e Moral atmosphere -.018 .146 -.012 .052 -.048 .074 .016 .049
H4f Parental involvement -.005 .145 .008 .052 .014 .073 -.037 .048
H4g School policies -.014 .127 -.014 .046 -.002 .064 -.011 .042

 

Macrosystem level
H5a Collectivism

Vertical collectivism .016 .139 .001 .050 .004 .070 .043 .046
Horizontal collectivism -.023 .133 -.034 .048 -.012 .067 -.035 .044

H5b Social disorganization -.002 .126 -.008 .045 -.009 .064 -.002 .042
H50 Collectivistic peergroup .019 .095 .022 .034 .011 .048 .013 .031

 

Adjusted R-square .429 .291 .416 .361

 

* Note: Variables found to be significant at least at p < .05 are in Bold.
** Note: Beta = Standardized Coefficients

Overall, the independent variables predicted roughly 30 — 40 % of
variance in the three different types of bullying behaviors. It is worth noting that
the literature review revealed that those variables included in this analysis were
all important predictors of bullying behaviors, but that this study found that many
of them were not significant predictors at least of some types of bullying. This is
partly because the previous empirical studies did not include all different levels of
variables in their analyses, but this study included them all in one model. Unlike
correlation results between the independent variables and the dependent
variables, the multivariate regression analysis results showed that further
investigation is necessary to see whether there is evidence of different paths of
causal influence among the independent variables, as would be suggested by

theory.

142

IV. SENSITIVITY ANALYSIS: ITCV

As mentioned in the section, Cross-Sectional Research Design, it is
important to pay attention to confounding variable(s), which may alter causal
inferences made based on statistical analyses without the unmeasured
confounding variable(s). In other words, careful investigation of the impact of
omitted confounding variable(s) will help to support a strong causal inference,
especially with outcomes from cross-sectional research design. Previously, it was
argued that prior propensity for bullying behavior could be an important
confounding variable for this ecological model. In this section, ITCV (Impact
Threshold for a Confounding Variable) is estimated for each relationship between
the dependent variables and the variables found to be significant in the
multivariate regression. Using this method, this study examines how large the
correlations between the confounding and the dependent I independent variables
would need to be to alter causal inferences that this study will draw based on
results of the statistical significance tests using cross-sectional non-experimental
data. In other words, this method will provide a range of correlations of the prior
propensity for bullying behavior with other variables, within which causal
inferences drawn from study findings will be treated as credible.

According to Frank (2000), the impact of the unobserved confounding

variable is calculated by the following equation.

143

.IcleCVzrm xr

y-cv

 

t2 —tJ3 + —d+t«/E

rm
(n—q—l) —(n-q—1> "I

 

 

=(,/(1—r:.)(1—r,:)> _

where, rm, is the correlation between x and an unobserved confounding variable,
rm, is the correlation between y and an unobserved confounding variable, r” is
the correlation between x and covariates z, ry.z is the correlation between y and
covariates z, t is the critical value of t, n is the total number of cases, q is the
number of parameters estimated, and d = t2+(n-q-1). The following table contains
the ITCV scores calculated from the equation for three different types of bullying
behaviors.

As already noted, previous bullying victimization was a strong and
significant predictor of all types of bullying behaviors. For relational bullying,
previous relational bullying victimization was the strongest predictor of relational
bullying. This inference can be altered when the impact of the confounding
variable, i.e., prior bullying propensity, is greater than .114 (ITCV=.114, see Table
19). The robustness to percent bias score (.747) indicates that there should be

about 75% bias to make the relationship become insignificant.

144

mmE o\o 9 wmmSmanom u m .m

 

 

8m. 2:. 2:. Re. 985% 8209880

«8. 80. NB moo. new. BF. «NF. 20. we. 2:. 8.. :o. .880
8m. 8m. 8?. Bo. Em. N2. of. 88. EN. RF. N? 3o. 82.833558 28998.8.
8o. 95. So. 80. m3. at. a: m8. 0.8 238 .6 8980

EN. 8:. B F. o 8. cm F. «8. 88. So. 88388-8:986 so .93

mom. 8 F. 8 F. :o. 258% 830a

as. mmo. N8. «8. 3858 .92: $088+

48. 80. So. So. 9.: Zoo ”835m 8288
NE. 8m. 88. 8o. 8s. 88. wow. 0:. mms. :8. own. E. 3829 8:88-50
:8. ma. :8. 8o. 8. mmm. mmm. Bo. EN. RF. OS. to. 88988 239 83:2.
8m. 3:. NS. «8. 8m. 8: 02. 8o. Em. 8N. RN. 8a. 88558
08. 84. 84. Br. 98. am. :8. mmo. 58. EN. 8N. 8o. 88858; 82:5 .802;
0:. men. RN. So. 8:888; 9.5.3 88>

NE. 86.. m 5. 4:. 8:858; 83:3 .mcozgmm

1 >96 20.x; >O._-_ m Rye REC >U._v_ mm >9¢ >9xk >Q._._

 

823m .8038

mc_>=:m .mn.o>

823m 383%

 

wmcoow >9: .2. mfimh

145

F un-seeking tendency also has a strong significant relationship with
relational bullying. The ITCV is .111, and the robustness to percent bias score
also shows that a strong causal inference can be made (Robustness to °/o bias
=.728)

In terms of verbal bullying behavior, previous verbal bullying victimization
and fun-seeking tendency have strong associations, and it would be difficult to
invalidate them due to the impact of unobserved prior propensity of bullying
behavior (ITCV = .097 and .110 respectively). The inference drawn from the
relationship between the teacher’s moral authority and verbal bullying can be
easily invalidated by the impact of the confounding variable (ITCV = .004), and
the same coercive conformity (ITCV = .003).

For physical bullying behavior, previous physical bullying victimization
and the fun-seeking tendency were found to have strong associations, that would
be difficult to invalidate by the impact of the omitted variable (ITCV = .197
and .093 respectively). The inferences for the relationships of physical bullying
with teacher’s attitude toward bullying and with gender can be easily invalidated
by the confounding variable (ITCV = .001 and .005 respectively).

Overall, each type of prior bullying victimization has the most significant
and non-spurious relationship with each corresponding type of bullying behavior.
Based on the ITCV results, fun-seeking tendency was also found to be a
significant predictor of relational and physical bullying behaviors. Those
relationships are the most certain that is that they are least likely to be spurious
due to prior propensity to bully others.

Finally, to compare impacts of the confounding variable (ITCV) with

146

impacts of covariates on relationships between types of prior victimizations and
bullying behaviors, impact k (= rsz ry-z) 0f each independent variable has been
calculated (see Table 20). Since the prior bullying victimization was one of the
most important predictors of bullying, the comparison is made for prior bullying
victimization as a predictor of primary interest. As mentioned previously, ITCV
indicates the necessary impact of a confounding variable to alter inference
between each type of bullying victimizations and each type of bullying behaviors.
As the results indicate, the prior physical bullying victimization has impact k=.150,
which is large enough to invalidate the inference for relational bullying. Also, it
has impact k=.317, which is also large enough to invalidate the inference for
verbal bullying. The impact magnitude of the prior physical bullying victimization
is three times greater than the impact of the confounding variable for verbal

bullying.

Table 20. Comparison between ITCV and Impact of Covariates

 

Types of Bullying

 

 

 

Relational Verbal Physical

ITCV (Prior bullying propensity)

Relational bullying victimization .114

Verbal bullying victimization .097

Physical bullying victimization .197
Impact k(= rx.zX ryz)

Physical bullying victimization .150 .317

Dominance .004 .000 .001

Attitude towards aggression .011 .014 .018

Fun-seeking tendency .012 .045 .055

Teacher’s attitude: Don’t like .000 .000 .001

Teacher’s moral authority .019 .011 .001

 

147

Table 20 continued

 

Power dynamic

Level of difference-acceptance
Coercion of public self
Parents-peers communication
Gender

Cooperative climate

.019
.066
.007
.016
.002
.012

.033
.004
.000
.025
.071
.016

.024
.005
.000
.034
.062
.006

 

148

V. STRUCTURAL EQUATION MODELING: BASIC MODEL

1. Confirmatory Factor Analysis

Before conducting SEM, the second-order confirmatory factor analysis
was carried out for all 8 constructs and their 34 indicators. Table 21 contains the
standardized regression weights (factor loadings), the squared multiple
correlation coefficients (R2), and the different types of indexes (chi-square, GFI5,
AGFIB, CFI7, and RMSEAB) for all variables.

Three steps were taken to conduct confirmatory factor analysis, 1) basic
CFA, 2) separated individual trait factor CFA, and 3) CFA with variables dropped.
The basic CFA included all variables that were found to be important in the
previous literature. Most factor loadings were in an acceptable range, but there
were several variables with low loading scores. Variable with low loading scores
are dominance, impulsivity, attitude toward aggression, fun-seeking tendency,
parents’ attitude (Don’t like bullying), teachers’ attitude (Don’t like bullying),
coercive conformity, cooperative climate, social disorganization, and peer group
collectivism. It is obvious that three items related to previous bullying
victimization load on a common factor, while the standardized regression weights

for individual tendency (4 items) are near zero. This indicates that separation of

 

5 GFI (the Goodness of Fit Index) tells what proportion of the variance in the
sample variance-covariance matrix is accounted for by the model.

6 AGFI (Adjusted GFI) is a GFI adjusted for the number of parameters in the
model.

7 CFI (Comparative Fit Index) is a result of comparison between the model and
the independence model.

8 RMSEA (Root Mean Square Error of Approximation) estimates lack of fit
compared to the saturated model.

149

the individual tendency factor from previous bullying victimization is necessary.
CFA was conducted again with separated individual trait factors, 1) the
individual traits of victimization, and 2) the individual traits of a tendency to bully.
By doing this, the squared multiple correlation coefficients (R2) increased
from .04, .01, and .04 to .12, .29, and .41 respectively for impulsivity, attitude
toward aggression, and fun-seeking tendency. Overall, the model improved
slightly, but the fit indexes are still not in acceptable ranges (although GFI (.900)
and RMSEA (.052) are in acceptable ranges). However, there were still variables
which had low standardized regression weights, such as dominance. Their
squared multiple correlation coefficients were near zero (ranging from .00 to .07,
see the numbers in bold in the R‘2 column of the separated CFA section in Table
20). Thus, another CFA was carried out with those variables dropped from the
analysis. Overall, although x2 is still significant, meaning that the null hypothesis
of good model fit should be rejected, the model improved greatly (GFI=.936,
AGFI=.917, CFI=.903, RMSEA=.047) and significantly (Ax2=938.257). Thus, this
study uses the measurement model developed from the final CFA for further

SEM.

Table 21. Confirmatory Factor Analysis

 

Basic Separated Dropped

 

Loadings R2 Loadings R2 Loadings R2
DV: Bullying Behaviors

Relational .75 .56 .74 .55 .74 .55
Verbal .94 .88 .94 .88 .94 .88
Physical .72 .52 .72 .52 .72 .52

IV: Individual Traits

 

150

Table 21 continued

 

IV:

Prior relational victim
Prior verbal victim
Prior physical victim
Dominance

Impulsivity

Attitude on aggression
Fun-seeking tendency
Family Interactions
Parents’ attitude: Like
Parents’ attitude: Care
Authoritarian parenting
Domestic abuse
Importance of education

IV: Teacher Interactions

IV:

IV:

IV:

Teachers’ attitude: Like
Teachers’ attitude: Care
Teachers’ intervention
Moral authority

: Peer Interactions

Power dynamics
Difference-acceptance
Coercive conformity
Pseudo friendship
Parental Communication
Parent-Teacher
Parent-Peer

School Climate
Academic standard
Cooperative climate
Moral atmosphere
Parental involvement
Policies, regulations
Community character
Vertical collectivism

.66
.90
.75
—.02a
.19
.12
.19

.10
.29
.65
.68
-.45

.09
.36
.44
.72

.64
.45
.11
.40

.88
.76

.49
.09
.65
.58
.58

-.30

.43
.80
.57
.00
.04
.01
.04

.01
.08
.42
.47
.20

.01
.13
.19
.52

.41
.20
.01
.16

.78
.57

.24
.01
.43
.33
.34

.09

.66
.93
.73
.07
.35
.54
.64

.11

.31

.64
.67
-.45

.11
.36
.41
.75

.69
.44
.10
.35

.88
.75

.49
.09
.65
.57
.58

-.30

.43
.86

.00
.12
.29
.41

.01
.10
.41
.45
.20

.01
.13
.17
.57

.48
.19
.01
.12

.78
.57

.24
.01
.43
.33
.34

.09

.66
.93
.73

.35
.54
.64

.29
.65
.67
-.45

.32
.41
.76

.70
.44

.34

.88
.75

.49
.66
.58

.58

-.34

.43
.86
.54

.12
.29
.41

.09
.42
.45
.26

.10
.17
.58

.49
.19

.12

.78
.57

.24
.43
.33

.34

.12

 

151

Table 21 continued

 

Horizontal collectivism
Social disorganization
Peer group collectivism

Chi-square

Degree of Freedom
sz

GFI

AGFI

CF I

RMSEA

-.78
.26
.26

2851.973

495

760.552* (df=8)

.862
.834
.767
.062

-.79

2091.421
487

l

.900
.878
.841
.052

-.64 .41

.09

1153.164
313
938.257* (df=174)
.936
.917
.903
.047

 

a: Factor loading for dominance in basic CFA was not significant.

2. Structural Equation Modeling

The basic theoretical model was previously presented in Figure 3. The

initial ecological prediction model which will be tested in this section is identical to

the one presented in Figure 3, except that the individual trait factor is divided. into

two factors, prior bullying victimization and individual tendency to bully. This

modification is consistent with the findings in the analysis of the measurement

model (CFA in the previous section). The newer version of the initial ecological

prediction model is presented in Figure 7, and it contains the standardized path

coefficients and R2.

152

Figure 7. Initial Ecological Prediction Model of Bullying Behavior

/ “\\
‘Macrosystwa

\,
(family 89 1,

\»_/ \\ \\

   
    

//\\
@9ka 007
o. 09* \
//——\ "1,, \ \OQ4
Family ‘1 \ 0‘11 =\
/W \ ' 1...
/_\:I -0.25 , _ , . ~
.10 /-0.51 006 0.2.3

,1n/dividual Traits:‘,6 7
\ Victimization A\

I
I

 

-
Peer School
Who/90 21 \\C\Iimat/

153

A. The Initial Ecological Prediction Model for Bullying (ME)

The model fit summary for this basic model is presented in Table 22, in
the row headed “ME Ecological Model.” Since the probability value of the chi-
square test is smaller than the .05 level (1400.114 at df=357), the null hypothesis
that the model fits the data should be rejected. Values on the GFI and AGFI were
acceptable for ME (in excess of .9), and the value on the RMSEA was acceptable

(lower than .05).

Table 22. Model Fit Summary for All Models (N=1238)

 

)8 df sz GFI AGFI CFI RMSEA

 

ME Ecological Model 1400.114 357 .925 .909 .880 .049
Mm Revised Model #1 1193.766 355 Yes“ .934 .919 .904 .044
MFT Full Theory Model 1176.749 355 Yesb .935 .921 .906 .043
Mg Revised Model #2 1163.363 357 Yes° .936 .922 .907 .043
Mm Revised Model #3 1138.318 355 Yesd .937 .923 .910 .042
MM Measurement Model 1153.164 313 .936 .917 .903 .047

 

Note: a) Ax2 = 206.348/ Adf= 2, b) M =17.o17/ Adf= 0,0) sz =13.386/
Adf = 2, d) sz = 25.045 / Adf = 2 (The critical Chi-square value with 2 df is
5.991 (p < .05».

There were five insignificant paths. Table 23 shows the standardized path
coefficients for the paths in different models. Those insignificant paths are
between the macrosystem and bullying, family SES and bullying, school climate
and bullying, family interactions and school climate, and teacher interaction and
prior bullying victimization. Although there were insignificant paths, overall 84% of

bullying behaviors has been explained by the initial theoretical model.

154

Table 23. Standardized Path Coefficients and R2 for All Models

 

 

 

 

 

 

 

 

ME MI MFT Ma Mr3

Bullying Behaviors (R2) .84 .89 .98 .78 .87

Macrosystem -.067

Exo: Family SES .041

Exo: School Climate -.065 .011 -.215***

Mesosystem .1 04*** .025

Micro: Family Interactions .235*** .225*** .234*** .243*** .250***

Micro: Teacher Interactions .258** .228** .250** .144* .324***

Micro: Peer Interactions -1.026** -1.230** -1.532* -.896** -1.029***

Individual Trait: Victimization .516*** .507*** .512*** .516*** .512***

Individual Trait: Tendency 1.666*** 1.841*** 2.160" 1.468*** 1.721 ***
Exo: School Climate (R2) .19 .45 .34 .35 .34

Macrosystem -. 558""r -.61 5*** -.580*** -.567***

Exo: Family SES -.093** -.106** -.104** -.110***

Mesosystem 090* .056 .082*

Micro: Family Interactions -.057

Micro: Teacher Interactions .525*** .449*** -.065

Micro: Peer Interactions .211** .347*** -.O16
Mesosystem (R2) .01 .03 .03

Exo: School Climate .071

Micro: Family Interactions -.17 *** -.172***
Micro: Family Interactions (R2) .01 .01 .01 .01

Exo: Family SES .089“ .090** .090** .089“
Micro: Teacher Interactions (R2) .48 .49 .54 .52 .58

Exo: School Climate .320*** .282*** .340***

Individual Trait: Tendency -.693*** -.700*** -.640*** -.623*** -.655***
Micro: Peer Interactions (R2) .78 .81 .84 .77 .79

Macrosystem .046 . 1 17**

 

155

Table 23 continued

 

Exo: School Climate .050
Individual Trait: Tendency .885*** .899*** .922*** .863*** 355*"

Individual Trait: Victimization (R2) .1 0 .10 .11 .1 1 .11

 

Mesosystem .115*** .115*** .102** .101**
Micro: Family Interactions -.251*** -.241*** -.243*** -.237*** -.235***
Micro: Teacher Interactions .040

 

Micro: Peer Interactions . 143* .098* .094* .104* .107*
Individual Trait: Tendency (R2) .26 .34 .34 .36 .34

Macrosystem .27 *** .258*** .250*** .199***

Mesosystem 211m 231*“ .209“... .210...

Micro: Family Interactions -.512*** -.468*** -.469*** -.471*** -.474***

 

Note: a) The standardized path coefficients, which were insignificant and which
were dropped in the subsequent model estimations, are in Bold.

b) The standardized path coefficients for paths, which were added based on
the results of MI (Modification Index) and theoretical consideration, are-in Italics.

The nomological validity of the ecological prediction model can be tested
by comparing chi-squares from both the ecological model and the measurement
model. An insignificant difference between the chi-square difference indicates
that the theoretical model is valid in accounting for the observed relationships
between the latent constructs (Anderson & Gerbing, 1988). To do the test, the
chi-square for the measurement model was subtracted from that of the ecological
model (1400.114 — 1153.164 = 246.95), and the difference between the two
degrees of freedom was 44 (357 — 313 = 44). This chi-square difference was
significant at p < .05. Thus, the result indicates that the initial ecological model
was not successful in accounting for the paths among the constructs. Therefore,
a modification of the model has been carried out to find a better-fitting model in

156

subsequent analysis steps.

B. Revised Model 1 (Mr1 )

Although model modification is necessary in some cases, there is always
inevitable danger in model modification. This danger stems basically from
implicitly changing relationships between the observed constructs, a practice that
results in a data-based model (Anderson & Gerbing, 1988; Hatcher, 1994; Kline,
1998). Since dropping paths is generally safer than adding paths, one cautious
way to begin the modification process is dropping paths that do not significantly
hurt the model’s fit (Bentler & Chou, 1987). Thus, this study begins the
modification process with dropping insignificant paths. To add paths that seem to
be theoretically and statistically plausible, this study relies on the MI (Modification
Index) based on careful theoretical elaboration. The MI results are presented in
Table 24. The values in the table suggest how much the chi-square values in
each model can be reduced by adding suggested paths between the observed

constructs.

157

Table 24. Results of Modification Indexes from All Models

 

 

 

 

 

 

Exogenous
F1 F2 F3 F4 F5 F6
School Climate 73.641 7.422 5.888
Family Interactions 4.793 6.516 35.264
ME Teacher Interactions 48.199
IT: \flctimization 11.396
IT: Tendency 34.444 28.269
Family Interactions 20.267
M” Teacher Interactions 16.297 6.452
Peer Interactions 4.081 6.445
IT: Victimization 5.933 4.551
Bullying Behaviors 4.021
MFT Mesosystem 19.749
Peer Interactions 4.821 5.581
IT: \fictimizations 5.823 5.050
Bullying Behaviors 4.569 7.587
School Climate 5.170
Mg; Teacher Interactions 4.985
Peer Interactions 6.598 5.659
IT: Victimizations 5.510 4.884

 

Note: a) F1=Macrosystem, F2=SES, F3=School Climate, F4=Mesosystem,
F 5=Family Interaction, F6=lndividual Traits: Victimization.
b) Mls (Modification Indexes), which were selected to add a path, are in

Bold.

Four insignificant paths were first dropped from the model. The path

between school climate and bullying behavior is kept in the next model because

the ecological theory of bullying suggests a direct effect and considers this

relationship to be important. Then, six new paths which seem to be theoretically

valid were selected out of the total 10 MI for addition in MI Revised Model #1.

Those paths are between school climate and the macrosystem (Ml=73.641),

158

school climate and family SES (Ml=7.422), family interactions and family SES
(Ml=6.516), prior victimization and the mesosystem (MI=11.396), the individual
tendency and the macrosystem (MI=34.444), and the individual tendency and the
mesosystem (MI=28.269). The ecological system theory suggests that all levels
of systems interact with each other to influence children’s behavior, and
especially the macro- and the mesosystem have distal but important influences
over children (Bronfenbrenner, 1979, 1989).

Model fit indexes are presented in Table 22, in the row headed “Mn
Revised Model #1 Overall, this model does not do well in accounting for
relationships between the observed constructs, because the chi-square value
(1193.766) is significant at 355 degrees of freedom. However, in SEM, obtaining
a nonsignificant chi-square value is very difficult with an especially large data set,
so many use other indexes as well to evaluate model fit (Hatcher, 1994; Kline,
1998). First, this model significantly improved from the initial ecological model.
The difference in the chi-square test between ME and Mn is 206.348, which is
significant at df=2. In addition, all other indexes indicate the model is a good fit to
the data (GFI=.934, AGFI=.919, CFI=.904, RMSEA=.044). However, there are
still three insignificant paths. They are between school climate and bullying, the
mesosystem and bullying, and the mesosystem and school climate. Overall, this
model explained significantly more variance, than the initial ecological model, in
bullying behavior (R2=.89), school climate (R2=.45), peer interactions (R2=.81),

and Individual traits (R2=.34).

159

C. The Full Theoretical Model (reciprocal relationship of climate, MFT)

In this model, a reciprocal relationship between school climate and some
exogenous variables has been added. Since SEM allows treating all of the latent
constructs as endogenous, and since the ecological system theory suggests a
significant contextual effect over individual behavior and micro- and
mesosystems, the paths between school climate and teacher interactions, school
climate and peer interactions, and school climate and the mesosystem have
been added to estimate model improvement.

Model fit indexes are presented in Table 22, in the row headed “MFT Full
Theory Model.” The chi-square value (1176.748) was significant at df=355,
meaning that the null hypothesis of a good model fit should be rejected. However,
compared to M”, this model improved significantly (Ax2 = 17.017at the same
level of df), and other model fit indexes indicated the model is a good fit to the
data (GFI=.935, AGFI=.921, CFI=.906, RMSEA=.043). Overall, this model
explained significantly more variance than the M”, in bullying behavior (R2=.98),
teacher interactions (R2=.54), and peer interactions (R2=.84).

The results of testing the reciprocal relationship between school climate
and the micro- and the mesosystems suggest that school climate significantly
influences only teacher interactions. Once a reciprocal relationship of school
climate was considered, the bidirectional path between peer interaction and
school climate became insignificant. In the subsequent modification, those

insignificant paths were dropped.

160

D. Revised Model 2 (Mg)

From the previous modification, paths between teacher interaction and
school climate, peer interaction and school climate, school climate and the
mesosystem, and school climate and peer interaction have been dropped. The
MI suggested possible paths between family interaction and the mesosystem
(Ml=19.749) and between the macrosystem and school climate (MI=4.821). Both
paths seem to be plausible based on the ecological system theory. According to
the theory, parents determine and influence types of interactions with children
within the family. Parents also determine levels of parents-teacher
communication and parent-peer communication. Thus, those two paths were
added in Mfg.

Model fit indexes are presented in Table 22, in the row headed “Mrz
Revised Model #2.” The chi-square value (1163.363) was significant at df=357,
meaning that the null hypothesis of a good model fit should be rejected. However,
compared to My, this model improved significantly (Ax2 = 13.386 at df=2), even
thought R2 for all of the endogenous variables have been decreased compared to
MFT. Other model fit indexes indicated the model is a good fit to the data

(GFI=.936, AGFI=.922, CFI=.907, RMSEA=.043).

E. Revised Model 3 (003)

In Mrz, there was one path which was not significant, between the
macrosystem and peer interaction. This path has been dropped in this model.
The MI results suggested expected significant paths between school climate and

bullying (Ml=7.587) and the mesosystem and school climate (Ml=5.170). These

161

two paths, which were dropped in M”, were added back in M3.

Model fit indexes are presented in Table 22, in the row headed “Mr3
Revised Model #3.” The chi-square value (1138.318) was significant at df=355,
meaning that the null hypothesis of a good model fit should be rejected. However,
compared to ~19, this model improved significantly (Ax2 = 25.045 at df=2). In
addition, R2 increased significantly for bullying behavior (R2=.87), teacher
interactions (R2=.58), and peer interactions (R2=.79). Other model fit indexes
indicated the model is a good fit to the data (GFI=.937, AGFI=.923, CF I=.910,
RMSEA=.042). Figure 8 summarizes the revised model #3. Overall, this final
model provided the best model fit indexes and since chi-square tests indicated
that models have been improved significantly, this study retains this model as the

“final” and “best-fitting” model.

162

Figure 8. Revised Model 3, Ecological Prediction Model of Bullying Behaviors

, «If ‘\»\,

\

/.. \
l.\Famin SES/L,
.\\‘//

\
0 . 097 II,
:5 Oi II1

E _0 23 Family \1, 1,
<\\ Lteractlons/ ‘1,-0.1 1

(Ifidividual TraitIs\.'1,‘\//0.51 17-0. 47} \ I11,

\ VIctimization \\ \L,
\\—/;< \\ -0.17 \
0.1 I \7

// \ ‘1,\0. 25
\1,,, \\ l I;@\\ ‘1 \

 

7 / 7.
011:, , Mesosystem \g 2' I87,
; \-.-_ - 0.08 1/ \
0121/ 581 032 $ Bullying
Teacher \, //1 T /
’ ’ gkteractlons/x/ ‘1 /' -1 ,03
/ 1 1,7,2, ' ./ \ / 1‘ -021
a] -O. 65 I
1 1 / / l f
//—\it /1.):1,./ Ali 0 34\ ,_.II_ 1.34
,11/dividual Traitsi‘ée H/ TPeerV \YV School \
”le 086 Wadi Ion \CLmiIE/
-0.57
0.2\\\ O'IZI \
\/\ /
Macrosyste/

163

VI. STRUCTURAL EQUATION MODELING: MULTIPLE GROUP ANALYSIS

In this section, multiple subgroup analysis was attempted to see if the
best fitting model identified is consistent across different subgroups, which were
categorized based on duration and visibility of bullying behaviors. As mentioned
in Chapter 2, bullying behaviors vary according to duration and visibility of the
behaviors. Specifically, physicality/severity of the behaviors is found to be related
to levels of persistence of the behaviors in Korea (KEDI, 1998; Kwon, 1999). In
addition, more serious bullying behaviors openly take overt forms of advertising
bullying behaviors to influence other peer groups (KEDI, 1998; Kwon, 1999).
Thus, it will be useful to see whether the ecological prediction model holds
consistent across different groups of bullies.

It has been suggested that there are possibly three types of bullying
behaviors in terms of duration and visibility, 1) chronic visible (wangtta group), 2)
chronic-invisible and non-chronic-visible (euntta group), and 3) non-chronic-
invisible bully groups (tta group) (see Chapter 2, Section 1. Definition of Bullying).
Thus, this study divides data into three subgroups accordingly. Each subgroup
has the same dependant variables, i.e., frequency of relational, verbal, and
physical bullying behaviors. Thus, this subdivision takes all of the severity,
duration, and visibility dimensions into account.

For duration, since persistence of bullying behaviors usually lasts a
semester, q52 (‘Have you been doing the above things to them during the entire
semester?) has been selected to categorize the data into chronic vs. non-chronic

groups. Simple frequency statistics showed that 159 students said they bullied

164

others for the entire semester. For visibility, a scale developed from questions 54
— 58 was used to categorize the data into covert vs. overt groups (mean=14.02,
range=5~25, see Table 8 and 10).

For the first step of multiple group analysis, data were divided into three
data sets. Thus, there were separate data sets for overt chronic bullies (wangtta
group, n=81), covert-chronic and overt-non-chronic bullies (euntta group, n=549),
and covert-non-chronic bullies (tta group, n=608). Second, the measurement
model used in the previous structural equation modeling was created for the
three groups, and the model was estimated. The free model of the measurement
model yielded the chi-square value 1998.470. The value is significant at df=939.
Next, the factor loadings (lambdas) were constrained across three groups. The
constrained model yielded the chi-square value 2075.697. The degree of
freedom was 975. The X2 different of these two model shows significant
difference (Ax2=77.227, df=36). This means that the measurement model is not
assumed to be equal across three groups of bullies. Consequently, further

multiple group analysis was not performed.

165

CHAPTER 5

DISCUSSION AND CONCLUSION

I. Summary and Discussion
II. Limitations and Recommendations
Ill. Conclusion and Policy Implications

I. SUMMARY AND DISCUSSION

1 . Multivariate Analysis

The multivariate analysis revealed that overall roughly 30 — 40 percent of
relational, verbal, and physical bullying behaviors have been explained by all of
the independent variables in the ecological model. Age was not found to be an
important predictor of bullying behaviors in South Korea. Unlike previous studies
of bullying in Western countries (Espelage et al., 2003; Harris & Petrie, 2003;
Nansel et al., 2001; Pellegrini & Long, 2002; Seals & Young, 2003), it seems that
a peak age of bullying for South Korean students is different from that of students
in other countries. Consistent with the previous findings (Ah et al., 2005; Baldry,
2003; Bentley & Li, 1995; Bosworth et al., 1999; Boulton et al., 2002; Kim, 2001;
Lee & Kaok, 2000; Naito & Gielen, 2005; Park, 2002), individual traits were found
to be the important predictors of bullying behaviors. More specifically and
interestingly, this study found that each type of prior bullying victimization
(relational, verbal, and physical) is the most important predictor of the same type
of bullying behavior. In addition, prior physical bullying victimization was

predictive of all three types of bullying behaviors. Among the individual

166

tendencies, dominance, attitude toward aggression and fun-seeking tendency
were found to be important predictors of all types of bullying behaviors. Unlike
the previous studies (Bosworth et al., 1999), impulsivity was found to be non-
predictive of bullying behaviors. It is also worth noting that this study empirically
supports Naito and Gielen’s (2005) suggestion that a fun-seeking tendency is an
important influence on children bullying others.

Only a few microsystem variables were found to be significant predictors
of bullying behaviors. Especially, unlike previous findings (Baldry, 2003; Batsche
& Knoff, 1994; Bowers et al., 1994; Christie-Mizell, 2003; Espelage et al., 2000;
Espelage & Swearer, 2003; Kim & Lee, 2000; Naito & Gielen, 2005; Shields &
Cicchetti, 2001; Stevens et al., 2002), family interactions, including parents’
attitude, parenting, domestic abuse, and importance of education, were found to
be insignificant predictors of any types of bullying behaviors. However, this
finding is altered somewhat by considering the results of the structural equation
modeling analyses. The effects of the family interactions are mediated by other
variables. This result seems to be more logical than the previous finding. Since
bullying occurs in school environments, family interactions would not have direct
impacts. Rather the family interactions should be mediated by children who
function as bridges between family and school (see Bronfenbrenner, 1979).

Unlike family interactions, teacher and peer interactions could have direct
impacts on bullying behaviors. However, this study found just weak effects of
teacher interactions (only teacher’s attitude: Don’t like bullying). The peer-related
variables --- power dynamic, difference-acceptance, and coercion of public self --

- were found to be important predictors of relational and verbal bullying behaviors.

167

Of the mesosystem variables, only parent-peer communication was found
to be an important predictor of bullying behaviors. The more severe the bullying
the greater the communication between parents and peers about bullying. This
finding might occur because parents talk to their children’s friends about bullying
in reaction to bullying of their child.

This study expected that there would be no relationship between family
SES and bullying behaviors, if the relationship was examined by simple linear
regression. Based on the literature review, this study expected that there would
be a U shape distribution of SES with bullying behaviors. However, both
scatterplot and mean difference tests showed that there was no U shape
distribution, and both bivariate and multivariate analyses suggested that there
was no direct relationship between family SES and bullying behaviors. However,
family SES was found to be indirectly related to bullying behavior through the
micro-, meso-, and exosystems. This finding will be further discussed in the next
section.

Most indicators of school climate were found to be insignificant in
predicting bullying behaviors, except the positive relationship between a
cooperative climate and verbal bullying. Like family interactions, school climate
also seems to be mediated by other factors, such as teacher and peer
interactions, and individual traits. This null finding is also plausible in that there
are children who do not bully others even though they are in the same school
climate habited by bullies. In addition, Bronfenbrenner (1979) suggested that an
exosystem is a playground for interactions between the individual and the micro-

and the mesosystem. As evidence, school climate was found to be mediated by

168

teacher interaction in SEM, although there was a weak but significant negative
relationship with bullying behaviors.

Finally, unlike the previous arguments about the effects of community and
cultural characteristics on bullying behaviors, this study found a null relationship
between them. According to Bronfenbrenner (1979), the macrosystem has
indirect distal influence over individual behaviors. Thus, it is also possible that the

effect of the factors in this level could be mediated by other constructs.

2. Structural Equation Modeling

Overall, the structural equation modeling of the ecological system theory
for bullying behaviors was found to be successful in accounting for multiple levels
of ecological variables and their relationships with bullying behaviors. Eighty
seven percent of bullying behaviors has been explained by the final ecological
model, compared to lower R2 values in multivariate analyses (R2=.291/.416/.361
for three types of bullying behaviors). Many insignificant relationships in the
multivariate regression were found to be important significant paths in SEM. In
addition, the contextual variables, such as school climate and the macrosystem,
were found to have distal indirect effects over individual bullying behaviors.

Prior bullying victimization and an individual tendency to bully have
significant strong direct effects on bullying behaviors. Also these factors mediate
the effects of family interactions and the mesosystem. This finding is consistent
with previous research results about effect of victimization on involvement in
delinquency. For instance, testing the general strain theory suggested by Agnew

(1981 ), a study found that violent victimization significantly increased later

169

involvement in delinquency (Hay & Evans, 2006). In this study, victimization was
viewed as a cause of delinquency, rather than an outcome. The relationship
between victimization and psychological problems, e.g., stress, also has been
investigated by several studies. These studies found that victimization has a
long-lasting psychological effect on children’s behaviors (Herrero, Estevez, &
Musitu, 2006; Wareham & Dembo, 2007). As Bronfenbrenner (1979) noted, this
study also found that individual tendency directly influenced interactions with
teachers and peers in schools. It is also worth noting that individual tendency has
been influenced by community and cultural characteristics (macrosystem).

As suggested by Bronfenbrenner (1979), the microsystem was important
in shaping children’s behaviors. Family interactions significantly influenced
individual victimization and the tendency to victimize. Also it has a significant
direct impact on bullying behaviors. Especially, family interactions are associated
with family SES, and family interactions are associated with parental
communication with teachers and peers. This result supports the argument that
the exosystem functions as a playground for the micro-and the mesosystem. The
findings support the conclusion that teacher interactions have been influenced by
an individual tendency to bully, the mesosystem, and school climate. Peer
interactions are heavily explained by individual tendencies to bully as well as the
macrosystem. The variable, peer interactions, has a direct positive relationship
with individual victimization, but a direct negative relationship with bullying
behaviors. This means, consistent with the previous findings, that children who
are more power-oriented, more coercive and less tolerant are more likely to bully

others, but less likely to be bullied by others.

170

Parental communication with teachers/peers had a direct statistical effect
on school climate. This is consistent with the conclusion that parental
involvement has a strong but indirect impact on children’s behaviors in schools.
However, as the ecological theory suggests, there was no direct relationship
between the mesosystem and bullying behaviors.

Finally, this model also suggested the importance of contextual factors as
influences on bullying behaviors. School climate had strong negative relationship
to bullying behaviors, and strong positive relationship to teacher interactions.
However, school climate did not have significant statistical effects on peer
interactions. This may be because there is a gap between students’ life and the
school system including teachers and other school climate variables. Korean
scholars have argued that although school climate has an indirect influence over
students’ behaviors, Korean students often perceive teachers and school policies
and rules as irrelevant to their lives (Park, 2003; Shin, 2000). They argued that
traditional educational philosophy, which emphasizes guiding students to become
good human beings, is not accepted by contemporary teachers and students, but
an emphasis on academic achievement has taken the place of the traditional
philosophy. Thus, they argued, students think that schools are the places where
they learn knowledge rather than the places where they mature and develop.
Consequently, it may be possible that Korean students think that school climate,
including academic standards, cooperativeness, and any anti-bullying policy,
does not directly affect what is actually going on among their friends.

A recent study also revealed a very similar result about partial effects of

school climate on students’ behaviors. Gottfredson and his colleagues found that

171

psychosocial climates “did not influence student victimization or delinquent
behavior” (Gottfredson, Gottfredson, Payne, & Gottfredson, 2005). Their
structural equation model did not produce significant paths from the psychosocial
school climate factor, which included organizational focus, morale, planning, and
administrative leadership. However, this study found an indirect path from school
climate to bullying through teacher interaction. By adding reciprocal relationship
between school climate (a type of exosystem) and microsystems, this study
found that school climate could influence children’s behaviors indirectly.
Likewise, community characteristics were found to be important indirect
influences on bullying behaviors. As Bronfenbrenner (1979) suggested, this study
empirically found that collectivistic community characteristic has a distal indirect
influence on children’s bullying behavior through individual traits, microsystem,
and exosystem. This finding provides empirical evidence for Naito and Gielen
(2005). Although different levels of collectivism have been compared across
different school districts, this study could not compare effects of collectivistic vs.
individualistic characteristics on bullying behavior due to the sampling frame.

Future study will benefit from conducting comparative study on this regard.

172

ll. LIMITATIONS AND RECOMMENDATIONS

There are several limitations in this study. First, although this study
attempted to overcome the limitations of cross-sectional data, there is still a
danger of drawing causal inference based on the results. For example, this study
expected that the mesosystem (parental communication with peers) would be
negatively related to bullying behaviors. However, the results suggested that
there was a significant positive relationship between them. Based on this result, it
is hard to say whether parental communication with peers increased bullying
behaviors. Instead, it would be more plausible to say that parental
communication has increased to react to increased bullying behaviors. In
addition, the R-square values were much greater than is normally found in
empirical studies. One possible reason for this high level of R-square values is
that the data are not longitudinal, and the R2 is inflated by reciprocal relationships
between systems. Thus, it is recommended that future ecological study of
bullying should collect longitudinal data for making stronger causal inferences.

Unlike previous studies on bullying in Western countries, it seems that a
peak age of bullying for South Korean students is different from that of students
in other countries. Unfortunately, this study included only middle school students,
so that finding a peak age for bullying behaviors among South Korean students is
impossible. In addition, because of limited resources, the current study does not
investigate the escalation of bullying behaviors into anti-social and violent
behaviors during puberty or early adulthood. Thus, it is recommended that further

study might benefit from including elementary and high school students as study

173

subjects.

Finally, this study intended to investigate whether the ecological model of
bullying was consistent across different groups of bullies, wangtta, euntta, and tta
groups. However, the initial step of multiple group analysis revealed that the
measurement model developed in this study was not equal for the different
groups, meaning that further comparison between path coefficients in models in
each group does not yield valid results (Kline, 1998). This result indicates that a

better measurement model is required to do further study of group differences.

174

llI. CONCLUSION AND POLICY IMPLICATIONS

Although bullying has not been a major research topic in South Korea as
well as the United States, it has been studied to some extent for several decades.
However, lack of empirical evidence for the ecological approach to bullying in
schools has been generating doubt about the utility of the ecological model. The
present study is unique in that it applies and tests the ecological system theory
for bullying. The results, overall, suggest the utility of the ecological approach and
that the ecological model of bullying accounted for a high portion of variance in
bullying behaviors. Specifically, this study found that individual traits, notably prior
bullying victimization, are most important in predicting bullying behaviors. The
analysis results also suggested that children’s immediate surroundings, including
family, teachers and peers, have influence on children’s behaviors. As the
ecological system theory suggested, school and social contexts also have
important and significant impacts over children’s behaviors.

Overall, these findings may provide some ideas for effective interventions
into schools and families. First, early evaluation of children’s school experiences
as well as their tendency to bully others may help parents and teachers develop
effective intervention to limit victimization and the transition from victim status to
offender status. However, emphasis should not be given only to individual
characteristics. According to this study, parents and their involvement and
teachers and their attitude toward bullying were also important predictors of the
bullying phenomenon. A recent antibullying intervention study has been carried

out in the Netherlands (Fekkes, Pijpers, & Verloove-Vanhorick, 2006). This study

175

found that a school-based antibullying policy could reduce bullying behavior, but
the policy should be continued to maintain the lowered level of bullying behavior.
Furthermore, an extensive review of previous antibullying intervention programs
found that school-based interventions with multiple disciplines do significantly
reduce bullying in schools. Thus, a school policy that formally increases parental
communication with teachers and school administrators would benefit students.
Educational and/or training opportunities for teachers about bullying could also
benefit students. Above all, the most important finding from this study is that our
children, as a center of their world, are influenced by all of their surroundings, so
that, instead of blaming deviant children, the rest of the world should pay more

attention in developing healthier environments for our future.

176

APPENDICES

177

 

High Level Social
Disorganization Areas
9859

¥H$

9

9
3
¥
9

g
50
E

0

LL.

Co

“fl

C

ox
50

mo
0:
E

at

Co
E

8999

MEMWE
5992

lo
2:.

a:

:5
%

 

Medium Level Social
Disorganization Areas

W¥W¥
W$MWE

LI

A._
m_.
u:

JXIE

W

79
E

NIL

E
9
.E
.J

K

We
._OD

”a
J

I

Mo
Io

IE
.5

.od
4

O

as
J

X

mo

0

If

.E
a

E
E
urn

.1._

.A

[PF

10
IE
.J

K

 

Disorganization Areas

Low Level Social

 

 

 

 

 

 

Appendix A. 232 Korean Governmental Districts Categorized into Three Levels of

Social Disorganization

mo

.Ir

E

a:

 

 

8
.58

IE

 

mo
.8

If

.E

.l

L
X

I...

1401' A}.

D
EM

g

g

 

 

 

 

 

 

 

 

Aa
ma

55

 

Ll

.XO
A:
E

a

 

Ir

3
9
E

1._

I

 

¥
E

J]:

K;

 

g
2
5

1

X._

 

L

La
__0
.E

.1._

_A

 

 

 

 

 

 

we
an
as
J

.A

 

?H

13

1._

 

8
2..
ud

.1._

I

 

A‘I%E%

178

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

S
a
m
MA
mm
8.9. H
mm imfiwigowaozuritbr 2. 2.9mrmouTMouTbTuTwlmTwamofl 9.20:5
Lmv ulnayoyopoflmwowmwl aw fiofiowamaxaéfiuxgoeacamooza #921
mm 9059905686th 6 tttt???????caflfl 17.17.71
HD MMHMMuTuTuTuTuT H__I uTuTuTuTWUnutuwnUnWWfioflofio fiofiofio
la
mm
0A
Sn
9.0
Vt
mm. 1 I 9 WT
mmwozaozaExTEomthmfl mm woumxnmxwfiwmw mflwomoifilw 3W5:
.m moca E a: m mo 8 :1 fie .20 Flo W M We W162 E m n 94 go 8 .5 we 1:2 .52 8 we
amnisxagxuxagxuxaxbca .a anaflzlwlwl flflflflflflmz mewumz
M D W 4.6 4.6 4.6 we 4.6 we we we we 4.6 who we we fio 4.6 we 4.6 4.6 we 4.6 we fie fie 70 7o 70 70
S
a
m
mA
can
o.m
sm
d.‘ ILI I LP LFLI ..F
mm 56858588094? .8 amzmomz mocowwMofldExT H.988
Lmu 870989229999 9 199:9 McMamamem 79:8
nwom E E 2.. a: fix. a: a: 22. Hi 25 5:. EL. .5 .E .6 IE .6 IE .5 .E W W I7I we .%
L D m m fie fio fie we 4.6 we 4.6 we fie 4.6 4.6 fie 4.6 we we 4.6 we we we fie 4.6 70 7o

 

179

 

 

Low Level Social
Disorganization Areas

Medium Level Social
Disorganization Areas

High Level Social
Disorganization Areas

 

 

 

 

 

 

 

0:1 07'.
0:1

 

9999 9999 9999
9999 9999 9999
9999 9999 999:
9999 929 9919
9999 9 9

 

* Note: 28 districts are omitted from this categorization due to missing information
about population mobility and density.

180

 

Appendix B. Survey Questionnaire - Students

Questionnaire about School Life

 

 

This survey is about things that happen in your school, and the researcher wants
to understand how kids treat each other in school.

This is an anonymous questionnaire. This means that you don’t have to write
your name or let us know who you are. The results of this survey will be used
solely for academic research purposes, and any information collected from this
survey will be kept confidential.

There is no right or wrong answer. If you are not sure of the answer, please do
your best to select one of the choices. However, if you do not want to answer a
question for some reason, that is OK.

If you have any question regarding this survey, or if you want to receive results of
this study, please contact this researcher with the following contact information:

Chang-Hun Lee

School of Criminal Justice

Michigan State University

560 Baker Hall

East Lansing, MI 48823

Tel: 517-525-2827, Email: Leechan7@msu.edu

Thank you for your participation.

 

I. The following questions ask about your demographic information.

1. Are you a boy or a girl? 1) boy 2) girl
2. What is your grade level? 1) 7th 2) 8th 3) 9th
3. How old are you? ( )-year-o|d

4. How do you think of your family’s socioeconomic status?
1) Upper class 2) Upper-Middle class 3) Middle class 4) Lower-Middle
class 5) Lower class

181

 

For the following two questions, select on of the following options.

 

 

 

 

 

 

 

 

 

 

 

A Unemployed
B Skilled worker, Group 1 Janitor, Daily Workers (e.g., construction worker),
wage worker Handyman

C Group 2 Carpenter, Cook, Employed Farmer, Hair-stylist,
Store Clerk

D Group 3 Self-employed Farmer, Skilled industrial worker,
Taxi or bus driver

E Self-employer Owner of wholesale / retail store, Owner of
restaurant, Self-employed taxi driver

F White-collar worker Businessman, Teller, Low-ranked government
officials, Elementary school teacher

G Manager Middle or high school teachers, Nurse, High skill
industrial worker, Small company owner,
Managers in company, Middle-ranked government
officials, Owner of Farm

H Professional Medical doctor, Judge, Prosecutor, Defense

attorney, Professor, Owner of big company,
Minister, Artist, High-ranked government officials

 

 

 

House wife

 

 

5. What is your father’s job? (

6. What is your mother’s job? (

7. What is your father’s educational level?
1) No education
2) Elementary school graduate
3) Middle school graduate

4) High school graduate

5) 2 Year College graduate
6) 4 Year University graduate
7) Graduate school or more

182

 

8. What is your mother’s educational level?
1) No education
2) Elementary school graduate
3) Middle school graduate
4) High school graduate
5) 2 Year College graduate
6) 4 Year University graduate
7) Graduate school or more

9. What do you think is the total amount of your parents’ monthly income?

1) 500,000 won or lower

2) 500,000 — 1,000,000 won

3) 1,000,000 — 1,500,000 won

4) 1,500,000 - 2,000,000 won

5) 2,000,000 - 2,500,000 won

6) 2,500,000 — 3,000,000 won

7) More than 3,000,000 won

10. What is your average grade level from your last mid-tenn examination?

1) 90 - 100

2) 80 - 90

3) 7O — 80

4) 60 — 7O

5) 50 — 60

6) 40 — 50

7) 3O - 40

8) 3O — or lower

11. What is your level of academic achievement in your class?

1) Upper 10% of your class
2) Upper 20% of your class
3) Upper 30% of your class
4) Middle of your class

5) Lower 30% of your class
6) Lower 20% of your class
7) Lower 10% of your class

183

12. What is your academic rank in your class from the last mid-term exam?

( )

II. The following questions ask about experiences in school

 

I have experienced the Never 1 or 2 Three Four More
following things in my class in times times times
this school year.

(tick (V) for each category)

 

 

 

 

 

 

Relational:

 

13. rumors about me

 

14. being purposely left out of
things

 

15. being ignored

 

16. being left out of
conversations

 

 

 

 

 

 

Verbal:

 

17. being teased

 

18. being taunted

 

19. being threaten

 

20. others called me names

 

21. others made sexual
comments about me

 

22. others talked about my
physical defects

 

 

 

 

 

 

Physical:

 

23. kicked

 

24. hit & punched

 

25. cut with sharp objects

 

26. deliberately pushed

 

 

27. gotten bones broken

 

 

 

 

 

 

184

 

 

Duration:

 

 

 

 

 

 

 

 

 

 

 

 

agree or disagree?

28. How long have you been Less For 2 For 4 For 2 For 4 More

experiencing the above than 2 weeks weeks months months

things? weeks ( ) ( ) ( ) ( ) ( )
( )

29. Have you been 1) yes 2) no

experiencing the above things

during the entire semester?

30. Have you experienced 1) yes 2) no

these things in an earlier

grade?

Visibility:

For the different things you Disagree Somewhat neutral Mostly agree

have experienced, do you disagree agree

 

31. They have done the above
things so they are hidden.

 

32. They have done the above
things so others see them.

 

 

 

 

 

 

 

 

The following questions ask about your general actions in school.

D (Disagree), SD (Somewhat disagree). N (Neutral), MA (Mostly Agree). and A (Agree)

 

Tick (V) for each category:

DSDN

MAA

 

33. I try to treat all my classmates equally.

 

34. I try to be as open as I can to all my classmates.

 

 

classmates.

35. I want to be considered as a fair person by my

 

 

 

 

 

 

IV.

The following questions ask about actions in school.

 

mave done the following

1 Never I1or2

Three

Four

I More 1

 

185

 

 

things to a classmate in my
class in this school year.
(tick (V) for each category)

 

times

 

times

 

 

times

 

 

Relational:

 

36. spread rumors

 

37. purposely left a classmate
out of things

 

38. ignored a classmate

 

39. left a classmate out of
conversations

 

 

 

 

 

 

Verbal:

 

40. teased a classmate

 

41. taunted a classmate

 

42. threatened a classmate

 

43. called a classmate names

 

44. made sexual comments to
make a classmate
uncomfortable

 

45. talked about physical
defects

 

 

 

 

 

 

Physical:

 

46. kicking

 

47. hitting & punching

 

48. cutting with sharp objects

 

49. deliberately pushing

 

50. breaking bones

 

 

 

 

 

 

Duration:

 

 

51. For classmates you did
these things to the most, how
long have you been doing the
above things to them?

 

Less
than 2
weeks

( )

For2
Weeks

For4
weeks

For2

For4

months

months

(

 

)

 

(

)

 

(

)

 

(

)

 

More

 

186

 

 

52. Have you been doing the
above things to them during
the entire semester?

1) yes

2) no

 

53. Did you do these things to
them in an earlier grade?

 

1) yes

 

2) no

 

Visibility:

 

For the different things you did,
do you agree or disagree?

Disagree

Somewhat
disagree

Neutral

Mostly
agree

Agree

 

54. I try to hide my behavior
from other students.

 

55. I don’t care what other
students think about my
behaviors.

 

56. I try to show my behaviors
to other students because they
like what I do.

 

57. l have done the above
things so others see them.

 

58. l have done the above
things so they are hidden.

 

 

 

 

 

 

Involvement:

 

59. I never have been involved
with the above things in this
school year.

 

60. l have never done the
above things, but I don’t feel
bad for the person who they are
done to.

 

61. I have never done the
above things, but I think I
understand why others do such
things.

 

 

62. l have initiated the above

 

 

 

 

 

 

187

 

 

 

things, and am fully involved
with such things.

 

 

 

 

 

 

V. The following questions ask about reasons for the actions.

D (Disagree), SD (Somewhat disagree), N (Neutral), MA (Mostly Agree), and A (Agree)

 

Tick(V)for each category: D SD N MA A

 

Individual Traits: Dominance

 

63. It is difficult to refuse when my friends ask me to do
a favor.

 

64. I usually try to avoid confrontation even ifl am right.

 

65. I usually try to observe rules and regulations.

 

 

 

 

 

 

66. I hate to lose in an argument.

 

Individual Traits: Impulsivity

 

67. It is difficult for me to sit still during the class.

 

68. I start things but have a hard time finishing them.

 

69. I do things without thinking or planning.

 

70. I need to use a lot of self-control to keep out of
trouble.

 

 

 

 

 

 

Individual Traits: Attitude towards aggression

 

71. If I walked away from a fight, I’d be a coward.

 

72. It is okay to hit someone who hits you first.

 

73. If a kid teases me, I usually cannot get him/her to
stop unless I hit him/her.

 

 

 

 

 

 

74. If I refuse to fight, my friends will think I’m afraid.

 

Individual Traits: Fun seeking tendency

 

75. I did the above things to the person for fun.

 

76. My friends enjoy it, because they think I am having
fun with them.

 

 

77. It is fun to watch Wangtta students go through
these things.

 

 

 

 

 

 

188

 

 

 

78. There is nothing wrong with these things because
they are just for fun.

 

Microsystem variables: Family interactions

 

Perceptions of parents’ attitude to wards bullying:

 

79. I think my parents do not like me teasing other
students.

 

80. I think my parents do not like me hitting other
students.

 

81. I think my parents do not like me making fun of
other students.

 

82. I think my parents do not care about me teasing
classmates in school.

 

83. I think my parents do not care about me hitting
classmates in school.

 

84. I think my parents do not care about me making fun
of classmates in school.

 

Perceptions and experiences of authoritarian parenting:

 

85. My parents use physical discipline for punishments.

 

86. I think my parents want to have control over almost
every aspect of my life.

 

Child’s experiences and witness of domestic abuse:

 

87. My parents sometimes hit me.

 

88. I saw my father hit my mother at home.

 

89. I saw my mother hit my father at home.

 

90. I think my parents sometimes verbally abuse me.

 

Importance of education:

 

91. I think my parents emphasize educational
achievement too much so that I feel too much
pressure.

 

92. I think my parents believe that academic success is
the most important thing in my life.

 

Microsystem variables: Teacher interactions

 

 

Perception of teachers’ attitude towards bullying:

 

189

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

93. I think my teacher does not like me teasing other
students.

 

94. I think my teacher does not like me hitting other
students.

 

95. I think my teacher does not like me making fun of
other students.

 

96. I think my teacher does not care about kids teasing
classmates in school.

 

97. I think my teacher does not care about kids hitting
classmates in school.

 

98. I think my teacher does not care about kids leaving
out classmates in school.

 

Perception of efl'ectiveness of teacher’s intervention:

 

99. I think my teacher is not good at preventing kids
from teasing, hitting, and ostracizing their classmates
in school.

 

100. I think my teacher is not aware of what is going on
among students, especially whether there is ostracism,
teasing, hitting, and other ways of getting at a
classmate.

 

Teacher’s moral authority:

 

101. I don’t care about what my teacher says about
morality.

 

102. I think my classmates do not care about what my
teacher says about how classmates treat each other.

 

103. I think my teachers’ moral standard is good for
their generation, not for my generation.

 

Microsystem variables: Peer interactions

 

Power dynamic:

 

104. I believe getting power over other kids is important
to survive in school.

 

105. I respect those classmates who have their own
power to control others in the class.

 

 

106. I think it is OK that there is a group of students

 

190

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

who have power to control other classmates.

 

107. I believe physical strength is important to have
since it gives me power to control other classmates.

 

Microsystem variables: Peer interactions

 

Level of difference-acceptance:

 

108. My friends do not like to talk and hang out with
those who are different from them in terms of clothing
style, hair style, and other things. '

 

109. I do not like to talk and hang out with others who
behave differently and who have different tastes in
hobbies, clothing style, etc.

 

110. I would not have my current friends in school if we
could not find any similarity among us.

 

111. I think my close friends will not accept a new friend
if he/she doesn’t share any similarity with us.

 

Microsystem variables: Peer interactions

 

Coercive conformity:

 

112. I believe that to remain friends, it is important to
behave according to how my friends think of me.

 

113. I really don’t like to see someone act like they are
beautiful, kind, or smart.

 

114. I think, if someone pretends to be beautiful, kind,
or smart, my friends will not hang out with him/her.

 

115. I really want to correct someone’s behavior if
he/she is not acting the way others think he/she
should.

 

Microsystem variables: Peer interactions

 

Pseudo friendship:

 

 

116. I think my friends in class will not help me if I am in
very serious trouble, such as severely banning others,
damaging others’ property, or being arrested for my
actions.

 

191

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

117. I think my friends in class will not help me if I am
bullied by other students in the class.

 

118. I think having friends who are bullying others will
help me not to be bullied.

 

119. I think my friends in class will not protect me from
other kids because they could be victimized.

 

 

 

 

 

 

N (Never), 0 (Once), A (A few times), M (More than a few times), V (Very often)

 

How many times did the following things happen this N O A M V
school year?
Tick (V) for each category:

 

 

 

 

 

 

Mesosystem variables: Parents’ communication with teachers

 

120. How often in this school year did your parents and
teachers talk to each other to share information about
students teasing others?

 

121. How often in this school year did your parents and
teachers talk to each other to share information about
students hitting others?

 

122. How often in this school year did your parents and
teachers talk to each other to share information about
students making fun of others?

 

 

 

 

 

 

Mesosystem variables: Parents’ communication with peers

 

123. How often in this school year did your parents and
friends talk to each other to share information about
students teasing others?

 

124. How often in this school year did your parents and
friends talk to each other to share information about
students hitting others?

 

125. How often in this school year did your parents and
friends talk to each other to share information about
students making fun of others?

 

 

 

 

 

 

 

D (Disagree), SD (Somewhat disagree), N (Neutral), MA (Mostly Agree), and A (Agree)

 

192

 

 

Tick (V) for each category:

 

 

SD

 

 

MA

 

Exosystem variables: School climate

 

Academic standards & General Image of School:

 

126. I think my school sets academic standards very
high so that I should feel proud of it.

 

127. Overall, I am proud of being a part of my school.

 

128. I think my school values academic achievement.

 

 

 

 

 

Competitive vs. cooperative school climate:

 

129. In my school, winning is a more important goal
than cooperation among students.

 

130. In my school, academic success is more valued
than good friendships.

 

 

 

 

 

Students- Teachers Cohesion

 

131. Teachers in my school do care what we are doing
and how we are doing in school.

 

132. I feel that my teacher is a very close friend, so that
I can bring any personal problem to him or her.

 

133. Teachers and students in my school work very
closely to make our school better.

 

 

 

 

 

Perception of moral atmosphere:

 

134. I don’t do smoking, stealing, fighting, or teasing
because I am afraid of being punished by other
students.

 

135. In my school, students have a chance to get
together and discuss problems and solutions for the
problems.

 

136. In my school, wrong doings are punished
appropriately.

 

 

 

 

 

Parental involvement in school boards:

 

137. I think my school gets a lot of parental
involvement in school affairs.

 

138. I think my parents have a say in decision making
about school policies, rules, and programs.

 

 

 

 

 

 

School policies, rules, programs:

 

193

 

 

 

 

 

 

 

 

139. I think my school clearly set forth anti-bullying
policies and rules.

 

140. I think my school is eager to develop anti-bullying
programs.

 

141. I think my school’s anti-bullying efforts are useless
and not helpful for prevention of bullying.

 

 

 

 

Macrosystem variables:

 

Individualism: (VI: Vertical Individualism, HI: Horizontal Individualism)

 

142 (VI). In my neighborhood, it is important to win in
any competition.

 

143 (VI). In my neighborhood, it is important to do
better than any others in every thing.

 

144 (VI). I enjoy being in a very competitive
environment.

 

145 (VI). I work harder when I see others achieve
better than I do.

 

146 (HI). In my neighborhood, it is important to pursue
my own individuality.

 

147 (HI). In my neighborhood, I rely on myself rather
than others.

 

148 (HI). It is important for me to develop an
independent personality.

 

 

 

 

149 (HI) I am proud of having my own personality.

 

Collectivism: (VC: Vertical Collectivism, HC: Horizontal Collectivism)

 

150 (VC). In my neighborhood, it is important to
respect decisions made by the group.

 

151 (VC). In my neighborhood, it is important to
respect the majority’s opinions in the group.

 

152 (VC). I surrender my interests to group interests.

 

153 (VC). It is important for me to support my family
even if I have to sacrifice myself.

 

154 (HC). In my neighborhood, I will be happy when
others around me are happy.

 

 

155 (HC). In my neighborhood, it is my pleasure to be

 

 

 

 

194

 

 

 

 

 

 

 

 

with others.

 

156 (HC). It is important for me to get along with
others.

 

157 (HC). I will be very happy when my friend earns an
award.

 

 

 

 

Levels of Collectivism of Peer Group

 

158. It is important for maintaining friendship to respect
decisions made by my friends.

 

159. Sometimes I have to surrender my interests to my
friends’ interests to keep our friendship.

 

160. My friends believe that friendship should be kept
at any sacrifice.

 

161. My friends will be happy when I am happy.

 

 

 

 

162. My friends will be happy when I earn an award.

 

Social Disorganization:

 

163. My neighbors do not care what my friends do in
this area.

 

164. Teenagers in my neighborhood are out of control.

 

 

165. It is difficult for kids to make friends in my
neighborhood.

 

 

 

 

******************Thank you for your participation*****************

195

 

 

 

 

 

 

 

Appendix C. Survey Questionnaire - Teacher

Questionnaire about Class and School

 

 

This survey is about the class that you are in charge of and the school that you
currently work in. This researcher wants to know how you perceive students in
your class, teachers, rules and programs in your school.

This is an anonymous questionnaire. This means that you don’t have to write

your name or let us know who you are. The results of this survey will be used

solely for academic research purpose, and any information collected from this
survey will be kept confidential.

If you have any question regarding this survey, or if you want to receive results of
this study, please contact this researcher using the following contact information:

Chang-Hun Lee

School of Criminal Justice

Michigan State University

560 Baker Hall

East Lansing, MI 48823

Tel: 517-525-2827, Email: Leechan7@msu.edu

Thank you for your participation.

 

I. Questions about the class

4. How many students are there in your classroom?
Male ( ) Female ( ) Total ( )
5. What is your class's academic rank in your school?

( out of the total classes)

196

 

Please evaluate your class on the following aspects.

D (Disagree), SD (Somewhat disagree), N (Neutral), MA (Mostly Agree), and A (Agree)

 

Tick (V) for each category:

In Isol NjMAlA

 

Commitment to Education:

 

3. Students in my class think academic achievement is
the most important thing.

 

4. Students in my class believe entering college is the
most important goal they have.

 

5. All students in my class want to enter college-prep
high schools.

 

 

 

 

 

 

Cohesion:

 

6. Students in my class work together as a team to
prepare school events, such as athletic competition.

 

7. Students in my class work together to clean their
classroom.

 

8. Students in my class take care of each other.

 

 

 

 

 

 

Safety:

 

9. I believe my classroom is a safe placefor students.

 

10. Students cannot harm other students at least in my
classroom.

 

11. I believe my classroom is well supervised, so that it
provides a safe environment for my students.

 

 

 

 

 

 

Punitive Climate:

 

12. Students in my class want to prohibit delinquent
behaviors by peer pressure.

 

13. Students in my class report to me if there is any
behavior that harms other students.

 

14. l have seen that students in my class punish others
students for their delinquent behavior.

 

 

 

 

 

 

 

 

197

 

 

Moral Authority:

 

15. I believe students in my class care what I say about
morality.

 

16. l have a say in how students treat each other.

 

17. My moral standard is well accepted and practiced
among students in my class.

 

Bullying problems:

 

18. I am well aware of students teasing other
classmates in my class.

 

19. I am well aware of students hitting other classmate
in my class.

 

20. I am well aware of students leaving out other
classmate in my class.

 

21. I know exactly who teases, hits, or leaves out other
classmates in my class.

 

22. I know exactly who is teased, hit, or left out by
others students.

 

Number of Delinquent Students:

 

23. How many delinquent students you think you have
in your class?

 

24. How many students have been suspended in your
class this semester?

 

 

25. How many students you think are out of your
control?

 

ll. Questions about the school

26.What is your student population in your school?

Male ( ) Female ( ) Total (

198

 

 

 

 

 

 

 

 

 

 

 

 

27. Is there any school policy or rule prescribing what constitutes bullying

behaviors and what are punishments for those behaviors in your school?

28. If yes, when was the policy or rule set forth?

29. Is there any school training program for teachers to reduce bullying in

school?

30. If yes, when was the program established?

Please evaluate your school for the following aspects.

D (Disagree), SD (Somewhat disagree), N (Neutral), MA (Mostly Agree), and A (Agree)

 

Tick (V) for each category:

In [solN [MAlA

 

Commitment to Education:

 

31. Students in my school think academic achievement
is the most important thing.

 

32. Students in my school believe entering college is
the most important goal they have.

 

33. All students in my school want to enter college-prep
high schools.

 

 

 

 

 

 

Safety:

 

34. I believe my school is a safe place for students.

 

35. Students cannot harm other students at least in my
school

 

36. I believe my school is well supervised, so that it
provides a safe environment for my students.

 

 

 

 

 

 

 

Punitive Climate:

 

199

 

 

37. Students in my school want to prohibit delinquent
behaviors by peer pressure.

 

38. Students in my school report to teachers if there is
any behavior that harms other students.

 

39. l have seen that students in my school punish
others students for their delinquent behavior.

 

Moral Authority:

 

40. Students in my school care what teachers say
about morality.

 

41. Teachers have a say in how students treat each
other.

 

42. Teachers’ moral standard is well accepted and
practiced among students in my school.

 

Bullying problems:

 

43. Teachers are well aware of students’ behaviors,
such as teasing, hitting, and leaving out other
classmates, in my school.

 

44. Teachers know exactly who teases, hits, or leaves
out other classmates in my school.

 

45. Teachers know exactly who is teased, hit, or left out
by others students.

 

Smooth Administration:

 

46. Simple, non-time consuming procedures exist for
the acquisition and use of resources.

 

47. Teachers and administrators get along at this
school

 

48. Resources are well allocated to administer school
policies and programs.

 

Staff Morale:

 

 

49. Our problems in this school are so big that it is
unrealistic to expect teachers to make much of a dent

 

200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

in them.

 

50. The teaching faculty is frustrated.

 

51. Teachers are satisfied with what they are doing.

 

Clarity of School Rules:

 

52. Everyone in this school knows the rules for
behaviors, such as teasing, hitting or leaving out other
students.

 

53. Teachers teach students about the rules for
behaviors, such as teasing, hitting or leaving out other
students, and what they are expected.

 

54. The principal is firm with regard to the rules for
behaviors, such as teasing, hitting or leaving out other
students.

 

Fairness of School Rules:

 

55. The rules for behaviors, such as teasing, hitting or
leaving out other students, are fair.

 

56. The principal is fair with regard to the rules for
behaviors, such as teasing, hitting or leaving out other
students.

 

57. The punishment for breaking the rules is the same
no matter who students are.

 

Effectiveness of Rules, Policies, and Programs:

 

58. The current school rules for behaviors, such as
teasing, hitting or leaving out other students, are
effective in reducing the behaviors.

 

59. The current school policies for behaviors, such as
teasing, hitting or leaving out other students, are
effective in reducing the behaviors.

 

 

60. The programs for teachers are helpful to reduce
bullying in school.

 

201

 

 

 

 

 

 

BIBLIOGRAPHY

202

Agnew, R. (1992). Foundation for a general strain theory of crime and
delinquency. Criminology, 30(1), 47.

Agnew, R. (2001). Juvenile delinquency: causes and control. Los Angeles, Calif:
Roxbury Pub. Co.

Ah, Y., w. Jeong, & T. Cha. (2005). 999 9i [499 7l2‘sll 9 EH 2% 9 al *lfil ’9
.991 [The study on the actual conditions of bullying and the psychosocial
factors affecting bullying behavior]. 9999.1719— <5? [Joumal of Fish
and Marine Science of Education], 17(3), 390-403.

Ahmad, Y., & P. K. Smith. (1990). Behavioral measures: Bullying in schools. Child
Psychology and Psychiatry. 2, 26-27.

Ahn, H. (2002). (VJ/#7} %ll40// U/fl gig‘toi/ Jr??? gflél'lA study ofoutfita
thing that ’Iiime' affects ’Wangtta']. 411:8 LH 2313?. [Saemyung University].

Akers, R. L. (1996). Is differential association/social learning cultural deviance
theory? Criminology, 34(2), 229.

Akers, R. L. (2000). Criminological Theories: Introduction, Evaluation, and
Application.Los Angeles, CA: Roxbury Publishing Company.

Andershed, H., M. Kerr, & H. Stattin. (2001). Bullying in school and violence on
the streets: Are the same people involved? Joumal of Scandinavian
Studies in Criminology and Crime Prevention, 2(1), 31.

Anderson, C. S. (1982). The search for school climate: A review of the research.
Review of Educational Research, 5_2_(3), 368 - 420.

Anderson, J. C., & D. W. Gerbing. (1988). Structural equation modeling in
practice: A review and recommended two-step approach. Psychological
Bulletin403, 411—423.

 

Anderson, M., J. Kaufman, T. R. Simon, L. Barrios, L. Paulozzi, G. Ryan, R.
Hammond, W. Modzeleski, T. Feucht, & L. Potter. (2001). School-

203

associated violent deaths in the United States, 1994-1999. Journal of the
American MedicalAssociatiom286, 2695-2702.

 

Ando, M., T. Asakura, & B. Simons-Morton. (2005). Psychosocial influences on
physical, verbal, and indirect bullying among Japanese early adolescents.
The Journal of Early Adolescence, 25(3), 268.

Andreou, E., & P. Metallidou. (2004). The relationship of academic and social
cognition to behavior in bullying situations among Greek primary school
children. Educational Psychology. 24(1), 27.

Baldry, A. C. (2003). Bullying in schools and exposure to domestic violence. Child
Abuse & Neglect, 27, 713-732.

Batsche, G. M., & H. M. Knoff. (1994). Bullies and their victims: Understanding a

pervasive problem in the schools. School Psycho/oatI Review, 23(2), 165-
174.

Bauman, S., &A. D. Rio. (2006). Preservice teachers' responses to bullying
scenarios: Comparing physical, verbal, and relational bullying. Journal of
Educational Psychology. 98(1), 219.

Bentler, P. M., & C. P. Chou. (1987). Practical issues in structural modeling.
Sociological Methods & Research, 16(1), 78-117.

Bentley, K. M., &A. K. F. Li. (1995). Bully and victim problems in elementary
schools and students’ beliefs about aggression. Qanadiag Joumal of
School Psychology 11, 1 53-165.

Bjorkqvist, K., K. M. Lagerspetz, &A. Kaukiainen. (1992). Do girls manipulate
and boys fight? Aggressive Behavior, 18, 117-127.

Bollmer, J. M., R. Milich, M. J. Harris, & M. A. Maras. (2005). Afriend in need:

The role of friendship quality as a protective factor in peer victimization
and bullying. Journal of Interpersonal Violence, 20(6), 701.

204

Bosworh, K., D. L. Espelage, & T. R. Simon. (1999). Factors associated with
bullying behavior in middle school students. Jouma/ of Early Adolescence,
1_9(3), 341-362.

Bosworth, K., D. L. Espelage, & T. R. Simon. (1999). Factors associated with
bullying behavior in middle school students. Journal of Eariy Adolescence,
12(3), 341-362.

Boulton, M. J., & P. K. Smith. (1994). Bully/victim problems in middle-school
children: Stability, self-perceived competence, peer perceptions and peer
acceptance. British Journal of Developmental Psychology, 12, 315-329.

Boulton, M. J., M. Trueman, C. Chau, C. Whitehand, & K. Amatya. (1999).
Concurrent and longitudinal links between friendship and peer
victimization: Implications for befriending interventions. Joumal of
Adolescence, 22, 461-466.

Boulton, M. J., M. Trueman, & l. Flemington. (2002). Associations between
secondary school pupils' definitions of bullying, attitudes towards bullying,
and tendencies to engage in bullying: Age and sex differences.
Educational Studies, 28(4), 353.

Bowers, L., P. K. Smith, & V. Binney. (1994). Perceived family relationships of
bullies, victims and bully/victims in middle childhood. Joumal of Social and
Personal Relationships. 11, 215-232.

Bronfenbrenner, U. (1979). The Ecology of Human Development: Experiments by
Nature and Design.Cambridge, MS: Harvard University Press.

Bronfenbrenner, U. (1989). Ecological systems theory. In R. Vasta (Ed), Annals
of Child Development (Vol. 6, pp. 197-249). Greenwich, CT: JAI Press.

Buhs, E. S., G. W. Ladd, & S. L. Herald. (2006). Peer exclusion and victimization:
Processes that mediate the relation between peer group rejection and
children's classroom engagement and achievement? Journal of
Educational Psychology. 98(1), 1.

205

Bursik, R. J. J. (1988). Social disorganization and theories of crime and
delinquency: Problems and prospects. Criminology, 26, 519-551.

Cairns, R. B., & B. D. Cairns. (1991). Social cognition and social networks: A
developmental perspective. In D. J. Pepler & K. H. Rubin (Eds), T_he__
Development and Treatment of Childhood Aggression (pp. 249-278).
Hillsdale, NJ: Lawrence Erlbaum Associates.

Cairns, R. B., B. D. Cairns, H. Neckerman, S. D. Gest, & J.-L. Gariepy. (1988).
Social networks and aggressive behavior: Peer support or peer rejection?
Developmental Psychology, 24, 815-823.

Canchola, J. A., T. B. Neilands, & J. A. Catania. (2002). lmputation Strategies for
Sexual Orientation using SAS PROC MI.Cary, NC: SAS Institute, Inc.

Carney, J. V. (2005). Factor structure of the bullying situations identification (BSI)
instrument. Journal of School Violence, 4(3), 77.

Chan, J. H., R. Myron, & M. Crawshaw. (2005). The efficacy of non-anonymous
measures of bullying. School Psychology Intematiorfl 26(4), 443.

Choi, W. (2003). 783119 119191 4131 38352—1 ‘11-? [Social structural analysis of
cause of adolescence suicide], iffdfiiaax/Zéiii’ 519/ [Conference of
Korean Social Welfare Policy]: 5&5? 7'8 ink—L61 71191531. [Youth Professional
Institute of Korea].

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

Christie-Mizell, C. A. (2003). Bullying: The consequences of interparental discord
and child’s self-concept. Family Process, 42(2), 237-251.

Chu, H. (1978). The Korean learner in an American school. In T. T. Ladu (Ed),
Teaching for Cross-Cultural UnderstandingArlington, VA: Arlington Public
Schools.

206

Collins, K., & R. Bell. (1996). Peer perceptions of aggression and bullying
behavior in primary schools in Northern Ireland. Annals of the New York
Academ of Sciences, 794(September), 336-338.

Coloroso, B. (2003). The Bully, the Bullied, and the Bystander. New York: Harper-
Resource.

Cornell, D. G., & K. Brockenbrough. (2004). Identification of bullies and victims: A
comparison of methods. Journal of School Violence, 3(2/3), 63.

Cowie, H. (1998). Perspectives of teachers and pupils on the experience of peer
support against bullying. Educational Research and Evaluation, 4, 108-
125.

Crothers, L. M., & E. M. Levinson. (2004). Assessment of bullying: A review of
methods and instruments. Jouma/ of Counseling and Development, 82(4),
496-503.

Curtner-Smith, M. E., A. M. Culp, C. Scheib, K. Owen, A. Tllley, M. Murphy, L.
Parkman, & P. W. Coleman. (2006). Mothers' parenting and young
economically disadvantaged children's relational and overt bullying.
Journal of Child and Family Studies, 15(2), 177.

Dahlberg, L. L., S. B. Toal, & C. B. Behrens. (1998). Measuring violence-related
attitudes, beliefs, and behaviors among youths: A compendium of
assessment tools.At|anta, GA: Genters for Disease Control and
Prevention, National Center for Injury Prevention and Control.

Dill, E. J., E. M. Vernberg, P. Fonagy, S. W. Twemlow, & B. K. Gamm. (2004).
Negative affect in victimized children: The roles of social withdrawal, peer
rejection, and attitudes toward bullying. Journal of Abnormal Child
Psychology. 32(2), 159. '

Dishion, T. J., D. C. French, & G. R. Patterson. (1995). The Development and
Ecology of Antisocial Behavior. In D. Cicchetti & D. Cohen (Eds),
Developmental Psychopathology, Volume 2: Risk, Disorder: and

207

Adaptation (Vol. 2, pp. 421-471). New York: A Wiley-Interscience
Publication.

Duncan, G. J., & G. Kalton. (1987). Issues of design and analysis of surveys
across time. Intemationa/ Statistical Review, 55(1), 97-117.

Elliott, D. 8., & V. Hanrvin. (1974). Delinquency and Dropout.Lexington, MS: D.C.
Heath.

Elliott, D. S., D. Huizinga, & S. S. Ageton. (1985). Explaining delinquency and
drug use.Thousand Oaks, CA: Sage.

Espelage, D. L., K. Bosworh, & T. R. Simon. (2000). Examining the social context
of bullying behaviors in early adolescence. Journal of Counseling and
Development. 78(3), 326-333.

Espelage, D. L., M. K. Holt, & R. R. Henkel. (2003). Examination of peer-group
contextual effects on aggression during early adolescence. Child
Development, 74(1), 205-220.

Espelage, D. L., & S. M. Swearer. (2003). Research on school bullying and
victimization: What have we learned and where do we go from here?
School Psychology Review, 32(3), 365-383.

Fagan, J., & D. L. Wilkinson. (1998). Social contexts and functions of adolescent
violence. In D. 8. Elliott, B. A. Hamburg & K. R. Williams (Eds), Violence
in American Schools: A New Perspective (pp. 55-93). Cambridge, UK:
Cambridge University Press.

Farrington, D. P. (1993). Understanding and preventing bullying. Crime and
Justice, 17, 381-458.

 

Farrington, D. P. (Ed). (1986). Age and crime (Vol. 7). Chicago, IL: University of
Chicago Press.

Farver, J. M., & Y. L. Shin. (1997). Social pretend play in Korean- and Anglo-
208

American preschoolers. Child Development, 68(3), 544-556.

Fekkes, M., F. l. M. Pijpers, A. M. Fredriks, T. Vogels, & S. P. Verloove-Vanhorick.
(2006). Do bullied children get ill, or do ill children get bullied? A
prospective cohort study on the relationship between bullying and health-
related symptoms. Pediatrics, 117(5), 1568-1575.

Fekkes, M., F. l. M. Pijpers, & S. P. Verloove-Vanhorick. (2006). Effects of
antibullying school program on bullying and health complaints. Archives of

 

Pediatrics & Adolescent Medicine, 160(6), 638.

Flores, R. L. (2005). Developmental aspects of school violence. In F. L. Denmark,
H. H. Krauss, R. W. Wesner, E. Midlarsky & U. P. Gielen (Eds), Violence
in Schools: Cross-National and Cross-Cultural Perspectives (pp. 75-84).
New York: Springer.

Flouri, E., & A. Buchanan. (2003). The role of mother involvement and father
involvement in adolescent bullying behavior. Journal of Interpersonal
Violence, 18(6), 634.

Frank, K. A. (2000). Impact of a confounding variable on a regression coefficient.
Sociological Methods & Research, 29(2), 147-194.

Gest, S. D., S. Graham-Bermann, & W. W. Hartup. (2001). Peer experience:
Common and unique features of number of friendships, social network
centrality, and sociometric status. Social Development, 10(1), 23-40.

Gladstone, G. L., G. B. Parker, & G. S. Malhi. (2006). Do bullied children become
anxious and depressed adults? A cross-sectional investigation of the

correlates of bullying and anxious depress. Joumal of Nervous and Mental
DiseaseJ 94(3), 201 .

Gollob, H. F., & C. S. Reichardt. (1987). Taking account of time lags in causal
models. Child Development, 58, 80-92.

Gollob, H. F., & C. S. Reichardt. (1991). Interpreting and estimating indirect
209

effects assuming time lags really matter. In L. M. Collins & J. L. Horn
(Eds), Best Methods for the Analysis of Change (pp. 243-259).
Washington, DC: American Psychology Association.

Gottfredson, D. C. (1986). An empirical test of school-based environmental and
individual interventions to reduce the risk of delinquent behavior.
Criminology, 24(4), 705-.

Gottfredson, D. C. (2001). Schools and delinquencyCambridge, UK; New York:
Cambridge University Press.

Gottfredson, G., D. C. Gottfredson, A. A. Payne, & N. Gottfredson. (2005). School
climate predictors of school disorder: Results from a national study of
delinquency prevention in schools. Journal of Research in Crime and
Delinquency, 42(4), 412-444.

Gottfredson, G D., & D. C. Gottfredson. (1985). Victimization in schools.New
York: Plenum Press.

Graham, 8., & J. Juvonen. (1998). A social cognitive perspective of peer
aggression and victimization. Annals of Child Development, 13, 23-70.

Griffore, R. J., & L. A. Phenice. (2001). The Language of Human Ecology: A
General Systems PersgectiveDubuque: Kendall/Hunt Publishing.

Harris, 8., & G. F. Petrie. (2003). Bullying: The Bullies, the Victims, the
Bystanders.Lanham: The Scarecrow Press.

Haselager, G. J. T., W. W. Hartup, C. F. M. v. Lieshout, & J. M. A. Riksen-
Walraven. (1998). Similarities between friends and nonfriends in middle
childhood. Child Development, 69(4), 1198-1208.

Hatcher, L. (1994). A Step-by-Step Approach to Using SAS for Factor Analysis
and Structural Equation ModelingCary, NC: SAS Institute Inc.

Hay, C., & M. M. Evans. (2006). Violent victimization and involvement in

210

delinquency: Examining predictions from general strain theory. Journal of
Criminal Justice, 34(3), 261.

Haynie, D. L. (2001). Delinquent peers revisited: Does network structure matter?
American Journal of Sociology, 106(4), 1013-1057.

Haynie, D. L., T. Nansel, P. Eithel, A. D. Crump, K. Saylor, K. Yu, & B. Simons-
Morton. (2001). Bullies, victims, and bully/victims: Distinct groups of at-risk
youth. Journal of Early Adolescence, 21(1), 29-49.

Heinemann, P. P. (1973). Mobbning: Gruppvald blant barn oqvokane [Bu/lying:
Group violence among children and adultslStockholm: Natur och Kultur.

Herrero, J., E. Estevez, & G. Musitu. (2006). The relationships of adolescent
school-related deviant behavior and victimization with psychological
distress: Testing a general model of the mediational role of parents and
teachers across groups of gender and age. Journal of Adolescence, 29(5),
671.

Hodges, E. V. E., M. Boivin, F. Vitaro, & W. M. Bukowski. (1999). The power of
friendship: Protection against an escalating cycle of peer victimization.
Developmental Psychology. 35(1), 94-101.

Hodges, E. V. E., M. J. Malone, & D. G. Perry. (1997). Individual risk and social
risk as interacting determinants of victimization in the peer group.
Developmental Psychology, 33(6), 1032-1039.

Hofstede, G. (1991). Cultures and organizations: Software of the mind. London:
McGraw-Hill.

Hoshino, K. (2001). Higaishya to Kagaishyatono Kankei [Relations between bully
and victim]. ln Y. Morita (Ed), Iiime no Kokusai Hikaku Kenkyu [Cross-
national Study of Bullying] (pp. 73-92). Tokyo: Kaneko Shyobou.

Hu, L. T., & P. M. Bentler. (1995). Evaluating model fit. In R. H. Hoyle (Ed),
Structural Equation Modeling (pp. 76-99). Thousand Oaks, CA: Sage.

211

Huizinga, D., & D. S. Elliott. (1986). Reassessing the reliability and validity of
self-report delinquency measures. Journal of Quantitative Criminology,
2(4), 293-327.

Huttunen, A., C. Salmivalli, & K. M. J. Lagerspetz. (1996). Friendship networks
and bullying in schools. Annals of the New York Academy of Sciences,
M(September), 355-359.

Juvonen, J., S. Graham, & M. Schuster. (2003). Bullying among young
adolescents: The strong, the weak, and the troubled. Pediatrics, 112(6),
1231.

 

KDE. (1999). 791%} “319% 011901 91 2151:11le 0,3? [A study on bullying and
prevention policy]. In K. D. 0. Education (Ed):§}i‘5§ifi7ll ”9%! [Korean
Institute for Youth Development].

KEDI. (1998). 999/ 991 (9’9 III-9:9 -9 999) 999/ 99 91? [A
study on bullying among students] (No. 98-19): Korean Educational
Development Institute.

Keith, S., & M. E. Martin. (2005). Cyber-bullying: Creating a culture of respect in
a cyber world. Rec/aiming Children and Youth, 13(4), 224-228.

Kepenekci, Y. K., & S. Cinkir. (2006). Bullying among Turkish high school
students. Child Abuse & Neglect, 30(2), 193.

Kim, G- (1996) EVE/{7.4712 $077419] 5589‘ HZé’Ef-Q/ 9’5} ‘79”: $295147? :51
”if 0,1"? [Morality and Distinction between Inside-Outside Groups withtin
Social Relationship: Cultural Comparison Study]. E90019) 91.1—17- [JoongAng
University], Seoul.

 

Kim, H. (2001). 9.4.99 99 9999 9.1199 9901] 99 0.3-11: 999
El, 993,351, mafia?) 9.} 71311/315111‘21‘3 %-¢:}9_E [The bullying and
school-related variables]. 35./f. E] 9}" 103-F [Korean Journal of Youth

Studiesl, 8(2), 119-142.

212

Kim, H., &H. Lee (2000) 999-9 9999-9 9973901 9161999 .19-l
“55.135. % 31595-148 71.19] ”ll—11% 253/99% [A study on predictors of
bullying victimization among students: Focus on comparison between
elementary, middle, and high schools]. JLIT 7" {773/9}? [Journal of
Educational Psychology], 1 5(1 ), 1 83-210.

Kim, J. U. (1991). The Koreans: Their mind and behavior.Seoul, Korea: Kyobo
Centre Co.

Kim, U., & S. Choi. (1994). Individualism, collectivism, and child development: A
Korean perspective. In P. M. Greenfield & R. R. Cocking (Eds), Cw
Cultural Roots of Minority Child Development (pp. 227-258). Hillsdale, NJ:
Lawrence Erlbaum Associates.

Kim, U., H. C. Triandis, C. Kagitcibasi, S.-C. Choi, & G. Yoon (Eds). (1994).
Individualism and Collectivism: Theory, method, and applications (Vol. 18).
Thousand Oaks, CA: Sage.

Kim, W. (2004). 59E]: 9113], 9.9], 919101] 532?} %if@ 3179} [Wangttaz
definition, status, reasons]. 9999 511T he Korean Joumal of
Counseling], 5(2), 451-472.

Kim, Y., Y. Koh, & B. Leventhal. (2005). School bullying and suicidal risk in
Korean middle school students. Pediatrics, 115(2), 357-363.

Kim, Y., & H. Park. (1997). 9.9 9 9777- 171999 99./EX]. III-.999- 90/9.
“/5 ’3 ’5’5‘}? (010/9:a [A study on bullying among youths: Bullies and
Bullied].Seoul: 59 3116919191 731201 [Youth Communication Plaza].

King, M. P. (2001). Cross-sectional and Longitudinal Research Designs Issues In
the Studies of Human Development. from
http://wwwgraduateresearchcom/Kinghtm

Kinlock, T. W., R. J. Battjes, & M. S. Gordon. (2004). Factors associated with
criminal severity among adolescents entering substance abuse treatment.
Joumal of Drug Issues, 34(2), 293-319.

213

Kline, R. B. (1998). Principles and Practice of Structural Equation Modeli:ng.New
York, NY: The Guilford Press.

KNSO. (2005). 2005 {34--3E Li 2%: 2I/ L2005 Youth Statistics]:Korean National
Statistical Office.

Kupersmidt, J. B., M. E. DeRosier, & C. P. Petterson. (1995). Similarity as the
basis for children's friendships: The roles of sociometric status, aggressive
and withdrawn behavior, academic achievement and demographic
characteristics. Journal of Social and Personal Relationships, 12(3), 439-
452.

Kwon, J. (1999). 3.11-H 91 ’9] {191%‘19' ‘33:?901‘9 9] 7H 91 [Methodological review
of school bullying research]. The Joumal of Educational Research, 15, 45-
56.

Laub, J. H., & J. L. Lauritsen. (1998). The interdependence of school violence
with neighborhood and family conditions. In D. S. Elliott, B. A. Hamburg &
K. R. Williams (Eds), Violence in American Schools: A New Perspective
(pp. 127-158). Cambridge, UK: Cambridge University Press.

Leary, M. R., R. M. Kowalski, L. Smith, & S. Phillips. (2003). Teasing, rejection,
and violence: Case studies of the school shootings. Aggressive Behavior:
_2_9, 202-214.

Lease, A. M., R. M. McFall, & R. J. Viken. (2003). Distance from peers in the
group's perceived organizational structure: Relation to individual
characteristics. The Joumal of Early Adolescence, 23(2), 194-216.

Lee, c. (2000). 999 79;)..79/99/ 92: 999 4.99 99 «2/92/ :2/0/
£14] [Analysis of differences irmersonality traits and sociometric status of
bully/victim]. 9%47-92 CH 914?. [Korean National University of Education].

Lee, C., & G. Kaok. (2000). 9%? 027419/ 13’ 5‘ [511% 5%]: 9’ 1:779} 5’31 [Bullying in
schools: Status and CharactersJSeoulz 791 {vi-Fat [Jipmoondang].

214

Lee, D. (2004). 7:343: 52’ age! *5 65h? 5'5 Iii—33:3 3’ 5!! flaw/1% 5’5’9/ {W
[Relationship between bullying and youth culture and peer conformity].
Seokang University, Seoul.

 

Lee, E. H., & K. W. Lee. (1987). The Korean mother’s socialization process for
children (in Korean).Seoul, Korea: Ehwa Women's University.

Lee, H. (2003). Ki] ‘3 El%%‘ 91 gap] [Hit BOW} [Current status of bullying and
preventive policy]. 472%51-51‘ [Educational Development]. 2(60), 83-94.

Lee, S. (1999). %§i*§ BEER alga ('%erl')9] 041%991 [Predicting peer
rejection of middle-school students with ordered probit analysis]. iffdffi?
52/ bit [Korean Journal of Social Welfare]. 37(4), 357-379.

Lee, S., C. Byun, & E. Jin. (1969). £5553} {9’27” {75‘ gm} QM] £2 5‘
LStandardized Personality Evaluation Test]. *1 % [Seoul]: $530.} 35. *3 03 :rlfi:
[Chung-Ang Aptitude Publishing].

Lee, 3., & M. Kim. (2000). 7'3 fizki WEED] 3+ 7% 3:] $219] 99337)] oil 3321:]
‘23-? [A study of correlation between the alienating behaviors of the youth
and domestic factors]. Journal of Research, 27(1), 315-334.

Liepe-Levinson, K., & M. H. Levinson. (2005). Ageneral semantics approach to
school-age bullying. et Cetera, 62(1), 4-16.

Long, J. D., &A. D. Pellegrini. (2003). Studying change in dominance and
bullying with linear mixed model. School Psychology Review. 32(3), 401-
417.

MacCallum, R. C., & J. T. Austin. (2000). Applications of structural equation
modeling in psychological research. Annual Review of Psychology. 51,
201-226.

Maday, B. C., & L. B. Szalay. (1976). Psychological correlates of family
socialization in the United States and Korea. In T. Williams (Ed.),
PsychologicalAnthropology.The Hague: Mouton.

215

Masataka, N. (1998). liime o yurusu shinri [Psychology of Allowing
Bullyinngokyo: lwanami Shyoten.

Matsueda, R. L., & K. Anderson. (1998). The dynamics of delinquent peers and
delinquent behavior. Criminology, 36(2), 269.

Matsueda, R. L., & K. Heimer. (1987). Race, family structure, and delinquency: A
test of differential association and social control theories. American
Sociological Review, 52(December), 826-840.

Menesini, E., E. Melan, & B. Pignatti. (2000). lnteractional styles of bullies and
victims observed in a competitive and a cooperative setting. The Journal
of Genetic Psychology. 161(3), 261 -281.

Moon, s. (2002). Aiii b5 Bl Z—i 215199.91 91 5:15:15 >8 Er [4&9 9] 92119] 531%
[Problems and causes of youth bullying as sociopathological
phenomenon]. A}? #2] 0.3 51 [The Journal of Social Welfare], 11, 1-21.

Moriatry, L. (1999). The conceptualization and Operationalization of the
intervening dimensions of social disorganization. In M. L. Dantzker (Ed.),
Readings for Research Methods in Criminology and Criminal
Justice.Woburn, MA: Butterworth-Heinemann.

Morita, Y., & K. Kiyonaga. (1994). Iiime: Kyositsu no yamai [Bu/lying: The
Classroom CrisisI.Tokyo: Kaneko Shyobou.

Morrison, B. (2001). Bullying, violence and alienation (pp. 1-15): Centre for
Restorative Justice, Research School of Social Science, Australian
National University.

Morrison, B. (2002). Bullying and victimization in schools: A restorative iustice
approach (No. 219). Canberra ACT: Australian Institute of Criminology.

Mouttapa, M., T. Valente, P. Gallaher, L. A. Rohrbach, & J. B. Unger. (2004).
Social network predictors of bullying and victimization. Adolescence.
@054), 315-335.

216

Naito, T. (1990). Moral education in Japanese public schools. Moral Education
Forum 15(2), 27-36.

 

Naito, T., & U. P. Gielen. (2005). Bullying and Ijime in Japanese Schools. In F. L.
Denmark, H. H. Krauss, R. W. Wesner, E. Midlarsky & U. P. Gielen (Eds),
Violence in Schools: Cross-National and Cross-Cultural Perspectives (pp.
169-190). New York: Springer.

Nansel, T., M. Overpeck, R. S. Pilla, W. J. Ruan, B. Simons-Morton, & P. Scheidt.
(2001). Bullying behaviors among US youth: Prevalence and association I
with psychosocial adjustment. Joumal of the American Medical
Association, 285(16), 2094-2100.

Natvig, G. K., G. Albrektsen, & U. Qvarnstrom. (2001). School-related stress
experience as a risk factor for bullying behavior. Journal of Youth and
Adolescence, 30(5), 561-575.

Naylor, P., & H. Cowie. (1999). The effectiveness of peer support systems in
challenging school bullying: The perspectives and experiences of teachers
and pupils. Joumal ofAdo/escence, 22, 467-479.

Nesdale, D., & M. Naito. (2005). Individualism - Collectivism and the attitudes to
school bullying of Japanese and Australian students. Journal of Cross -
Cultural Psychology, 36(5), 537.

Olweus, D. (1978). Aggression in the Schools: Bullies and Whipping
Boys.Washington DC: Hemisphere Press (Wiley).

Olweus, D. (1991). Bully/victim problems among schoolchildren: Basic facts and
effects of a school based intervention program. In D. J. Pepler & K. H.
Rubin (Eds), The Development and Treatment of Childhood Aggression
(pp. 411-448). Hillsdale, NJ: Lawrence Erlbaum Associates.

Olweus, D. (1993). Bullying at School: What we know and what we can
ioCambridge, MA: Blackwell.

217

Olweus, D. (1995). Bullying and peer abuse at school: Facts and interventions.
Current Directions in Psychological Science: A Journal of the American
Psychological Society, 4, 196-200.

Olweus, D. (2003). A profile of bullying. Educational Leadership, 60(6), 12-17.

Pakaslahti, L., & L. Keltikangas-Jarvinen. (2001). Peer-attributed prosocial
behavior among aggressive/preferred, aggressive/non-preferred, non-
aggressive/preferred and non-aggressive/non-preferred adolescents.
Personality and Individual Differences, 30, 903-916.

Park, 5 (2002) 3%5-1311011 4194 flfiuf%%i(%ml) ‘23—? [Bullying(Wangtta)
study in elementary schools]. gala-2763127? [Open Education
Practice Study]. 5, 145-166.

Park, H. S., K. G. Schepp, E. H. Jang, & H. Y. Koo. (2006). Predictors of suicidal
ideation among high school students by gender in South Korea. The
Journal of School Health, 76(5), 181-188.

Park, J. (2003). 7331951 73 Emléau‘ 91 3:19.134 EH 511%“?! [The causes and the
settlements of teenagers mobbing]. 375-43274? i5??? 7'1 [Social Science
Research]. 10, 355-369.

 

Park, K., & H- 300- (1998). 331%“? 35%; (ii/5* W5?! Er! 377% gl) Efé‘ol/ {4?
£177 [A study on Wagqtta (Bullying) among studentsl.Seoul: Korean
Institute of Educational Development.

 

Pellegrini, A. D., & J. D. Long. (2002). A longitudinal study of bullying, dominance,
and victimization during the transition from primary school through
secondary school. The British Journal of Developmental Psychology. 20,
259-280.

Portes, A. (1998). Social capital: Its origins and applications in modern sociology.
Annual Review of Sociology, 24, 1-24.

Portes, A. (2000). The two meanings of social capital. Sociological Forum, 15(1),
218

1-12.

Poulin, F., A. H. N. Cillessen, J. A. Hubbard, J. D. Coie, K. A. Dodge, & D.
Schwartz. (1997). Children's friends and behavioral similarity in two social
contexts. Social Development, 6(2), 224-236.

Rigby, K. (1996). Bullying in Schools: and what to do about it. London, England:
Jessica Kingsley Publishers.

Rigby, K. (1998). The relationship between reported health and involvement in
bully/victim problems among male and female secondary schoolchildren.
Journal of Health Psychology. 3(4), 465-476.

Rigby, K. (2004). Addressing bullying in schools: Theoretical perspectives and
their implications. School Psychology International. 215(3), 287-300.

Rigby, K. (2005). Why do some children bully at school? The contributions of
negative attitudes towards victims and the perceived expectations of
friends, parents, and teachers. School Psychology International, 26(2),
147.

Rigby, K., & I. Cox. (1996). The contribution of bullying at school and low self-
esteem to acts of delinquency among Australian teenagers. Personality
and Individual Differences, 21 (4), 609-612.

Rigby, K., & P. T. Slee. (1993). Dimensions of interpersonal relation among
Australian children and implications for psychological well-being. The
Journal of Social Psychology. 133(1), 3343.

Rivers, |., & P. K. Smith. (1994). Types of bullying behavior and their correlates.
Aggressive Behavior, 20, 359-368.

Rodkin, P. C., & T. W. Farmer. (2000). Heterogeneity of popular boys: Antisocial
and prosocial configurations. Developmental Psychology, 36(1), 14-24.

Rodkin, P. C., & E. V. E. Hodges. (2003). Bullies and victims in the peer ecology:
219

Four questions for psychologists and school professionals. School
Psychology Review, 32(3), 384-400.

Rosenbaum, P. R. (1986). Dropping out of high school in the United States: An
observational study. Journal of Educational Statistics, 11(3), 207-224.

Rubin, D. B. (1976). Inference and missing data. Biometrika, 63, 467-474.

Rubin, D. B. (1987). Multiple lmputation for Nonresponse in Surveys.New York:
John Wiley & Sons, Inc.

Salmivalli, C., A. Huttunen, & K. M. Lagerspetz. (1997). Peer networks and
bullying in schools. Scandinavian Joumal of Psychology, 38, 305-312.

Salmivalli, C., K. M. Lagerspetz, K. Bjorkqvist, K. Osterman, &A. Kaukiainen.
(1996). Bullying as a group process: Participant roles and their relations to
social status within the class. Aggressive Behavior, 22, 1-15.

Salmivalli, C., M. Lappalainen, & K. M. J. Lagerspetz. (1998). Stability and
change of behavior in connection with bullying in schools: A two-year
follow-up. Aggressive Behavior. 2_4, 205-218.

Sampson, R. J., & W. B. Groves. (1989). Community structure and crime: Testing
social-disorganization theory. American Journal of Sociology. 94(4), 774-
802.

Sampson, R. J., S. W. Raudenbush, & F. Earls. (1997). Neighborhoods and
violent crime: A multilevel study of collective efficacy. Science, 277, 918-

924.

SAS. (2002). SAS/STAT Software Enhancements: Multiple lmputation for Missipg
Data.Cary, NC: SAS Institute.

Schafer, J. L. (1997). Analysis of Incomplete Multivariate Data.New York:
Chapman and Hall.

220

Scott, J. (2000). Social Network Analysis: A Handbook (2nd ed). Thousand Oaks,
CA: Sage.

Seals, D., & J. Young. (2003). Bullying and victimization: Prevalence and
relationship to gender, grade level, ethnicity, self-esteem, and depression.
Adolescence, 38(152), 735-747.

Seo,J.-B.,&J.-l. Han. (2004). :35] ELI—Eat. iii/slot] EH22] olEZzl 31% [A
semantic analysis of bullying]. 417w??— grog—7— [The Journal of Educational
Researchl. 2, 117-134.

Sexton-Radek, K. (Ed.). (2005). Violence in Schools: Issues, Consequences, and
Expressions.Westport, Connecticut: Praeger.

Shaw, C. R., & H. D. McKay. (1969). Juvenile delinquency and urban areas; a
study of rates of delinguency in relation to differential characteristics of
local communities in American cities (Rev. ed.). Chicago: University of
Chicago Press.

Shields, A., & D. Cicchetti. (2001). Parental maltreatment and emotion
dysregulation as risk factors for bullying and victimization in middle
childhood. Journal of Clinical Child Psychology. 30(3), 349.

Shim, J. (2004). ado/I Elle} 7i? a’ Elf/l #11713" a”. 5917/2? HIE [A study of
pupils' perception on Korean bullying (Wangttall. Kyunghee University,
SeouL

Shin. Y. (2000). awe gran/a %’(%‘19‘L0/L£v‘€f amaz- x/z/z/a-g- ea
21%: [A comparison study in bullying of the middle school students: Focus
on social environment]. 2600111113137. [Joongang University].

Short, J. F., & F. l. Nye. (1957). Reported behavior as a criterion of deviant
behavior. Social Pmblems.j(3), 207-213.

Simpson, R., & C. Cohen. (2004). Dangerous work: The gendered nature of
bullying in the context of higher education. Gender, Work and

 

221

Organization, 11(2), 163.

Singelis, T. M., H. C. Triandis, D. D. Bhawuk, & M. Gelfand. (1995). Horizontal
and vertical dimensions of individualism and collectivism: Atheoretical and
measurement refinement. Cross-Cultural Research, 29, 240-275.

Singleton, R. A. J., & B. C. Straits. (1999). Approaches to Social Research (3rd
ed.). New York: Oxford University Press.

Slee, P. T. (1994). Situational and interpersonal correlates of anxiety associated
with peer victimization. Child Psychiatry and Human Develmment, 25(2),
97-1 07.

Smith, P. K., & P. Brain. (2000). Bullying in schools: Lessons from two decades of
research. Aggressive Behavior. 26, 1-9.

Smith, P. K., H. Cowie, R. F. Olafsson, &A. P. D. Liefooghe. (2002). Definitions of
bullying: A comparison of terms used, and age and gender differences, in
a fourteen-country international comparison. Child Development, 73(4),
1119-1133.

Sobel, M. (1995). Causal inferences in the social and behavioral sciences. In G.
Arminger, C. Clogg & M. Sobel (Eds), _I_-I_apdbook of Statistical Modeling
for the Social and Behavioral Sciences.New York: Plenum Press.

Sobel, M. (1996). An introduction to causal inference. Sociological Methods &
Research, 24, 353-379.

Spivak, H., & D. Prothrow-Stith. (2001). The need to address bullying -An
important component of violence prevention. Journal of the American
Medical Association. 285(16), 21 31 -21 32.

Stevens, V., l. D. Bourdeaudhuij, & P. V. Oost. (2002). Relationship of the family

environment to children's involvement in bully/victim problems at school.
Jouma] of Youth and Adolescence, 31(6), 419-428.

222

Sullivan, K., M. Cleary, & G. Sullivan. (2004). Bullying in Secondary Schools:
What It Looks Like and How to Manage lt.Thousand Oaks, CA: Sage.

Sutherland, E. H., & D. R. Cressey. (1970). Criminology (8th ed.). Philadelphia:
Lippincott.

Swearer, S. M., & B. Doll. (2001). Bullying in schools; An ecological framework.
Journal of Emotional Abuse, 2, 7-23.

Tagiuri, R. (1968). The Concept of Organizational Climate. InR. Tagiuri & G. H.
Litwin (Eds.), Organizational Climate: Exploration of A Concept.Boston:
Harvard University.

Tai, M. (2001). Ijime taiou to sonokouka [Treatments for Bullying and Their
Effects]. In Y. Morita (Ed), Iiime no kokosai hikaku kenkyu [Cross-national
Study of Bullying] (pp. 123-144). Tokyo: Kaneko Shyobou.

Thornberry, T. P., & M. D. Krohn. (2000). The Self-Report Method for Measuring
Delinquency and Crime. In D. Duffee (Ed.), Measurement m Analysis of
Crime and Justice (pp. 33-83). Rockville: National Institute of
Justice/NCJRS.

Toit, S. d., & G. Mels. (2005). Supplementary Notes on Multiple
ImputationLincolnwood, IL: Scientific Software International, Inc.

Triandis, H. C. (1995). Individualism and CollectivismBoulder, CO: Westview.

Twemlow, S. W., P. Fonagy, & F. C. Sacco. (2001). An innovative
psychodynamically influenced approach to reduce school violence.
Journal of the American Academy of Child and Adolescent Psychiatry.
4_0(3), 377-379.

Unnever, J. D., & D. G. Cornell. (2003). Bullying, self-control, and ADHD. Journal
of Interpersonal Violence. 18(2), 129-148.

Verlinden, S., M. Hersen, & J. Thomas. (2000). Risk factors in school shootings.
223

Clinical Psychology Review, 20(1), 3-56.

Vold, G. B., T. J. Bernard, & J. B. Snipes. (1998). Theoretical Criminology (4th
ed.). New York: Oxford University Press.

Wareham, J., & R. Dembo. (2007). A longitudinal study of psychological
functioning among juvenile offenders: A latent growth model analysis.

Criminal Justice and Behavior. 34(2), 259.

West, S. G., J. F. Finch, & P. J. Curran. (1995). Structural equation models with
non-normal variables: Problems and remedies. In R. H. Hoyle (Ed.),
Structural Equation Modeling (pp. 56-75). Thousand Oaks, CA: Sage.

Wolfgang, M. E., & F. Ferracuti. (1981). The Subculture of Violence.Beverly Hills,
CA: Sage.
Wolke, D., S. Woods, K. Stanford, & H. Schulz. (2001). Bullying and victimization

of primary school children in England and Germany: Prevalence and
school factors. British Joumal of Psychology. 92, 673-696.

Wong, D. S. W. (2004). School bullying and tackling strategies in Hong Kong.
International Journal of Offender Therapy and Comparative Criminology,

5&5), 537.

Xie, H., R. B. Cairns, & B. D. Cairns. (1999). Social networks and configurations

in inner-city schools: Aggression, popularity, and implications for students
with EBD. Joumal of Emotional and Behavioral Disorders, 7(3), 147-155.

Yang, J. (2005). {552:5}qu 96m} ‘f-al‘*§ 8.9101] CW} 03,-? [A study on causal
factors of bullying among middle school students]. {fill-iv fr he Journal

 

of Korean Education]. 32(2), 31-52.

Yang, S., J. Kim, S. Kim, I. Shin, & J. Yoon. (2006). Bullying and victimization
behaviors in boys and girls at South Korean primary schools. Journal of
the American Academy of Child and Adolescent Psychiatry. 45(1), 69.

224

Young, R., & H. Sweeting. (2004). Adolescent bullying, relationships,
psychological well-being, and gender-atypical behavior: A gender
diagnosticity approach. Sex Roles, 50(7/8), 525.

 

225

   

ulili‘iiljiijjljj][iii]